i

Carbon Isotope Ratios and Composition of Fatty Acids: Tags and Trophic Markers in Pelagic Organisms by Ruben Jelmar Veefkind Doctorandus, Utrecht University, 1997 A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY in the School of Earth and Ocean Sciences We accept this dissertation as conforming to the required standard

Dr. M.J. Whiticar, Supervisor (School of Earth and Ocean Sciences)

Dr. D.L. Mackas, Departmental Member (School of Earth and Ocean Sciences)

Dr. V. Tunnicliffe, Departmental Member (School of Earth and Ocean Sciences)

Dr. L.A. Hobson, Outside Member (Department of Biology)

Dr. J.N.C. Whyte, Additional Member (Department of Fisheries and Oceans, Canada)

Dr. M.A. Teece, External Examiner (Department of Chemistry, College of Environmental Science and Forestry, State University of New York) © Ruben Jelmar Veefkind, 2003 University of Victoria All rights reserved. This dissertation may not be reproduced in whole or in part, by photocopying or other means, without the permission of the author.

ii

Supervisor: Dr. Michael J. Whiticar

Abstract Understanding the movement and feeding habits of marine animals is crucial when managing their populations. The molecular, and stable carbon isotope composition of fatty acids from an organism provides time-integrated information on its dietary intake. Hence, when spatial differences in the quality of seston exist it should be able to trace these differences up into higher trophic level organisms. The presented study evaluates the applicability of

13

C/12C ratios of individual fatty acids, as natural tags and dietary

markers in marine pelagic organisms. In addition, the use of

13

C/12C ratios of bulk

sample, as well as fatty acid composition data in examining the movement, and diet of animals are further explored. Samples of particulate organic matter, zooplankton, larval fish and juvenile salmon collected during three cruises off the west coast of Vancouver Island were analyzed. The fatty acid composition, stable carbon isotope ratio of either bulk sample, or individual fatty acids could typically distinguish samples collected in continental shelf waters from off-shelf samples. The differences in fatty acid composition between the adjoining food webs seem to be mainly caused by the different contribution of diatom-derived material to the base of the food web. The higher 13C/12C ratios found in the diatom-richer seston in shelf waters were not simply caused by the higher contribution of diatoms. Instead, stable carbon isotope data on individual fatty acids indicate that growth conditions favouring diatom growth caused 13C-enrichment in algae other than diatoms as well. The relative abundance of polyunsaturated fatty acids, such as docosahexaenoic acid (22:6n-3), were found to increase with trophic level. Whereas the abundance of saturated, and monounsaturated fatty acids was higher in organisms from lower trophic levels. This suggests that the fatty acid composition may be a useful trophic level indicator. However, literature data indicate that these trends observed in seston,

iii zooplankton, larval fish and juvenile salmon, do not hold for larger organisms and adult life stages. The variation in

13

C/12C ratios of individual fatty acids from almost 200 samples

from 3 cruises were compared. A large range (typically about 7‰) in the δ13C values of fatty acids is observed within single samples. The variation in δ13C between the individual fatty acids was found to be reproducible, independent of the quality of seston, and in accordance with patterns reported by other studies. This suggests the presence of common underlying mechanisms, most likely biosynthetic effects, producing the semipredictable offsets between the δ13C of fatty acids. The factors identified here as having potentially the largest impact on the δ13C seem to be desaturation, different timing of lipid class synthesis during the growth cycle of autotrophs, and perhaps also the proportion of PUFAs synthesized via an alternative (polyketide synthase -catalyzed) pathway. The δ13C of essential fatty acids in zooplankton and larval fish did not prove to reflect the δ13C of the same fatty acids in the seston better than other, non-essential fatty acids. As natural tags, δ13C values of the bulk and/or the fatty acid composition were found to be similarly succesful. However, when an animal moves into an area with isotopically distinct food, an unusual difference between the δ13C values of fatty acids that exhibit different turnover rates can be an indication for recent diet shift. When the various turnover rates are well constrained an estimate of the timing of the diet switch may be possible.

Examiners:

Dr. M.J. Whiticar, Supervisor (School of Earth and Ocean Sciences)

Dr. D.L. Mackas, Departmental Member (School of Earth and Ocean Sciences)

iv

Dr. V. Tunnicliffe, Departmental Member (School of Earth and Ocean Sciences)

Dr. L.A. Hobson, Outside Member (Department of Biology)

Dr. J.N.C. Whyte, Additional Member (Department of Fisheries and Oceans, Canada)

Dr. M.A. Teece, External Examiner (Department of Chemistry, College of Environmental Science and Forestry, State University of New York)

v

Table of Contents Abstract............................................................................................................................... ii Table of Contents ................................................................................................................v List of Tables .................................................................................................................... xii List of Figures ................................................................................................................. xiii Acknowledgements........................................................................................................ xviii 1. Introduction and objectives ........................................................................................... 1 1.1 Statement of problem ............................................................................................. 1 1.2 Objectives................................................................................................................. 3 1.3 Outline of thesis....................................................................................................... 4 1.4 References................................................................................................................ 5 2. Spatial food web characterization, and identification of movement between distinct pelagic food webs using the molecular, and stable carbon isotope composition of fatty acids and bulk sample. ..................................................................................................... 10 2.1 Abstract.................................................................................................................. 10 2.2 Introduction........................................................................................................... 11 2.3 Methods.................................................................................................................. 13 2.3.1 Collection, ship-handling and preparation of samples..................................... 13 2.3.2 Preparation of fatty acid methyl esters............................................................. 15 2.3.3 Chromatographic analysis................................................................................ 16 2.3.4 Gas chromatography-isotope ratio mass spectrometry (GC-IRMS)................ 17 2.3.4a GC-IRMS on fatty acid methyl esters........................................................ 19 2.3.4b δ13C correction for methyl group addition................................................. 22 2.3.4c Measurement of δ13C of methanol (of methanolic KOH).......................... 23 2.3.5 Isotope ratio mass spectrometry on bulk samples............................................ 24 2.3.6 Comparison of analytical accuracy and natural variability.............................. 24

vi 2.3.7 Multivariate Data analyses............................................................................... 26 2.3.7a Discriminant analysis ................................................................................. 26 2.3.7b Misclassification- and error rates............................................................... 27 2.3.7c Contribution to separation by each variable............................................... 28 2.3.7d Selection of Variables ................................................................................ 28 2.3.7e Data handling and estimation of missing values........................................ 30 2.4 Results .................................................................................................................... 30 2.4.1 Shelf - off shelf classification .......................................................................... 30 2.4.1a Fatty acids .................................................................................................. 30 2.4.1b Bulk stable carbon isotope composition .................................................... 33 2.4.1c Stable carbon isotope ratios of individual fatty acids ................................ 34 2.4.2 Comparison of techniques................................................................................ 35 2.4.3 Identification of transported or moved animals ............................................... 38 2.4.4. May 1998 LC4 and LC9 zooplankton ............................................................ 38 2.5 Discussion............................................................................................................... 41 2.5.1 Shelf-off shelf difference ................................................................................. 41 2.5.2 Stable carbon isotope composition of fatty acids ............................................ 43 2.5.3 Turnover in different organisms ...................................................................... 44 2.5.4 Carbon turnover in different tissues and biochemical fractions ...................... 44 2.5.5 Turnover of different fatty acids ...................................................................... 45 2.5.6 Conceptual model ............................................................................................ 46 2.5.7 Application and evaluation .............................................................................. 53 2.5.7a Recognizing a change in diet ..................................................................... 53 2.5.7b Tag longevity ............................................................................................. 54 2.5.7c Evaluation .................................................................................................. 55 2.6 Conclusions............................................................................................................ 56 2.7 References.............................................................................................................. 56 Appendix...................................................................................................................... 64

vii 3. Regional and Temporal Patterns in the Fatty Acid- and Stable Carbon Isotope Composition of Seston off the Coast of Vancouver Island, Canada.............................. 68 3.1 Abstract.................................................................................................................. 68 3.2 Introduction........................................................................................................... 69 3.3 Methods.................................................................................................................. 71 3.3.1 Physical parameters ......................................................................................... 71 3.3.2 Nutrients and chlorophyll a.............................................................................. 73 3.3.3 Phytoplankton identification............................................................................ 73 3.3.4 δ13C of dissolved inorganic carbon.................................................................. 73 3.3.5 Principal component analysis .......................................................................... 74 3.4 Results .................................................................................................................... 75 3.4.1 Phytoplankton taxonomy ................................................................................. 75 3.4.2 Linking fatty acid abundance to taxonomy...................................................... 77 3.4.3 PCA of fatty acid abundance data.................................................................... 78 3.4.4 Cross-shelf trends in δ13C, 16:2n-4 abundance and chl a ................................ 81 3.4.5 Nutrients........................................................................................................... 86 3.4.6 Physical parameters ......................................................................................... 86 3.4.7 Upwelling index............................................................................................... 90 3.4.8 δ13C of POM off the Washington coast ........................................................... 91 3.4.9 δ13C of DIC ...................................................................................................... 92 3.4.10 Correlation matrix.......................................................................................... 93 3.5 Discussion............................................................................................................... 94 3.5.1 The 16:2n-4 fatty acid as a diatom indicator ................................................... 94 3.5.2 Influence of land-derived POM ....................................................................... 96 3.5.3 Diatom effect?.................................................................................................. 96 3.5.4 Diatom favourable conditions........................................................................ 100 3.5.5 Other taxa....................................................................................................... 102 3.5.6 Growth rate .................................................................................................... 104 3.5.7 Cell Size ......................................................................................................... 105 3.5.8 [CO2(aq)] and active uptake of inorganic carbon .......................................... 105

viii 3.5.9 Differences between cruises .......................................................................... 108 3.5.10 Endeavour Segment POM............................................................................ 110 3.5.11 Synthesis and application............................................................................. 112 3.6 Conclusions.......................................................................................................... 114 3.7 References............................................................................................................ 115 Appendix.................................................................................................................... 125 4. Stable Carbon Isotope Ratios of Individual Fatty Acids in Marine Pelagic Organisms....................................................................................................................... 126 4.1 Abstract................................................................................................................ 126 4.2 Introduction......................................................................................................... 127 4.3 Methods................................................................................................................ 129 4.3.1 Sample collection and preparation................................................................. 129 4.3.2 Fatty acid extraction, methyl ester preparation and isotope ratio measurements ................................................................................................................................. 130 4.4 Results and discussion ........................................................................................ 130 4.4.1 Trophic transfer.............................................................................................. 130 4.4.2 Bulk – fatty acid δ13C difference ................................................................... 133 4.4.3 Differences in δ13C among individual fatty acids.......................................... 138 4.4.4 Source effect .................................................................................................. 141 4.4.5 Effect of preferential catabolism.................................................................... 142 4.4.6 Effect of different timing of lipid class synthesis .......................................... 144 4.4.7 Biosynthetic effects........................................................................................ 147 4.4.7a Hydrolysis of fatty acyl-ACP................................................................... 148 4.4.7b Desaturation ............................................................................................. 148 4.4.7c Elongation ................................................................................................ 151 4.4.7d The polyketide synthase pathway ............................................................ 153 4.4.8 Practical implications..................................................................................... 154 4.5 Conclusions.......................................................................................................... 155

ix 4.6 References............................................................................................................ 156 5. The Fatty Acid Composition of Marine Pelagic Organisms off Vancouver Island, Canada; Nature and Nurture ........................................................................................ 164 5.1 Abstract................................................................................................................ 164 5.2 Introduction......................................................................................................... 165 5.3 Methods................................................................................................................ 167 5.3.1 Sample collection........................................................................................... 167 5.3.2 Fatty acid methyl ester preparation and analyses .......................................... 167 5.3.3 δ15N measurements ........................................................................................ 167 5.3.4 Multivariate statistics ..................................................................................... 168 5.4 Results and Discussion........................................................................................ 169 5.4.1 Spatial variation in fatty acid composition .................................................... 169 5.4.2 Differences between trophic groups .............................................................. 173 5.4.3 Fatty acid proxy for trophic level?................................................................. 177 5.4.4 Literature data ................................................................................................ 178 5.4.5 Terrestrial animals and size ........................................................................... 183 5.4.6 Marine mammals; evolutionary control? ....................................................... 185 5.4.7 Fatty acids as trophic markers........................................................................ 187 5.5 Conclusions.......................................................................................................... 189 5.6 References............................................................................................................ 189 6. The Effect of Starvation on the Relative Abundance and δ13C of Fatty Acids in Rotifers (Brachionus plicatilis) ..................................................................................... 197 6.1 Abstract................................................................................................................ 197 6.2 Introduction......................................................................................................... 198 6.3 Methods................................................................................................................ 199 6.3.1 Experimental conditions and procedure......................................................... 199 6.3.3 T-iso samples and culture conditions............................................................. 200 6.3.4 Fatty acid methyl ester preparation and δ13C measurements......................... 201

x 6.4 Results .................................................................................................................. 201 6.4.1 Fatty acid composition of T-Iso and rotifers.................................................. 201 6.4.2 Effect of starvation on fatty acid profile ........................................................ 202 6.4.3 Unusual variation in stable carbon isotope ratios .......................................... 207 6.4.4 Effect of starvation on stable carbon isotope composition ............................ 208 6.4.5 Fecundity........................................................................................................ 209 6.5 Discussion............................................................................................................. 210 6.5.1 Decrease in total fatty acid concentration...................................................... 210 6.5.2 Selective mobilization of fatty acids.............................................................. 213 6.5.3 Retroconversion? ........................................................................................... 213 6.5.4 Shift in T-Iso δ13C before and after start of rotifer starvation ....................... 214 6.5.5 Difficulties for interpretation ......................................................................... 216 6.5.6 Observed changes in δ13C.............................................................................. 216 6.5.7 Conceptual model .......................................................................................... 217 6.5.8 Differential 13C-enrichment ........................................................................... 222 6.5.9 Recommendations.......................................................................................... 223 6.6 Conclusions.......................................................................................................... 223 6.7 References............................................................................................................ 224 Appendix.................................................................................................................... 228 7. Juvenile Salmon; a natural diet switch experiment ................................................. 231 7.1 Abstract................................................................................................................ 231 7.2 Introduction......................................................................................................... 232 7.3 Methods................................................................................................................ 233 7.3.1 Sample collection and location of sampling .................................................. 233 7.3.2 Handling and processing of samples.............................................................. 234 7.3.3 Fatty acid methyl ester preparation and isotope ratio measurements ............ 236 7.3.4 Multivariate analyses ..................................................................................... 236 7.4 Results and discussion ........................................................................................ 236

xi 7.4.1 Stable carbon- and nitrogen isotope composition.......................................... 236 7.4.2 δ13C and δ15N of gut contents ........................................................................ 239 7.4.3 Hypotheses for origin of conspicuous sockeye.............................................. 241 7.4.4 Fatty acid signatures ...................................................................................... 243 7.4.5 Origin of conspicuous sockeye ...................................................................... 245 7.4.6 δ13C difference between bulk and individual fatty acids ............................... 251 7.4.7 Distinguishing species with their fatty acid signature ................................... 251 7.5 Conclusions.......................................................................................................... 254 7.6 References............................................................................................................ 255 8. Conclusions ................................................................................................................ 259 8.1 Fatty acid composition........................................................................................ 259 8.2 Shelf - off shelf δ13C difference in POM ........................................................... 259 8.3 δ13C of individual fatty acids.............................................................................. 260 8.4 Application and evaluation ................................................................................ 261 9 Recommendations and outlook .................................................................................. 262 9.1 Polyketide Synthase pathway............................................................................. 262 9.2 n-3 PUFAs and human health........................................................................... 262 9.3 Evolutionary role and function of docosahexaenoic acid................................ 263 9.4 δ13C of fatty acids in algae.................................................................................. 264 9.5 Diet switch experiment ....................................................................................... 265 9.6 Lipid – lipid-free matter δ13C difference as trophic level indicator............... 266 9.7 Corroboration of chl a – δ13C relationship....................................................... 266 9.8 References............................................................................................................ 267

xii

List of Tables Table 2.1. Analytical precision estimates of fatty acid abundance, and δ13C of fatty acid measurements. .............................................................................................................. 25 Table 2.2. Results of discriminant analysis on combinations of fatty acids selected by stepwise discriminant analysis (top), and iterative search for lowest leave-one-out error rate (bottom)................................................................................................................ 31 Table 2.3. Comparison of misclassification/error rates for classification procedures using the 3 types of data........................................................................................................ 35 Table 2.4. List of samples identified by at least one classification procedure as consisting of organisms that moved off the shelf (first 9 samples), or to shelf waters (last 5 samples). .................................................................................................................... 38 Table 3.1. Correlation coefficients (R) and squared correlation coefficients (R2) for correlations between the abundance of fatty acids in 22 POM samples and [log(nano/diatom)]. .............................................................................................................. 78 Table 5.1. Results of discriminant analysis on shelf-off shelf groups and north-south groups, using combinations of fatty acids selected by stepwise discriminant analysis. ....... 168 Table 5.2. Ratios of the abundances of unsaturated over saturated fatty acids, and PUFAs over saturated and monounsaturated fatty acids. The abundance of the fatty acids from the n-3 and n-6 series, and their ratio. The abundance of 22:6n-3 (DHA). The ratios of 22:5n-3 (DPA) over 14:0, DHA over 14:0 and DHA over 20:5n-3 (EPA). .... 176 Table 5.3. References used measurements on whole organisms or muscle tissue of wild marine organisms (plot in Figure 5.8).......................................................................... 180 Table 6.1. Relative abundance of fatty acids in rotifers before and during starvation. ....... 203

xiii

List of Figures Figure 2.1. Map of study area with sample locations. ......................................................... 13 Figure 2.2. Gas chromatogram (FID trace) of fatty acid methyl esters derived from crab larvae sampled at station LC4 in May 1998 (a). The mass-44 ion current as a function of time, and the instantaneous ratio of the m/z-45 and m/z-44 ion currents (b).......................................................................................................................................... 18 Figure 2.3. The structures of eicosapentaenoic- (20:5n-3) and arachidonic acid (20:4n-6). .............................................................................................................................. 19 Figure 2.4. Frequency distribution of the discriminant scores of fatty acid data of POM, zooplankton and larval fish samples from all 3 cruises (79 shelf-, and 118 off shelf samples)........................................................................................................................ 32 Figure 2.5. Box and whisker plots of bulk stable carbon isotope composition of shelfand off-shelf samples plotted per cruise. .............................................................................. 34 Figure 2.6. Frequency distributions of discriminant scores of fatty acid δ13C data. ........... 36 Figure 2.7. Misclassification rates for samples of each cruise, trophic group and shelf break position plotted for the 3 classifications. .................................................................... 37 Figure 2.8. Match between misclassified samples as identified by discriminant analysis on the 3 data sets. ................................................................................................... 37 Figure 2.9. Stable carbon isotope composition of fatty acids from POM and zooplankton collected at station LC4 and LC9 in May ’98. ................................................. 39 Figure 2.10. Partial fatty acid composition of zooplankton (a) and POM (b) collected at LC4 and LC9 in May 1998. .............................................................................................. 40 Figure 2.11. Results of conceptual model, showing the modelled δ13CFA and fatty acid abundance in a euphausiid furcilia after a diet switch as a function of time. ...................... 48 Figure 2.12. The modeled difference between the δ13C of the 20:5n-3 and 18:3n-3 fatty acid over time. .............................................................................................................. 50 Figure 2.13. A normal probability plot (or qq plot) of the standardized quantiles of the δ13C differences between 20:5n-3 and 18:3n-3 fatty acids in zooplankton..................... 51 Figure 3.1. Map of study area with sample locations. ......................................................... 72 Figure 3.2. Pie charts showing the abundance, and proportion of different phytoplankton groups per station. ........................................................................................ 76

xiv Figure 3.3. The abundance of 16:2n-4 and several fatty acid ratios (or the logarithm of the ratio) measured for POM samples plot against [log(nano/diatom)].......................... 79 Figure 3.4. Loading plot and score plot of principal component analysis of POM sampled during all three cruises........................................................................................... 80 Figure 3.5. Trends in the 16:2n-4 fatty acid abundance, and δ13C of POM with distance offshore. .................................................................................................................. 82 Figure 3.6. The δ13C of POM (bulk sample) plotted against the abundance of the 16:2n4 fatty acid (a) and the chlorophyll a concentration in surface water (b)................... 83 Figure 3.7. Bulk POM δ13C plotted against the abundance of the 22:6n-3 and 16:2n-4 fatty acids.............................................................................................................................. 84 Figure 3.8. The 16:2n-4 fatty acid abundance plotted against the δ13C measured for the 18:4n-3 and 20:5n-3 fatty acids...................................................................................... 84 Figure 3.9. Maps of the chlorophyll a concentrations and the δ13C of POM....................... 85 Figure 3.10. Maps of the silicate- (Si(OH)4), and nitrate (NO3- + NO2-) concentrations....................................................................................................................... 87 Figure 3.11. Maps of the temperature, and salinity measured at 5m depth......................... 88 Figure 3.12. The δ13C of POM from all three cruises plotted against salinity, temperature, and the distance to the shelf-break (200m iso-bath), as well as the salinity plotted versus the distance to the shelf-break. ......................................................... 89 Figure 3.13. Box and whisker plots of the mixed layer depth in shelf- and off-shelf waters, plotted per cruise...................................................................................................... 90 Figure 3.14. Bakun Upwelling Index values for 48ºN 125ºW generated by the PFEL. ....... 91 Figure 3.15. Unpublished data provided by Drs. Brian Fry and Timothy Bates: δ13C measurements of POM and DIC from water collected off the Washington plotted against the distance offshore. Next to the bulk POM, also >20µm net POM δ13C is shown. ................................................................................................................................... 92 Figure 3.16. δ13C of POM plotted against the δ13C of DIC collected both at 5m depth in July 1999........................................................................................................................... 93 Figure 3.17. The 16:2n-4 fatty acid abundance in POM samples plot against the 16:1n-7/16:0 fatty acid ratio (a), and discriminant score (b)............................................... 95 Figure 3.18. Mixing curves produced when mixing a diatom end-member and a nondiatom end-member in which the same fatty acid supposedly has a δ13C value of -20‰ and -30‰, respectively. ........................................................................................................ 99

xv Figure 3.19. Mixed layer depth plotted versus the δ13C of bulk POM collected during May 1998 and May 1999, as well as the mixed layer depths observed in May 1998 against the offshore distance to the shelf-break (200m isobath), and salinity...................... 110 Figure 3.20. Fatty acid profile of POM from water (5m depth) above Endeavour Segment (station 2). . ............................................................................................................ 111 Figure 4.1. Differences in δ13C (∆δ13C) of fatty acids from zooplankton and larval fish with those in POM from the same location........................................................................... 131 Figure 4.2. Stable carbon isotope composition of fatty acids and bulk from organisms all collected at station LG3 in May 1998.............................................................................. 133 Figure 4.3. Comparison of the δ13C values of fatty acids (and bulk) from zooplankton and larval fish collected at stations in continental shelf waters (LBP2 and LG3) with those collected off shelf (LBP7 and LG7). ............................................................................ 134 Figure 4.4. Differences between the δ13C of fatty acids and the bulk of the sample for each of the trophic groups. ................................................................................................... 139 Figure 4.5. The median δ13C of fatty acids normalized against the 14:0 fatty acid and the weighted average, or total fatty acids, plotted for each of the trophic groups. .............. 140 Figure 4.6. Comparison with fatty acid δ13C values obtained by other authors. ................. 141 Figure 4.7. The median δ13C of fatty acids normalized to the δ13C of the bulk of 15 POM samples from three pre-selected groups: POM believed to be richest in diatoms, poorest in diatoms, and with the most bacterial matter (15 samples in each group). .......... 143 Figure 4.8. Mechanistic scheme for the desaturation of a fatty acid. .................................. 149 Figure 4.9. The median δ13C of fatty acids, normalized against the 14:0 fatty acid δ13C, plotted for the C18-series against the number of double bonds present in the respective fatty acids (a). In addition, the median δ13C of fatty acids, normalized against the 14:0 fatty acid δ13C, plotted for the (n-3)-series against the number of carbons present in the respective fatty acids (b)................................................................... 152 Figure 5.1. Loading plot and score plot of principal component analysis of fatty acid abundance data from cultured algae (derived from the literature). ..................................... 170 Figure 5.2. POM samples plotted on the first and second principal component axes that were produced by the principal component analysis on the algal culture data (Fig. 5.1). .............................................................................................................................. 171 Figure 5.3. Score-plot of two discriminant functions that optimally separate shelfand off shelf samples, and samples taken from northern and southern stations................... 172

xvi Figure 5.4. Loading plot and score plot of principal component analysis of fatty acid abundance data from POM, zooplankton and larval fish collected during three separate cruises. ................................................................................................................... 174 Figure 5.5. The relative abundance of the 14:0 (a) and 22:5n-3 (b) fatty acids plotted against that of 22:6n-3 (DHA). ............................................................................................. 175 Figure 5.6. The relative abundance of the 20:5n-3 fatty acid (EPA) plotted against that of 22:6n-3 (DHA)........................................................................................................... 177 Figure 5.7. The relative abundance of DHA (22:6n-3) plotted against the (bulk) δ15N of the same samples (a), showing an increase in DHA (%) per trophic level. The lower plots (b and c) show an increase of the DHA/14:0 and DHA/EPA fatty acid ratios with δ15N, respectively. ................................................................................................................. 179 Figure 5.8. Data derived from the literature (see Table 5.3 for references). The relative abundance of the 14:0 (a) and 20:5n-3 (b) fatty acids are plotted against that of 22:6n-3.............................................................................................................................. 182 Figure 6.1. Fatty acid profiles of Tahitian Isochrysis galbana and of the rotifers (Brachionus plicatilis) before and after 72 hours of starvation............................................ 202 Figure 6.2. Percent loss of concentration of the various fatty acids from 72 hours of starvation. The error bars represent ± 1 standard deviation, calculated from measurements on 3 batches of starved rotifers. Negative numbers indicate a rise in concentration. ....................................................................................................................... 205 Figure 6.3. Concentration measurements (triplicates) of four fatty acids plotted against time of starvation...................................................................................................... 206 Figure 6.4. Concentration and the δ13C of the total fatty acids (weighted average, with fatty acid abundance as weighting factor) plotted against time starved....................... 206 Figure 6.5. The δ13C of individual fatty acids in rotifers before and 2 hours into starvation compared to δ13C values of the same fatty acids in T-Iso sampled two days after the commencement of starvation study......................................................................... 207 Figure 6.6. Bulk δ13C measurements of rotifers plot against time of starvation. ................. 208 Figure 6.7. δ13C measurements during the 72 hour time series plotted for each fatty acid, with symbol size increasing with time of starvation..................................................... 209 Figure 6.8. Examples of the variation in δ13C of fatty acids during 72 hours of starvation. ............................................................................................................................. 210 Figure 6.9. Change in δ13C of fatty acids from 2 hours after the commencement (t=2 h) to 72 hours of starvation (t=72 h). ................................................................................... 211 Figure 6.10. Fecundity (average number of eggs per rotifer) during the 72 hours of starvation. ............................................................................................................................. 212

xvii Figure 6.11. Measured values of total fatty acid concentration compared to modeled amount of total fatty acid (top). Modeled δ13C of total fatty acids plot with the measured δ13C of total fatty acids during 72 hours of starvation of rotifers (bottom). ........ 219 Figure 6.12. Sensitivity of modeled δ13C of the total fatty acids to different parameters. ........................................................................................................................... 221 Figure 7.1 Map with juvenile salmon sample locations from May 1998 cruise (R988155). .................................................................................................................................... 234 Figure 7.2. Map with juvenile salmon sample locations from May 1999 and June 1999 cruises (HS9913 and HS9914, respectively)................................................................ 235 Figure 7.3. The δ13C of bulk muscle tissue of juvenile salmon and the δ13C of their potential marine prey collected in shelf waters (water depth < 200m). ............................... 237 Figure 7.4. The δ13C plotted against the δ15N of bulk muscle tissue of juvenile salmon from the May 1998 and May 1999 cruises............................................................................ 238 Figure 7.5. The δ13C and the δ15N of bulk muscle tissue of juvenile salmon from the May 1998 and May 1999 cruises plotted against their fork length. ..................................... 239 Figure 7.6. Comparison of the δ13C and δ15N values of gut contents with values for muscle tissue from the same juvenile salmon........................................................................ 240 Figure 7.7. The change in stable carbon, and nitrogen isotope composition of the juvenile salmon over time as proposed by three hypotheses................................................. 242 Figure 7.8. Loading plot and score plot of principal component analysis on the fatty acid abundance data from juvenile salmon from all three cruises. ...................................... 244 Figure 7.9. The ratio of the abundance of n-6 over n-3 fatty acids in muscle tissue of juvenile salmon plot against the δ13C of the same tissue. ..................................................... 245 Figure 7.10. Loading plot and score plot of principal component analysis on the fatty acid abundance data of sockeye salmon from all three cruises............................................ 246 Figure 7.11. The δ13C of muscle tissue of the “conspicuous sockeye salmon” from the May 1998 plot against their fork length and weight. ............................................................ 247 Figure 7.12. Differences between the δ13C of fatty acids and the bulk of muscle tissue of juvenile salmon. ................................................................................................................ 252 Figure 7.13. Results of discriminant analysis on fatty acid abundance data of muscle tissue of sockeye, chinook and coho salmon from all three cruises, showing separation between the four species. ...................................................................................................... 253

xviii

Acknowledgements I’d like to express my gratitude to my supervisor Michael Whiticar, who graciously invited me to join his lab when we met on the Dutch island Texel. He has not only given me the freedom to develop as a researcher, but at the same time kept an open door for whenever I had any burning questions, or to throw me a joke about wooden shoes. I am indebted to Ian Perry who joined me and showed how to do things on my first sampling trip ever on a research vessel. He read and corrected this thesis while not being on my supervisory committee. I thank him for his helpful discussions, hospitality and unstoppable kindness. This work would also not have been possible without Ian Whyte, who taught me “the Tao of Fatty Acids” in a Scottish accent. His supervision and lessons in life are greatly appreciated and remembered. I would like to thank Norma Ginther for her help in the Chemistry Lab at the Pacific Biological Station in Nanaimo, and Paul Eby at the Biogeochemistry Facility (UVic), who made me feel stupid around mass spectrometers, but still helped out. I am also indebted to Magnus Eek, not only for technical assistance, but also for the many discussions about isotope ratios of carbon in algae (yes, someone in E-Hut studied E-hux). Nicky Haigh’s help with counting T-Iso cells and her vocal entertainment at the lab in Nanaimo are also greatly appreciated and fondly remembered. Many thanks to Shannon Harris who shared her nutrient, chlorophyll and phytoplankton count data with me. Also, Drs. Brian Fry and Tim Bates are thanked for providing δ13C data of POM and DIC from waters off the Washington coast. Drs. Kim Hyatt and Paul Rankin offered juvenile salmon outmigration timing data. Special thanks go out as well to Alex Bec who showed me his fatty acid δ13C data, and shared French wine and Pastisse with me till deep in the night. The samples analyzed for this work would not have been taken if it weren’t for the help of the crews of the CCGS J.P. Tully and the CCGS W.E. Ricker. Especially the assistance of Doug Yelland, Doug Moore and John Morris is greatly appreciated. Paul Callow is thanked for providing me with a program that helped me process the fatty acid data faster. I am also indebted to Dr. Francis Zwiers for his patience and helpfulness while teaching me about multivariate statistics. The scientific discussions with Drs. Martin Kainz, Brian Fry and Michael Crawford have also improved this work. Not only would I like to thank my supervisory committee for correcting my thesis, but also my dad. Special thanks to Mark Teece, my external examiner who flew over from New York and still calls me “Ruben Romero” (now Dr. Romero), after a Flamenco guitarist advertised during a conference in Santa Fe, New Mexico. My time in Victoria has been made unforgettable and will be cherished deeply because of the many great friends I have met here. Sorry, I cannot mention all of you, but let me mention Kumar Ramachandran and Michael Riedel for the countless lunches in the geophysics lab that nourished body, mind and soul. A tremendous source of support and fun were my roommates, of which I especially give warm thanks to David Mate, Gloria Lopez, Sean Bailey and Paul Flueck. How lucky I am to have shared so many good times with you. We were one big family. Family here in Canada has provided me the love and support that I can’t even begin to express. My aunt Marco, Rod, and my cousins Eamon and Raffi here in Victoria integrated me into their family, like I was a son or brother to them. Thank you so much….. Last, but not least, I want to thank my mom, dad, Victor, my brother, and also Sheila for supporting me all the way. Thanks for believing in me. Thanks for the many postcards, mom. Your love was felt from far away.

xix

“Sit down before fact as a little child, be prepared to give up every preconceived notion, follow humbly wherever and to whatever abysses nature leads, or you shall learn nothing. I have only begun to learn content and peace of mind since I have resolved at all risks to do this.”

- T.H. Huxley, from letter to Charles Kingsley, September 23, 1860

1

1. Introduction and objectives

1.1 Statement of problem Tracing the movement of organisms and their prey are critical components in the conservation efforts and management of marine populations. Studies investigating the migration or movement of animals traditionally rely on marking or tagging, followed by recapture. Additionally, radio- and satellite transmitters have allowed researchers to study migratory routes in great detail. The disadvantage of physical tags is that re-capture is necessary, which may imply that large numbers of individuals will have to be tagged for the study to be successful. Furthermore, tags in general and transmitters in particular can typically only be used on animals that are relatively large. More recently, characterization of stocks using DNA has become a useful method to track the source of animals (Ball et al., 1988; Wenink et al., 1994; Beacham et al., 2000; Whitler et al., 2000). In addition, fatty acid- (Smith et al., 1996), elemental(Gillanders, 2002; Jessop et al., 2002) and stable isotope composition measurements (reviewed by Hobson, 1999) have been employed in order to trace movements of organisms. To be able to apply these methods, the existence and knowledge of the regional variation in the respective compositions of the animals and their diet is required. Due to their applicability on small organisms, and because no extensive release and recapture programs are necessary, these tools can be a very useful addition to the use of tags and transmitters in movement studies. The continental shelf off the West Coast of Vancouver Island is a highly productive fishing region, supporting commercial fisheries of, for example, salmon, herring, hake, cod, sablefish and shellfish. The total yield of the various stocks has undergone large fluctuations during the last century (Hollowed and Wooster, 1995; Rothschild, 1995; Henderson and Graham, 1998; Finney et al., 2002). A large number of studies have shown that many changes in marine ecosystems are linked to shifts in ocean climate (Beamish and Bouillon, 1993, 1995; Hare and Francis, 1995; Mantua et al., 1997; Noakes

2 et al., 1998; Welch et al., 1998). However, the exact mechanisms coupling the physical changes to the observed fluctuations in fish stocks have not entirely been uncovered (Verity et al., 2002). One of the processes that need further elucidation is the importance of advective washout of shelf organisms and cross-shelf exchange between shelf, and oceanic food webs. Perry et al. (1999) found that off southwestern Vancouver Island the particulate organic matter (POM) in shelf waters was enriched in 13C relative to the adjoining waters of the deeper ocean. The difference in

13

C/12C ratios in the POM persisted through

zooplankton to larval fish. Hence, Perry et al. (1999) demonstrated the potential of using stable carbon isotope ratios to differentiate between pelagic food webs from the adjacent shelf- and deep ocean water masses. The 13C/12C ratio of a heterotroph is related to that of its diet (DeNiro and Epstein, 1978; Peterson and Fry, 1987; Post, 2002). Therefore, measurements of stable carbon isotope ratios (in combination with 15N/14N ratios) have been employed to determine the relative contribution of various food sources with distinct isotope signatures (Kline et al., 1993; Ben-David et al., 1997a, b; Whitledge and Rabeni, 1997; Szepanski et al., 1999; Ben-David and Schell, 2001; Phillips and Koch, 2002). Several problems arise when interpreting

13

C/12C ratio data of measurements on animal tissue. Due to variation in

condition and diet of the animals, the proportions of the various biochemical fractions may vary. For example, changes in the proportion of lipids, which are depleted in

13

C

relative to other constituents (Abelson & Hoering, 1961; Park & Epstein, 1961; Parker, 1964; DeNiro and Epstein, 1977, 1978), will affect the carbon isotope signature. Additionally, differences in efficiency of assimilation of dietary components may result in inaccurate estimates of the contribution from different food sources (Gannes et al., 1997; Ben-David and Schell, 2001). A problem arising when using non-lethal methods (i.e., sampling of adipose tissue, feathers, hair etc.) is the fact that the 13C/12C ratio of the sample obtained may not be representative of the whole animal. This is partly due to the different contribution of the various compound classes. Some of the aforementioned problems may be avoided by comparing the stable carbon isotope ratios of individual compounds measured in the animal as well as its

3 potential prey. This study focuses on the use of 13C/12C ratios of individual fatty acids as dietary markers and “natural tags” that could be used in the studies of movement of organisms. Lipids are relatively easily extracted, identified and quantified as compared to other major biochemical fractions, such as proteins and carbohydrates. Fatty acids in particular are effective in this study because they are present in every living cell and display great structural diversity (Sargent et al., 1988). The fatty acid composition of organisms is influenced by their diet, and therefore it provides time-integrated information on the dietary history of the animal. Indeed, the use of fatty acids as trophic markers has been shown successful (Lee et al., 1971; Fraser et al., 1989; Graeve et al., 1994; Desvilettes et al., 1994; Smith et al., 1996; St. John and Lund, 1996; Napolitano et al., 1997; FalkPetersen et al., 2000; Virtue et al., 2000). Some fatty acids that are essential for the structural integrity of the membranes cannot be synthesized in sufficient amount, or not at all, by the organism. These fatty acids are known as “essential fatty acids”, and for most animals include many of the n-3 and n-6 fatty acids (also known as omega-3 and -6 fatty acids) (Cook, 1996). Because a minimum of additional synthesis and modification can be expected during the transfer through the food web, the 13C/12C ratios of the essential fatty acids may be good trophic markers. Stable carbon isotope ratios of individual fatty acids have not been applied as natural tags in marine organisms before, and assessing their transfer through more than one trophic linkage is equally new. Here, the use of

13

C/12C ratios of individual fatty

acids in ecological studies are assessed and compared with the use of the fatty acid composition-and bulk isotope data. Also, the use of all three of these types of measurements in tandem is explored. Samples of POM, zooplankton, larval fish and juvenile salmon collected off the west coast of Vancouver Island were analyzed.

1.2 Objectives The two primary objectives of this study were:

4 •

To evaluate the applicability of 13C/12C ratios of individual fatty acids as natural tags in marine pelagic organisms

•

To assess the use of 13C/12C ratios of fatty acids in the elucidation of trophic links and determination of the dietary history of marine organisms

In order to meet these primary objectives the following secondary objectives were set: •

To map and understand the spatial differences in

13

C/12C ratios and fatty acid

composition of POM (at the base of the food webs) off the west coast of Vancouver Island under different oceanographic conditions •

To advance the understanding of the variation in

13

C content of individual fatty

acids and factors influencing their 13C/12C ratio during trophic transfer •

To compare the applicability of the stable carbon isotope ratio of individual fatty acids with the use of

13

C/12C ratios of bulk sample as well as fatty acid

composition data in ecological studies

1.3 Outline of thesis The following six chapters are written as stand-alone papers. Further background on the relevant topics is introduced in each chapter. After these six chapters the general conclusions will follow, together with recommendations and outlook for future research. Chapter 2 describes the methods and procedures followed in detail. Additionally, it evaluates the applicability of

13

C/12C measurements on individual fatty acids and

compares it with the use of bulk stable carbon isotope ratios and fatty acid composition data as natural tags. Focus is on the use of these techniques in the investigation of movement of organisms between the shelf- and adjacent open ocean water masses. In order to use the molecular- and stable carbon isotope composition of organisms as natural tags, it first has to be established whether indeed large enough differences exist between regions. Additionally, to predict the future use of these techniques in particular

5 areas it is important to understand why such differences in fatty acid- and isotope compositions exist. Therefore, in Chapter 3 the spatial and temporal variation in stable carbon isotope ratios and fatty acid composition of POM are described. These variations are compared with parameters such as temperature, salinity, nutrient- and chlorophyll a concentrations, and phytoplankton species composition. In Chapter 4 the stable carbon isotope ratios of individual fatty acids from POM, zooplankton, larval fish and juvenile salmon are compared. Additionally, the potential mechanisms causing the differences in

13

C/12C ratios between individual fatty acids are

assessed and evaluated. Chapter 5 explores the variations observed in the fatty acid composition of the various trophic groups. The data obtained for this study are compared to data collected from the literature. Additionally, the use of fatty acids as trophic markers is discussed. As fatty acids are transferred to higher trophic levels a great portion is lost via catabolism. In Chapter 6 a starvation experiment using marine rotifers to study the effect of catabolism on the 13C/12C ratios of fatty acids is described. In Chapter 7 the molecular- and stable carbon isotope composition of fatty acids, as well as the 13C/12C and 15N/14N ratios of bulk muscle tissue from juvenile salmon are reported. The migration of juvenile salmon to the ocean offers an ideal case study of a natural diet shift. This chapter investigates the effects of the switch from the freshwaterto the marine diet on the fatty acid composition and on the 13C/12C ratio of the bulk and individual fatty acids. Chapter 8 summarizes the conclusions of this thesis. Chapter 9 provides recommendations and an outlook for future studies.

1.4 References Abelson, P.H. and Hoering, T.C. 1961. Carbon isotope fractionation in formation of amino acids by photosynthetic organisms. Proceedings of the National Academy of Science U.S.A. 47, 623-632.

6

Ball, R.M., Freeman, S., James, F.C., Bermingham, E. and Avise, J.C. 1988. Phylogeographic population structure of red-winged blackbirds assessed by mitochondrial DNA. Proceedings of the National Academy of Sciences, USA 85, 1558-1562. Beacham, T.D., Le, K.D., Raap, M.R., Hyatt, K., Luedke, W. and Withler, R.E. 2000. Microsatellite DNA variation and estimation of stock composition of sockeye salmon, Oncorhynchus nerka, in Barkley Sound, British Columbia. Fishery Bulletin 98, 14-24. Beamish, R.J. and Bouillon, D.R. 1993. Pacific salmon production trends in relation to climate. Canadian Journal of Fisheries and Aquatic Sciences 50, 1002-1016. Beamish, R.J. and Bouillon, D.R. 1995. Marine fish production trends off the Pacific coast of Canada and the United States. In: R.J. Beamish (ed.) Climate change and northern fish populations. Canadian Special Publication of Fisheries and Aquatic Sciences 121, p. 585-591. Ben-David, M., Flynn, R.W. and Schell, D.M. 1997a. Annual and seasonal changes in diets of martens: evidence from stable isotope analysis. Oecologia 111, 280-291. Ben-David, M., Hanley, T.A., Klein, D.R. and Schell, D.M. 1997b. Seasonal changes in diets of coastal and riverine mink: the role of spawning Pacific salmon. Canadian Journal of Zoology 75, 803-811. Ben-David, M. and Schell, D.M. 2001. Mixing models in analyses of diet using multiple stable isotopes, a response. Oecologia 127, 180-184. Cook, H.W. 1996. Fatty acid desaturation and chain elongation in eukaryotes. In: D.E. Vance and J.E. Vance (Eds.) Biochemistry of Lipids, Lipoproteins and Membranes, New Comprehensive Biochemistry 31, Elsevier Science, Amsterdam, p. 363-389. DeNiro, M.J. and Epstein, S. 1977. Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197, 261-263. DeNiro, M.J. and Epstein, S. 1978. Influence of diet on the distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta 42, 495-506. Desvilettes, Ch., Boudier, G., Breton, J.C. and Combrouze, Ph. 1994. Fatty acids as organic markers for the sudy of trophic relationships in littoral cladoceran communities of a pond. Journal of Plankton Research 16, 643-659. Falk-Petersen, S., Hagen, W., Kattner, G., Clarke, A., Sargent, J. R. 2000. Lipids, trophic relationships, and biodiversity in Arctic and Antarctic krill. Canadian Journal of Fisheries and Aquatic Sciences 57, Supplement 3, 178-191.

7 Finney, B.P., Gregory-Eaves, I., Douglas, M.S.V. and Smol, J.P. 2002. Fisheries productivity in the northeastern Pacific Ocean over the past 2,200 years. Nature 416, 729-733. Fraser, A.J., Sargent, J.R., Gamble, J.C. and Seaton, D.D. 1989. Formation and transfer of fatty acids in an enclosed marine chain comprising phytoplankton, zooplankton and herring (Clupea harengus L.) larvae. Marine Chemistry 27, 1-18. Gannes, L.Z., O’Brien, D.M. and del Rio, C.M. 1997. Stable isotopes in animal ecology: Assumptions, caveats, and a call for more laboratory experiments. Ecology 78, 1271-1276. Gillanders, B.M. 2002. Temporal and spatial variability in elemental composition of otoliths: implications for determining stock identity and connectivity of populations. Canadian Journal of Fisheries and Aquatic Sciences 59, 669-679. Graeve, M., Kattner, G. and Hagen, W. 1994. Diet-induced changes in the fatty acid composition of Arctic herbivorous copepods: experimental evidence of trophic markers. Journal of experimental marine biology and ecology 182, 97-110. Hare, S.R. and Francis, R.C. 1995. Climate change and salmon production in the northeast Pacific ocean. In: R.J. Beamish (ed.) Climate change and northern fish populations. Canadian Special Publication of Fisheries and Aquatic Sciences 121, p. 357-372. Henderson, M.A. and Graham, C.C. 1998. History and status of Pacific salmon in British Columbia. North Pacific Anadromous Fish Commission Bulletin 1, 13-22. Hobson, K.A. 1999. Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia, 120, 314-326. Hollowed, A.B. and Wooster, W.S. 1995. Decadal-scale variations in the eastern subarctic Pacific: II. Response of northeast Pacific fish stocks. In: R.J. Beamish (ed.) Climate change and northern fish populations. Canadian Special Publication of Fisheries and Aquatic Sciences 121, p.373-385. Jessop, B.M., Shiao, J.C., Iizuka, Y. and Tzeng, W.N. 2002. Migratory behaviour and habitat use by American eels Anguilla rostrata as revealed by otolith microchemistry. Marine Ecology Progress Series 233, 217-229. Kline, T.C. Jr., Goering, J.J., Mathisen, O.A., Poe, P.H., Parker, P.L. and Scalan, R.S. 1993. Recycling of elements transported upstream by runs of Pacific salmon. II. δ15N and δ13C evidence in the Kvichak River watershed, Bristol Bay, southwestern Alaska. Canadian Journal of Fisheries and Aquatic Sciences 50, 2350-2365. Lee, R.F. Nevenzel, J.C. and Pfaffenhöfer, G.A. 1971. Importance of wax esters and other lipids in the marine food chain: phytoplankton and copepods. Marine biology 9, 99-108.

8 Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M. and Francis, R.C. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society 78, 10691079. Napolitano, G.E., Pollero, R.J., Gayoso, A.M., MacDonald, B.A. and Thompson, R.J. 1997. Fatty acids as trophic markers of phytoplankton blooms in the Bahía Blanca estuary (Buenos Aires, Argentina) and in Trinity Bay (Newfoundland, Canada). Biochemical Systematics and Ecology 25, 739-755. Noakes, D.J., Beamish, R.J., Klyashtorin, L. and McFarlane, G.A. 1998. On the coherence of salmon abundance trends and environmental factors. North Pacific Anadromous Fish Commission Bulletin 1, 454463. Park R. and Epstein, S. 1961. Metabolic fractionation of 13C and 12C in plants. Plant Physiology 36, 133138. Parker, P.L. 1964. The biogeochemistry of the stable isotopes of carbon in a marine bay. Geochimica et Cosmochimica Acta 28, 1155-1164. Perry, R.I., Thompson, P.A., Mackas, D.L., Harrison, P.J. and Yelland, D. 1999. Stable carbon isotopes as pelagic food web tracers in adjacent shelf and slope regions off British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 56, 2477-2486. Peterson B.J. and Fry B. 1987. Stable isotopes in ecosystem studies. Annual review of ecology and systematics 18, 293-320. Phillips, D.L. and Koch, P.L. 2002. Incorporating concentration dependence in stable isotope mixing models. Oecologia 130, 114-125. Post, D.M. 2002. Using stable isotopes to estimate trophic position: models, methods and assumptions. Ecology 83, 703-718. Rothschild, B.J. 1995. Fishstock fluctuations as indicators of multidecadal fluctuations in the biological productivity of the ocean. In: R.J. Beamish (ed.) Climate change and northern fish populations. Canadian Special Publication of Fisheries and Aquatic Sciences 121, p. 201-209. Sargent, J.R., Parks, R.J., Mueller-Harvey, I. and Henderson, R.J. 1988. Lipid biomarkers in marine ecology. In: M.A. Sliegh (ed) Microbes in the sea. Ellis Horwood Ltd., Chichester, U.K., p.119-138. Smith, R.J., Hobson, K.A., Koopman, H.N. and Lavigne, D.M. 1996. Distinguishing between populations of fresh- and salt-water harbour seals (Phoca vitulina) using stable-isotope ratios and fatty acid profiles. Canadian Journal of Fisheries and Aquatic Sciences 53, 272-279. St. John, M.A. and Lund, T. 1996. Lipid biomarkers: linking the utilization of frontal plankton biomass to enhanced condition of juvenile North Sea cod. Marine Ecology Progress Series 131, 75-85.

9

Szepanski, M.M., Ben-David, M. and Van Ballenberghe, V. 1999. Assessment of anadromous salmon resources in the diet of the Alexander Archipelago wolf using stable isotope analysis. Oecologia 120, 327335. Verity, P.G., Smetacek, V. and Smayda, T.J. 2002. Status, trends and the future of the marine pelagic ecosystem. Environmental Conservation 29, 207-237. Virtue, P., Mayzaud, P., Albessard, E. and Nichols, P. 2000. Use of fatty acids as dietary indicators in northern krill, Meganyctiphanes norvegica, from northeastern Atlantic, Kattegat, and Mediterranean waters. Canadian Journal of Fisheries and aquatic Sciences 57, Supplement 3, 104-114. Welch, D.W., Ishida, Y. and Nagasawa, K. 1998. Thermal limits and ocean migrations of sockeye salmon (Oncorhynchus nerka): long-term consequences of global warming. Canadian Journal of Fisheries and Aquatic Sciences 55, 937-948. Wenink, P.W., Baker, A.J. and Tilanus, M.G.J. 1994. Mitochondrial control-region sequences in two shorebird species, the turnstone and the dunlin, and their utility in population genetics studies. Molecular and Biological Evolution 11, 22-31. Withler, R.E., Le, K.D., Nelson, R.J., Miller, K.M. and Beacham, T.D. 2000. Intact genetic structure and high levels of genetic diversity in bottlenecked sockeye salmon (Oncorhynchus nerka) populations of the Fraser River, British Columbia, Canada. Canadian Journal of Fisheries and Aquatic Sciences 57, 19851998. Whitledge, G.W. and Rabeni, C.F. 1997. Energy sources and ecological role of crayfishes in an Ozark stream: insights from stable isotopes and gut analysis. Canadian Journal of Fisheries and Aquatic Sciences 54, 2555-2563.

10

2. Spatial food web characterization, and identification of movement between distinct pelagic food webs using the molecular, and stable carbon isotope composition of fatty acids and bulk sample.

2.1 Abstract Three types of data: 1) stable carbon isotope ratios of whole samples, 2) stable carbon isotope ratios of individual fatty acids, and 3) fatty acid abundance data, were evaluated on their ability to identify marine pelagic animals that moved from one region to another. In order to make these identifications it first had to be established that food webs of the two regions can be differentiated and characterized by the techniques applied. Samples of particulate organic matter (POM), zooplankton and larval fish were collected in May '98, May '99 and July '99 on multiple shelf to off-shelf transects off the west coast of Vancouver Island. Shelf and off-shelf food webs in May 1998, and to a lesser extent in 1999, could be distinguished on the basis of the relative concentrations of fatty acids, as well as the

13

C/12C ratio of whole samples and individual fatty acids.

Discriminant analysis and variable selection techniques were employed to make this distinction. When using fatty acid abundance data or carbon isotope ratios of fatty acids, the overall success rates of assigning samples to their respective shelf or off shelf origin was estimated to be around 85%. Utilizing stable carbon isotope ratios of the whole sample resulted in a successful classification for approximately 80% of the samples. After animals have moved to a region with different food type, the various biochemical fractions in the organism are expected to lose the old dietary signature at different rates. Some indication of this effect was found in animals that are hypothesized to have switched between shelf- and off shelf waters. An unusual difference between the lipid and protein carbon isotope compositions, as a result of the “disequilibrium” between the animal tissue and the new diet, is therefore tentatively proposed as a tool to confirm a

11 recent dietary shift. Additionally, it was demonstrated that after a change in diet the 13

C/12C ratio adjusts to present diet values at different rates for each of the individual

fatty acids. These results show the potential use of this effect to obtain time constraints on animal movement between two dietary regimes. A conceptual model was constructed to estimate the time before the fatty acid

13

C/12C ratio signature of the previous diet is

obscured due to metabolic turnover and growth.

2.2 Introduction Migration studies often utilize tags and transmitters to follow the movement of an animal. When the studied organisms are too small or too numerous for such approaches, other techniques have to be invoked. Given a regional variation in diet composition, fatty acid compositions can help to identify the origin of an organism (e.g., Castell et al., 1995; St. John and Lund, 1996). The relative abundances of fatty acids in the food intake greatly influence the fatty acid composition of a heterotroph (e.g., Frolov et al., 1991; Graeve et al., 1994; Ederington et al., 1995). Hence, the analysis of the fatty acid composition of animals has also been proven useful as trophic markers (e.g., Fraser et al., 1989; Desvilettes et al., 1994; St. John and Lund, 1996; Napolitano et al., 1997; Virtue et al., 2000). Few fatty acids are known to be unique to a single group of species. Therefore, ratios of two or more fatty acids and multivariate techniques are often used to establish trophic relationships. Some success has also been achieved by the use of stable isotope ratios as natural tags in migratory movement of animals (Fry, 1981; Hesslein et al., 1991; Hansson et al., 1997; Rubenstein et al., 2002). Such studies capitalize on regional variability and significant differences in the isotope composition of dietary constituents. Several factors can complicate the interpretation of stable carbon isotope data obtained from the traditional whole sample analysis. For example, changes in the relative contribution of the various biochemical fractions (e.g., proteins, carbohydrates and lipids). This is due to differences in the isotope composition of the different compound classes (Abelson and Hoering, 1961; DeNiro and Epstein, 1978). To circumvent the

12 limitations of measurements on whole samples, a growing number of researchers study trophic linkages with 13C/12C ratio measurements of individual compounds (e.g., Fang et al., 1993; Abrajano et al., 1994; Canuel et al., 1995; Pond et al., 1997; Trust Hammer et al., 1998; Boschker et al., 1999; Fantle et al., 1999). These compounds include fatty acids, sterols and amino acids. To measure the isotope ratio information contained in individual compounds a technique has been developed using gas chromatography-isotope ratio mass spectrometry (GC-IRMS) (Hayes et al., 1990) In this chapter Elemental Analyzer-IRMS, GC-IRMS, and fatty acid profile analysis are evaluated in their ability to identify animals that actively moved or were transported into a region with distinct food characteristics. In order to make such a distinction it first was established that spatially separated food webs can indeed be differentiated and characterized with these techniques. Perry et al. (1999) found that during Spring 1992 the pelagic food web in shelf waters were more

13

C enriched than organisms collected off-shelf of Vancouver Island.

In the present study, bulk stable carbon isotope ratio measurements on seston, zooplankton and larval fish, also determined by Perry et al. (1999), were complemented with δ13C measurements of individual fatty acids, and determination of fatty acid profiles. The analyzed samples were collected on three cruises (during May 1998, May and July 1999) off the west coast of Vancouver Island, B.C., Canada. The isotope composition of animals that moved into a region with a different food quality, will be adjusting, and eventually assume the same value as the new diet. Hence, the initial tag is lost after some time and is replaced by a tag for the new region. The period of disequilibrium can be investigated in more detail with δ13C data of the individual fatty acids. Application of these data in placing time constraints on the movement of an animal is also discussed.

13

2.3 Methods 2.3.1 Collection, ship-handling and preparation of samples Samples of POM, zooplankton and larval fish were collected on multiple on-shelf to offshelf transects off the west coast of Vancouver Island during three surveys in 1998 and 1999 (Figure 2.1). All sampling was done on-board the C.S.S. John P. Tully during the cruises: IOS9810 (May 12-24, 1998), IOS9911 (May 04-12, 1999) and IOS9928 (June 30

Figure 2.1. Map of study area. At all stations POM was collected, and at stations indicated by the closed circles zooplankton and larval fish was also collected (no larval fish on CS-line). Open circles with vertical line represent stations sampled in May ’98, horizontal: May ’99, and diagonal lines: July ’99. P10 was only sampled in May ’98, P8 only in May ’99, and samples were only collected at ER2 during July ’99. The 200m-isobath is defined as the shelf-break and is indicated with a thick black line.

14 – July 09, 1999). POM samples were collected by utilizing the “Sea-Loop”, a clean sea-water system which pumps unfiltered water from the intake at the bow (ca. 5m depth) to the wet lab of the ship. The water was passed through a 142mm diameter pre-combusted (450˚C for >2h.) Gelman A/E glass fibre filter (1µm effective pore diameter). No pre-filter was used before the glass fibre filter. As a result of the single filter method one or more zooplankton animals were occasionally found on the filter. All filters were visually checked on-board the ship and when zooplankton animals were observed they were carefully removed from the filter with forceps. Subsequently, the filters were folded and wrapped in pre-combusted aluminium foil and kept frozen in zip-lock bags at approximately -20˚C during the cruise.

Zooplankton was collected by a bongo plankton net towed obliquely (at a vessel speed of 1 knot) from 50m depth to the surface. Each sampler had a 0.25m2 mouth opening and a nominal mesh size of 236µm. During the 1998 cruise, in addition to using a bongo net, a ringnet with a 0.25m2 mouth opening and 100µm mesh was towed vertically from 50m water depth to the surface. All samples were first screened through a 425µm- and then a 212µm mesh netting to produce three size fractions (two fractions in 1999). Each fraction was rinsed numerous times with filtered (1µm) seawater until no phytoplankton “contamination” was visible with the naked eye. When a high abundance of one species was found, some individuals from that species were picked out to form a single species sample. All zooplankton samples were rinsed with an isotonic solution of ammonium formate in distilled water to eliminate sea salt before storing the samples in zip-lock bags at -20ºC on-board the ship.

Larval fish were collected at night from the surface by towing a neuston sampler (500µm mesh) at a speed of 3 knots for 15 minutes. The fish were sorted by species, length measurements were taken, and then stored frozen at -20ºC.

15

On land, all samples were transferred to an Ultra-Cool freezer in which they were kept at -80ºC until further processing. Samples were freeze-dried within a few months after collection at sea and were ground to a fine, homogeneous powder with pestle and mortar. The powdered samples were kept in screw cap vials at -80ºC until usage for the fatty acid methyl ester preparation (generally within weeks after freeze drying).

2.3.2 Preparation of fatty acid methyl esters The procedure of Whyte (1988) was used for the in situ saponification of the samples and methylation of the fatty acids. For clarity, the procedure followed is outlined below.

When available, at least 20mg of freeze-dried and powdered sample was added to a 5 ml Reacti-Therm Vial to which a known quantity (typically around 60µg) of heneicosanoic acid (21:0) had been added in advance as an internal standard. The sample containing vials were evacuated in a vacuum oven at room temperature with the cap loosely fitted to the vials. After evacuation the oven was vented with nitrogen and the vials were tightly sealed with Mininert valves. Through the valve septum 1 ml of 0.5M methanolic potassium hydroxide was added with a glass syringe. The mixtures were stirred with a vortex mixer, placed in a heating block (Reacti-Therm module), and left at 85ºC for 30 minutes. When cooled down, 1ml of hexane was added, the contents were stirred, and after the mixture had settled the upper hexane layer was carefully withdrawn with a glass syringe and discarded. This step was repeated one or more times to ensure no unsaponified organic material was left behind. Boron trifluoride-methanol (2ml) was added as an esterification reagent and, after mixing, the vials were kept in the heating block at 85ºC for 15 minutes, and then cooled for about 10-15 minutes. 1ml hexane and 0.5ml of saturated aqueous sodium chloride (to increase the density- and solubility difference between the two layers) were then added, the mixture was stirred and briefly centrifuged, after which the lower aqueous layer was

16 taken up by a glass syringe and discarded. The solution left behind, was then washed with (1.5-2ml) saturated aqueous sodium bicarbonate (washings were discarded). The hexane layer, containing the fatty acid methyl esters (FAMEs) was pipetted off and transferred into a 1ml vial, which was covered with aluminium foil and evaporated in the vacuum oven (at room temperature). After venting the oven with nitrogen the FAMEs were transferred again in hexane into another, pre-weighed, tapered vial fitted loosely with a Teflon-lined crimp cap with septum. Upon evaporation of the hexane and nitrogen venting in the vacuum oven, the caps were crimped on tightly, and the vials weighed again. The FAMEs were dissolved in an adequate amount of ethyl acetate (generally about 100µl per 1mg of FAME) for later chromatographic analysis.

The FAME preparation procedure described above was employed for the zooplankton and fish samples. In the case of the POM samples, the method was modified slightly. To obtain sufficient material, a quarter of the glass fibre filter was cut up in small squares (0.1-0.2cm2) and added to the vial. Since not all of the sample would be immersed in the methanolic KOH by adding only 1ml, more was added. Because the addition of BF3 methanol would overfill the vial due to the extra methanolic KOH, the vial was opened and placed under a gentle, constant stream of nitrogen till the liquid level was reduced sufficiently by evaporation. The cap was then screwed on tightly again and the rest of the procedure followed as for the zooplankton and fish samples.

2.3.3 Chromatographic analysis The individual FAMEs dissolved in ethyl acetate were separated by gas liquid chromatography, using a Hewlett-Packard 5890 gas chromatograph (GC) fitted with a Supelcowax 10 fused silica capillary column (30m, 0.32mm ID, 0.25µm film thickness). The GC was equipped with a flame ionisation detector and a split injector. Helium was the carrier gas and a 1:5 split ratio was used. When samples were introduced, the oven temperature was set at 180ºC and held isothermal for the first 35 minutes of the run. Then

17 the temperature was ramped up to 240ºC at 2ºC/min and kept at there for 25 minutes. A typical chromatogram produced under these conditions is shown in Figure 2.2a. A Hewlett Packard 3393A integrator (interfaced with a computer) was connected to the GC. Peaks were quantified as percentage of total area (minus internal standard). Peaks of less than 0.2% of the total area were not included in the fatty acid profile. When the weight of the sample analyzed was known, which was not the case for POM, concentrations of the analytes could be inferred by comparing the peak areas with the area of the internal standard. Peaks were identified by comparison with the cod liver oil lab standard (with known composition), which was run in concert with the samples, and by comparison with equivalent chain length values from the literature, e.g., Ackman (1986). Standard shorthand nomenclature (L:Bn-X) is used for the identified fatty acids, where L is the number of carbon atoms, B the number of double bonds, and n-X denotes the number of carbon atoms from the double bond in the terminal region of the molecule (assuming methylene-interrupted cis-double bonds). For example, the 20:5n-3 fatty acid has 20 carbon atoms, 5 double bonds, with the double bond closest to the methyl-end situated between the carbon atoms at positions 17 and 18. For illustration, the structures of eicosapentaenoic- (20:5n-3) and arachidonic acid (20:4n-6) are drawn in Figure 2.3. To test the reproducibility of relative abundances, i.e., percentage of total fatty acids, 10 sub-samples with varying weights were separately prepared and measured. No trends were observed with weight analyzed, showing that accurate measurements could be made on only 2mg of dry homogenized sample. The standard deviations of relative abundances vary between 0.007 to 0.19 percentage point for the different fatty acids measured (Table 2.1).

2.3.4 Gas chromatography-isotope ratio mass spectrometry (GCIRMS)

18

a

b

Figure 2.2. Gas chromatogram (FID trace) of fatty acid methyl esters derived from crab larvae sampled at station LC4 in May 1998, using the Supelcowax 10 column (a). The peaks in the lower half (b) represent CO2 from combusted fatty acid methyl esters, which were separated on the SPB-PUFA column (on GC-IRMS). The lowest trace shows the mass-44 ion current as a function of time. Above that, the instantaneous ratio of the m/z-45 and m/z-44 ion currents is shown. *: Peak was cut off at the top.

19

2.3.4a GC-IRMS on fatty acid methyl esters Stable carbon isotope ratios of the FAMEs were measured with a Varian 3400 gas chromatograph coupled to a Finnigan MAT 252 isotope ratio mass spectrometer (GCIRMS). The GC was equipped with a split injector and an SPB-PUFA column (30m, 0.25mm ID, 0.2µm film thickness; Supelco), which has a polyalkylene glycol stationary phase. Helium was used as a carrier gas and the column head pressure was kept at 25 psi with a split ratio of approximately 1:6.

20:5n-3 n=20

18

17

19

15

14

16

12

11

13

9

8

10

6

5

7

O

3 4

2

1

OH

20:4n-6 19 n=20

17 18

15

14

16

12 13

11

9

8

10

6 7

5

O

3 4

2

1

OH

Figure 2.3. The structures of eicosapentaenoic- (20:5n-3) and arachidonic acid (20:4n-6). The numbering of the carbon atoms is indicated. The last carbon is referred to as “n” (or “ω”). The most abundant unsaturated fatty acids have double bonds in the cis-configuration, at three carbon intervals. Therefore, these fatty acids can be described by indicating the number of carbon atoms, the number of double bonds, and the position of the double bond closest to the terminal methyl group (n). For example, the 20:5n-3 fatty acid has 20 carbon atoms, 5 double bonds, with the double bond closest to the methyl group at n-3.

The stable carbon isotope composition of the individual fatty acids and whole samples are expressed as δ13C values, which are defined as parts per thousand or per mil (‰) differences from an international standard: 13

δ13C =

C / 12 C sample −13 C / 12 C s tan dard 13

C / 12 C s tan dard

× 1000

(2.1)

The international standard used as a reference is the Peedee belemnite (PDB) standard, which has a 13C/12C ratio of 0.0112372 (Craig, 1957).

20 C19 and C20 n-alkanes were co-injected with the FAMEs and used as internal isotopic references. Additionally, several CO2 spikes with known δ13C values were introduced into the IRMS at the beginning, during, and end of all runs. The two n-alkane standards had been combusted off-line with copper oxide (at 550ºC, overnight), and their δ13C values were measured with conventional dual-inlet IRMS. The δ13C values of the FAMEs were calculated by integrating the individual m/z (mass to charge ratio) 44, 45 and 46 ion currents of the CO2 peaks derived from the combustion of the GC effluent, and calibrating them against those of the introduced CO2 spikes with a known

13

C/12C ratio. Natural CO2 consists predominantly of the common

isotopes (12C16O16O), with much lower abundances of molecules containing one of the minor isotopes (e.g., 13C16O16O, 12C16O18O etc.). The probability of molecules containing more than one rare isotope is negligible. Singly charged ions of CO2 will therefore give a large ion beam corresponding to m/z = 44 (12C16O16O) and two minor ion beams with m/z = 45 (13C16O16O or

12

C16O17O) and m/z = 46 (12C16O18O or

13

C17O16O). Hence, the

measured m/z 45/44 ratios have to be corrected for these interferences in order to obtain the true 13C/12C ratio of the CO2. In CO2 from natural samples approximately 6% of the m/z 45 peak is due to 17O and about 0.2% of the peak at m/z 46 is from CO2 containing both 13C and 17O. n-Alkane standards were used as an additional check for any drift between runs and also served to estimate the accuracy and precision of the GC-IRMS. Unfortunately the nC20 standard frequently coeluted with a small peak of an unknown compound in the FAME mixture. Therefore, the standard could not be used in every case as a good indicator for the analytical precision. The n-C19 standard was usually free of coelution of compounds with any appreciable peak size. For this standard, the standard deviation of the measured δ13C values was 0.25‰ (n=255). The mean n-C19 δ13C of the 255 measurements made by GC-IRMS was 0.04‰ higher than the corresponding δ13C determined by off-line combustion and dual-inlet IRMS.

In the lower part of Figure 2.2b the m/z 44 ion current signal is shown as a function of time. The upper portion of Figure 2.2b depicts the instantaneous ratio of the m/z 45 and

21 m/z 44 ion currents. As can be observed in the figure the m/z 45/44 ratio shows a typical bi-directional swing during the elution of a compound, which is caused by an isotope effect during the separation of the FAMEs in the GC-column (Hayes et al., 1990). Due to this effect, compounds that are not fully separated are contaminated by either 12C or 13C from a shouldering peak, depending on its position. Therefore, measured δ13C values often had to be omitted because they lacked the required separation between peaks. As can be seen in Figure 2.2b, especially 18:0 and 18:1 FAME isomers, and the 20:1n11/20:1n-9 doublet are troublesome. The numbers for the 20:1 isomer doublet were left out altogether due to poor separation. Values obtained for the 18:0 and 18:1 FAME isomers were only kept when they were all separated by “peak valleys” down to a maximum 0.4 Volts away from the baseline (e.g., the separation of the 18:0 and 18:1 isomers in Figure 2.2 was not sufficient). This threshold was determined empirically by comparing the reproducibility of values obtained from samples injected at different concentrations. At different concentrations the peak widths are different, resulting in varying degrees of separation when different concentrations of the FAME mixture are injected into the GC. Introducing too much of a compound into the GC-IRMS overloads the amplifier, while too small an amount will also decreases the precision of the δ13C value obtained due to a lower signal to noise ratio. Therefore, occasionally a sample had to be run at two different concentrations in order to get reliable measurements of both the highly abundant compounds and those present in low amounts. Hence, data for some samples consist of merged measurements from two (or more) runs.

The average precision of measurements on each of the individual FAMEs was calculated from the mean of the standard deviations obtained from the samples run more than once (63 replicates). This result is shown in Table 2.1. The mean standard deviations are conservative precision estimates, since runs were often repeated in order to improve on the chromatography and to run with more desirable concentrations. Therefore, these repeated measurements are not always true replicates. Moreover, as can be seen in Table

22 2.1, with only one exception the median of the standard deviations is always lower than the mean, indicating that the distributions are skewed towards smaller values.

2.3.4b δ13C correction for methyl group addition In order to calculate the true δ13C of the fatty acids, the values obtained for the FAMEs have to be corrected for the extra carbon added during the methylation of the fatty acids. Previous studies found that no isotopic fractionation occurred during derivatization (Abrajano et al., 1994; Pond et al., 2000). Therefore, the original δ13C of the fatty acid (δ13CFA) can be back-calculated using a linear mass balance equation:

δ 13CFAME = f ⋅ δ 13CFA + (1 − f ) ⋅ δ 13CMeOH ,

(2.2)

with

f =

n , n +1

(2.3)

then

δ 13CFA =

( n + 1) ⋅ δ 13CFAME − δ 13CMeOH , n

(2.4)

where δ13CFAME and δ13CMeOH are the δ13C values of the measured fatty acid methyl ester and methanol used during derivatization, respectively. δ13CFA represents the δ13C of the fatty acid prior to methylation, and n is the number of carbon atoms in the (nonmethylated) fatty acid. For this study, by mistake, the δ13C of the methanol used for the preparation of the methanolic KOH was measured (see below), and the δ13C values obtained for the FAMEs were corrected using this δ13C value (-31.64‰). Although this methanol was present during methylation, it is mostly the methanol of BF3-MeOH that will be present in the

23 FAMEs. δ13C values of -41.8, -44.8, -46, -47.5, -50.1 and -53.3‰ have been reported for the carbon in BF3-methanol by Pond et al. (1997a, b; 2000), Fang et al. (1993), MacAvoy et al. (2002), Ballentine et al. (1996), Abrajano et al. (1994) and Pond et al. (1995), respectively. Therefore, the correction that was applied in this thesis is most likely too small. It can be calculated by using equation 2.4 that when a δ13C value of -50‰ for the methanol is used, a C14 fatty acid will be 1.3‰, and a C22 fatty acid will be 0.8‰ more enriched in 13C with respect to the corrected values reported in this thesis (0.6 and 0.4‰ for C14 and C22, respectively, when δ13C of MeOH is -40‰). The δ13C differences between the individual fatty acids, therefore, will not change substantially (i.e., max. 0.5‰ change in difference between C14 and C22). For fatty acids with the same carbon number (e.g., the C18-series) the correction will be the same, and the δ13C differences between the individual fatty acids will not change at all. However, the possible adjustment of the δ13C values should be borne in mind throughout this thesis.

2.3.4c Measurement of δ13C of methanol (of methanolic KOH) The δ13C of the methanol used for the preparation of the methanolic KOH was measured with an elemental analyzer (EA) coupled with the Finnigan MAT 252 isotope ratio mass spectrometer (see below). Right before the start of a run, 2.0 or 4.0 µL of MeOH was injected into a smooth walled tin capsule, the opening was quickly squeezed and folded shut with forceps, and then the capsule was dropped in the EA immediately. Huang et al. (1999) found isotope effects (α) of 1.00031 and 1.00065 during the evaporation of trichloroethene and dichloromethane, respectively. Therefore, some fractionation due to preferential evaporation of the

12

C-methanol had been anticipated. However, during the

measurements of the δ13C of the methanol such an isotope effect was not observed. The δ13C measurements from 13 runs displayed a standard deviation of only 0.14‰, even though it was clear from the relatively high variation in the yields that different degrees of evaporation had occurred (lowest yield was about 10% of the highest yield). The δ13C of the methanol (Aristar, from B.D.H. Chemicals Ltd.) was found to be –31.64‰ (±0.14, n=13).

24

2.3.5 Isotope ratio mass spectrometry on bulk samples The stable carbon isotope composition of an (unextracted) aliquot of the freeze-dried and powdered whole sample is referred to as the bulk stable carbon isotope composition in this thesis. Depending on the sample, 0.4 to 1 mg of freeze-dried and powdered sample was weighed out and placed into a (pre-combusted) silver capsule. In the case of POM samples, a sub-sample of the glass fibre filter was cut into 0.2 cm2 pieces and placed into the silver capsule. One or two drops of (pre-extracted) 1 N HCl was added to the weighed samples in order to remove inorganic carbon. Samples were vented overnight with the sample tray lid loosely fitted, then left in a desiccator for at least 3 days before analysis. The stable carbon isotope ratio was measured using a Carlo Erba elemental analyzer interfaced with the Finnigan MAT 252 isotope ratio mass spectrometer. CO2 gas with a known δ13C (vs PDB) was used as a reference during each run, and an acetanilide standard (δ13C vs PDB = -30.54‰) was run 4-6 times during a 50-sample session. The standard deviation of the repeated acetanilide runs was typically below 0.1‰. However, δ13C values of actual samples were less reproducible due to inhomogeneities in the sample material, and standard deviations of 0.2 and 0.3‰ were commonly observed.

2.3.6 Comparison of analytical accuracy and natural variability Measurements were done on 7 single crab megalopa samples (Cancer magister, Dungeness crab) from the same station (LC4), as a test for natural variability. The results of these measurements are listed in Table 2.1. The variability in the fatty acid composition of the individual crab larva samples is clearly higher (standard deviation [SD] up to 1.5 percentage point) than the small variance introduced by preparation and measuring techniques (SD up to 0.2 percentage point). However, the standard deviation generally remained below one percentage point. The variability in δ13C of the fatty acids from the single crab larva samples, on the other hand, is often not far from the measurement reproducibility estimates.

25 Table 2.1. Standard deviations of fatty acid composition analyses on separately processed samples of pairs of 2, 5, 10, 20 and 40mg aliquots of homogenized crab larvae (measurements in % of total). On the right: the mean (and median) of the standard deviations of repeated GC-IRMS measurements (in ‰ vs PDB). The reproducibility of both methods is compared with the variability (standard deviation) of measurements on 7 individual crab larvae (from different location). *Replicates of POM, zooplankton, larval fish, juvenile salmon and rotifers. -:Abundance was less than 0.2% of total. --:No reliable number available.

Fatty acid 14:0 iso 15:0 15:0 16:0 16:1n-7 16:1n-5 iso 17:0 16:2n-4 16:3n-4 17:0 16:4n-3 16:4n-1 18:0 18:1n-9 18:1n-7 18:1n-5 18:2n-6 18:2n-4 18:3n-6 18:3n-3 18:4n-3 20:0 20:1n-11 20:1n-9 20:1n-7 20:2n-6 20:4n-6 20:3n-3 20:4n-3 20:5n-3 22:1n-11 22:1n-9 21:5n-3 22:5n-6 22:5n-3 22:6n-3

Fatty acid composition analyses Crab larvae Individual crab replicates (n=10) larvae (n=7) SD SD 0.048 0.274 0.042 0.010 0.094 0.153 0.484 0.066 0.569 0.007 0.029 0.008 0.077 0.044 0.011 0.052 0.021 0.062 0.077 0.049 0.241 0.112 1.521 0.035 0.231 0.010 0.060 0.017 0.142 0.040 0.011 0.209 0.018 0.327 0.014 0.069 0.904 0.028 0.219 0.026 0.159 0.047 0.215 0.024 0.096 0.030 0.043 0.010 0.026 0.182 1.209 0.114 1.436 0.022 0.078 0.044 0.048 0.011 0.052 0.009 0.057 0.190 0.486

δ13C of individual fatty acid analyses All replicates Individual crab larvae (n=7) (n=63)* Mean of SD (median) SD 0.26 (0.17) 0.29 0.56 (0.32) -0.44 (0.37) 0.58 0.26 (0.19) 0.29 0.37 (0.29) 0.49 ----0.55 (0.50) -0.65 (0.54) -0.50 (0.34) 0.30 0.38 (0.30) 0.38 0.40 (0.37) 0.53 0.49 (0.38) 0.63 0.37 (0.29) 0.45 0.42 (0.38) 0.78 --0.33 (0.25) 0.47 0.70 (0.62) 0.81 --0.43 (0.28) 0.69 0.48 (0.41) 0.47 --------0.49 (0.39) 0.45 0.41 (0.34) 0.41 --0.59 (0.60) 0.32 0.39 (0.29) 0.29 0.59 (0.33) 0.66 --0.37 (0.20) ---0.38 (0.30) 0.57 0.33 (0.24) 0.22

26

2.3.7 Multivariate Data analyses 2.3.7a Discriminant analysis Discriminant analysis is a multivariate statistical technique that describes group separation. The analysis will find a weighted composite of the observed variables, often called the discriminant- or canonical variate. This new axis has the property that the samples of the predefined groups have minimal overlap. Subsequently, a classification rule is constructed in order to classify unknown future samples into the groups. This rule can also be used to assess the extent of the separation and rate of misclassification in the present data set. For this study, two-group linear discriminant analysis (LDA) was used to find the optimum separation between samples collected in waters on the continental shelf (water depth at station <200m) and samples from off the shelf (water depth >200m). For LDA no distributional assumptions are made, though the analysis works best on elliptical distributions such as the multivariate normal distribution (Hand, 1997). An assumption that is made is that the variance-covariance matrices of the groups are equal, an assumption that is rarely satisfied (Lachenbruch, 1975). Quadratic discriminant analysis (QDA) does not make this assumption. However, this method is known to produce results that are not robust in cases of non-normality (Lachenbruch, 1975), where multicollinearity (high correlation between several variables) exists, and when the sample size is small relative to the number of variables (Seber, 1984). The variance-covariance matrices of the groups in the datasets used in this study were not equal (Box’s M-test, P<0.01), but it was found that the classification rules obtained with QDA were indeed less stable when testing for robustness (see below). Hence, Fisher’s (1936) LDA was used. Discriminant analysis was performed on data of the separate cruises for the δ13CBulk and δ13CFA data. But for the fatty acid abundance data, the analysis was performed on the whole data set. This was done to minimize overfitting the data, since more variables were necessary when using the fatty acid data (see below). Additionally, a general classification rule for the 3 cruises has the advantage that it can be more useful for future sampling trips.

27

2.3.7b Misclassification- and error rates The misclassification rate (or error rate) and the robustness of the classification was tested in a few ways. One method is commonly referred to as re-substitution. All samples were submitted to the linear discriminant function and classified according to the obtained classification rule. After each sample had been assigned to either the shelf- or the off-shelf group (i.e., samples collected at stations with a water depth < 200m or > 200m, respectively), the percentage of misclassified samples was calculated. This method generally underestimates the error rate for future samples, since the discriminant analysis has been performed on the same data set. Another, more robust, way of estimating the error rate of future samples is the leave-one-out method (Lachenbruch and Mickey, 1968). All but one sample is used to determine the classification rule, and this rule is then used to classify the omitted observation. This is repeated for each sample, so that every observation has been classified by a discriminant function and classification rule based on the other n-1 samples. As a third measure of estimating error rates, the leave-one-out method was modified and five samples instead of just one were consecutively omitted and classified. However, in this case the order in which the samples are initially arranged will affect the estimation. Therefore, the whole procedure was repeated a thousand times, each time with a new randomly picked sequence of samples. After the thousand runs, the average error rate was determined and reported as the leave-5-out error rate. In addition to the leave-one-out and leave-5-out procedures, a final test was used to assess the classification performance of the three techniques. One third of the datasets (i.e. one third of the shelf and off shelf groups each) was taken out as a validation sample set. Discriminant analysis was performed on the two thirds left over (the training set), and subsequently the determined classification rule and discriminant function was applied to the validation set. This procedure was repeated a thousand times with new, randomly

28 picked, validation sets and a mean misclassification rate was calculated. This error rate was reported as the mean 1/3 sub-set error rate. Here, when a sample is reported as “misclassified” or “misclassification rates” are given this was determined by re-substitution. “Error rates” were calculated by the other three methods (leave-one-out, leave-5-out, and mean 1/3 sub-set procedures).

2.3.7c Contribution to separation by each variable Standardized coefficients of the discriminant function were calculated to assess the joint contribution of the variables to the separation. These coefficients are calculated so that they apply to the standardized, scale free, variables (observations are subtracted by the mean, and divided by the standard deviation), and therefore reflect the actual contribution to the separation of the groups by the selected variables (see Rencher, 1995). To estimate the significance of the extra separation provided by a variable yr in addition to the separation due to the other variables, partial F values were calculated (Rencher, 1995):

F = ( v − p + 1)

Tp2 − Tp2−1 v + Tp2−1

,

(2.5)

where Tp2 is the two-sample Hotelling T2 ,i.e., the multivariate equivalent of the tstatistic, with all p variables, Tp2−1 is the T2-statistic with all variables except variable yr, and v = n1 + n2 – 2 (n1 and n2 are the number of samples in group 1 and 2 respectively). The F-statistic is distributed as F1,v-p+1.

2.3.7d Selection of Variables It is well known that a reduction in the number of variables can lead to lower error rates in the classification analysis (see for example Seber, 1984; Hand, 1997). Moreover, the robustness of the classification function is increased when the number of variables is

29 reduced (e.g., see Lachenbruch, 1975; Seber, 1984). Various statistical tools that help find the optimal selection of variables exist. One method used here is often termed stepwise discriminant analysis. In the stepwise selection approach used here, the first step entails a search for the variable with the maximum (partial) F-value for separation (or minimum Wilks’ Λ ). During each next step the variable that maximizes the partial Fstatistic will be joined with the already selected variables. Additionally, the previously picked variables are re-examined and a variable is left out again when it fails to exceed a preset threshold F-value. In the present study, with the stepwise discriminant analysis procedure on 197 samples 8 fatty acids were selected from a dataset containing abundance data of 32 fatty acids, using a re-examination threshold of F=3. This combination of fatty acids produced the lowest error rates when used in the classification analysis. Using any other combination of F-value and number of fatty acids resulted in higher error rates.

With the enhanced computing speed of personal computers, however, it has become increasingly viable to simply test all of the combinations of variables, to find the undisputedly best subset of variables. This more time consuming but also more thorough, procedure was tested here for the fatty acid abundance dataset containing all 197 samples from the three cruises, and compared with the stepwise discriminant analysis approach. For all combinations of 30 out of the 32 pre-selected fatty acids, the leave-one-out error rate was determined. The combination of 30 fatty acids with the lowest error rate was selected for further analysis. When several combinations had equal error rates, the combination with the lowest misclassification by re-substitution was picked. In cases that still more than one combination was left over, the fatty acids resulting in the highest Hotelling’s T2 were chosen. This variable selection procedure was repeated, every step reducing the number of selected fatty acids by two, to minimize computing time. When a group of 20 fatty acids that best separated the shelf from off-shelf samples was found, no more fatty acids were omitted from the set of variables being searched. Therefore, every possible combination of any number of fatty acids from the selected 20 was tested on its performance in assigning the samples to the correct group.

30

2.3.7e Data handling and estimation of missing values When fatty acids were not detected or the abundance was less than 0.2% of the total, a zero was given for that fatty acid in the data matrix. For the GC-IRMS data matrix, it is not possible to fill in a zero for missing data. Ten fatty acids with the least missing values were chosen from the data set, but missing values still made up about 10% of the matrix. Therefore, recognizing that this is a compromise, an estimation was made for the missing values. This estimation could be made since it was found that the δ13C values of the fatty acids, to a certain extent, had predictable offsets from each other (e.g., see average zooplankton δ13CFA profile in Figure2.8). Hence, the whole data set average difference (∆δ) between the δ13C value of the 14:0 fatty acid (δ13C14:0) and the missing fatty acid δ13C was added to the δ13C14:0 of the same sample. In the few cases that the δ13C14:0 is also missing, the average ∆δ13CFA-16:0 was added to the δ13C16:0 for that sample to estimate the missing δ13CFA value. All matrix computations, statistical tests and calculations for the conceptual model (see discussion) were performed by using the software package Matlab (version 4.2b, The Mathworks Inc., Natick, MA, U.S.A.).

2.4 Results 2.4.1 Shelf - off shelf classification 2.4.1a Fatty acids The stepwise discriminant analysis procedure on all 197 samples resulted in a minimum error rate when 8 fatty acids were selected with a re-examination threshold of F=3. The amount of samples that were classified as originating from the shelf, but were collected off the shelf and vice versa, accounted for 14.7%, and the error rate was estimated at almost 16% of the total samples (Table 2.2).

31 Table 2.2. Results of discriminant analysis on combinations of fatty acids selected by stepwise discriminant analysis (top), and iterative search for lowest leave-one-out error rate (bottom). The coefficients of the standardized discriminant function and the partial F-test results are a measure of separation contributed by the individual fatty acids (in the presence of the others). *: df: 1, n-p-1.

Forward selection with stepwise discriminant analysis (n=197, p=8) Fatty acid

14:0

Discr. Funct. Standardized Discr. Funct.

0.238

1.729

-1.479

0.298

-0.616

-2.063

2.539

1.170

1.378

0.536

-0.764

0.904

-0.791

-0.584

0.691

0.622

Part.F-values*

18.97

4.54

8.51

15.66

10.57

6.75

9.24

5.41

16:1n5 16:2n4 18:1n9 18:1n7

Re-substitution Leave-1-out Error rate (%) 14.7 15.7

20:0

20:2n6 20:4n3

Leave-5-out Mean-1/3-subset 15.9 16.7

Iterative search for combination resulting in lowest leave-1-out error rate (n=197, p=10) Fatty acid

14:0

iso 15:0

15:0

16:1n7 16:1n5 16:2n4 18:1n9 18:1n7 20:4n3 22:6n3

Discr. Funct.

0.253

-0.830

0.984

-0.173

2.391

-0.318

0.300

-0.476

1.225

0.024

Standardized Discr. Funct.

1.460

-0.455

0.439

-0.892

0.741

-0.164

0.909

-0.611

0.651

0.227

Part.F-values

10.77

1.62

3.72

4.62

7.90

0.18

15.30

5.70

6.12

0.30

Re-substitution Leave-1-out Error rate (%) 13.7 14.2

Leave-5-out Mean-1/3-subset 14.6 17.7

With the iterative search for a combination of fatty acids that had the lowest leaveone-out error rate, a group of 10 fatty acids was found to best separate shelf samples from samples originating from off the shelf (Table 2.2). This combination performed slightly better at classifying all samples, decreasing the misclassifications to 13.6% and leave-5out error rate estimate to 14.8% (frequency distribution shown in Figure 2.4). However, on the most conservative test, where the discriminant function was determined on a training set and then tested on a validation set selected from the data (one third of the total set), the group of 8 fatty acids selected by the stepwise discriminant analysis procedure performed better. This combination performed 1% better than the group of 10 fatty acids found by the iterative search (see mean 1/3 sub-set error rate, Table 2.2). Since the procedure is less time consuming and available in some statistical packages (i.e., more accessible), the rest of this chapter will focus on the results of the group of 8 fatty acids selected with stepwise discriminant analysis.

32

18 16

Number of samples

14 12 10 8

Fatty acids used: 14:0 16:1n-5 16:2n-4 18:1n-9 18:1n-7 20:0 20:2n-6 20:4n-3

Shelf Off shelf

6 4 2 0 -2

0

2

4

6

8

10

Discriminant score Figure 2.4. Frequency distribution of the discriminant scores of fatty acid data of POM, zooplankton and larval fish samples from all 3 cruises (79 shelf-, and 118 off shelf samples). The combination of 8 fatty acids (selected with stepwise discriminant analysis) used for the discriminant analysis are listed in the figure.

It is possible to estimate the contribution of the individual fatty acids to the separation of the groups. For each fatty acid, both the standardized discriminant function coefficient and the score of the partial F-test for added separation is given in Table 2.2. However, these values do not reflect the ability of the individual fatty acids to separate the sample groups, but strictly show the contribution to group separation in the presence of the other fatty acids. For example, when the abundance of a fatty acid correlates highly to that of another selected fatty acid, it will not contribute as much to the sample group separation as when the other fatty acid was not in the selection. When single fatty acids are used for the classification, the 16:2n-4 fatty acid has the lowest misclassification rate (32.3%), followed by 16:4n-1 and 18:1n-7 (33.8%).

A high success rate was attained when discriminating shelf from off shelf samples collected on the May ’98 and July ’99 cruises (see bottom of Figure 2.7). Only about 1

33 out of 12 (May ’98) or 8 samples (July ’99) was misclassified. Classification of the samples collected on the May ’99 cruise, however, was less successful with almost 24% of the samples misclassified. An even lower misclassification rate could have been attained with the fatty acid abundance data when separate classification rules were applied for each cruise. The most conservative estimate of the error rate (mean 1/3 subset error rate) obtained when discriminant analysis was done on the fatty acid data of the separate cruises was 15.5%. This is 1.2% better than the classification with the combination of 8 fatty acids, selected by stepwise discriminant analysis, used on the whole dataset.

2.4.1b Bulk stable carbon isotope composition As can be seen in Figure 2.5, a large difference exists between the stable carbon isotope composition of organisms caught on the shelf and samples collected off the continental shelf during the May ’98 cruise. The shelf food web was more enriched in

13

C and had

δ13C values that on average were almost 5‰ higher. The mean offset between samples from stations P10 and LC4 is almost 9‰. Only 1 POM and 3 zooplankton samples misclassify when discriminant analysis is performed on the vector of bulk δ13C values of the May 1998 data set (n=57). In May of 1999 there was much more overlap in the bulk stable carbon isotope compositions of samples from the shelf and off shelf environments (Figure 2.5). Some of the overlap can be attributed to samples of station LG3. POM from this station had a δ13C value of almost -23‰, an unusually low value for shelf POM in this region. Of the 8 shelf zooplankton samples measured, 3 originated from station LG3, and these 3 samples had an average of about -21‰. Almost 31% of the 55 samples of the May 1999 dataset were misclassified on the basis of their bulk stable carbon isotope composition. POM samples collected off the continental shelf in July 1999 were significantly more depleted in 13C than the shelf POM samples (P<0.05, two-sample t-test). However, again there was considerable overlap (Figure 2.5), causing 1 out of 3 samples to be misclassified.

34 Using only the bulk isotope ratios, the misclassifications in the data set containing samples of all 3 cruises, amounted to 21% of the 138 samples measured.

-16

Off shelf

Shelf

-16

13

Bulk δ C (‰ vs. PDB)

May 1998

Off shelf

May 1999

Shelf

-16

-18

-18

-18

-20

-20

-20

-22

-22

-22

-24

-24

-24

-26

-26

-26 POM Zooplankt. Larval fish -28 -28 -28 POM Zoo L.FishPOM Zoo L.Fish POM Zoo L.FishPOM Zoo L.Fish

Off shelf

Shelf

July 1999

POM

POM

Figure 2.5. Box and whisker plots of bulk stable carbon isotope composition of shelf- (right) and off shelf samples (left), plotted per cruise. The black line within a box is the median, the upper and lower box edges define the 75th and 25th percentiles, and the whiskers represent the 95th and 5th percentiles. The black squares indicate the mean, and asterisks mark the maximum and minimum values. The δ13C values are plotted for POM, zooplankton and larval fish (x-axes), except for the July 1999 cruise, for which only POM samples were measured.

2.4.1c Stable carbon isotope ratios of individual fatty acids The stable carbon isotope composition of 10 fatty acids in each sample were initially used in the discriminant analysis. However, best classification results were obtained by using only 1, 2 or 3 fatty acids for the May 1999, July 1999 and May 1998 cruises respectively. This is probably due to the fact that the δ13C values of the 10 fatty acids are all positively correlated, with R2 values ranging from 0.40 to 0.81. The 3 fatty acids in the 3 combinations that performed best in the separation (14:0, 16:0 and 22:6n-3) are the fatty acids for which the best analytical precision was estimated (see Table 2.1). In accordance with classification on the basis of bulk δ13C values, classification was most successful for the May 1998 cruise samples. As much as 91 percent of the samples

35 were correctly assigned to the shelf and off shelf groups, whereas the success rate for the May 1999 and July 1999 cruises was somewhat lower at 82 and 84 percent respectively (Figure 2.6).

2.4.2 Comparison of techniques Of the three kinds of data tested, stable carbon isotope composition of the bulk sample was the least successful overall in predicting the shelf or off shelf origin of the sample (79% correctly classified). With the two multivariate datasets, approximately 86% (δ13CFA) and 85% (fatty acid abundance) of the samples was correctly assigned to the shelf and off shelf groups (Table 2.3).

Table 2.3. Comparison of misclassification/error rates for classification procedures using the 3 types of data. *: determined by re-substitution.

% of total (n) misclassified* Leave-1-out error rate (%) Leave-5-out error rate (%) Mean 1/3 sub-set error rate (%)

Bulk δ13C (n=138) 21.0 22.5 20.8 20.4

δ13CFA (n=139) 13.7 15.8 15.7 16.2

Fatty acids (n=197) 14.7 15.7 15.9 16.7

To correct for any overfitting of the data and bias due to the variable selection procedure, 3 different estimations of error rates were made. Results of these tests are shown in Table 2.3. The classification methods invoking variable selection procedures showed, indeed, to be over 2% lower in their misclassification rate than the more robust estimations of the error rate. However, classification with δ13CFA and fatty acid abundance data still proved to be more successful. Figure 2.7 shows how the misclassifications for each technique are distributed. The proportion of shelf samples misclassified is always lower than the fraction of off shelf samples. In addition, the percentage of larval fish not correctly classified is lower than the percentages of POM and zooplankton misclassified. The July 1999 samples show very different misclassification rates for the different techniques. However, it should be noted that the comparison between the three techniques is perhaps less informative for the July

36 1999 samples, since the δ13CFA and δ13CBulk analyses were only carried out on 24 POM samples.

5 4

Classification rule

July 1999*

Off shelf sample Shelf sample

3 2 1

Number of samples

0 8 6

22

23

24

25

26

27

28

May 1999

4 2 0 24 10 8

25

26

27

28

29

30

May 1998

6 4 2 0 40

45

50

55

60

Discriminant score Figure 2.6. Frequency distributions of discriminant scores of fatty acid δ13C data. For data of each cruise a different discriminant function and classification rule (line through the middle) was set up. Shelf samples to the right, and off shelf sample to the left of the line are “misclassified”. *: of the July 1999 samples only POM was analyzed.

The match between misclassified samples, as determined by the three methods, was investigated. A dataset containing all samples that were analyzed with all three techniques (n=121) was used for this comparison (Figure 2.8). As can be seen in Figure 2.8, close to 80% of the misclassifications determined with the bulk δ13C and fatty acid data was confirmed by at least one other method. Of the samples that were misclassified with the δ13CFA, 87.5% of the data matched with another classification procedure.

37

13

Bulk δ C classification 13

δ CFA classification Fatty acid classification

Off shelf samples

30

Shelf samples

% of each category misclassified

20 10 0 30

Zooplankton

POM

20

Larval Fish

10 0 30

May 1999

20

May 1998

* July 1999 *

10 0

Category

Figure 2.7. Misclassification rates for samples of each cruise, trophic group and shelf break position plotted for the 3 classifications. *: for July only 24 POM samples were analyzed for δ13CFA and δ13CBulk.

24

Match with both other methods 13 Match with bulk δ C 13 Match with δ CFA Match with fatty acid No match

% of total (n =121)* misclassified

22 20 18 16 14 12 10 8 6 4 2 0

13

Bulk δ C

13

δ CFA

Fatty acids

Classification method Figure 2.8. Match between misclassified samples as identified by discriminant analysis on the 3 data sets. *: includes only samples which were analyzed with all 3 techniques

38

2.4.3 Identification of transported or moved animals When an animal collected off the shelf is classified as a shelf sample, or vice versa, it is not necessarily implied that it moved from the one region into the other. For example, shelf-like conditions can occur away from the continental shelf and POM found off the shelf can therefore have a composition that resembles compositions as found in shelf waters. Thus, animals feeding in these “shelf-like off shelf waters” would be “correctly misclassified”, and may not have moved at all. In order to identify animals that actually moved from off shelf- to shelf waters, or in opposite direction, zooplankton and larval fish samples from stations with misclassified POM were left out. Additionally, all POM samples were left out. The misclassified samples that are left over are the animals that tentatively moved from the shelf into off shelf waters, and vice versa. Animals identified as such, by one or more techniques, are listed in Table 2.4.

Table 2.4. List of samples identified by at least one classification procedure as consisting of organisms that moved off the shelf (first 9 samples), or to shelf waters (last 5 samples). *: Although the last 3 samples were misclassified as shelf samples with the two isotope data sets, since POM from LG3 was also misclassified, the organisms were not hypothesized to have moved. Misclassified with: Region Off shelf Off shelf Off shelf Off shelf Off shelf Off shelf Off shelf Off shelf Off shelf Shelf Shelf Shelf Shelf Shelf

Group Zoopl. Zoopl. Zoopl. Zoopl. Zoopl. Zoopl. L. Fish L. Fish L. Fish Zoopl. Zoopl. Zoopl. Zoopl. Zoopl.

Cruise May '98 May '98 May '99 May '99 May '99 July '99 May '99 May '99 July '99 May '98 May '99 May '99 May '99 May '99

Station LC9 LC9 LC9 LG7 LC12 LG7 LC9 LG7 LG7 LBP2 LBP2 LG3 LG3 LG3

Description Euphausiid furciliae Zoo >425µm Zoo >425µm Euphausiids Euphausiids Polychaete (43mm) 4 un-ID (12-24 mm) 3 un-ID (15-25mm) Larval flatfish (30mm) Zoo 212-425µm Calanoid copepods Hyperid amphipods Zoo >425µm Zoo 212-425µm

2.4.4. May 1998 LC4 and LC9 zooplankton

13

δ CBulk Yes Yes Yes Yes Yes Yes No Yes (Yes)* (Yes) (Yes)

δ13CFA No Yes No Yes No (Yes) (Yes) (Yes)

Fatty acids No Yes No Yes Yes Yes No No Yes Yes No Yes Yes Yes

39 In Figure 2.9 the δ13C values of 14 fatty acids derived from zooplankton and POM collected in May 1998 at stations LC4 and LC9 are shown. The stable carbon isotope composition of fatty acids from zooplankton collected at LC4, generally corresponds to isotope compositions of fatty acids found in POM from the same station. Similarly, δ13C values of fatty acids in zooplankton smaller than 425µm sampled at LC9, are accordant with values found for fatty acids in POM from LC9 (Figure 2.9). In contrast, zooplankton bigger than 425µm, mostly consisting of euphausiid furciliae, and separately isolated euphausiid furciliae collected at LC9 were found to have fatty acids with mostly intermediate δ13C values. Not all fatty acids from the LC9 euphausiid furciliae show a similar difference in δ13C with other zooplankton from LC9. Notably, the stable carbon isotope ratio of the 18:2n-6, 18:3n-3 and 18:4n-3 fatty acids of the LC9 euphausiids are remarkably similar to other zooplankton from the same station.

LC4 Zooplankton (n=3) LC9 Zoo. < 425µm (n=2) LC9 Euphausiid furciliae LC9 Zoo. > 425µm LC4 POM LC9 POM

-18 -20

13

δ C of fatty acid (‰)

-22 -24 -26 -28 -30 -32 -34

22:6n-3

20:5n-3

20:4n-6

18:4n-3

18:3n-3

18:2n-6

18:1n-7

18:1n-9

18:0

17:0

16:1n-7

16:0

15:0

14:0

-36

Fatty acid

Figure 2.9. Stable carbon isotope composition of fatty acids from POM and zooplankton collected at station LC4 and LC9 in May ’98. Error bars represent ± 1 SD.

40 Figure 2.10 shows that the fatty acid composition of the LC9 euphausiid furciliae can also be regarded as an intermediate between the compositions of LC4 and LC9 zooplankton. The C18 unsaturated fatty acids are more abundant in LC9 POM, and the 16:1n-7 and 20:5n-3 fatty acids are more dominant in POM from LC4 (Figure 2.10b). A similar pattern is reflected in the zooplankton. The 18:2n-6, 18:3n-3 and 18:4n-3 fatty acids show a high relative difference between the two stations in both zooplankton and POM samples; with LC9 being more enriched in those particular fatty acids.

30

a

LC4 Euphausiid furciliae LC9 Euphausiid furciliae LC9 Zoopl. < 425µm

25 20

10 5

18:3n-3

18:4n-3

20:4n-6

20:5n-3

22:6n-3

18:4n-3

20:4n-6

20:5n-3

22:6n-3

18:2n-6

18:1n-7

18:1n-9

18:0

17:0

16:1n-7

16:0

b

18:3n-3

30

15:0

0 14:0

Fatty acid abundance (% of total)

15

LC4 POM LC9 POM

25 20 15 10 5

18:2n-6

18:1n-7

18:1n-9

18:0

17:0

16:1n-7

16:0

15:0

14:0

0

Fatty acid Figure 2.10. Partial fatty acid composition (same fatty acids as in Figures 2.9 are shown) of zooplankton (a) and POM (b) collected at LC4 and LC9 in May 1998.

41

2.5 Discussion 2.5.1 Shelf-off shelf difference As shown here by the data obtained with three different techniques, POM from the shelf environment is typically quite different from POM off the shelf, and this difference persists in zooplankton and larval fish. This result further substantiates the results of Perry et al. (1999) who reported the stable carbon isotope ratios of pelagic organisms off the southwest coast of Vancouver Island. Nutrients and phytoplankton biomass are known to be much higher along and inshore of the continental shelf break than over the deeper waters in the studied area (Mackas, 1992). Current directions are on average alongshore and parallel to shelf break bathymetric contours, however occasional crossisobath flow features such as eddies, meanders and “filaments”, can occur (Mackas and Yelland, 1999). In spring and summer the prevailing winds are from the north along the west coast of Vancouver Island, and wind driven upwelling occurs during these months (Freeland et al., 1984). A more complete discussion of the observed shelf-off shelf differences in fatty acid and stable carbon isotope composition of POM will follow in the next chapter. However, some mechanisms will be discussed here briefly. The fatty acid composition of POM is influenced by phytoplankton species composition (e.g. Sargent et al., 1985; Napolitano et al., 1997; Pond et al., 1998), and growth conditions can affect the fatty acid profile of individual species (Roessler, 1990; Thompson et al., 1992). Additionally, the presence of bacteria and dead material, such as faecal pellets and plant debris, can alter the fatty acid signature of POM (Reemtsma and Ittekkot, 1992; Hama, 1999; Canuel, 2001). Large contrasts in fatty acid composition between vertically-mixed regions and more stratified water columns have been observed by St. John and Lund (1996). Diatoms tend to be the dominant phytoplankton group in weakly stratified waters, and species of the classes Chrysophyceae, Haptophyceae and Dinophyceae dominate the phytoplankton community in more stratified water columns (Margalef, 1978; Cushing, 1989). Typically, diatoms have a much higher 16:1n-7/16:0 ratio than other phytoplankton classes, a

42 relatively high abundance of 20:5n-3 and C16 unsaturated fatty acids (see e.g. Pohl and Zurheide, 1979; Volkman et al., 1989). In contrast, Chrysophyceae, Haptophyceae and Dinophyceae have more C18 unsaturated fatty acids (see e.g. Pohl and Zurheide, 1979; Volkman et al., 1981; Volkman et al., 1989; Mansour et al., 1999). To some extent the ratios such as 16:1n-7/16:0, 16:1n-7/18:1n-9 and 20:5n-3/18:4n3 are successful in separating the shelf- and off shelf food webs of this study. The last two ratios used in conjunction displayed particularly good discriminating power. However, to get successful classification the dataset had to be sorted into smaller subsets, and grouped by trophic groups for separate years. This was necessary because the ratio values separating individuals from the two environments are markedly different for each trophic group, and varied from survey to survey. The non-conservative behaviour of the ratios through the food web may be explained by preferential uptake of unsaturated fatty acids (e.g. Neal et al., 1986; Harvey et al., 1987; Pond et al., 1995), and possibly due to some portion of the fatty acids being biosynthesized by the animals. With the use of multivariate analysis, it was found that a successful general classification rule for all investigated trophic groups could be established.

The difference in stable carbon isotope composition between the shelf and off shelf POM can be the result of many factors. The most promising candidates in this case include: 1) differences in growth rate (Fry and Wainright, 1991; Laws et al., 1995), with higher growth rates on the shelf, due to higher nutrient supply- and recycling rates, 2) a species- and/or cell size effect (Fry and Wainright, 1991; Popp et al., 1998), with diatoms and species with bigger cell size found more on the shelf and being more enriched in 13C, 3) a CO2 (aq) concentration difference (e.g. Rau et al., 1989; Hinga et al., 1994), the CO2 concentration in shelf waters could potentially have been depressed due to enhanced biological drawdown of inorganic carbon, 4) a more common use of active bicarbonate uptake and utilization by phytoplankton in an upwelling environment (Tortell et al., 2000; Rau et al., 2001). None of these four mechanisms can be ruled out with the data presented here. A further discussion will follow in Chapter 3.

43 In particular, the fatty acid data indicates that the shelf-off shelf demarcation is a distinction between diatom-rich- and diatom-poor waters. The results suggest that the transition in molecular and isotopic characteristics of the pelagic food webs occurs across the shelf-break. This coincidence can be of practical use in monitoring exchange between the shelf and off shelf food webs. Whether the differences observed persist throughout the year, or break down during autumn and winter is of special interest when larger animals with lower carbon turnover rates are investigated. Phytoplankton studies in open coastal waters off the west coast of Vancouver Island are limited to only a few studies (Denman et al., 1981; Mackas and Sefton, 1982; Forbes and Denman, 1991; MacIsaac et al. 1991; Taylor and Haigh, 1996). Three of these studies included some stations off the continental shelf. No study sampled between October and April. To my knowledge, sediment trap data are only available for slope, and open ocean stations. Peña et al. (1999) collected sediment trap data from a station just over 50km off the shelf of Vancouver Island, and from two more offshore stations. These researchers reported that mean values of the atomic ratios Sibiogenic/Corganic and Sibiogenic/Ccarbonate were higher at the station closest inshore for every season. Additionally, they found that in the time period studied (Sept. 1994-1995), the organic matter was consistently 13C enriched for the most inshore station. The only exception is a single data point in October. Unfortunately data are lacking for the inshore station from October through March. Hence, there is indeed some indication that inshore-offshore differences persist throughout the year, but the available evidence is parsimonious. In the Chapter 3 the temporal and spatial variation of the fatty acid, and stable carbon isotope composition of seston will be described in more detail, and compared to various physical parameters. Also, the potential mechanisms explaining the observed variations will be more closely examined.

2.5.2 Stable carbon isotope composition of fatty acids As can be seen in Figure 2.9, substantial differences occur between the δ13C of individual fatty acids within a sample. A range of 9-10‰ in the δ13C values is quite common. A

44 large range in carbon isotope compositions of fatty acids within the sample is known (see e.g. Parker, 1964; Fang et al., 1993; Murphy and Abrajano, 1994; Pond et al., 2000). Fang et al. (1993) pointed out that these differences are most likely caused by fractionation during the synthesis of fatty acids. They proposed that kinetic isotope effects during desaturation and elongation resulted in a depletion in the heavier isotope. Results here also indicate that the differences occur during synthesis. The fact that shelf zooplankton had a different diet than zooplankton collected off the shelf, did not really affect the observed pattern. These data are in line with the hypothesis that 13C depletion occurs during desaturation. However, elongation seems to have had the opposite effect. More data and discussion about this will follow in chapter 4.

2.5.3 Turnover in different organisms Short-term variability in isotope composition of primary producers can be expected to smooth out in higher trophic level animals due to a lower turnover of carbon in these organisms. In the present study the percentage of larval fish that were misclassified was considerably lower than for other taxa when using any of the three data sets (see Figure 2.7). This could be interpreted as a result of lower carbon turnover rates in larval fish, making the stable carbon isotope signature less sensitive to recent changes in diet. It is known, however, that larval fish can exhibit very high growth rates. For example, first feeding (2.6mm) and settlement size (4-8mm) red drum larvae (Sciaenops ocellatus) were found to gain 7 times their initial weight in 10 days under culture conditions (Herzka and Holt, 2000). Such high growth rates will result in a rapid dilution of the original carbon pool. However, all of the larval fish collected during the three cruises for this study were larger than 8mm and can be expected to have slower growth rates than the fish of the aforementioned study.

2.5.4 Carbon turnover in different tissues and biochemical fractions Next to the wide range of turnover rates of carbon for different organisms, a considerable range in turnover times has been observed for various animal tissues. Tieszen et al.

45 (1983) measured the stable carbon isotope composition of hair, brain, muscle, fat and liver tissues of gerbils after a diet switch, and Hobson and Clark (1992) analyzed bone collagen, muscle, blood and liver from quails. Liver tissue was found to lose its original diet δ13C-value faster than muscle tissue in gerbils and quails. Half-life values were found to be 6.4 vs. 27.6 days for gerbils, and 2.6 vs. 12.4 days for quails, respectively. However, such a difference in turnover rate between the two tissues was not found in broad whitefish (Hesslein et al., 1993). Analogous to the variation in turnover times of different tissues, differences can be expected to occur between the various biochemical fractions. Roman (1991) fed the copepod Acartia tonsa

14

C-labelled algae, and his pulse-chase experiments showed that

carbon in proteins had a “much lower turnover rate and was thus conserved more than carbon in polysaccharides and lipids”. Similarly, in starvation studies proteins were found to be more conserved than other compound classes, e.g., in a prawn: Whyte et al. (1986); rotifer: Frolov and Pankov (1992); humans: Stryer (1988). When an animal switches from a 13C-enriched to a 13C-depleted diet, and fatty acids from the new diet replace the fatty acids from the old diet quicker than the proteins in the animal, a bigger δ13CFA to protein δ13C (δ13CProtein) difference is expected. Similarly, smaller than average δ13CFA-δ13CProtein differences are expected when an animal moved recently into waters with more

13

C enriched food. Hence, it is suggested here that the

δ13CFA-δ13CProtein difference can potentially serve as an indicator for a recent change in diet. For such an application, it is of paramount importance that a good estimate of the average δ13CFA-δ13CProtein is determined for organisms that are well equilibrated with their diet.

2.5.5 Turnover of different fatty acids Within the lipid fraction, differences in turnover rates for the various lipid classes are also expected. Triacylglycerols and wax esters are used as energy storage, whereas polar lipids, such as phospholipids and glycolipids, are known to be important building blocks of cell membranes. Polar lipids are therefore likely to be replaced at a slower rate.

46 In addition, it is argued here that when a change in diet occurs, each fatty acid can be expected to have a different turnover time. The turnover rate of a compound depends on three main factors: 1) the rate of new supply (assimilation and/or synthesis), 2) the rate of removal (metabolism and/or catabolism), and 3) the size of the pool to turnover (amount present before diet change). Consequently, when a fatty acid was present in low amounts before the diet change, and supplied in greater amounts via the new diet, a faster turnover is expected than for the reverse scenario. POM from the LC9 station had a higher abundance of C18 unsaturated fatty acids than LC4 POM in May 1998 (Figure 2.10b). This difference is also reflected in zooplankton collected at those stations (Figure 2.10a). Therefore, the C18 unsaturates are expected to turnover faster in animals switching from an LC4 to an LC9 diet. It was found that the δ13C-values of the C18 unsaturated fatty acids in the euphasiid furciliae which are hypothesized to have moved from the shelf to LC9, indeed show a close resemblance to values in other organisms collected at LC9. In contrast, the 20:5n-3 fatty acid which was found in relatively higher abundance in the shelf food web, is still found to be off by 3‰ (Figure 2.11). Conversely, for the May ’99 LBP2 copepods the 18:4n-3 fatty acid is expected to turnover slower because of its lower concentration in the shelf diet. Indeed, the δ13C of 18:4n-3 was still found to be closer to the LBP7 (off shelf) value, than the δ13C of 20:5n-3, which was found in higher abundance in POM on the shelf.

2.5.6 Conceptual model To illustrate the use of the different equilibration times for different fatty acids after a diet change, a simple linear mixing model was constructed. The model takes into account the three factors that influence the turnover time, as identified in the previous section: rate of new supply, rate of removal and size of the pool. Since in this discussion will be focussed on essential fatty acids, the rate of new supply is assumed to be equal to the rate of assimilation, i.e., proportional to diet intake. Synthesis of these essential fatty acids in the animal is considered negligible. The rate of removal (after assimilation) of the essential fatty acids is assumed to be equal to the rate at which they are catabolized. The initial size of the pool is the amount of the fatty acid estimated to be present in the organism before

47 the diet switch. The δ13CFA is calculated for each time increment using a mass balance equation: δ 13CFA (t ) = f (t ) ⋅ δ new + [1 − f (t )] ⋅ δ old

(2.6)

where f(t) is the fraction of the fatty acid derived from the new diet (as a function of time), δnew is the δ13CFA value of the new diet, and δold is the δ13CFA of the previous diet. The f(t) is calculated with the use of 7 other equations described in the Appendix of this chapter. The δ13CFA curves were modeled for the 18:3n-3 and 20:5n-3 fatty acids found in euphausiid furciliae collected at station LC9 in May 1998. For simplicity it was assumed that they had been feeding at LC4 and subsequently, at t=0, moved instantaneously to LC9. The modeled δ13CFA curves for 18:3n-3 and 20:5n-3 for a period of 10 weeks after the diet shift are shown in Figure 2.11. In about 18 days the 20:5n-3 fatty acid in the modeled euphausiid larva has attained 13

a δ C similar to that of the measured value of the LC9 euphausiid (-27.2‰), and the δ13C of 18:3n-3 is also within analytical error range of the LC9 euphausiid value. Hence, if the modeled scenario would apply to the euphausiid furciliae collected at LC9, the larvae would have been estimated to have moved and shifted to the off shelf LC9 diet around 18 days prior to sampling. As can be observed in Figure 2.11, this estimate is quite sensitive to the amount of lipid assimilated and catabolized per day. When incorporation rates (rassrcat) of 5% of body lipid per day are used, and catabolization is still held at 67% of the assimilated amount (see appendix for discussion), the model estimates the shift occurred about 10 days before sampling. Alternatively, when an incorporation rate of 1% of body lipid is used, it would take 55 days. This last scenario would result, however, in a very slow weight gain of a 100%, i.e., in about 100 days, and is too long because the furciliae were probably less than 55 days old. The study by Huntley and Brinton (1991) suggests a 100% gain in about 30-40 days for Euphausia superba furcilia stages. When the incorporation rate is adjusted to 3% of body lipid incorporated per day (see Figure 2.11,

48 rass-rcat=3 line) the growth rate corresponds better with the field observations of Huntley and Brinton (1991). If it is assumed that the euphausiid furciliae moved from station LC4 to LC9 in 18 days, they would have needed to have a velocity with an average cross-shore component of 3.4 cm/s away from the coast (perpendicular to the coastline). Such a speed is not in contradiction with average cross-shore velocities (from 16 cm/s towards, to 39 cm/s away from the coast) reported by Mackas and Yelland (1999) for surface-layer Loran drifter buoys off the west coast of Vancouver Island.

Figure 2.11. Results of conceptual model, showing the modelled δ13CFA (left) and fatty acid abundance (right) in a euphausiid furcilia after a diet switch as a function of time. The grey areas represent the analytical precision (± 1 SD of 63 replicates, Table 2.1). The three lines show the sensitivity of the model to the rate of incorporation (rass-rcat, in µg/d), while the proportion of the assimilated fatty acids being catabolized is kept stable at 67%. See Appendix for discussion on model and values used. The circles, squares and triangles represent real measured values (see Figs. 2.9 & 2.10).

49 Winds were generally equatorward and favorable to seaward Ekman transport prior to, and during the 1998 survey (Lu, 1999 and Lu et al., in press). The hypothesis that Euphausiid furciliae moved seaward off Vancouver Island during May 1998 has been corroborated by Lu (1999) and Lu et al. (in press). They found that Euphausiid larvae were located seaward of adults in May 1998. This was due to the cumulative effects of wind-driven (and vertically-sheared) cross-shore transport on the surface-dwelling larvae versus diel vertical migratory late juveniles and adults (Lu, 1999 and Lu et al., in press).

Figure 2.12 shows the modeled difference over time between the δ13C values of the 20:5n-3 and 18:3n-3 fatty acids, respectively (∆δ13C20:5n-3 – 18:3n-3). Due to the difference in turnover time of the two fatty acids the ∆δ13C20:5n-3 – 18:3n-3 first rapidly increases and then steadily decreases to the new steady state value. Alternatively, when an animal shifts to a more

13

∆δ13C20:5n-3

C-enriched (shelf-) diet, with relatively more 20:5n-3 and less 18:3n-3, the – 18:3n-3

will increase over time after an initial quick drop. When the

assimilation- and catabolization rates are well constrained the timing of the diet shift can be estimated (Figure 2.12). For the LC9 euphausiid furcilliae, used as an example here, the diet switch is estimated at 20 days before sampling when using the best estimate of rass-rcat, i.e., 3% of body lipid incorporated per day (with 67% of the assimilated fatty acids catabolized). As illustrated by Figure 2.12, an organism that recently switched to a diet with a different isotope ratio should exhibit anomalous differences between compounds with dissimilar turnover rates. Therefore, such an unusual difference can potentially be used as an indicator for a recent diet switch. However, some caution is advised, since after a very recent diet shift this phenomenon cannot be observed yet (Figure 2.12). In that case, however, both compounds will both have a very different isotope ratio in the current diet, as compared to those present in the organism itself.

δ13C20:5n-3 - δ13C18:3n-3 difference (‰)

50

5.5 5.0

ra ss r cat =1

4.5 4.0 3.5

ra

3.0

ra

ss

2.5

ss

-r c

at

-r c

at

=3

=5

2.0 1.5

0

10

20

30

40

50

60

70

Time since diet change (days) Figure 2.12. The modeled difference between the δ13C of the 20:5n-3 and 18:3n-3 fatty acid over time (see Fig. 2.11). The dashed line indicates the value found for the May 1998 LC9 euphausiid furciliae.

To test the applicability of anomalous ∆δ13C values as indicators for diet change, ∆δ13C20:5n-3 – 18:3n-3 values of the measured zooplankton samples were plotted in a normal probability plot, or quantile-quantile plot (Figure 2.13). Samples previously identified as consisting of organisms that moved into off shelf waters (see Table 2.4) are expected to plot at the high end of the distribution. This is because of the higher turnover rate of the 18:3n-3 fatty acid it will shift to more negative δ13C values quicker than the 20:5n-3 fatty acid, and the 13C/12C ratio of the latter is usually higher to begin with. Animals collected in shelf waters, hypothesized to have moved from further offshore, are also anticipated to have unusually high ∆δ13C20:5n-3 – 18:3n-3 values. In this case 20:5n-3 becomes more rapidly enriched in 13C than the 18:3n-3 fatty acid, thus increasing the ∆δ13C20:5n-3 – 18:3n-3. When all samples of the population plot close to the lines that represent values from an ideal normal distribution (see Figure 2.13), the samples are normally (or Gaussian) distributed. Outliers will plot as extremes and depart from the straight line. Because of the difference in mean ∆δ13C20:5n-3 – 18:3n-3, the shelf and off shelf samples were plotted separately. Unfortunately, only 8 of the 14 samples suspected to consist of organisms that

51 moved from shelf- to off shelf waters, or vice versa, were analyzed with GC-IRMS. Of those 8 samples, 6 had δ13C values available for both the 20:5n-3 and 18:3n-3 fatty acids.

6

1

5

5

4

6

2 3

3 2

4

1

5

∆δ C20:5n-3 - 18:3n-3 values (‰)

0

13

13

Ordered ∆δ C20:5n-3 - 18:3n-3 values (‰)

6

-1

Off shelf zooplankton Shelf zooplankton

-2 -2

-1

0

1

2

Theoretical Quantiles of Standard Normal (z-score)

4 3 2 1 0 -1

Off shelf zoopl. Shelf zoopl.

-2 0

1

2

3

4

5

6

7

Number of samples

Figure 2.13. A normal probability plot (or qq plot) of the standardized quantiles of the δ13C differences between 20:5n-3 and 18:3n-3 fatty acids in zooplankton (left). The straight lines represent a normal (Gaussian) distribution with the same mean and standard deviation as estimated for the two (shelf and off shelf) populations. Deviations from the line are deviations from normality. The x-axis is expressed in standard deviations away from the mean (z-score). Open symbols represent samples that were suspected to have switched between shelf and off shelf environment (Table 2.4). 1=euphausiids, at LG7, May 1999. 2=zoopl.>425µm, LC9, May 1998. 3=euphausiid furciliae, LC9, May 1998. 4=zoopl.212425µm, LBP2, May 1998. 5=zoopl.>425µm, LG3, May 1999. 6=zoopl.212-425µm, LG3, May 1999. The plot on the right-hand side shows the frequency distribution of the δ13C differences between 20:5n3 and 18:3n-3 fatty acids in zooplankton from the shelf, and off shelf environments.

The departures from the normal distribution (Figure 2.13) are not large enough to reject the hypothesis (H0) that both populations are normally distributed (H0 true with P>0.2, Lilliefors test statistic: variation of Kolmogorov-Smirnov test, accounting for estimation of the mean and variance). Though not classified as outliers, the 3 off shelf samples suspected to have come from shelf waters all rank in the top 4 highest ∆δ13C20:5n3 – 18:3n-3

values in the off shelf population (see Fig. 2.13). When these 3 samples are left

out from the population when calculating the mean and variance, the values would be 1.9, 2.7 and 3 standard deviations away from the mean. Therefore, a recent diet shift by these

52 organisms, possibly due to movement from the shelf- to the off shelf environment, appears to be confirmed.

The zooplankton from shelf-station LBP2 (May 1998) that was hypothesized to have moved (see Table 2.4), has the second lowest ∆δ13C20:5n-3

– 18:3n-3

value of the

samples collected in shelf waters (see Fig 2.13). This may indicate that the movement of the zooplankton into shelf waters was very recent, so that the ∆δ13C20:5n-3 – 18:3n-3 value could not yet have changed substantially. This would be in line with the fact that the ∆δ13C20:5n-3 – 18:3n-3 value is just above the average value for off-shelf samples. It is also important to note that the initial rise to high ∆δ13C20:5n-3 – 18:3n-3 values will not be as steep as in Figure 2.12 in this case. This is because the difference in the relative abundance in 20:5n-3 between the shelf and off shelf environments is not as high as for 18:3n-3 in the Line-C example used for Figure 2.11 and 2.12. Hence, turnover rates will also differ less. Due to the narrow continental shelf in the north, the distance between the station LBP2 and waters away from the shelf is about 10 km (Figure 2.1). Cross-shore transport with a velocity of less than 12 cm/s would be able bring the respective zooplankton from off the shelf to station LBP2 in 1 day. This is well within the range of cross-shore water movement velocities reported by Mackas and Yelland (1999). Therefore, it is not unlikely that the zooplankton (with size between 212 and 425µm) very recently moved to LBP2 from off-shelf waters. The 2 samples from LG3 (May 1999) that with the use of the fatty acid abundance data were predicted to have moved into the shelf area, do not display extreme ∆δ13C20:5n-3 – 18:3n-3

values. This is in accordance with the classification of the POM at LG3 as “off

shelf POM”, using the δ13CBulk and δ13CFA data. They probably attained more off shelf characteristics because of the “off shelf-like POM” at the base of the food web. Therefore, these organisms either moved with their food source or did not move from an off shelf location.

53

2.5.7 Application and evaluation 2.5.7a Recognizing a change in diet As explained anomalous isotope ratios, or abnormal differences between tissues or individual compounds with different turnover rates can indicate that an organism has recently shifted its diet. However, a change in diet can occur because of several reasons. A diet shift can occur due to: (1) a bottom-up effect, or shift in isotope signature by the diet itself (for example, when the δ13C of the phytoplankton changed because of changing physical- and or nutrient conditions, i.e., when a species shift or bloom took place). (2) Movement of the organism into an area where the dietary constituents have a different isotope ratio. Finally, (3) a changed food preference by the organism due to ontogenetic change (for example, a juvenile switching to bigger prey animals). It should be possible to recognize all three scenarios with the help of stable isotope ratio- and/or fatty acid data. The bottom-up effect will result in an unusually large or small difference between organisms from different trophic levels. Differences in isotope ratios between organisms that are one trophic level apart are around 0-1‰ for carbon and ca. 3.4‰ for nitrogen (Michener and Schell, 1994; Post, 2002). The organisms constituting a food web will have different turnover rates of carbon and nitrogen, with a general trend of slower turnover in animals at higher trophic levels. Therefore, when a recent decrease in the δ13C or δ15N of phytoplankton has taken place the trophic steps are associated with bigger differences in the isotope ratios. Similarly, a recent increase results in a decrease (or reversal) of the δ13C or δ15N differences between trophic levels. As an example, O’Reilly et al. (2002) showed that in an African lake the δ15N had a negative trend with trophic level, instead of a positive slope with 3.4‰ steps per trophic level. This phenomenon was caused by a flux of 15N-enriched nitrate recently taken up by the phytoplankton, and the result of different temporal integration of nitrogen at the different trophic levels. Movement of an animal into a region with different diet isotope composition will not result in an effect on trophic differences in the whole food web. Rather, the isotope composition of the moved animal(s) will not fit in the overall pattern of isotope ratios of

54 the other analyzed organisms of the food web. Alternatively, when animals have moved with the water and its food sources, the local food web isotope signature will be discordant with the isotope signature of the surrounding regional food web. The diet shift due to a changed preference (e.g., due to ontogenetic change) will not likely result in a change in isotope ratio that can easily be recognized. The shift in isotope ratio may be too small and too gradual, but the fatty acid signature may change quite significantly. For example a change from herbivory to carnivory should result in a significant change in the fatty acid signature of the animal (see for example, Cripps and Atkinson, 2000). When using isotope ratios to determine the trophic position of an animal, recent diet shifts due to a bottom-up effect or movement will result in a flawed estimation when the two end-member diets are significantly different. Determining the differences between tissues, biochemical fractions and individual compounds with different turnover rates will help to confirm a recent diet shift. Animals that are confirmed to have switched diet can then be left out of the dataset when determining the trophic position of a species.

2.5.7b Tag longevity The modelling exercise in this chapter demonstrates how timing of movement or diet shift can be estimated in a well-constrained system. The δ13C of various fatty acids and their differences can potentially be used as a “clock”. The precision of the estimates depends on the knowledge of carbon replacement rates within the organism, which unfortunately are often not well known. Additionally, the δ13C values of the fatty acids before the diet shift need to be well constrained. My model indicates that in a matter of weeks the old diet signal will be lost in the zooplankton considered (see Figure 2.11). When using the best estimate of rass-rcat=3, in 7 weeks the 20:5n-3 fatty acid in the euphausiid furciliae reaches a δ13C value that is 1 per mil away from the diet. As discussed, other biochemical fractions such as proteins may turnover even slower. It should be noted, however, that the longevity of the old diet signature can vary considerably as a wide range of growth rates is found in marine organisms. The bulk stable carbon isotope composition of rapidly-growing red drum larvae (Sciaenops

55 ocellatus) was estimated to stabilize within 10 days after a dietary shift (Herzka and Holt, 2000). Frazer et al. (1997) reared stage F6 Euphausia superba Dana at 1.5 and –1.5ºC and found linear growth, with a 100% weight gain in approximately 7 and 10 weeks respectively. The δ13C values had not stabilized yet within the 10 week experiment, and they found that nitrogen had an even slower turnover rate than carbon. Tissue of 2.5 year old broad whitefish took almost a full year to stabilize (Hesslein et al., 1993). Additionally, the isotope signature of the old diet is lost earlier when the difference between the two diet end-members is small. In that case the animal will adopt δ13C values falling within the range measured for organisms in the new area earlier. This is especially true when the within-food web variability of δ13C values is already high.

2.5.7c Evaluation As outlined before, it is possible to identify animals that have recently moved into an area using a fatty acid abundances and stable carbon isotope ratios. The use of fatty acids or isotope signatures as a natural tag depends on how well the food webs of different areas can be characterized and distinguished from each other. Of the three techniques (bulk IRMS, GC-IRMS, and fatty acid profile analysis) tested here on their ability to discriminate between shelf and off shelf samples, none is clearly superior. Using the bulk δ13C data resulted generally in the lowest success rates in classifying the shelf- and off shelf samples. Of the two other datasets, fatty acid δ13C data seemed to perform slightly better in correctly assigning the samples to the respective groups, than fatty acid abundance data (Table 2.3). It should be noted, however, that the classification with fatty acid abundance data is improved when separate discriminant functions and classification rules are determined for each cruise. As mentioned, the most conservative estimation of the error rate (mean-1/3-subset) would then be about 1.2% lower, making the fatty acid abundance data most successful in correctly classifying the samples. Although the bulk δ13C data resulted in the lowest success rate in classifying the samples, from a practical point of view bulk IRMS should not be dismissed. Because of the considerably lower consumption of time and analytical resources by bulk δ13C analyses, measurements can be done on a larger amount of samples in less time. An

56 initial “screening” of the samples by bulk δ13C analysis is suggested, which can then be followed by GC-IRMS and fatty acid profile analysis on the samples for which more information is required. As discussed, also bulk protein and lipid stable carbon isotope ratios may be useful for confirmation of a diet shift. This chapter demonstrates that it is in particular the combination of the three techniques that can be useful for tracking diet change and movement of aquatic- and, perhaps, even non-aquatic organisms.

2.6 Conclusions On the basis of fatty acid composition, stable carbon isotope ratio of either bulk sample, or individual fatty acids, samples of POM, zooplankton and larval fish from the continental shelf off Vancouver Island could generally be distinguished from specimens collected over deeper waters. In the present study it was found that the fatty acid composition and fatty acid δ13C data performed slightly better in discriminating between samples from the two environments than bulk δ13C values. When animals move between regions with different food sources, the various biochemical fractions in the organism will retain the old dietary signal on different time scales. Therefore, I propose that an unusual lipid-protein difference in carbon isotope composition can be a confirmation of a recent diet shift. Additionally, I suggest that after a diet change the 13C/12C ratio adjusts to present diet values at different rates for each of the individual fatty acids. Results presented here demonstrate the potential use of this effect to obtain time constraints on animal movement between two dietary regimes.

2.7 References Abelson, P.H. and Hoering, T.C. 1961. Carbon isotope fractionation in formation of amino acids by photosynthetic organisms. Proceedings of the National Academy of science U.S. 47, 623-632. Abrajano, T.A. Jr, Murphy, D.E., Fang, J., Comet, P. and Brooks, J.M. 1994. 13C/12C ratios in individual fatty acids of marine mytilids with and without bacterial symbionts. Organic Geochemistry 21, 611-617.

57

Ackman, R.G., 1986. WCOT (capillary) gas-liquid chromatography. In: R.J. Hamilton and J.B. Rossell (Eds.), Analysis of Oils and Fats. Elsevier, New York, pp. 137-206. Ballentine, D.C., Macko, S.A., Turekian, V.C., Gilhooly, W.P. and Martincigh, B. 1996. Compound specific isotope analysis of fatty acids and polycyclic aromatic hydrocarbons in aerosols: Implications for biomass burning. Organic Geochemistry 25, 97-104. Canuel, E.A. 2001. Relations between river flow, primary production and fatty acid composition of particulate organic matter in San Francisco and Chesapeake Bays: a multivariate approach. Organic Geochemistry 32, 563-583. Canuel, E.A., Cloern, J.E., Ringelberg, D.B., Guckert, J.B., and Rau, G.H. 1995. Molecular and isotopic tracers used to examine sources of organic matter and its incorporation into the food webs of San Francisco Bay. Limnology and Oceanography 40, 67-81. Castell. J.D., Boston, L.D., Miller, R.J. and Kenchington, T. 1995. The potential identification of the geographic origin of lobster eggs from various wild stocks based on fatty acid composition. Canadian Journal of Fisheries and Aquatic Sciences 52, 1135-1140. Craig, H. 1957. Isotopic standards for carbon and oxygen and correction factors for mass-spectrometric analysis of carbon dioxide. Geochimica et Cosmochimica Acta 12, 133-149. Cripps, G.C. and Atkinson, A. 2000. Fatty acid composition as an indicator of carnivory in Antarctic krill, Euphausia superba. Canadian Journal of Fisheries and Aquatic Sciences 57 (supplement 3), 31-37. Cushing, D.H. 1989. A difference in structure between ecosystems in strongly stratified waters and in those that are only weakly startified. Journal of plankton research 11, 1-13. DeNiro, M.J. and Epstein, S. 1977. Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197, 261-263. DeNiro, M.J. and Epstein, S. 1978. Influence of diet on the distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta 42, 495-506. Denman, K.L., Mackas, D.L., Freeland, H.J., Austin, M.J. and Hill, S.H. 1981. Persistent upwelling and mesosclae zones of high productivity off the west coast of Vancouver Island, Canada. In: Coastal upwelling (ed. F.A. Richards). American Geophysical Union, Washington, D.C. pp. 514-521. Desvilettes, Ch., Boudier, G., Breton, J.C. and Combrouze, Ph. 1994. Fatty acids as organic markers for the sudy of trophic relationships in littoral cladoceran communities of a pond. Journal of Plankton Research 16, 643-659.

58 Ederington, M.C., McManus, G.B. and Harvey, H.R. 1995. Trophic transfer of fatty acids, sterols, and a triterpenoid alcohol between bacteria, a ciliate, and the copepod Acartia tonsa. Limnology and Oceanography 40, 860-867. Fang, J., Abrajano, T.A., Comet, P., Brooks, J.M. and MacDonald, I. 1993. Gulf of Mexico hydrocarbon seep communities: XI – carbon isotopic fractionation during fatty acid biosynthesis of seep organisms and its implications for chemosynthetic processes. Chemical geology 109, 271-279. Fantle, M.S., Dittel, A.I., Schwalm, S.M., Epifanio, C.E. and Fogel, M.L. 1999. A food web analysis of the juvenile blue crab, Callinectes sapidus, using stable isotopes in whole animals and individual amino acids. Oecologia 120, 416-426. Fisher, R.A. 1936. The use of multiple measurements in taxonomic problems. Annals of Eugenics 7, 179188. Forbes, J.R. and Denman, K.L. 1991. Distribution of Nitzschia pungens in coastal waters of British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 48, 960-967. Fraser, A.J., Sargent, J.R., Gamble, J.C. and Seaton, D.D. 1989. Formation and transfer of fatty acids in an enclosed marine chain comprising phytoplankton, zooplankton and herring (Clupea harengus L.) larvae. Marine Chemistry 27, 1-18. Frazer, T.K., Ross, R.M., Quetin, L.B. and Montoya, J.P. 1997. Turnover of carbon and nitrogen during growth of larval krill, Euphausia superba Dana: a stable isotope approach. Journal of Experimental Marine Biology and Ecology 212, 259-275. Freeland, H.J., Crawford, W.R. and Thomson, R.E. 1984. Currents along the Pacific coast of Canada. Atmosphere-ocean 22, 151-172. Frolov, A.V., Pankov, A.V., Geradze, K.N., Pankova, S.A. and Spektorova, L.V. 1991. Influence of the biochemical composition of food on the biochemical composition of the rotifer Brachionus plicatilis. Aquaculture 97, 181-202. Frolov, A.V. and Pankov, A.V. 1992. The effects of starvation on the biochemical composition of the rotifer Brachionus plicatilis. Journal of the Marine Biological Association of the United Kingdom 72, 343356. Fry, B. 1981. Natural stable carbon isotope tag traces Texas shrimp migrations. Fishery Bulletin 79, 337345. Fry B. and Wainright, S.C. 1991. Diatom sources of 13C-rich carbon in marine food webs. Marine Ecology Progress Series 76, 149-157.

59 Graeve, M., Kattner, G. and Hagen, W. 1994. Diet-induced changes in the fatty acid composition of Arctic herbivorous copepods: experimental evidence of trophic markers. Journal of experimental marine biology and ecology 182, 97-110. Hama, T. 1999. Fatty acid composition of particulate matter and photosynthetic products in subarctic and subtropical Pacific. Journal of Plankton Research 21, 1355-1372. Hand, D.J. 1997. Construction and assessment of classification rules. Wiley, New York. Hansson, S., Hobbie, J.E., Elmgren, R., Larsson, U., Fry, B. and Johansson, S. 1997. The stable nitrogen isotope ratio as a marker of food-web interactions and fish migration. Ecology 78, 2249-2257. Harvey, H.R., Eglinton, G., O’Hara, S.C.M. and Corner, E.D.S. 1987. Biotransformation and assimilation of dietary lipids by Calanus feeding on a dinoflagellate. Geochimica et Cosmochimica Acta 51, 3031-3040. Hayes, J.M., Freeman, K.H., Popp, B.N., and Hoham, C.H. 1990. Compound-specific isotopic analyses: A novel tool for reconstruction of ancient biogeochemical processes. In: Advances in Organic Geochemistry 1989 (eds. B. Durand and F. Behar). Organic Geochemistry 16, 1115-1128. Herzka, S.Z. and Holt, G.J. 2000. Changes in isotopic composition of red drum (Sciaenops ocellatus) larvae in response to dietary shifts: potential applications to settlement studies. Canadian Journal of Fisheries and Aquatic Sciences 57, 137-147. Hesslein, R.H., Capel, M.J., Fox, D.E. and Hallard, K.A. 1991. Stable isotopes of sulfur, carbon, and nitrogen as indicators of trophic level and fish migration in the lower Mackenzie River basin, Canada. Canadian Journal of Fisheries and Aquatic Sciences 48, 2258-2265. Hesslein, R.H., Hallard, K.A. and Ramlal, P. 1993. Replacement of sulfur, carbon, and nitrogen in tissue of growing broad whitefish (Coregonus nasus) in response to a change in diet traced by δ34S, δ13C, and δ15N. Canadian Journal of Fisheries and Aquatic Sciences 50, 2071-2076. Hinga, K.R., Arthur, M.A., Pilson, M.E.Q., Whitaker, D. 1994. Carbon isotope fractionation by marine phytoplankton in culture: The effects of CO2 concentration, pH, temperature, and species. Global Biogeochemical Cycles 8, 91-102. Hobson, K.A. and Clark, R.G. 1992. Assessing avian diets using stable isotopes I: turnover of tissues. The Condor 94, 181-188.

13

C in

Huang, L., Sturchio, N.C., Abrajano, T., Heraty, L.J. and Holt, B.D. 1999. Carbon and chlorine isotope fractionation of chlorinated aliphatic hydrocarbons by evaporation. Organic Geochemistry 30, 777-785. Huntley, M. and Brinton, E. 1991. Mesoscale variation in growth and early development of Euphausia superba Dana in the western Bransfield Strait region. Deep-Sea Research 38, 1213-1240.

60

Lachenbruch, P.A. 1975. Discriminant analysis. Hafner Press, New York. Lachenbruch, P.A. and Mickey, M.R. 1968. Estimation of error rates in discriminant analysis. Technometrics 10, 1-11. Laws, E.A., Popp, B.N., Bidigare, R.R., Kennicutt, M.C. and Macko, S.A. 1995. Dependence of phytoplankton carbon isotopic composition on growth rate and [CO2]aq: Theoretical considerations and experimental results. Geochimica et Cosmochimica Acta 59, 1131-1138. Lu, B.W. 1999. Zooplankton distribution along the shelf break off Vancouver Island and its relationship with currents. MSc. thesis, University of Victoria, Canada, 80pp. Lu, B.-W., Mackas, D.L. and Moore, D.F. 2003 in press. Cross-shore separation of adult and juvenile euphausiids in a shelf-break alongshore current. Submitted to Progress in Oceanography. MacAvoy, S.E., Macko, S.A., Joye, S.B. 2002. Fatty acid carbon isotope signatures in chemosynthetic mussels and tube worms from gulf of Mexico hydrocarbon seep communities. Chemical Geology 185, 1-8. MacIsaac, E.A., Gollner, M.C. and Forbes, J.R. 1991. Phytoplankton distribution and productivity in Barkley Sound and Alberni Inlet: April 1987 survey for the marine survival of salmon program. Canadian Technical Report on Fisheries and Aquatic Science No. 1802. Mackas, D.L. 1992. Seasonal cycle of zooplankton off southwestern British Columbia: 1979-89. Canadian Journal of Fisheries and Aquatic Sciences 49, 903-921. Mackas, D.L. and Sefton, H.A. 1982. Plankton species assemblages off southern Vancouver Island: geographic patterns and temporal variability. Journal of marine research 40, 1173-1200. Mackas, D.L. and Yelland, D.R. 1999. Horizontal flux of nutrients and plankton across and along the British Columbia continental margin. Deep-Sea Research II 46, 2941-2967. Mansour, M.P., Volkman, J.K., Jackson, A.E. and Blackburn, S.I. 1999. The fatty acid and sterol composition of five marine dinoflagellates. Journal of phycology 35, 710-720. Margalef, R. 1978. Life-forms of phytoplankton as survival alternatives in an unstable environment. Oceanologica Acta 1, 493-509. Michener, R.H. and Schell, D.M. 1994. Stable isotope ratios as tracers in marine aquatic food webs. In: K. Lajtha and R.H. Michener (eds.), Stable isotopes in ecology and environmental science. Blackwell Scientific, p. 138-157.

61 Murphy, D.E. and Abrajano, T.A. Jr. 1994. Carbon isotope compositions of fatty acids in mussels from Newfoundland estuaries. Estuarine, coastal and shelf science 39, 261-272. Napolitano, G.E., Pollero, R.J., Gayoso, A.M., MacDonald, B.A. and Thompson, R.J. 1997. Fatty acids as trophic markers of phytoplankton blooms in the Bahía Blanca estuary (Buenos Aires, Argentina) and in Trinity Bay (Newfoundland, Canada). Biochemical Systematics and Ecology 25, 739-755. Neal, A.C., Prahl, F.G., Eglinton, G., O’Hara, S.C.M. and Corner, E.D.S. 1986. Lipid changes during a planktonic feeding sequence involving unicellular algae, Elminius nauplii and adult Calanus. Journal of the Marine Biological Association of the United Kingdom 66, 1-13. O’Reilly, C.M., Hecky, R.E., Cohen, A.S. and Plisnier, P.-D. 2002. Interpreting stable isotopes in food webs: Recognizing the role of time averaging at different trophic levels. Limnology and Oceanography 47, 306-309. Parker, P.L. 1964. The biogeochemistry of the stable isotopes of carbon in a marine bay. Geochimica et Cosmochimica Acta 28, 1155-1164. Peña, M.A., Denman, K.L., Calvert, S.E., Thomson, R.E. and Forbes, J.R. 1999. The seasonal cycle in sinking particle fluxes off Vancouver Island, British Columbia. Deep-Sea Research II 46, 2969-2992. Perry, R.I., Thompson, P.A., Mackas, D.L., Harrison, P.J. and Yelland, D. 1999. Stable carbon isotopes as pelagic food web tracers in adjacent shelf and slope regions off British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 56, 2477-2486. Pohl, P. and Zuhrheide, F. 1979. Fatty acids and lipids of marine algae and the control of their biosynthesis by environmental factors. In: Hoppe, H.A., Levring, T. and Tanaka, Y. (eds) Marine algae in pharmaceutical science. Walter de Gruyer, Berlin, 473-523. Pond, C.M., Mattacks, C.A., Gilmour, I., Johnston, M.A., Pillinger, C.T. and Prestrud, P. 1995. Chemical and carbon isotopic composition of fatty acids in adipose tissue as indicators of dietary history in wild arctic foxes (Alopex lagopus) on Svalbard. Journal of Zoology, London 236, 611-623. Pond, D.W., Priddle, J., Sargent, J.R. and Watkins, J.L. 1995. Laboratory studies of assimilation and egestion of algal lipid by Antarctic krill – methods and initial results. Journal of Experimental Marine Biology and Ecology 187, 253-268. Pond, D.W., Dixon, D.R., Bell, M.V., Fallick, A.E. and Sargent, J.R. 1997a. Occurrence of 16:2(n-4) and 18:2(n-4) fatty acids in the lipids of the hydrothermal vent shrimps Rimicaris exoculata and Alvinocaris markensis: nutritional and trophic implications. Marine Ecology Progress Series 156, 167-174. Pond, D.W., Segonzac, M., Bell, M.V., Dixon, D.R., Fallick, A.E. and Sargent, J.R. 1997b. Lipid and lipid carbon stable isotope composition of the hydrothermal vent shrimp Mirocaris fortunata: evidence for nutritional dependence on photosynthetically fixed carbon. Marine Ecology Progress Series 157, 221-231.

62

Pond, D.W., Bell, M.V., Harris, R.P. and Sargent, J.R. 1998. Microplanktonic polyunsaturated fatty acid markers: a mesocosm trial. Estuarine, Coastal and Shelf Science 46 (Supplement A), 61-67. Pond, D.W., Sargent, J.R., Fallick, A.E., Allen, C., Bell, M.V., and Dixon, D.R. 2000. δ13C values of lipids from phototrophic zone microplankton and bathypelagic shrimps at the Azores sector of the Mid-Atlantic Ridge. Deep-Sea Research I 47, 121-136. Popp, B.N., Laws, E.A., Bidigare R.R., Dore, J.E., Hanson, K.L. and Wakeham, S.G. 1998. Effect of phytoplankton cell geometry on carbon isotopic fractionation Geochimica et Cosmochimica Acta 62, 69-77. Post, D.M. 2002. Using stable isotopes to estimate trophic position: models, methods and assumptions. Ecology 83, 703-718. Rau, G.H., Takahashi, T. and Des Marais, D.J. 1989. Latitudinal variations in plankton δ13C: implications for CO2 and productivity in past oceans. Nature 341, 516-518. Rau, G.H., Chavez, F.P. and Friederich, G.E. 2001. Plankton 13C/12C variations in Monterey Bay, California: evidence of non-diffusive inorganic carbon uptake by phytoplankton in an upwelling environment. Deep-Sea Research I 48, 79-94. Reemtsma, T. and Ittekkot, V. 1992. Determination of factors controlling the fatty acid composition of settling particles in the water column by principal-component analysis and their quantitative assessment by multiple regression. Organic geochemistry 18, 121-129. Rencher, A.C. 1995. Methods of Multivariate Analysis. Wiley, New York, pp. 627. Roessler, P.G. 1990. Environmental control of glycerolipid metabolism in microalgae: commercial implications and future research directions. Journal of phycology 26, 393-399. Roman, M.R. 1991. Pathways of carbon incorporation in marine copepods: Effects of developmental stage and food quantity. Limnology and Oceanography 36, 796-807. Rubenstein, D.R., Chamberlain, C.P., Holmes, R.T., Ayres, M.P., Waldbauer, J.R., Graves, G.R. and Tuross, N.C. 2002. Linking breeding and wintering ranges of a migratory songbird using stable isotopes. Science 295, 1062-1065. Sargent, J.R., Eilertsen, H.C., Falk-Petersen, S. and Taasen, J.P. 1985. Carbon assimilation and lipid production in phytoplankton in northern Norwegian fjords. Marine Biology 85, 109-116. Seber, G.A.F. 1984. Multivariate observations. Wiley, New York.

63 St. John, M.A. and Lund, T. 1996. Lipid biomarkers: linking the utilization of frontal plankton biomass to enhanced condition of juvenile North Sea cod. Marine Ecology Progress Series 131, 75-85. Stryer, L. 1988. Biochemistry. Third edition. W.H. Freeman and company, New York. Taylor, F.J.R. and Haigh, R. 1996. Spatial and temporal distributions of microplankton during the summers of 1992-1993 in Barkley Sound, British Columbia, with emphasis on harmful species. Canadian Journal of Fisheries and Aquatic Sciences 53, 2310-2322. Thompson, P.A., Guo, M., Harrison, P.J. and Whyte, J.N.C. 1992. Effects of variation in temperature. II. On the fatty acid composition of eight species of marine phytoplankton. Journal of phycology 28, 488-497. Tieszen, L.L., Boutton, T.W., Tesdahl, K.G. and Slade, N.A. 1983. Fractionation and turnover of stable carbon isotopes in animal tissues: Implications for δ13C analysis of diet. Oecologia 57, 32-37. Tortell, P.D., Rau, G.H. and Morel, F.M.M. 2000. Inorganic carbon acquisition in coastal Pacific phytoplankton communities. Limnology and Oceanography 45, 1485-1500. Trust Hammer, B., Fogel, M.L. and Hoering, T.C. 1998. Stable carbon isotope ratios of fatty acids in seagrass and redhead ducks. Chemical Geology 152, 29-41. Virtue, P., Mayzaud, P., Albessard, E. and Nichols, P. 2000. Use of fatty acids as dietary indicators in northern krill, Meganyctiphanes norvegica, from northeastern Atlantic, Kattegat, and Mediterranean waters. Canadian Journal of Fisheries and aquatic Sciences 57 (Suppl. 3), 104-114. Volkman, J.K., Smith, D.J., Eglington, G., Forsberg, T.E.V. and Corner, E.D.S. 1981. Sterol and fatty acid composition of four marine haptophycean algae. Journal of the Marine Biological Association of the United Kingdom 61, 509-527. Volkman, J.K., Jeffrey, S.W., Nichols, P.D., Rogers, G.I. and Garland, C.D. 1989. Fatty acid and lipid composition of 10 species of microalgae used in mariculture. Journal of experimental marine biology and ecology 128, 219-240. Whyte, J.N.C. 1988. Fatty acid profiles from direct methanolysis of lipids in tissue of cultured species. Aquaculture 75, 193-203. Whyte, J.N.C., Englar, J.R., Carswell, B.L. and Medic, K.E. 1986. Influence of starvation and subsequent feeding on body composition and energy reserves in the prawn Pandalus platyceros. Canadian Journal of Fisheries and aquatic Sciences 43, 1142-1148.

64

Appendix This appendix describes the conceptual model used in section 2.5.6 to estimate the δ13C of fatty acids after a diet switch. The stable carbon isotope composition of a fatty acid (δ13CFA) after a diet switch depends mostly on the proportions of the fatty acid pool originating from the old- and of the new diet. For essential fatty acids, there are no secondary biosynthetic effects, so in determining the δ13CFA, a linear mass-balance equation will suffice: δ 13CFA (t ) = f (t ) ⋅ δ new + [1 − f (t )] ⋅ δ old

(1)

where f(t) is the fraction (between 0 and 1) of fatty acid derived from the new diet (as a function of time), δnew is the δ13CFA value of the new diet, and δold is the δ13CFA of the previous diet. To calculate f(t) one needs to know the amount of the fatty acid from the new diet (wFAnew) and the total of that particular fatty acid (old and new) (wFA), f (t ) = wFAnew (t ) /[ wFA (t )]

(2)

for which wFAnew and wFA are calculated as follows: wFAnew (t ) = wFAnew (t − 1) + wass + [ f (t − 1) ⋅ wcat (t )]

(3)

wFA (t ) = wFA (t − 1) + wass (t ) − wcat (t )

(4)

where f(t-1) is the fraction of the fatty acid from the new diet one time unit ago, and is zero at the start of the model run. Similarly, wFAnew(t-1) is set at zero at t=0, and the value of wFA(t-1) is given before the beginning of the model run. wass(t) and wcat(t) are the amounts of fatty acid assimilated and catabolized for each unit of time. They are calculated as follows:

65

wass = [( f New vs Old ⋅ FA%old ) / 100] ⋅ rass

(5)

wcat (t ) = [ wFA (t − 1) / wTot (t )] ⋅ rcat

(6)

where fNew vs Old is the factor of increase in fatty acid abundance in the new diet (will be 2 when the abundance is 2 times higher in new diet, and will be 0.5 when 2 times lower in new diet). FA%old is the abundance of the fatty acid in old diet (in % of total fatty acids), rass and rcat are the rates of assimilation and catabolization of total fatty acids, respectively (e.g. µg/day). wFA(t) was given in equation 4, and wTot is the total amount of fatty acids, obtained with: wTot (t ) = wTot (t − 1) + rass − rcat

(7)

in which a value for wTot(t-1) needs to be given at t=0. Then, abundance of the fatty acid over time, FA%(t), (as also plotted in Figure 2.11) can be calculated with:

FA%(t ) = [ wFA (t ) / wTot (t )] ⋅ 100

(8)

Starting values given to the model were estimates made for a single euphausid furcilia. The following values were used in the model: •

The total weight of fatty acids (wTot) is set at 100µg. This number is chosen so that it equals 10% of a larvae with a dry weight of 1mg.

•

For the abundance of the fatty acid before diet change (FA%old) 0.3% is given for 18:3n-3, and 31.3% is put in as a starting abundance for 20:5n-3. These numbers are based on the fatty acid composition of euphausiid furciliae collected at LC4 during the same cruise (Figure 2.10a).

66 •

The abundance of the fatty acid in the new diet, expressed as fraction of the old diet abundance (fnew vs old) is 5.2, and 0.5 for 18:3n-3 and 20:5n-3 respectively. The abundance of 18:3n-3 in POM from LC9 was found to be 5.2 times higher than in POM collected at LC4. Compared to LC4 POM, 20:5n-3 was half as abundant as in POM from LC9 (see Figure 2.10b). It should be noted that for simplicity no feeding selectivity is assumed.

•

The amount of total fatty acids assimilated (rass) per day was set at 9µg, which amounts to 3µg being incorporated when two thirds of the assimilated fatty acids are catabolized (see below). This value, resulting in a daily incorporation rate of 3% of the body lipid, is in the range of values found for the marine copepod Acartia tonsa (stages C3-C4) in a study by Roman (1991). He found that 0.6% and 4.7% of body lipid was incorporated daily by the copepods feeding the 14Clabelled diatom Thalassiosira weissflogii at concentrations of 29 and 146µg C L1

, respectively. However, the daily incorporated 3% body lipid is higher than the

below 1% values calculated from field estimates by Huntley and Brinton (1991) for Euphausia superba larvae. Therefore, the 3% of the initial weight used here may be on the high side, and result in a lower, more conservative estimate of the tag longevity. •

The amount of total fatty acids catabolized (rcat) per day was estimated at two thirds (~67%) of the amount of assimilated fatty acids. This value is in agreement with Huntley and Brinton’s (1991) numbers for Euphausia superba furcilae (stages F2-F4), and somewhat higher than values obtained by Pond et al. (1995). The immature Euphausia superba Dana in the Pond et al. (1995) study, catabolized 51% of the assimilated lipid of the 14C-labelled diatom Thalassiosira, and 61% when fed Isochrysis.

•

The δ13C values chosen for 18:3n-3 and 20:5n-3 from the old diet (δold) are – 26.5‰, and –23.5‰, respectively. These values are chosen as to correspond with the δ13C values of the fatty acids in LC4 euphausiids (Figure 2.9), which represent a time-integrated δ13C value of the food organisms at LC4. Taking

67 these values also obviates the need for any correction for isotope fractionation between animal and diet. •

Similarly, for the δ13C values of the new diet (δnew), δ13C values of the fatty acids in LC9 zooplankton were chosen (see Figure 2.9). Thus, for 18:3n-3 and for 20:5n-3 the values -31.5‰ and -30‰ were given, respectively.

Note that growth of the pool of fatty acids will occur in the model when the quantity of total fatty acids assimilated exceeds the amount being catabolized. Since the amounts of total fatty acids catabolized (rcat) and assimilated per day (rass) are held constant, growth will be linear. For practical reasons it was assumed that the loss due to catabolism (wcat) is simply proportional to the relative abundance of a fatty acid (see equation 6). Therefore, wcat goes up or down, depending on whether the abundance of the investigated fatty acid in the present diet is higher or lower than before. Another assumption made here is that the amount of food (fatty acid) intake stays the same as before the diet switch. This assumption is made in equation 5, but can be easily adjusted by multiplying fNew vs Old with a factor that represents the proportion of the new intake with respect to the previous food intake (i.e. when twice as much food is ingested in the new environment this factor will be 2).

68

3. Regional and Temporal Patterns in the Fatty Acid- and Stable Carbon Isotope Composition of Seston off the Coast of Vancouver Island, Canada.

3.1 Abstract In the previous chapter it was shown that the shelf and off-shelf food webs could typically be distinguished from each other with the use of bulk stable carbon isotope ratios,

13

C/12C ratios of individual fatty acids, and fatty acid composition data. The

quality of the food is the driving factor for the differences observed in higher trophic level organisms. Therefore, particulate organic matter (POM) was collected off the west coast of Vancouver Island from water at 5m depth on cruises during May 1998, May 1999 and July 1999 (covering an El Niño – La Niña transition). Bulk POM δ13C measurements and fatty acid abundance data were compared with each other, and with temperature, salinity, chlorophyll a, macro-nutrient, and phytoplankton taxonomy data. Additionally, 13C/12C ratio measurements of individual fatty acids were measured. With principal component analysis of the fatty acid abundance data it was found that POM could be characterized by the relative contributions of diatom-, dinoflagellate-, nanoflagellate-, or detritus/bacterial-derived matter. The 16:2n-4 fatty acid abundance appeared to be the best predictor for the relative amount of diatom-derived matter. The differences observed between the fatty acid composition of the shelf and off shelf food webs seem to be mostly driven by the different contribution of diatom-derived material in POM. In May 1998 the difference between the δ13C of POM from stations close to the coast and an open ocean station was as much as 9‰. Such a steep gradient in δ13C had also been observed by Fry and Bates off the Washington coast (unpublished data presented here). The measured variable that correlates best with the δ13C of POM is the abundance of the 16:2n-4 fatty acid. With the help of

13

C/12C ratio measurements of

69 individual fatty acids it was concluded here that the higher δ13C values of more diatomrich POM is not explained by a higher contribution of

13

C-enriched diatom-derived

carbon. Instead, this trend must have been caused by the fact that conditions favourable for diatom dominance are leading to 13C-enrichment in all phytoplankton. It is discussed whether the difference in fatty acid composition and δ13C of animals feeding close to shore and individuals feeding further off shore can be expected to be a common phenomenon.

3.2 Introduction In Chapter 2 it was shown that with the molecular, and stable carbon isotope composition of fatty acids, or the

13

C/12C ratio of bulk samples, on- and off-shelf food webs could

typically be distinguished. The differences in composition of the food webs from the two environments are the result of differences in the quality of the seston. It is therefore crucial to map the spatial pattern of the stable carbon isotope, and fatty acid composition of the particulate organic matter (POM). Additionally, to apply fatty acid and δ13C data as a food web tracer elsewhere, or in the future, it is important to understand the processes that cause the spatial variation in POM composition. Such an understanding will also aid in interpreting the δ13C record of organic matter in sedimentary samples, and thus gain insight into past conditions. Few studies have investigated the spatial variation of the fatty acid composition of POM in the marine environments. The characteristics of the fatty acid profiles of many algal classes that contribute to marine seston are well known (Sargent et al., 1988). The spatial variation of the fatty acid composition of POM is mostly a function of the relative abundance of the various algal classes, terrestrial sources, bacteria and dead organic matter (Reemtsma and Ittekot, 1992; Canuel, 2001). To know how the fatty acid composition varies within the studied region is not only of interest for food web tracer studies, but it can also be of interest for studies on the nutritional quality of seston available to, for example, zooplankton.

70 A large number of studies have reported δ13C measurements of POM from the open ocean (see, e.g., the compilations by Goericke and Fry, 1994; Rau et al., 1997) and from estuarine settings (reviewed by Fry and Sherr, 1984; see also Canuel et al., 1995, 1997). However, not many researchers investigated the spatial variability in

13

C/12C ratios of

POM from the transition between the continental shelf and open ocean settings. In most open ocean studies a correlation between the carbon isotope fractionation in phytoplankton and CO2(aq) concentrations was found (e.g., Rau et al., 1989, 1997; Francois et al., 1993; Bentaleb et al., 1998). In most coastal settings, on the other hand, the CO2(aq) concentration did not appear to be a key factor in determining the δ13C of phytoplankton and POM (Pancost et al., 1997, 1999; Tortell et al., 2000; Rau et al., 2001). Other factors such as growth rate, cell size (Kopczyńska et al., 1995; Pancost et al., 1997) and active inorganic carbon uptake mechanisms (Tortell et al., 2000; Pancost et al., 1997, 1999; Rau et al., 2001) were found to play a dominant role. For this study POM was collected off the west coast of Vancouver Island from water at 5m depth on cruises during May 1998, May 1999 and July 1999 (see Figure 3.1). The orientation of the currents off the coast is on average alongshore, but their direction changes seasonally. During the summer, persistent northwesterly (equatorward) coastal winds cause offshore advection in the surface Ekman layer, thus favouring upwelling. The upwelling off Vancouver Island during the summer results in a surface intensified southeastward directed flow over the outer portion of the shelf (Thomson and Ware, 1996). This is in the opposite direction of the Vancouver Island coastal current, which is poleward flowing year-round over the innermost portion of the shelf. Especially during spring and summer the phytoplankton biomass and nutrient availability is higher in surface waters above the continental margin than in the adjoining open ocean (Mackas, 1992). The two years, 1998 and 1999, which were sampled for this study had quite different physical oceanographical conditions. In May 1998 the west coast of Vancouver Island was still experiencing El Niño conditions, and in 1999 La Niña conditions existed. This was reflected by anomalously high sea surface temperatures (SST) in 1998 (ca. 1 °C above the 1990-96 mean), and low SSTs in the spring and summer of 1999 (ca. 1 or 2°C lower than the 1990-96 mean) (DFO science, 2000).

71 In the present chapter it is investigated whether the difference between the δ13C of bulk POM from the shelf and off shelf environments, as already observed by Perry et al. (1999), persisted under the different conditions encountered on each of the three cruises. Spatial patterns of the fatty acid, and bulk stable carbon isotope data are compared with temperature, salinity, chlorophyll a, and nutrient concentrations, and algal taxonomy data. The various factors influencing the

13

C/12C ratios in POM are discussed in light of the

patterns observed.

3.3 Methods POM was collected on multiple cross shelf transects off the west coast of Vancouver Island during three surveys in 1998 and 1999 (Figure 3.1). All sampling was done onboard the C.S.S. John P. Tully during the cruises: IOS9810 (May 12-24, 1998), IOS9911 (May 04-12, 1999) and IOS9928 (June 30 – July 09, 1999). Details of the sampling procedure were described in Chapter 2. For the procedures followed for the fatty acid profile analysis and stable carbon isotope ratio measurements please also refer to Chapter 2.

3.3.1 Physical parameters Vertical profiles of temperature and salinity were obtained using a Seabird® model SBE 911+ CTD serial# 0443 (May 1998), and serial# 0437 (May, July 1999). The conductivity, temperature and depth (CTD) data from the Institute of Ocean Sciences (IOS, Sidney, B.C., Canada) were provided by Douglas R. Yelland.

The mixed layer depth (MLD) is considered a quasi-homogeneous region in the upper ocean with little variation in temperature and density with depth (Kara et al., 2000). Here the mixed layer depth was chosen as the bottom of the vertically uniform layer. The definition used here for a “quasi-homogeneous” layer was a surface water body with temperatures that deviate less than 0.8ºC from a surface reference value. This

72 definition was adopted from the study of Kara et al. (2000). The average MLDs calculated here varied little from those determined by Harris (2001), who used the method using deviations in σt as described by Levitus (1982). Here, the average MLD for shelf and off shelf waters in May 1998 was determined at 14 and 24m, respectively. For the same regions, at the same time, Harris (2001) reported 13 and 24m.

Figure 3.1. Map of study area. At all stations POM was collected, and at stations indicated by the closed circles zooplankton and larval fish was also collected (no larval fish on CS-line). Open circles with vertical line represent stations sampled in May ’98, horizontal: May ’99, and diagonal lines: July ’99. P10 was only sampled in May ’98, P8 only in May ’99, and samples were only collected at ER2 during July ’99. The 200m-isobath is defined as the shelf-break and is indicated with a thick black line.

73

3.3.2 Nutrients and chlorophyll a Concentrations of nitrate plus nitrite (NO3- + NO2-), soluble reactive phosphate (HPO42-), silicic acid (Si(OH)4), and chlorophyll a were measured and provided by Shannon Harris (University of British Columbia). Materials and methods used for these measurements have been reported by Harris (2001). The combined nitrate and nitrite concentrations are reported here as nitrate.

3.3.3 Phytoplankton identification Water samples for phytoplankton identification were taken at 5m depth with 10L PVC Niskin bottles mounted on an instrumented rosette sampler. The samples were kept in 250ml glass bottles and fixed with neutral Lugol’s iodine solution and stored in the dark until analysis. Samples taken in 1999 were collected in 500ml glass bottles and fixed with acidic Lugol’s iodine solution. At some stations a second sample was fixed with a formalin solution. Depending on cell density 10, 25 or 50ml was settled in a counting chamber for at least 12 hours. Cells were counted with the use of an inverted microscope, following procedures outlined by Utermöhl (1958). The counts were converted into cell per litre. The identification and enumeration of the May 1998 samples was performed by Rowan Haigh at the University of British Columbia. The analysis of 1999 samples was done by Jennifer Putland at the Institute of Ocean Science (IOS), Sidney, Canada. The cells were grouped into the major taxa and a few, mostly diatoms, were identified to the species level.

3.3.4 δ13C of dissolved inorganic carbon For the dissolved inorganic carbon (DIC) analyses, samples of water were collected from 5m depth in Niskin bottles. The water was transferred on-deck to a syringe equipped with a syringe filter (0.2 µm pore size). The water was pushed through the filter into a 125ml glass Wheaton bottle till completely filled. A butyl-rubber cap, with a syringe needle

74 penetrated through it, was pressed onto the Wheaton bottle. The syringe needle allowed excess water to squirt out while the cap was being put on, leaving no headspace in the bottle. After the needle was carefully removed the cap was secured with an aluminium crimp-seal. Samples were then stored in the dark at 5ºC till further processing. The DIC was extracted at the Institute of Ocean Science (Sidney, Canada) with a NIWA DIC Extraction System. The seawater was removed from the Wheaton bottles under an N2 atmosphere and mixed with concentrated phosphoric acid and then placed on the extraction line. The released CO2 was purified by high vacuum and cryogenic removal and subsequently sealed in breakseals. The stable carbon isotope composition of the CO2 was analyzed at the Biogeochemistry Facility, University of Victoria, using the dual inlet on a Finnigan MAT 252 mass spectrometer. The IOS standard-1 (δ13C=-1.06‰ vs PDB: November, 2001) was used as a reference. Each sample was measured against the reference in 10 blocks. The standard deviations of these δ13C measurements varied between 0.006 and 0.019‰. However, no water sample replicates were measured. Therefore, a precision estimate for the whole procedure, including the DIC extraction is not available.

3.3.5 Principal component analysis Principal component analysis (PCA) was used on the fatty acid abundance data as an exploratory and dimension-reducing technique. PCA finds new axes (the principal components), which are linear combinations of the original variables. The first principal component accounts for the largest amount of variance present in the data set, the second for the largest amount of residual variance unexplained by the first, and so on. The transformation of the coordinate system is in fact an orthogonal rotation in p-space (with p, the number of variables). The samples can be plotted in the new coordinate system, which is often referred to as a score-plot. Generally, already most of the structure in the data set is revealed by plotting the samples in the sub-space of the first two principal components, because of the high percentage of the total variance captured by component

75 1 and 2. Hence, a scatter plot of the first few principal components is often useful to check for groupings of the samples and to discover outliers. The large range in variances of the abundances of the different fatty acids in the data set used here would let the components be dominated by fatty acids with a high variance. Fatty acids with a low abundance and variance, but with possibly high discriminating power between groups, will contribute little if no form of standardization is used. Therefore, it was chosen here to standardize the data set by using the correlation matrix instead of the variance-covariance matrix in the PCA (see also Meglen, 1992). The weights assigned to the original variables for calculation of the component scores are generally called the “loadings” of the variables. Since the correlation matrix was used, the loadings are a measure of the correlation between the variables and the principal component. Matrix computations for PCA and calculations for the mixing model (see discussion) were performed using the software package Matlab (version 4.2b, The Mathworks Inc., Natick, MA, U.S.A.).

3.4 Results 3.4.1 Phytoplankton taxonomy In May 1998 diatoms, mostly Chaetoceros spp. and Skeletonema costatum, dominated the phytoplankton community at the shelf stations and LBP7 (see Figure 3.2). Water samples taken in May and July 1999 from both the shelf and the off shelf stations were dominated by nanoflagellates. However, fatty acid data indicates that during May 1999, stations not sampled for identification, in the north on the CS- and BP-lines, and more inshore on the other lines (C1, C2, J2, L2) were in fact rich in diatoms (data will be shown in section 3.4.3). Approximately 24% of the cells at Station LBP2 (May 1999) were identified as diatoms, with Leptocylindricus danicus and Dactyliosolen (or Rhizosolenia) fragilissimus as dominant species. Dinoflagellates never dominated the assemblage in terms of cell numbers, but they occurred in higher numbers at all stations during the July 1999 cruise compared to other

76

LBP2

July 1999

LBP7

LBP2 LBP7 LG3

LG3 LG7

LG7 LC9

ER3

LC4

LC9 LC12

LC12

LC4

ER03

LBP2

May 1999

LBP7

LBP2 LBP7 LG3

LG3

LG7 P8

LG7

P8

LC9

LC12

May 1998

LC4 LC9 LC12

LC4

LBP2

LBP7

Nanoflagellates

LBP2 LBP7

Diatoms Dinoflagellates Ciliates Other

LG3

LG3 LC4

P10

Total # of cells/L (million)

LC4

P10 LC9

LB16

LC9 LB16

C1 C1

8 6 4 2 0

Figure 3.2. Pie charts showing the abundance of different phytoplankton groups per station. The height of the pie indicates the total number of cells per litre, and the portions reflect the proportion of the total number cells for each group (see legend). Sample locations are indicated with stars.

77 cruises. Samples taken from shelf water during that time had consistently higher numbers of large dinoflagellates (>20µm) than off shelf samples. Ciliates and choanoflagellates were also counted, but their numbers were generally low. In May 1999 the relative abundance of choanoflagellates was somewhat higher than during the other cruises (0.7 – 5.7% versus 0 –1.4% of total number of cells). At P10 in May 1998 the phytoplankton assemblage was dominated by unidentified, perfectly spherical, picoplankton cells. Additionally, an abundance of about 100,000 cells per litre of Phaeocystis sp. (mostly non-motile forms) was found at P10. Phytoplankton at the Endeavour segment of the Juan de Fuca Ridge (station ER2) was dominated by cells classified as “flagellates and coccoid cells <20µm”, followed by cryptomonads and small (<20µm) dinoflagellates (78, 13 and 6% of total number of cells, respectively).

3.4.2 Linking fatty acid abundance to taxonomy Because diatoms are often much larger than nanoflagellates, the relative abundance in terms of numbers (Figure 3.2) does not give a good representation of the biomass. Unfortunately, especially the counts of the phytoplankton of 1999 were not detailed enough to make estimates of the amount of carbon per litre for the different phytoplankton classes. As a numerical expression of flagellate over diatom dominance the logarithm of the nanoflagellate/diatom cell number ratio was computed. Subsequently the fatty acid data were compared to this parameter. The fatty acids of which the abundances correlate best with log(nano/diatom) are all fatty acids that are known to be common in diatoms (Table 3.1). Of all fatty acids the 16:2n-4 fatty acid correlates best (i.e., negatively) with the log(nano/diatom) parameter (R2=0.61, n=22; Figure 3.3), followed by 16:1n-7, 16:4n-1 and 20:5n-3 (R2s of 0.42, 0.39 and 0.38, respectively; Table 3.1). The 16:1n-7/16:0 ratio, often used as diatom marker (St. John and Lund, 1996), only shows a correlation with an R2 of 0.33 (Figure 3.3). Out of all possible ratios, the fatty acid ratio exhibiting the best correlation with the log(nano/diatom) parameter is 16:2n-4/18:2n-6 (R2=0.66, n=22; when using log(16:2n-

78 4/18:2n-6): R2=0.64, n=21; Figure 3.3). Log(18:1n-9/20:5n-3) also correlates well with log(nano/diatom) (R2=0.56, n=22; Figure 3.3).

Table 3.1. Correlation coefficients (R) and squared correlation coefficients (R2) for correlations between the abundance of fatty acids in 22 POM samples and the logarithm of the ratio of concentrations of nanoflagellate, and diatom cells [log(nano/diatom)] in water samples from the same station, and from the same depth (5m). Only the fatty acids that showed a significant correlation (P<0.05; F-test; df=1, 20) with log(nano/diatom) are shown. Fatty acids printed in bold letters showed correlations with log(nano/diatom) for which the probability of R=0 is smaller than 0.01 (P<0.01; F-test; df=1, 20).

Fatty acid 16:2n-4 16:1n-7 16:4n-1 20:5n-3 16:3n-4 16:1n-5 18:1n-9 22:5n-3 20:0 iso 15:0

Correlation coefficients of correlation with log(nano/diatom) R R2 -0.78 0.61 -0.65 0.42 -0.63 0.39 -0.61 0.38 -0.54 0.29 0.53 0.28 0.49 0.24 -0.48 0.23 -0.46 0.21 0.46 0.21

3.4.3 PCA of fatty acid abundance data Principal component analysis of the POM fatty acid abundance data shows several apparent features (Figure 3.4). In each of the quadrants fatty acids that are known to be associated with each other form a group together (Figure 3.4a). The C16-unsaturates and the 20:5n-3, known to occur at higher levels in diatoms (Chuecas and Riley, 1969; Wood, 1974; Pohl and Zuhrheide, 1979; Volkman et al., 1989; Dunstan et al., 1994), plot in the lower right quadrant. All saturated fatty acids, including the iso 15:0, together with the 18:1n-7 fatty acid can be found in the lower left quadrant. The odd- and branched (iso-) fatty acids are often reported to occur in higher relative amounts in bacteria (Leo and Parker, 1966; Johns et al., 1977; reviewed by Sargent et al., 1988). Substantial amounts of 18:1n-7 have also

79 been found in bacteria (Perry et al., 1979; Volkman et al., 1980; Teece et al., 1999). Dead organic matter gets enriched with saturated fatty acids due to higher rates of photooxidation reactions for unsaturated fatty acids (Kieber et al., 1997), but also due to preferential use of unsaturated fatty acids by zooplankton (Prahl et al., 1984; Harvey et al., 1987) and microbial populations (Rhead et al., 1971).

3.5

2.5 2.0 1.5 1.0 0.5 0.0 -0.5 0.0

2

R = 0.33 n=22

3.0

log (nanos / diatoms)

3.0

Log (nanos / diatoms)

3.5

2

R = 0.61 n=22

2.5 2.0 1.5 1.0 0.5 0.0 -0.5

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0.2

0.4

0.6

16:2n-4 abundance (% of tot. fatty acids)

Log (nanos / diatoms)

3.0

3.5

2

R = 0.64 n=21

3.0

Log (nanos / diatoms)

3.5

2.5 2.0 1.5 1.0 0.5 0.0 -0.5 0.0

0.8

1.0

1.2

16:1n-7/16:0 2

R = 0.56 n=22

2.5 2.0 1.5 1.0 0.5 0.0 -0.5

0.2

0.4

0.6

Log (18:2n-6 / 16:2n-4)

0.8

1.0

-0.6 -0.4 -0.2

0.0

0.2

0.4

0.6

0.8

1.0

Log (18:1n-9 / 20:5n-3)

Figure 3.3. The abundance of 16:2n-4 (top-left), and several fatty acid ratios (or the logarithm of the ratio) measured for POM samples plot against the logarithm of the ratio of concentrations of nanoflagellate, and diatom cells [log(nano/diatom)] determined for water samples from the same stations, and from the same depth (5m). The squared correlation coefficient (R2) and the number of samples used are shown in each plot.

The top left quadrant exhibits C18-fatty acids typically found at higher concentrations in nanoflagellates such as the Cryptophyceae and Prymnesiophyceae (Chuecas and Riley, 1969; Pohl and Zuhrheide, 1979; Viso and Marty, 1993).

80

ell a

22:6n3

Fla g

18:3n3 18:2n6

20:1

22:5n3

20:4n3 18:1n9

0.0 -0.1

16:4n3

16:1n5

20:2n6

16:0

22:1

s/b

17:0 15:0

-0.3

16:1n9

16:1n7

18:1n5

-0.2

16:2n4

iso 17:0

20:0

14:0

al eri

-0.3

20:5n3 16:4n1 16:3n4

21:5n3

18:0 18:1n7 iso 15:0

ac t

-0.2

20:4n6 22:5n6

-0.1

0.0

0.1

0.2

Di at o ms

0.1

tritu De

PC-2 (15.3% of variance)

20:3n3

s te lla ge

0.3 0.2

18:4n3

tes

a

fla no Di

0.4

0.3

0.4

PC-1 (27.7% of variance) 10

b

May 1998 off shelf May 1999 off shelf July 1999 off shelf May 1998 shelf May 1999 shelf July 1999 shelf

C2

8

PC-2 (15.3% of variance)

B7 B3

6 4 2 0 -2 -4 -6

ER2

G5 J5 BP9 BP2 E1 G3 G7 J4 L3 C4 P8 C12 BP7 J6 K5 CS1 L2 CS3 C9 L8 C7 J2 BP6 P6 BP4 C10C9 C8 J1 C3 J2 C11 P10 C1 E9 G9 BP4 C9 C4 C1 BP3 B11 G7 C6 BP7BP2 C12 CS7 C2 BP4 C12 BP2 CS1 G7 G3 G1 G5 CS1 B16 C4 G3 BP3 CS5 J9

BP7

G7

-8

-6

-4

-2

0

2

4

6

8

10

PC-1 (27.7% of variance) Figure 3.4. Loading plot (top) and score plot (bottom) of principal component analysis of POM sampled during all three cruises. The fatty acids plotting within the envelopes shown in the loading plot (top) are fatty acids typically found in higher amounts in the type of seston indicated in each quadrant. The labels in the score plot indicate from which station the samples originate (see Figure 3.1 for locations).

81 Finally, docosahexaenoic acid (22:6n-3), which is known to be more abundant in dinoflagellates (Chuecas and Riley, 1969; Pohl and Zuhrheide, 1979; Mansour et al., 1999), plots in the top right corner. The distribution of the samples in the same space of PC-1 and PC-2 is shown in Figure 3.4b. The majority of the samples collected in shelf waters plot on the right side (i.e., the “diatom- and dinoflagellates-quadrants”), and the majority of the off shelf samples plot on the left (i.e., in the “flagellate- and detritus/bacterial-quadrant”). However, there are quite a lot of exceptions. It should be noted, though, that all off shelf samples that do plot in the “diatom-quadrant” are from stations from the north (BP- and CS-line). Another feature to note is that it is mostly July 1999 samples that fill the “dinoflagellates-quadrant”, and hardly any samples collected in July occupy the “diatomquadrant”.

3.4.4 Cross-shelf trends in δ13C, 16:2n-4 abundance and chl a Figure 3.5 shows the δ13C of POM collected from a depth of 5m and the abundance of the 16:2n-4 fatty acid in the same samples plotted against the distance from shore. There is a general trend to lower 13C/12C ratios away from shore (Figure 3.5). On all lines high δ13C values (-17 to -21‰) occur within 10-30km from shore. Then values often decline more rapidly when moving further away from land, which may be followed by a stabilization of the values even further from shore. This pattern is clearest for line C, which extends furthest offshore. As shown in section 3.4.2, the 16:2n-4 fatty acid apprears to be a good indicator for the relative amount of diatoms. Though less obvious, a similar trend can be observed for the 16:2n-4 fatty acid abundance data. Both the 13C/12C ratios and the relative amount of 16:2n-4 were generally higher in the POM collected in May 1999, with the exception of the mid-section of line G. In May 1998 the difference between the δ13C of POM from stations close to the coast and the open ocean station P10 was as much as 9‰.

82

POM δ13C

2.0

-18

1.5

-20

1.0

-22

0.5

-24

0.0

Line BP 100

80

60

40

20

-26

0

2.0

-18

1.5

-20

δ 13C (‰)

16:2n-4 abundance (% of total)

16:2n-4 fatty acid

1.0 0.5 0.0

NW Shelf-break Line BP 100

80

60

40

100

80

60

40

20

0

-22 -24

Line G 100

80

60

40

20

0

-26

2.0

-18

1.5

-20

1.0

-22

0.5

-24

0.0

May 1998 May 1999 July 1999

Line C 100

80

60

40

20

0

-26

Line G 20

0

SE Line C 100

80

60

40

20

0

Distance offshore (km) Figure 3.5. Trends in the 16:2n-4 fatty acid abundance (left) and δ13C of POM (right) with distance offshore. Samples on lines C (bottom), G (middle), and BP (top) for all three cruises are plotted (for location of lines see Figure 3.1). The vertical line in the plots indicates the position of the shelf-break (200m iso-bath). The plots are placed in geographical order, with the BP-line the most northwestern line.

As could already be observed in Figure 3.5 the δ13C of POM correlates with the 16:2n-4 fatty acid abundance (Figure 3.6a and Figure 3.7). The positive correlation (R2=0.69, n=68) shows that when there is a higher relative contribution of the 16:2n-4 fatty acid to the seston it is more enriched in 13C. The same was true for the 22:6n-3 fatty acid, except in shelf samples collected in May 1998 (see Figure 3.7). Not only the δ13C of

83 bulk POM, but also the δ13C of individual fatty acids such as 18:4n-3 and 20:5n-3 correlate well with the 16:2n-4 abundance (Figure 3.8).

-18

-16

a δ13C of bulk POM (‰)

δ13C of bulk POM (‰)

-16

-20 -22 -24 -26 -28 0.0

0.4

0.8

1.2

1.6

2.0

2.4

16:2n-4 abundance (% of total fatty acids)

b

-18 -20 -22 -24

Shelf Off shelf

-26 -28

0

2

4

6

8 3

Chl a concentration (mg/m )

Figure 3.6. The δ13C of POM (bulk sample) plotted against the abundance of the 16:2n4 fatty acid (a) and the chlorophyll a concentration in surface water (b). Samples taken from water on the shelf, and off the shelf are plotted with different symbols (see legend).

A correlation also exists between the chlorophyll a (chl a) concentration found in the surface water and the δ13C of the POM at 5m water depth (Figure 3.6b). In contrast to the 16:2n-4 abundance, chl a concentrations do not correlate in a linear fashion to the δ13C of POM. At chl a concentrations higher than about 4 mg/m3 a further increase does not correspond with rising 13C/12C ratios (Fig. 3.6b) and δ13C values remain between –17 and -19‰. The spatial variation of the stable carbon isotope ratios of POM and surface water chl a concentrations is shown in Figure 3.9. The overall pattern shows high chl a concentrations and δ13C values in shelf waters, and lower values for both parameters off the shelf. There are some notable exceptions, however. During both May cruises shelf stations on line G displayed low levels chl a. Especially in 1998 these low concentrations of chl a did not coincide with low

13

C/12C ratios. On the other hand, the Cape Scott

region, sampled more extensively in May 1999, showed both high amounts of chl a and high δ13C values. Finally, in July 1999 POM samples from the inner part of line B and a sample taken in the Juan de Fuca Strait have low 13C/12C ratios, whereas chl a levels were high.

84

-16

-16

-18

-18

-20

-20

-22

-22

-24

-24

-26 -28

Bulk POM δ13C (‰)

July 1999

high 18:4n-3

-26 -28

-18

-18

-20

-20

-22

-22

-24

-24

-26

May 1999

-28

July 1999

high 18:4n-3

-26 -28

-18

Off shelf

-18

-20

Shelf

-20

-22

-22

-24

-24

-26 -28

high 22:6n-3

May 1998 0

2

4

6

8

10

12

-26

May 1998

-28 0.0

14

22:6n-3 abundance (% of tot. fatty acids)

0.4

0.8

1.2

1.6

2.0

2.4

16:2n-4 abundance (% of tot. fatty acids)

Figure 3.7. Bulk POM δ13C plotted against the abundance of the 22:6n-3 and 16:2n-4 fatty acids.

-24

-20

a

δ13C of 20:5n-3 fatty acid (‰)

13

δ C of 18:4n-3 fatty acid (‰)

-22

-26 -28 -30 -32 -34 -36 0.0

0.5

1.0

1.5

2.0

2.5

16:2n-4 abundance (% of total fatty acids)

-22

b

-24 -26 -28 -30 -32 -34 0.0

0.5

1.0

1.5

2.0

2.5

16:2n-4 abundance (% of total fatty acids)

Figure 3.8. The 16:2n-4 fatty acid abundance plotted against the δ13C measured for the 18:4n-3 and 20:5n-3 fatty acids.

85

δ C of POM 13

Chlorophyll a Chl a concentr. at surface (mg/m3)

δ13C of POM from 5m depth (‰)

8.0 4.3 2.3 1.3 0.7 0.4 0.2

ne Li

log scale

BP

ne Li e Lin

July 1999

ne Li

ER 2

C

0.4 0.2

log scale

ne Li

ne Li

CS

BP

P08

ne Li ne Li

May 1999

ne Li

C

Chl a concentr. 3 at surface (mg/m )

G ne Li

May 1999

C

δ C of POM from 5m depth (‰) 13

8.0 4.3 2.3 1.3 0.7 0.4 0.2

ne Li

CS

log scale

ne Li

BP ne Li

May 1998

ne Li

J

ne Li

P10

G ne Li

-18.0 -19.0 -20.0 -21.0 -22.0 -23.0 -24.0 -25.0 -26.0

BP

P08

G ne Li

C

δ13C of POM from 5m depth (‰)

8.0 4.3 2.3 1.3 0.7

CS

G ne Li

July 1999

Chl a concentr. at surface (mg/m3)

ne Li

BP

G ne Li

-18.0 -19.0 -20.0 -21.0 -22.0 -23.0 -24.0 -25.0 -26.0

-18.0 -19.0 -20.0 -21.0 -22.0 -23.0 -24.0 -25.0 -26.0

BP ne Li

C

May 1998

J

ne Li

G ne Li

C

Figure 3.9. Maps of the chlorophyll a concentrations (left) and the δ13C of POM (right). The shelf break is indicated with a black line on the maps. Small numbers in the maps indicate water depths.

86 In July 1999 a POM sample was taken (at 5m depth) from water above Endeavour Segment (underway from 47°57.05 N, 129°05.72 W to 47°52.55 N, 129°08.29 W). Three measurements in this sample of the bulk POM δ13C had an average value of –33.2‰ (SD=0.09‰). The δ13C of individual fatty acids ranged from about –35 to -45‰. Although the validity of the numbers is not questioned here, this sample is left out from most plots due to its anomalous nature.

3.4.5 Nutrients The spatial pattern of nutrient concentrations (Figure 3.10) does not show much resemblance to the patterns observed for chlorophyll a and the δ13C values for POM (Figure 3.9). Nitrate levels were very low to undetectable in surface waters at most stations in May 1998. Such low nitrate concentrations were only sporadically encountered during the other two cruises (Figure 3.10). In May 1998 silicate concentrations may have been limiting phytoplankton growth at a few stations such as LC4 and the inland stations of line G. However, during May and July of 1999 silicatereplete conditions seemed to be prevalent in the entire study area. Phosphate (not shown here) was, unlike nitrate, hardly depleted down to undetectable levels in May 1998. During the other two cruises phosphate concentrations showed a distribution comparable to that shown here for nitrate.

3.4.6 Physical parameters Temperature and salinity measurements at 5m depth are shown in Figure 3.11. Most striking is the large temperature difference between May 1998 and 1999. Although the May 1999 cruise was only a week earlier the water was 2 to 3 °C colder than in May 1998, clearly illustrating the El Niño and La Niña sequence. In July 1999 surface waters were warmer (also warmer than May 1998), but colder water was still present in the southeast, close to the Juan de Fuca Strait (see Figure 3.11). No relationship between temperature and δ13C (Figure 3.12) or chl a was observed.

87

Nitrate

Silica Si(OH)4 concentr. at <6m depth (µM)

-

NO3 concentr. at <6m depth (µM)

15.0 4.3

3.2

1.2

1.0

0.3

0.3

0.1

0.1

log scale

log scale

July 1999

July 1999 Si(OH)4 concentr. at <6m depth (µM)

-

NO3 concentr. at <6m depth (µM)

15.0 4.3

10.0 3.2

1.2

1.0

0.3

0.3

0.1

0.1

log scale

log scale

May 1999

May 1999 Si(OH)4 concentr. at <6m depth (µM)

May 1998

10.0

-

NO3 concentr. at <6m depth (µM)

15.0 4.3

10.0 3.2

1.2

1.0

0.3

0.3

0.1

0.1

log scale

log scale

May 1998

Figure 3.10. Maps of the silicate- (Si(OH)4) (left), and nitrate (NO3- + NO2-) (right) concentrations. A logarithmic scale was used for the contours in order to distinguish the lowest concentrations better (see legend).

88

Salinity

Temperature Temperature at 5m depth (°C)

Salinity (psu) at 5m depth

13.5 12.9

32.25

12.3

32.00

11.7

31.75

11.1

31.50

10.5

31.25

9.9

31.00

9.3

30.75

8.7

30.50

July 1999

July 1999 Temperature at 5m depth (°C)

13.5

32.50

12.9

32.25

12.3

32.00

11.7

31.75

11.1

31.50

10.5

31.25

9.9

31.00 30.75

9.3

30.50

8.7

May 1999

May 1999 Temperature at 5m depth (°C)

Salinity (psu) at 5m depth

13.5 12.9

May 1998

32.50

32.50 32.25

12.3

32.00

11.7

31.75

11.1

31.50

10.5

31.25

9.9

31.00

9.3

30.75

8.7

30.50

May 1998

Figure 3.11. Maps of the temperature (left), and salinity (right) measured at 5m depth.

-16

-16

-18

-18

Bulk δ C of POM (‰)

-20 -22 -24

13

13

Bulk δ C of POM (‰)

89

-26 -28

30.0

30.5

31.0

31.5

32.0

-20 -22 -24 -26 -28

32.5

8

9

Salinity (psu)

10

11

12

13

14

15

Temperature (°C)

-16 32.5 32.0 -20

Salinity (psu)

13

Bulk δ C of POM (‰)

-18

-22 -24

31.5 31.0 30.5

May 1998 May 1999 July 1999

-26 30.0 -28

-50

0

50

100

150

Distance to shelf-break (km)

200

-50

0

50

100

150

200

Distance to shelf-break (km)

Figure 3.12. The δ13C of POM from all three cruises plotted against salinity (top-left), temperature (top-right), and the distance to the shelf-break (200m iso-bath). The distances to the shelf-break shown as positive numbers indicate a distance offshore, and negative numbers indicate the distance toward the coast from the shelf-break. In the lower right corner the salinity is plotted versus the distance to the shelf-break.

The salinity was generally lower in shelf waters than in waters seaward of the shelfbreak (Figures 3.11 and 3.12). Close to the coast the salinities were found to be somewhat higher during May 1998 than in May and July of 1999. Both the salinity and the offshore distance to the shelf-break show a negative correlation with the δ13C of POM (salinity vs δ13C: R2=0.37; dist. to shelf-break vs δ13C: R2=0.40, n=68; Figure 3.12).

90 Water on the continental shelf typically had a thinner mixed layer than off the shelf (see Figure 3.13). The mixed layer depth at the stations visited was generally deeper in May 1998 than during the other two cruises. Also, the disparity between the mixed layer depth in shelf- and off shelf waters was on average greater in May 1998 than found in 1999. The mixed layer depth was smallest in July 1999.

45

May 1998

May 1999

July 1999

Mixed layer depth (m)

40 35 30 25 20 15 10 5 0

Off

Shelf

Off

Shelf

Off

Shelf

Figure 3.13. Box and whisker plots of the mixed layer depth in shelf- (right) and off-shelf waters (left), plotted per cruise. The black line within a box is the median, the upper and lower box edges define the 75th and 25th percentiles, and the whiskers represent the 95th and 5th percentiles. The black squares indicate the mean, and asterisks mark the maximum and minimum values.

3.4.7 Upwelling index In Figure 3.14 daily upwelling index values for 48ºN 125ºW during 1998 and the first half of 1999 are shown. These values are generated by the Pacific Fisheries Environmental Laboratory (PFEL, http://www.pfeg.noaa.gov), and are essentially estimates of daily wind-induced offshore Ekman transport. The estimates are calculated with winds that are derived from six-hourly synoptic and monthly mean surface atmospheric pressure fields. As can be observed in Figure 3.14, the fall and winter months are characterized by strong downwelling events. This period is followed by a

91 transition-period, which usually occurs some time after the middle of February (Thomson and Ware, 1996). In 1998 the spring transition occurred in the last week of February, whereas in 1999 the transition took place about a month later at the very end of March (see Figure 3.14). A major upwelling event took place at the end of May, in between the May and July 1999 cruises.

Upwelling Index ((m sec )/100 metres)

200 100 0 -100 -200 -300 -400 -500 -600 -700

1998

1999

-800 0 30 60 90 120 150 180 210 240 270 300 330 360 30 60 90 120 150 180 210

Time (days) Figure 3.14. Bakun Upwelling Index values for 48ºN 125ºW generated by the PFEL (see text). Values are a daily average of wind-driven cross-shore transport computed from FNMOC six hourly surface pressure analyses (see text). Indices are in units of cubic metres per second along each 100m of coastline. Positive numbers indicate offshore transport (i.e., upwelling conditions) and negative numbers represent downwelling-favourable conditions, with transport in the opposite direction. Periods of sampling are indicated with grey.

3.4.8 δ13C of POM off the Washington coast In June 1989 Fry and Bates collected POM and DIC samples off the coast of Washington state (U.S.), along a line at 48º18’N. The δ13C values measured for POM show a variability and magnitude equivalent to the values obtained in the present study (see Figure 3.15; Fry and Bates, unpublished data). Their data also show a comparable decreasing pattern from shelf to off shelf, with high δ13C values close to the shore

92 associated with diatom blooms and chlorophyll a levels of 1-6µg/L. In addition to

2

13

δ C (‰)

0 -18 -20 -22 -24 DIC POM >20µm Bulk POM

-26 -28 250

200

Off shelf 150

100

Shelf 50

0

Distance offshore (km) Figure 3.15. Unpublished data provided by Drs. Brian Fry and Timothy Bates. The figure shows δ13C measurements of POM and DIC from water collected off the Washington coast along an east-west track at 48º18’N in June of 1989 (NOAA Cruise nr. AR8902). In addition to bulk POM, >20µm net POM was analyzed (see legend). The values are plotted against the distance offshore and the approximate position of the shelf break is indicated with a black line.

measurements of bulk POM, seston collected with a 20µm-mesh net was analyzed for its stable carbon isotope ratio. As can be observed in Figure 3.15, the POM >20µm was always more enriched in

13

C than the bulk POM samples. Although samples of the

different size fractions were not always taken at the same stations, the differences between δ13C values of bulk POM and POM>20µm seem to be in the range of 0.5 to 3.5‰. It should be noted that the values measured for the bulk and the larger seston follow the same trend (see Figure 3.15).

3.4.9 δ13C of DIC The stable carbon isotope ratio of DIC was measured only for water (from 5m depth) of 7 stations sampled in July 1999. During DIC extraction it was suspected that the CO2

93 extracted from the DIC sample taken at station LC4 was slightly exposed to lab air. No correlation was found between the δ13C of the DIC and POM (Figure 3.16). Neither did the data of Fry and Bates show that the large variation in stable carbon isotope ratios measured for POM could be explained by variations in the δ13C of the DIC (Figure 3.15). Moreover, if a correlation between the the δ13C of the DIC and the distance to shore exists, Figure 3.16 seems to indicate that the DIC became more depleted in

13

C when

moving toward the coast. 1.5

1.0

LC9

0.5

LC12

δ C of DIC (‰)

LG7

LBP7

0.0

1.0

LG3

LBP2

-0.5

0.5 0.0 -0.5

13

13

δ C of DIC (‰)

1.5

-1.0 -1.5 -24

LC4 -23

-22

-21

-20

-19

-18

-1.0 -1.5 120

δ13C of bulk POM (‰)

100

80

60

40

20

0

Distance to shore (km)

Figure 3.16. δ13C of DIC plotted against the δ13C of POM collected both at 5m depth in July 1999 (left). The data points are labelled with the station names. On the right the δ13C of DIC is plotted against the distance to shore. The CO2 extracted from the DIC sample taken at station LC4 may have been contaminated by lab air (see text).

3.4.10 Correlation matrix A correlation matrix is shown in the Appendix. The variables that correlate best with the δ13C of POM are fatty acid abundances. Chl a, distance to shore and salinity correlate reasonably well too (see Appendix). The water temperature and nutrient concentrations do not show any relationship with POM 13C/12C ratios.

94

3.5 Discussion 3.5.1 The 16:2n-4 fatty acid as a diatom indicator Of the fatty acids commonly found in marine seston, none is unique to a single algal class or even species (Pohl and Zuhrheide, 1979; Sargent et al., 1988; Reuss and Poulsen, 2002). However, ratios and the relative abundance of certain fatty acids can indicate whether certain algal classes or bacteria are dominant. C16-polyunsaturated fatty acids (PUFAs) are characteristically more abundant in diatoms, some C18-PUFAs are abundant in Prymnesiophyceae, Dinophyceae and Cryptophyceae, and branched fatty acids are indicative for the presence of bacteria (Chuecas and Riley, 1969; Pohl and Zuhrheide, 1979; Sargent et al., 1988). The 16:1n-7/16:0 ratio has been used to show diatom dominance (Viso and Marty, 1993; Napolitano et al., 1997; Reuss and Poulsen, 2002) and has successfully been applied as a food web tracer (St. John and Lund, 1996). Here the relative abundance of the 16:2n-4 fatty acid seems to be the best indicator for both diatom dominance as well as relative scarceness of flagellates (Figure 3.3). The 16:2n-4 occurs at higher levels in diatoms (Wood, 1974; Volkman et al., 1989; Budge and Parrish, 1999; Piveteau et al., 1999). As illustrated by Figure 3.17a the 16:2n-4 abundances correlate reasonably well (R2=0.59, n=68) with the 16:1n-7/16:0 ratio, which is more traditionally used as a diatom marker (see above). It should be noted that the log(nano/diatom) – 16:2n-4 correlation shows quite a bit of heteroskedasticity, with more scatter at lower log(nano/diatom) values (Figure 3.3). This can be explained by the difference in concentration of 16:2n-4 between diatom species, and variability in abundance of this fatty acid under different growth conditions (Opute, 1974; Budge and Parrish, 1999). This variability can easily amount to a factor of two. Hence, in this case, at low log(nano/diatom) values (i.e., high abundance of diatoms) a factor of two can amount to 0.8 percentage point difference in 16:2n-4 abundance, whereas at high log(nano/diatom) values this leads to only a spread of 0.2 percentage points (Figure 3.3). Therefore, it should be stressed that the relative amount of 16:2n-4 can only be used as a first approximation of diatom abundance.

95

In chapter 2 a discriminant function using the abundances of 8 different fatty acids was constructed to optimally predict the shelf- or off shelf origin of marine organisms. The discriminant scores of POM samples, using the same discriminant function derived in Chapter 2, correlate with the abundance of 16:2n-4 (Figure 3.17b). This relationship underscores the suggestion that the diatom over nanoflagellate dominance, and vice versa, drives a large part of the differences found in the fatty acid composition of

16:1n-7/16:0 ratio

1.6

a

1.2 0.8 0.4 0.0 0.0

0.5

1.0

1.5

2.0

2.5

16:2n-4 abundance (% of total fatty acids)

Shelf - off shelf discriminant score

organisms higher up in the food web.

8

b

Off shelf Shelf

6 4 2 0 -2 0.0

0.5

1.0

1.5

2.0

2.5

16:2n-4 abundance (% of total fatty acids)

Figure 3.17. The 16:2n-4 fatty acid abundance in POM samples plot against the 16:1n-7/16:0 fatty acid ratio (a), and discriminant score (b). The discriminant scores were computed using the discriminant function from chapter 2 that was found to best distinguish between organisms (i.e., not just POM) from shelf and off shelf waters.

Although better than any other fatty acid alone, by itself the 16:2n-4 fatty acid is unfortunately not suitable enough to be a single food web tracer. As outlined in Chapter 2, about 32% of the samples are wrongly classified when predicting their origin with the abundance of 16:2n-4 alone. This is likely due to the low amounts of this fatty acid taken up by most heterotrophs. For example, Piveteau et al. (1999) found that the 16:4n-1, 16:2n-4 and 16:3n-4 fatty acids were poorly accumulated in oysters, fed the diatom Skeletonema costatum. These authors suggested that oysters discriminate against these fatty acids. It is therefore imperative to use several ratios or preferably multivariate

96 techniques to properly distinguish between organisms from the shelf and off shelf environments using fatty acid data.

3.5.2 Influence of land-derived POM In particular, along line C the δ13C of POM appears to be more a function of the distance to shore than a stepwise shelf-off shelf disparity (Figure 3.5). Because a relationship exists between the distance from land and the 13C/12C ratio of POM it is reasonable to ask whether this was caused by input of organic carbon from land. Virtually no macrophyte fragments were observed on the filters with the naked eye. And, although the filters were not examined under the microscope, the data available do not support mixing with landderived material as a cause for the observed δ13C trend. This is because C3-plant dominated terrestrial material typically has δ13C values of around -26‰ (Fry and Sherr, 1984 and references therein), which is even more depleted in 13C than samples from most off shelf stations. Another possibility is that seagrasses like Zostera sp. contributed to POM collected close to the coast. Zostera and marsh plants like Spartina are known to exhibit high 13

C/12C ratios, with δ13C values of up to -7‰ (Fry and Sherr, 1984; Hemminga and

Mateo, 1996; Kharlamenko et al., 2001). However, these plants are very rich in linoleic(18:2n-6) and linolenic acid (18:3n-3) (Canuel et al., 1997; Kharlamenko et al., 2001; Lovell et al., 2001). In this study the average abundance of both fatty acids was, in fact, found to be significantly lower (two-sample t-test: P<0.05) in POM from the shelf environment than off the shelf. Therefore, it seems clear that no appreciable amounts of terrestrial plant- or seagrass-derived material contributed to the total POM sampled in this study. It certainly cannot explain the spatial patterns observed in the stable carbon isotope and fatty acid data.

3.5.3 Diatom effect? In the previous discussion it was established that the relative amount of the 16:2n-4 fatty acid is a reasonably good proxy for the abundance of diatom-derived material in POM.

97 Hence, the relationship of the 16:2n-4 abundance with the δ13C of POM (Figures 3.6 and 3.7) seems to suggest that the

13

C/12C ratio of seston is controlled by the proportion of

diatom derived carbon. Several authors have indeed found that diatoms are generally more enriched in 13C than other algal classes (Wong and Sackett, 1978; Falkowski, 1991; Fry and Wainright, 1991). However, close inspection of the data, leads to the conclusion that the δ13C of POM is not simply a function of the proportion of diatom derived carbon. As outlined before, both lab- and field studies show that diatoms exhibit very low amounts (generally 0-1% of total) of 18:4n-3, whereas this fatty acid is generally found at much higher levels in flagellates (see below). Therefore, if we assume that the contribution of diatoms to 18:4n-3 in the analyzed POM is negligible, the δ13C of the 18:4n-3 fatty acid can be regarded as a flagellate signal. Then, the fact that the δ13C of 18:4n-3 correlates with the 16:2n-4 abundance (Figure 3.8) rather suggests that diatoms seem to thrive better under growing conditions that lead to 13C-enrichment in other algal taxa (such as flagellates) as well. The previous inference was further tested as a few authors (Volkman et al., 1989; Thompson et al., 1992; Graeve et al., 1994) did report noticeable amounts of 18:4n-3 in mono-cultures of diatoms (i.e., 5.3, 3.7 and 3.9% of total fatty acids for their cultures of Thalassiosira pseudonana, Chaetoceros simplex and Thalassiosira Antarctica, respectively). In Figure 3.18 the theoretical δ13C values of a fatty acid are plotted against the end-member proportion when two sources with unequal (and equal) concentrations of that fatty acid are mixed. Formulas used are:

FAf a =

FA% a ⋅ Totf a FA% a ⋅ Totf a + FA% b ⋅ (1 − Totf a )

and

δ13C FA = FAf a ⋅ FAδ13Ca + (1 − FAf a ) ⋅ FAδ13C b

(3.1)

, with

(3.2)

FAfa is the fraction (between 1 and 0) of the fatty acid derived from the diatom endmember (a). FA%a and FA%b are the abundances (in % of total fatty acids) of the fatty acid in a and b, with b being the non-diatom end-member. Tot%a is the percentage of the

98 total amount of fatty acids in the sample derived from diatoms (a). FAδ13Ca and FAδ13Cb are the δ13C values of the fatty acids in end-members a and b, respectively. The median 18:4n-3 abundance for diatoms and flagellates in culture, computed from values reported in the literature, was found to be 0.45 and 10.1% of the total fatty acids, respectively (References used: Pohl and Zuhrheide, 1979 and refs. therein; Volkman et al., 1981, 1989; Ben-Amotz et al., 1987; Kattner and Brockmann, 1990; Thompson et al., 1992; Viso and Marty, 1993; Graeve et al., 1994; Mansour et al., 1999). For these estimations 36 values for diatoms, and 55 values were used for the flagellates (i.e., 21 Prymnesiophyte-, 12 cryptophyte- and 22 dinoflagellate samples). Hence, the 1:20 curve in Figure 3.18 would be the best estimate when mixing a diatom- and a flagellate end-member. This concave curve and the almost convex pattern shown in Figure 3.8a are obviously quite dissimilar. The dataset should cover almost the entire range of x-axis values from Figure 3.18. From the May 1998 phytoplankton counts Harris (2001) estimated that the percentage of the total biomass made up by diatoms was around 92% at stations LC4 and LBP2. The fatty acid data (Figure 3.4) suggest that some CSline stations in May 1999 may have been even more dominated by diatoms. Therefore, when we accept that the relative amount of 16:2n-4 is a good proxy for diatom abundance, the relationship between the 16:2n-4 abundance and the δ13C of 18:4n-3 is indeed not simply the result of mixing of 18:4n-3 from

13

C-enriched diatoms and

13

C-

depleted flagellates. Additionally, a similar, but weaker case could be made for the relationship between the δ13C of 20:5n-3 and the 16:2n-4 abundance, with 20:5n-3 on average being present at higher levels in diatoms. From the previous exercise it is clear that the spatial and temporal pattern of δ13C values of POM cannot simply be explained by mixing of different proportions of diatomand flagellate derived carbon. It may still be, however, that diatoms are at least somewhat more enriched in

13

C than other taxa. For instance, Kukert and Riebesell (1998) found

that in a Norwegian fjord the δ13C of the bigger size fraction of seston, dominated by diatoms, was always enriched in dominated by flagellates.

13

C by about 2‰ than the smaller size fraction

99 It is not straightforward to find out with the molecular data whether diatoms had higher 13C/12C ratios than other algal classes. This is because the difference between the δ13C values of, for example, a typical flagellate- and a diatom fatty acid, does not necessarily reflect the difference between the δ13C of flagellates and diatoms. Rather, the difference in δ13C values of the different fatty acids within a single sample gives information about the various biosynthetic pathways that each fatty acid has undergone (see also Chapter 4). As already suggested by other authors, both elongation and unsaturation reactions during the formation of the fatty acids discriminate between

13

C

and 12C (Fang et al., 1993).

-22 -24

1 1: 3 1: 5 1:

-26 -28

13

δ C of individual fatty acid (‰)

-20

-30

0

20

40

60

10 1: 20 1:

80

100

% of total fatty acids derived from diatoms Figure 3.18. Mixing curves produced when mixing a diatom end-member and a non-diatom end-member in which the same fatty acid supposedly has a δ13C value of -20‰ and -30‰, respectively. The 1:1 curve describes the scenario in which both end-members contain similar relative amounts of the particular fatty acid. The 1:3 curve describes the case in which the fatty acid is 3 times more abundant in the non-diatom end-member than in the diatom end member (i.e., % of total fatty acids is 3 times higher), and so on. The curves show how the δ13C of the particular fatty acid (δ13CFA) will vary when different proportions of the two end-members are mixed.

Despite the aforementioned complications a test can be done to see whether in fact diatoms were more enriched in 13C. This is possible by means of studying the difference between the δ13C of a typical non-diatom fatty acid (e.g., 18:4n-3) and a fatty acid that is common in all phytoplankton, and especially in diatoms (e.g., 16:1n-7). Of the 70 POM

100 samples available (from all 3 cruises), 15 samples believed to contain the most-, and 15 samples that contained the least amount of diatom-derived material were selected. In the diatom-rich samples a high proportion of the 16:1n-7 is expected to be derived from diatoms, whereas most of the 16:1n-7 in the diatom-poor samples is expected to originate from other algae. As stated before, only a marginal percentage of 18:4n-3 is expected to have a diatom origin, even in shelf waters. Therefore, the difference in δ13C values of the 16:1n-7 and the 18:4n-3 fatty acids in POM is expected to be larger in the 15 diatom rich samples if diatoms were more enriched in

13

C. Indeed, the average difference in δ13C

between 16:1n-7 and 18:4n-3 was significantly larger (P=0.05, two sample t-test) for the 15 diatom-rich samples than for the 15 diatom-poor samples. However, doing the same test using 16:0 or 20:5n-3 instead of 16:1n-7 did not result in significant differences. Therefore, with this data set no conclusive evidence was found to state that diatoms were in fact more enriched in 13C than other phytoplankton.

Thus, it is argued here that the higher δ13C values of more diatom-rich POM is not explained by the higher contribution of diatom-derived carbon, but by diatom favourable growth conditions that cause 13C-enrichment in all phytoplankton.

3.5.4 Diatom favourable conditions From the previous discussion it was concluded that diatoms (and hence, 16:2n-4) become more abundant in conditions that cause

13

C-enrichment in seston in general. To know

what diatom favourable conditions are may therefore give insight into what mechanisms drove up the δ13C values of POM sampled predominantly in the shelf habitat. Diatoms are able to out-compete most other species under non-limiting nutrient conditions (Thomas et al., 1978; Egge, 1998). For diatoms in particular, an adequate amount of silicate (i.e., above approx. 2µM) is critical in maintaining dominance in the phytoplankton community (Egge and Aksnes, 1992). Off the west coast of Vancouver Island several mechanisms are responsible for the high nutrient supply rates to shelf waters during the summer (April-September). Episodically upwelling occurs, bringing

101 deep nutrient rich water onto the shelf. More continuous enrichment is delivered by estuarine discharge from the Juan de Fuca Strait, mainly fed by water from the Fraser River and rivers emptying into Puget Sound. Additionally, nutrient rich water is supplied to surface waters on the central portion of the shelf off Juan de Fuca Strait by a cyclonic gyre interacting with a tributary of Juan de Fuca canyon (Freeland & Denman, 1982). Finally, tidal mixing may also provide a nutrient flux across the isopycnals, but Crawford and Dewey (1989) estimated its contribution is less than 10% of the nutrients provided by the estuarine outflow out of Juan de Fuca Strait. No correlation was discovered here between nutrient levels and diatom abundance data from all three cruises. Moreover, in May 1998 nitrate concentrations were often below the detection limit when diatoms were dominant on the shelf (cf. Figures 3.10 and 3.2). It is important to note, however, that the standing stock of phytoplankton is more a reflection of past nutrient conditions than the present. Additionally, in situ concentrations may not be a good indication of the flux, or supply rate of nutrients, which should ultimately be the determining factor for growth. It was found that diatom markers, such as the 16:2n-4 fatty acid, generally indicate that the highest relative amounts of diatoms occurred within a distance of 20-30km off the coast (Figure 3.5). The Vancouver Island Coastal Current, which flows northward within the first 20km inshore, is driven primarily by low salinity water emanating from the Juan de Fuca Strait (Freeland et al., 1984; Thomson et al., 1989). Crawford and Dewey (1989) estimated that the Juan de Fuca source provides a higher flux of nutrients than upwelling or tidal mixing. It seems likely, therefore, that the higher diatom abundance close to the coast (within 20-30km) was a result of the higher supply rate of nutrients provided via the Vancouver Island Coastal Current.

Another factor that has been found to favour diatoms is enhanced turbulence and mixing (Margalef, 1978). In experimental mesocosms Parsons et al. (1978) found shifts from diatom to flagellate-dominated communities during reduced vertical mixing. Intermittent dinoflagellate blooms are often found during upwelling relaxation periods (Blasco, 1977), and turbulence induced by winds >350cm s-1 was reported to disrupt

102 dinoflagellate blooms in the Dead Sea (Pollingher and Zemel, 1981). More mixing can be expected on the shelf due to the shallower water depth and frequent upwelling events. Especially the combination of shallower depth and complex topography can make a flow over the bottom and tidal mixing more effective stirring agents (Gill, 1982).

The mixed layer depths during the three cruises (Figure 3.13) extended deepest in May 1998, and was shallowest in July 1999. Diatoms were indeed most dominant in May 1998 when the mixed layer extended to greater depth on the shelf. In contrast, dinoflagellates were most abundant in July 1999 when shelf waters were more stratified.

3.5.5 Other taxa The δ13C of POM shows positive correlations with the abundances of typical diatom fatty acids (C16 series) and negative correlations with (C18) fatty acids which are known to occur more in flagellates (see Appendix). An exception is docosahexaenoic acid (22:6n3) which, especially in July 1999, shows a significant positive correlation (R2=0.64) with the δ13C of POM (Figure 3.7). As mentioned earlier, the 22:6n-3 fatty acid is particularly abundant in dinoflagellates. Phytoplankton counts were available for water from station LBP2 sampled in July 1999, for which the POM has the second highest abundance of 22:6n-3 (after LC2 of July 1999). In cell numbers, dinoflagellates only made up about 8% of the community (Figure 3.2). However, the proportion of dinoflagellates that is larger than 20µm was higher than for any other sample counted (i.e., about 3% of the total number of cells). Dinoflagellates can have a volume that is several orders of magnitude greater than nanoflagellates. Therefore, the contribution of dinoflagellates to the total biomass may in fact have exceeded that of nanoflagellates at LBP2 in July 1999. The highest abundance of dinoflagellates occurred in July 1999 (Figure 3.2). Samples taken from shelf water had consistently higher numbers of large dinoflagellates (>20µm) than off shelf samples. Nutrients, including silica, were abundant (Figure 3.10), but no high numbers of diatoms were found. Furthermore, the water column was more

103 stratified (shallower mixed layer) which is known to favour dinoflagellate growth (Smayda, 1997). With the compound specific isotope data (δ13C of 22:6n-3 relative to other fatty acids) no indication was found that the dinoflagellates were more enriched in

13

C than

other algae at the same stations. Just as is suggested here for diatoms, dinoflagellates (particularly the larger ones) seem to thrive better in conditions that led to more

13

C

enrichment in all phytoplankton. For the May 1998 shelf samples, however, a negative correlation was found between 22:6n-3 abundance and POM δ13C (Figure 3.7). This can perhaps be explained by the general low abundance of dinoflagellates and the fact that diatoms virtually dominated the biomass. Thus, in May 1998 higher levels of 22:6n-3 were perhaps more indicative of a decreased amount of diatoms rather than higher amounts of dinoflagellates, with the 22:6n-3 probably more originating from other flagellates.

Dominance of taxa other than diatoms or dinoflagellates caused significant departures from regression lines predicting the δ13C of POM from the abundance of 16:2n-4 and 22:6n-3 (Figure 3.7). POM from stations LB3 and LB7 exhibited anomalously high abundances of 18:4n-3 (8-9% of total fatty acids), a fatty acid known to occur at high concentrations in flagellates (see above). Chlorophyll a levels were not measured at these stations, but concentrations were found to be fairly high at stations nearby (3 and 9 mg/m3 for LB6 and LB8, respectively). Hence, these samples would probably also deviate from the relationship found between the δ13C of POM and chl a concentration (Figure 3.6b). Another deviation from the chl a – δ13C relationship is POM from LBP7 in July 1999. At this station a relatively high number of cryptomonads were identified (23% of total cells). The lower δ13C of POM at these stations is either caused by a difference in physiology of the dominant flagellate species or by the growth conditions leading up to the dominance of these species.

104

3.5.6 Growth rate On all surveys the shelf environment typically had higher surface chlorophyll a levels than off-shelf (Figure 3.9). Chl a concentrations correlate with δ13C (Figure 3.6b) and the abundance of diatom marker 16:2n-4 (not shown, but see Appendix). High chl a amounts can be explained either by high productivity (perhaps associated with higher growth rates), or by less grazing. Harris (2001) confirmed that higher productivity typically occurred in shelf waters, relative to off the shelf off Vancouver Island (May 1998 included). Therefore, in order to sustain the higher productivity, resources such as nutrients and light must generally have been more plentiful in shelf waters. Such conditions are also known to favour dominance of diatoms in phytoplankton communities (Thomas, et al., 1978; Egge, 1998). A correlation between chlorophyll a concentrations and δ13C of POM has been noted by Gu et al. (1999) in an Alaskan lake. Field studies near Antarctica found a relationship between particulate organic carbon (POC) concentration and δ13C (Kopczyńska et al., 1995 and also on one of the lines by Villinski et al., 2000). Additionally, in the Delaware estuary (Cifuentes et al., 1988), as well as in a controlled ecosystem enclosure in Saanich Inlet (Nakatsuka et al., 1992) primary productivity correlated well with the δ13C of suspended particulate matter. Nakatsuka et al. (1992) investigated the changes of the stable carbon isotope ratio of POM during a phytoplankton bloom. They found that neither the phytoplankton species composition, nor the δ13C of DIC, or the concentration of molecular CO2 could well explain the variations in δ13C. These authors suggested that increased growth rates were responsible for the higher δ13C values observed. Growth rates have also been shown to influence the δ13C of algae in controlled lab experiments (Fry and Wainright, 1991; Laws et al., 1995). During fast growth less CO2 is able to leak out of the cell due to the higher demand to supply ratio. This results in 13C accumulation inside the cell due to the preferential use of 12

C by ribulose 1,5 bisphosphate carboxylase, the carbon fixating enzyme. Hence, faster

growth can be expected to lead to higher δ13C values.

105

3.5.7 Cell Size Off the Washington coast the larger size fraction (>20µm) of POM, probably dominated by diatoms, was found to have consistently higher 13C/12C ratios than the bulk (Fry and Bates, unpublished: Figure 3.15). Similarly, Kukert and Riebesell (1998) reported that during a diatom bloom in a Norwegian fjord the larger cells (>20µm, dominated by the diatom Skeletonema costatum) had δ13C values that were about 2‰ higher than measured for the smaller size fraction. Rau et al. (1990) also found in the Mediterranean the POM of higher size classes was consistently enriched in 13C. Coastal areas and oceanic upwelling areas are typically higher in algal biomass with a dominance of larger species compared to open ocean waters (Malone, 1980). It has been shown that large algae can become dominant under fluctuating or patchy nutrient regimes (Turpin and Harrison, 1979; Sciandra, 1991). The larger storage capacity of large algae would make them compete better under fluctuating conditions than smaller algae (Stolte and Riegman, 1995). On the other hand, selective grazing by microzooplankton can also induce a higher abundance of larger cells (Riegman et al., 1993). Harris (2001) found that in shelf waters off the west coast of Vancouver Island during the spring, summer and fall of 1997 and 1998 on average 72% of the primary productivity was accounted for by algae larger than 5µm. Off the shelf the larger size fraction amounted to only 36% of the primary productivity. Cell size and geometry were found to have an effect on the fractionation of carbon isotopes during carbon fixation (Popp et al., 1998; Burkhardt et al., 1999).

3.5.8 [CO2(aq)] and active uptake of inorganic carbon The concentration of dissolved CO2 in (sea)water ([CO2(aq)]) can be influenced by several parameters. Temperature affects the solubility of CO2 (and other gases) such that more CO2 can dissolve in water at lower temperatures. The pH of the water also influences the [CO2(aq)] by affecting the equilibration between CO2, bicarbonate and carbonate. Hence, a higher pH will result in less CO2(aq) being available. High primary production rates, for example during blooms, are known to drive up the pH and lower the

106 availability of CO2, not only in lakes but also in the marine environment (Brussaard et al., 1996; Hobson et al., 2001 and refs. therein; Berman-Frank et al., 1998). Other processes that can increase the [CO2(aq)] include upwelling (Christensen, 1994), which brings water from intermediate depths to the surface and enhanced turbulence and bubble injection during storm events (Farmer et al., 1993). Several authors have shown that the CO2(aq) concentration influences the isotopic fractionation (εp) by phytoplankton, both in field- and lab studies (Degens et al., 1968; Rau et al., 1989, 1997; Freeman and Hayes, 1992; Francois et al., 1993; Hinga et al., 1994; Laws et al., 1995, 1997; Bentaleb et al., 1998; Burkhardt et al., 1999). However, in coastal, in particular upwelling regions the correlation between [CO2(aq)] and εp seems to break down (Pancost et al., 1997, 1999; Tortell et al., 2000; Rau et al., 2001). This insensitivity to CO2(aq) concentrations has been ascribed by active uptake of inorganic carbon (ci) becoming a more important carbon acquisition strategy in these settings (Rau et al., 2001). Many microorganisms possess mechanisms that concentrate CO2 at the carboxylation site in order to compensate for the low affinity for CO2 of ribulose bisphosphate carboxylase-oxygenase (RuBisCO) (Kaplan and Reinhold, 1999). For marine phytoplankton organisms the possession of a carbon concentrating mechanism (CCM) seems to be more the rule than an exception (Nimer et al., 1997; Moroney, 2001). Induction of a CCM has been shown to decrease cellular εp (Sharkey and Berry, 1985; Goericke et al., 1994; Laws et al., 1997; Keller and Morel, 1999; Fielding et al., 1998). There is mounting evidence that various marine phytoplankton species can use bicarbonate as a source of carbon (Fielding et al., 1998 and references therein; Larsson and Axelsson, 1999; Huertas et al., 2000; Young et al., 2001). The δ13C of HCO3- is, when in equilibrium with CO2, as much as 8-12‰ (depending on temperature) higher than the δ13C of CO2 (Mook et al., 1974; Zhang et al., 1995). However, carbonic anhydrase (CA), the enzyme that mediates the dehydration of HCO3- either inside or outside the cell, causes a fractionation of about 10‰, making the resulting CO2 virtually indistinguishable from free CO2 (Riebesell and Wolf-Gadrow, 1995). Nevertheless, a significant enrichment in

13

C can still be achieved when the CO2 obtained from HCO3-

(with the aid of CA) is utilized rapidly or quantitatively (Tortell et al., 2000). This can be

107 the case, for instance, due to CA activity in the immediate vicinity of RuBisCO. Additionally, the 13C enrichment of CCM using cells has been attributed to a decrease in CO2 leakage relative to the carbon fixation (Goericke et al., 1994). The effects of environmental parameters on the operation of CCMs has been reviewed by Beardall et al. (1998). The CO2(aq) concentration (which in turn is affected by, inter alia: temperature, biological productivity, pH; see discussion above) has been found to affect the rate of active uptake of inorganic carbon, with higher CCM activities at lower [CO2(aq)] (Beardall et al., 1998 and references therein). Burkhardt et al. (2001) measured CO2 and HCO3- uptake and in vivo activity of extracellular- (eCA) and intracellular carbonic anhydrase (iCA) in two marine diatoms acclimated to different CO2 partial pressures. Both species responded to decreasing CO2 supply with an increase in both eCA and iCA activity. In both diatoms CO2:HCO3- uptake ratios progressively decreased with decreasing CO2 concentrations (Burkhardt et al., 2001). Berman-Frank et al. (1998) found that over the course of a dinoflagellate bloom in a lake, as the pH increased and the [CO2(aq)] decreased, intracellular inorganic carbon concentrations and CA activity rose drastically. Light and nutrient levels are also important in regulating CCMs. A lower photon flux density can reduce CCM activity and capacity, and similarly, CCMs appear to be repressed in phosphate–limited cultures (Beardall et al., 1998). Another important nutrient for CCMs is zinc. Zinc is used in carbonic anhydrase, but can be replaced by Co or Cd in the active site (Lane and Morel, 2000). Therefore, when these trace metals are limiting, active inorganic carbon uptake will be reduced. Low nitrogen levels, however, may have a different effect on the rate of active uptake of inorganic carbon. In theory a microalga with a CCM would have a lower cost of CO2 assimilation than one relying on diffusive CO2 uptake only (Beardall et al., 1998). This is because diffusive CO2 entry does not saturate RuBisCO. Hence, the fixation rate per unit N invested in RuBisCO is lower than that in plants with a CCM where internal CO2 is saturating RuBisCO (Beardall et al., 1998). With the data available it is not possible to establish whether more active uptake of inorganic carbon was invoked by phytoplankton in the shelf region (giving rise to higher

108 δ13C values) or not. Neither were there any CO2 concentrations measured. However, when comparing shelf with off shelf waters, the shelf region appears more favourable to higher activity of CCMs. In shelf waters the primary productivity is generally higher than further offshore due to the higher nutrient flux to surface waters (as outlined before). This causes enhanced drawdown of CO2, which may promote the use of CCMs. Trace metals, such as zinc, are less likely to be limiting close to the coast. Additionally, the shallower mixed layer depths above the shelf may cause light to be available more continuously at a higher photon flux density. As discussed above, nitrate limitation in May 1998 may have even supported the competitiveness of species that can induce a CCM. On the other hand, water with a high [CO2(aq)] can be brought up to the surface during upwelling. Furthermore, Ianson et al. (in preparation) report high pCO2 levels in the Vancouver Island Coastal Current, even though a high primary productivity was observed in the same water. Therefore, during upwelling conditions the use of CCMs may not always be more advantageous as high pCO2 levels can exist in the photic zone. However, right after the upwelling event, when pCO2 levels decrease, CCMs can perhaps provide a greater competitive advantage.

3.5.9 Differences between cruises During the spring of 1998 the conditions in the North Pacific were still dominated by an El Niño event that had started in the spring of 1997 (DFO Ocean Status Report, 2000). As can be observed in Figure 3.11 the water temperatures at 5m depth were 2 to 3ºC higher in May 1998 compared to May 1999. Temperature and salinity differences between the two years caused the mixed layer depth to be much shallower in the winter of 1997/98 than the following year in the Gulf of Alaska (DFO Ocean Status Report, 2000). When the mixed layer is thin, weaker than normal exchange between shallow and deep water is possible, causing reduced winter entrainment of nutrients in the surface layer (Freeland et al., 1998). Additionally, the transition from predominant strong downwelling to more upwelling favourable conditions took place earlier in 1998 (Figure 3.14). This allowed high phytoplankton productivity to start sooner, and this, together with the lower amount

109 of nutrients entrained in the surface ocean during the winter, may have allowed the depletion of nitrate in May 1998 (Figure 3.10). As can be expected for the summer months, during the July 1999 cruise the surface water temperatures had increased and the mixed layer depth had been reduced. In particular the waters of the inner shelf had become less dominated by diatoms and dinoflagellates were more abundant. It is important to note that higher amounts of dinoflagellates and the concomitant lower abundance of diatoms did not result in lower δ13C values of POM in shelf waters in July 1999. However, seston from shelf waters with hypothesized higher nanoflagellate abundance (i.e., with high 18:4n-3) did have anomalously low 13C/12C ratios (Figure 3.7, July 1999). If temperature were a controlling factor of the δ13C of POM values should have been lowest in May 1999. In fact, δ13C values of POM collected in shelf waters during that cruise were found to be slightly higher than that collected during other cruises (Figures 3.7 and 2.5). The samples that are most enriched in 13C were collected in May 1999 at stations LG1 and LC2 (δ13C= -17.7 and –17.4‰, respectively). Both samples also contained high relative amounts of 16:2n-4, suggesting that diatom derived material dominated the POM. In May 1998 a big variation in 13C/12C ratios of POM samples was found (range of 9‰). This larger range of δ13C values was entirely driven by the lower values found for the off shelf samples collected during that cruise. The fact that the May 1998 off shelf POM was more depleted in 13C than during other cruises was most likely caused by two factors. Firstly, the water was more depleted in nutrients (Figure 3.10), probably resulting in lower growth rates and productivity. Secondly, the MLD was larger than during both other cruises (Figure 3.13), which reduced the daily average availability of light for the growing cells. Further examination of the data revealed a negative correlation between the mixed layer depth and the δ13C of POM (Figure 3.19). However, in 1999 such a correlation was not observed, and the MLD correlated also with the distance to the shelfbreak in 1998 (Figure 3.19). Therefore, a causal relationship between the higher MLDs and 13C-depletion of off-shelf POM in May 1998 remains a conjecture.

110

3.5.10 Endeavour Segment POM The δ13C value of –33.2‰ measured for the POM at Endeavour Segment is unusually low. The even lower range of δ13C values measured for the individual fatty acids (-35 to 45‰) indicates that the low bulk value was not caused by a higher than normal lipid content. Additionally, the proportion of unsaturated fatty acids seems to indicate that most likely the majority of the POM at that location consisted of fresh organic matter of algal origin (Figure 3.20). To my knowledge such a low δ13C value for marine POM has only been reported for samples taken from water with a temperature of about 0ºC, south of 60ºS (see Goericke and Fry, 1994 and references therein).

2

R = 0.50 n=22

2

R = 0.002 n=22

-17 -18

Bulk δ C of POM (‰)

-20 -22

-19 -20 -21

13

-24

13

Bulk δ C of POM (‰)

-18

-26

May 1998 May 1999

-28 0

5

10

15

May 1998 20

25

30

35

-22 -23 -24

40

May 1999 0

5

Mixed Layer Depth (m) 40

40

2

Mixed Layer Depth (m)

Mixed Layer Depth (m)

35

30 25 20 15 10

-100

May 1998 -50

0

50

100

15

20

25

30

35

40

Mixed Layer Depth (m)

R = 0.55 35 n=22

5

10

150

Distance to shelf-break (km)

200

2

R = 0.23 n=22

30 25 20 15 10 5 30.5

May 1998 31.0

31.5

32.0

32.5

33.0

Salinity (psu)

Figure 3.19. Mixed layer depth plotted versus the δ13C of bulk POM collected during May 1998 (top-left) and May 1999 (top-right). At the bottom, the mixed layer depths observed in May 1998 are plotted against the offshore distance to the shelf-break (200m isobath) (bottom-left), and salinity (bottom-right). Negative numbers shown for the distance to shelf-break indicate the distances shoreward from the shelf-break. The squared correlation coefficient and the number of samples used are shown in each plot.

111 It is tempting to explain the unusual δ13C value with an unusual mechanism. In fact, the sample was taken at an unusual place at an unusual time. The POM originated from water exactly above hydrothermal vents that were affected by an earthquake swarm about three weeks prior to sampling (Johnson et al., 2000). A tenfold increase in fluid output by the vents was observed by Johnson et al. (2000) for a period of at least 80 days, extending along the entire ridge segment. It does, however, still seem unlikely that CO2 and methane from the vents would have seriously affected concentrations 2200m higher up in the water column. This is because mixing with surrounding water will dilute the vent water quite rapidly. Lupton et al. (1985) found that already at 200m directly above the vent field only 0.01% of the water originated from the vents. Additionally, due to stratification of the water column, currents may transport most of the water enriched in methane and CO2 laterally rather than vertically (through isopycnal surfaces).

16

Relative abundance (% of total)

14 12 10 8 6 4 2

14:0 ? iso 15:0 15:0 ? 16:0 16:1n-9 ? 16:1n-7 16:1n-5 iso 17:0 16:2n-4 ? 16:3n-4 ? ? ? 18:0 18:1n-9 18:1n-7 18:1n-5 18:2n-6 ? 18:3n-3 18:4n-3 ? 20:0 20:1n-11 20:1n-9 20:1n-7 20:2n-6 20:3n-6? 20:4n-6 20:3n-3 20:4n-3 ? 20:5n-3 22:0 22:1n-11 ? 22:5n-3 ? 22:6n-3

0

Fatty acid Figure 3.20. Fatty acid profile of POM from water (5m depth) above Endeavour Segment (station 2).

112 Assuming the dissolved CO2 available had a “normal” δ13C value of about -8‰, a fractionation of about 25‰ was associated with the carbon fixation by algae growing in these waters. This is still within the limits of the fractionation possible for carboxylation reactions catalyzed by RuBisCO (Roeske and O’Leary, 1984). It can be speculated that any form of CCM had been disabled due to trace metal limitation (e.g. Zn, Co and Cd, see Lane and Morel, 2000) or perhaps a high CO2 concentration (for whatever reason). This, together with slow growth and non-limiting diffusion of CO2 into the cells could perhaps have led to the anomalously low δ13C values measured. Because no [CO2(aq)]or trace metal measurements were done, it remains an unresolved phenomenon.

3.5.11 Synthesis and application Data presented here show that the δ13C of POM displays a strong gradient with values decreasing 5-6‰ when moving only 80km away from the shore. A similar transition was found by Fry and Bates off the coast of Washington (unpublished data presented here in Figure 3.15). The higher

13

C/12C ratios close to the coast can be ascribed to several

processes. Coastal waters experience a higher flux of nutrients due to the proximity of terrestrial runoff, estuarine entrainment and, in these cases, episodic upwelling. The higher availability of nutrients can sustain higher primary productivity and phytoplankton growth rates. This will increase the

13

C/12C ratio of algal cells due to the growth rate

effect (see above discussion), but also by potentially raising the pH and reducing the [CO2(aq)] in surface waters. The reduced CO2(aq) concentration can either raise the δ13C of algae due to less CO2 entering the cell by diffusion, or by a resulting increase in use of CCMs. Additionally, growth rates are potentially reduced further offshore due to deeper mixed layer depths, causing cells to pass through low irradiance conditions regularly. As outlined, it is often found that the conditions close to the coast are often more favourable for phytoplankton with larger cell size. This effect may also contribute to the high δ13C values measured for POM collected close to the coast. The presented data could not confirm that diatoms were more enriched in 13C than species from other taxa. However, diatom favourable conditions appeared to occur much more often in the

113 proximity of the shore. Whereas nanoflagellates were dominant further offshore. This is also reflected in the fatty acid composition of the seston.

Whether the

13

C/12C ratio is a good food web tracer, that can distinguish between

animals that fed close to the coast or in the open ocean, essentially depends on the spatial patterns of the δ13C of POM. Above all, a clear difference and a steep gradient should exist between the two environments. From the data and discussion above it follows that a steep gradient exists and that, especially during spring and summer months, there are processes in place that help maintain higher δ13C values of POM close to the coast. Hence, the 13C/12C ratio can give a good indication of whether animals have been feeding mostly close to the coast or further offshore. However, several factors will have to be taken into account when using 13C/12C ratios of animals for this purpose. Some patchiness exists in the spatial distribution of the δ13C values of POM. This may not have to be a problem due to time-averaging of the body composition of the animals grazing on the POM. Though, sometimes more “shelf-like conditions” can persist for longer times off the shelf and vice versa. In the case of Vancouver Island’s coastal waters this seems to happen especially in the north where the continental shelf is very narrow. Therefore, it is necessary to always compare the δ13C of the animal with that of the POM from the same location. Only when the 13C/12C ratio of the animal is significantly different from that of the POM (and possibly any other organisms) can any movement of the animal be hypothesized. Naturally, it has to be taken into account that with each trophic step some enrichment in

13

C can occur (e.g., DeNiro and Epstein, 1978; Peterson and Fry, 1987;

France and Peters, 1997). It was found here that, in the region sampled, the chlorophyll a concentration correlated with the δ13C of POM. Most stations that had POM with relatively high δ13C values (-20 to -17‰) exhibited chl a concentrations of at least 2 mg/m3 (Figure 3.6b). Therefore, as a first assessment to find out whether two regions would have potentially different δ13C (and fatty acid) signals, SeaWiFS satellite imagery of ocean colour may in fact give a good indication.

114 The fatty acid composition is a powerful tool to distinguish food sources (GrahlNielsen and Mjaavatten, 1991; Iverson et al., 1997; Kirsch et al., 1998), and these data can complement the isotope data very well as a combined food web tracer (Chapter 2). For example, POM from the open ocean can have δ13C values that are quite similar to terrestrial organic matter, but the fatty acid composition can distinguish the two. Therefore, it is especially the combination of stable carbon isotope- and fatty acid data that may provide valuable insights in spatial exchange between food webs.

3.6 Conclusions PCA of fatty acid abundances in POM could characterize the samples due to varying contributions of diatom-, dinoflagellate-, nanoflagellate-, or detritus/bacterial-derived matter. It was found that the 16:2n-4 fatty acid is a good diatom marker. The differences observed in the fatty acid composition between the shelf- and the off shelf food webs (Chapter 2) seem to be mostly driven by the different contribution of diatom-derived material in the seston. There is no indication that land-derived organic matter significantly contributed to the POM sampled for this study. The measured variable that correlated best with the δ13C of POM was the abundance of the 16:2n-4 fatty acid. With the help of

13

C/12C ratio measurements of

individual fatty acids it is concluded that the δ13C – 16:2n-4 abundance relation was not caused by mixing of 13C-rich diatom- and 13C-poor non-diatom end-members. Instead, it is suggested that diatoms seem to thrive better in “13C-enriching conditions”. In particular, high nutrient supply rate (i.e., not necessarily high concentration) and a high daily average photon flux density are believed to be important factors in creating these conditions. During all three cruises the POM from shelf waters had higher

13

C/12C ratios than

further offshore. However, in May 1998 the difference was larger due to lower δ13C values measured in off shelf waters. The 13C-depletion in off-shelf POM is believed to be caused by lower nutrient availability and a deeper mixed layer, causing the phytoplankton to daily pass through deeper and darker water. These conditions may have resulted in

115 slower growth and reduced ability to employ carbon concentrating mechanisms, and hence, less discrimination against 13C during carbon fixation. It was argued that POM in continental shelf waters can be expected to be systematically more enriched in

13

C than POM from further offshore. The average cell

size and growth rate of phytoplankton are generally higher in shelf environments. Also, CCMs may be more employed by algae in this environment due to the higher nutrient flux, higher light availability due to a shallower mixed layer, and lower [CO2(aq)] during times of high productivity. For these reasons a difference in δ13C of animals feeding close to shore and individuals feeding further off shore are expected to be a common phenomenon.

3.7 References Beardall. J., Johnston, A. and Raven, J. 1998. Environmental regulation of CO2-concentrating mechanisms in microalgae. Canadian Journal of Botany 76, 1010-1017. Ben-Amotz, A., Fishler, R. and Schneller, A. 1987. Chemical composition of dietary species of marine unicellular algae and rotifers with emphasis on fatty acids. Marine Biology 95, 31-36. Bentaleb, I., Fontugne, M., Descolas-Gros, C., Girardin, C., Mariotti, A., Pierre, C., Brunet, C. and Poisson, A. 1998. Carbon isotopic fractionation by plankton in the southern Indian Ocean: relationship between δ13C of particulate organic carbon and dissolved carbon dioxide. Journal of Marine Systems 17, 39-58. Berman-Frank, I., Erez, J. and Kaplan, A. 1998. Changes in inorganic carbon uptake during the progression of a dinoflagellates bloom in a lake ecosystem. Canadian Journal of Botany 76, 1043-1051. Blasco, D. 1977. Red tide in the upwelling region of Baja California. Limnology and Oceanography 22, 255-263. Brussaard, C.P.D., Gast, G.J., Van Duyl, F.C. and Riegman, R. 1996. Impact of phytoplankton bloom magnitude on a pelagic microbial food web. Marine Ecology Progress Series 144, 211-221. Budge, S.M. and Parrish, C.C. 1999. Lipid class and fatty acid composition of Pseudo-nitzschia multiseries and Pseudo-nitzschia pungens and effects of lipolytic enzyme deactivation. Phytochemistry 52, 561-566.

116 Burkhardt, S., Riebesell, U. and Zondervan, I. 1999. Effects of growth rate, CO2 concentration, and cell size on the stable carbon isotope fractionation in marine phytoplankton. Geochimica et Cosmochimica Acta 63, 3729-3741. Burkhardt, S., Amoroso, G., Riebesell, U. and Sultemeyer, D. 2001. CO2 and HCO3- uptake in marine diatoms acclimated to different CO2 concentrations. Limnology and Oceanography 46, 1378-1391. Canuel, E.A., Cloern, J.E., Ringelberg, D.B., Guckert, J.B., and Rau, G.H. 1995. Molecular and isotopic tracers used to examine sources of organic matter and its incorporation into the food webs of San Francisco Bay. Limnology and Oceanography 40, 67-81. Canuel, E.A., Freeman, K.H. and Wakeham, S.G. 1997. Isotopic compositions of lipid biomarker compounds in estuarine plants and surface sediments. Limnology and Oceanography 42, 1570-1583. Canuel, E.A. 2001. Relations between river flow, primary production and fatty acid composition of particulate organic matter in San Francisco and Chesapeake Bays: a multivariate approach. Organic Geochemistry 32, 563-583. Christensen, J.P. 1994. Carbon export from continental shelves, denitrification and atmospheric carbon export. Continental Shelf Research 14, 547-576. Chuecas, L. and Riley, J.P. 1969. Component fatty acids of the total lipids of some marine phytoplankton. Journal of the Marine Biological Association of the United Kingdom 49, 97-116. Cifuentes, L.A., Sharp, J.H. and Fogel, M.L. 1988. Stable carbon and nitrogen isotope biogeochemistry in the Delaware estuary. Limnology and Oceanography 33, 1102-1115. Crawford, W.R. and Dewey, R.K. 1989. Turbulence and mixing: sources of nutrients on the Vancouver Island continental shelf. Atmosphere-Ocean 27, 428-442. Degens, E.T., Guillard, R.R.L., Sackett, W.M. and Hellebust, J.A. 1968. Metabolic fractionation of carbon isotopes in marine plankton – I. Temperature and respiration experiments. Deep-Sea Research 15, 1-9. DeNiro, M.J. and Epstein, S. 1978. Influence of diet on the distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta 42, 495-506. DFO Science, 2000. 1999 Pacific region state of the ocean. Ocean status report 2000/01, 38 pp.. Dunstan, G.A., Volkman, J.K., Barrett, S.M., Leroi, J.M. and Jeffrey, S.W. 1994. Essential polyunsaturated fatty acids from 14 species of diatom (Bacillariophyceae). Phytochemistry 35, 155-161. Egge, J.K. 1998. Are diatoms poor competitors at low phosphate concentrations? Journal of Marine Systems 16, 191-198.

117

Egge, J.K. and Aksnes, D.L. 1992. Silicate as regulating nutrient in phytoplankton competition. Marine Ecology Progress Series 83, 281-289. Falkowski, P.G. 1991. Species variability in the fractionation of Journal of Plankton Research (Suppl.) 13, 21-28.

13

C and

12

C by marine phytoplankton.

Fang, J., Abrajano, T.A., Comet, P., Brooks, J.M. and MacDonald, I. 1993. Gulf of Mexico hydrocarbon seep communities: XI – carbon isotopic fractionation during fatty acid biosynthesis of seep organisms and its implications for chemosynthetic processes. Chemical geology 109, 271-279. Farmer, D.M., McNeil, C.L. and Johnson, B.D. 1993. Evidence for the importance of bubbles in increasing air-sea gas flux. Nature 361, 620-623. Fielding, A.S., Turpin, D.H., Guy, R.D., Calvert, S.E., Crawford, D.W. and Harrison, P.J. 1998. Influence of the carbon concentrating mechanism on carbon stable isotope discrimination by the marine diatom Thalassiosira pseudonana. Canadian Journal of Botany 76, 1098-1103. France, R.L. and Peters, R.H. 1997. Ecosystem differences in the trophic enrichment of 13C in aquatic food webs. Canadian Journal of Fisheries and Aquatic Sciences 54, 1255-1258. Francois, R., Altabet, M.A. and Goericke, R. 1993. Changes in the δ13C of surface water particulate organic matter across the subtropical convergence in the SW Indian Ocean. Global Biogeochemical Cycles 7, 627644. Freeland, H.J. and Denman, K.L. 1982. A topographically controlled upwelling center off southern Vancouver Island. Journal of marine research 40, 1069-1093. Freeland, H.J., Crawford, W.R. and Thomson, R.E. 1984. Currents along the Pacific coast of Canada. Atmosphere-ocean 22, 151-172. Freeland, H., Denman, K., Wong, C.S., Whitney, F. and Jacques, R. 1998. Evidence of change in the winter mixed layer in the Northeast Pacific Ocean. Deep-Sea Research I 44, 2117-2129. Freeman, K.H. and Hayes, J.M. 1992. Fractionation of carbon isotopes by phytoplankton and estimates of ancient CO2 levels. Global Biogeochemical Cycles 6, 185-198. Fry, B. and Sherr, E.B. 1984. δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contributions in Marine Science 27, 13-47. Fry B. and Wainright, S.C. 1991. Diatom sources of 13C-rich carbon in marine food webs. Marine Ecology Progress Series 76, 149-157.

118

Gill, A.E. 1982. Atmosphere-ocean dynamics. International Geophysics Series 30, Academic Press, San Diego, CA, p. 426. Goericke, R. and Fry, B. 1994. Variations of marine plankton δ13C with latitude, temperature, and dissolved CO2 in the world ocean. Global Biogeochemical Cycles 8, 85-90. Goericke, R., Montoya, J.P. and Fry, B. 1994. Physiology of isotope fractionation in algae and cyanobacteria. In: K. Lajtha and B. Michener (eds.), Stable Isotopes in Ecology and Environmental Science. Blackwell Scientific Publications, Oxford, p.187-221. Graeve, M., Kattner, G. and Hagen, W. 1994. Diet-induced changes in the fatty acid composition of Arctic herbivorous copepods: experimental evidence of trophic markers. Journal of experimental marine biology and ecology 182, 97-110. Grahl-Nielsen, O. and Mjaavatten, O. 1991. Dietary influence on fatty acid composition of blubber fat of seals as determined by biopsy: a multivariate approach. Marine Biology 110, 59-64. Gu, B., Alexander, V. and Schell, D.M. Seasonal and interannual variability of plankton carbon isotope ratios in a subarctic lake. Freshwater Biology 42, 417-426. Harris, S.L. 2001. Size-fractionated chlorophyll and primary productivity and nutrient distributions off the west coast of Vancouver Island. M.Sc. thesis, University of British Columbia, 196 pp. Harvey, H.R., Eglinton, G., O’Hara, S.C.M. and Corner, E.D.S. 1987. Biotransformation and assimilation of dietary lipids by Calanus feeding on a dinoflagellate. Geochimica et Cosmochimica Acta 51, 3031-3040. Hemminga, M.A. and Mateo, M.A. 1996. Stable carbon isotopes in seagrasses: variability in ratios and use in ecological studies. Marine Ecology Progress Series 140, 285-298. Hinga, K.R., Arthur, M.A., Pilson, M.E.Q. and Whitaker, D. 1994. Carbon isotope fractionation by marine phytoplankton in culture: The effects of CO2 concentration, pH, temperature, and species. Global Biogeochemical Cycles 8, 91-102. Hobson, L.A., Hanson, C.E. and Holeton, C. 2001. An ecological basis for extracellular carbonic anhydrase in marine unicellular algae. Journal of phycology 37, 717-723. Huertas, I.E., Espie, G.S., Colman, B. and Lubian, L.M. 2000. Light-dependent bicarbonate uptake and CO2 efflux in the marine microalga Nannochloropsis gaditana. Planta 211, 43-49. Ianson, D., Harris, S. Allen, S.E., Orians, K., Varela, D.E. and Wong, C.S. The inorganic carbon system in the upwelling region west of Vancouver Island, Canada. In preparation.

119 Iverson S.J., Frost K.J. and Lowry L.F. 1997. Fatty acid signatures reveal fine scale structure of foraging distribution of harbor seals and their prey in Prince William Sound, Alaska. Marine Ecology Progress Series 151, 255-271. Johns, R.B., Perry, G.J. and Jackson, K.S. 1977. Contribution of bacterial lipids to recent marine sediments. Estuarine and Coastal Marine Science 5, 521-529. Johnson, H.P., Hutnak, M., Dziak, R.P., Fox, C.G., Urcuyo, I., Cowen, J.P., Nabelek, J. and Fisher, C. 2000. Earthquake-induced changes in a hydrothermal system on the Juan de Fuca mid-ocean ridge. Nature 407, 174-177. Kaplan, A. and Reinhold, L. 1999. CO2 concentrating mechanisms in photosynthetic microorganisms. Annual review of plant physiology and plant molecular biology 50, 539-570. Kara, A.B., Rochford, P.A. and Hurlburt, H.E. 2000. An optimal definition for ocean mixed layer depth. Journal of Geophysical Research 105, 16803-16821. Kattner, G. and Brockmann, U.H. 1990. Particulate and dissolved fatty acids in an enclosure containing a unialgal Skeletonema costatum (Greve.) Cleve culture. Journal of Experimental Marine Biology and Ecology 141, 1-13. Keller, K. and Morel, F.M.M. 1999. A model of carbon isotopic fractionation and active carbon uptake in phytoplankton. Marine Ecology Progress Series 182, 295-298. Kharlamenko, V.I., Kiyashko, S.I., Imbs, A.B. and Vyshkvartzev, D.I. 2001. Identification of food sources of invertebrates from the seagrass Zostera marina community using carbon and sulfur stable isotope ratio and fatty acid analyses. Marine Ecology Progress Series 220, 103-117. Kieber, R.J., Hydro, L.H. and Seaton, P.J. 1997. Photooxidation of triglycerides and fatty acids in seawater: Implication toward the formation of marine humic substances. Limnology and Oceanography 42, 14541462. Kirsch, P.E., Iverson, S.J., Bowen, W.D., Kerr, S.R. and Ackman, R.G. 1998. Dietary effects on the fatty acid signature of whole Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 55, 1378-1386. Kopczyńska, E.E., Goeyens, L., Semeneh, M. and Dehairs, F. 1995. Phytoplankton composition and cell carbon distribution in Prydz Bay, Antarctica: relation to organic particulate matter and its δ13C values. Journal of Plankton Research 17, 685-707. Kukert, H. and Riebesell, U. 1998. Phytoplankton carbon isotope fractionation during a diatom spring bloom in a Norwegian fjord. Marine Ecology Progress Series 173, 127-137.

120 Lane, T.W. and Morel, F.M.M. 2000. Regulation of carbonic anhydrase expression by zinc, cobalt, and carbon dioxide in the marine diatom Thalassiosira weissflogii. Plant Physiology 123, 345-352. Larsson C. and Axelsson, L. 1999. Bicarbonate uptake and utilization in marine macroalgae. European Journal of Phycology 34, 79-86. Laws, E.A., Popp, B.N., Bidigare, R.R., Kennicutt, M.C. and Macko, S.A. 1995. Dependence of phytoplankton carbon isotopic composition on growth rate and [CO2]aq: Theoretical considerations and experimental results. Geochimica et Cosmochimica Acta 59, 1131-1138. Laws, E.A., Bidigare, R.R. and Popp, B.N. 1997. Effect of growth rate and CO2 concentration on carbon isotopic fractionation by the marine diatom Phaeodactylum tricornutum. Limnology and Oceanography 42, 1552-1560. Leo, R.G. and Parker, P.L. 1966. Branched chain fatty acids in sediments. Science 152, 649-650. Levitus, S. 1982. Climatological atlas of the world ocean. NOAA Professional Paper 13, U.S. Government Printing Office, Washington, D.C., 173 pp. Lovell, C.R., Bagwell, C.E., Czákó, M., Márton, L., Piceno, Y.M. and Ringelberg, D.B. 2001. Stability of a rhizosphere microbial community exposed to natural and manipulated environmental variability. FEMS Microbiology Ecology 38, 69-76. Lupton, J.E., Delaney, J.R., Johnson, H.P. and Tivey, M.K. 1985. Entrainment and vertical transport of deep-ocean water by buoyant hydrothermal plumes. Nature 316, 621-623. Mackas, D.L. 1992. Seasonal cycle of zooplankton off southwestern British Columbia: 1979-89. Canadian Journal of Fisheries and Aquatic Sciences 49, 903-921. Malone, T.C. 1980. Algal size. In: I. Morris (ed.), Physiological ecology of phytoplankton, Blackwell Scientific Publications, Oxford, p. 433-463. Mansour, M.P., Volkman, J.K., Jackson, A.E. and Blackburn, S.I. 1999. The fatty acid and sterol composition of five marine dinoflagellates. Journal of Phycology 35, 710-720. Margalef, R. 1978. Life-forms of phytoplankton as survival alternatives in an unstable environment. Oceanologica Acta 1, 493-509. Meglen, R.R. 1992. Examining large databases: a chemometric approach using principal component analysis. Marine Chemistry 39, 217-237. Mook, W.G., Bommerson, J.C. and Staverman, W.M. 1974. Carbon isotope fractionation between dissolved bicarbonate and gaseous carbon dioxide. Earth and Planetary Science Letters 22, 169-176.

121

Moroney, J.V. 2001. Carbon concentrating mechanisms in aquatic photosynthetic organisms: A report on CCM 2001. Journal of Phycology 37, 928-931. Nakatsuda, T., Handa, N., Wada, E. and Wong, C.S. 1992. The dynamic changes of stable isotope ratios of carbon and nitrogen in suspended and sedimented particulate organic matter during a phytoplankton bloom. Journal of Marine Research 50, 267-296. Napolitano, G.E., Pollero, R.J., Gayoso, A.M., MacDonald, B.A. and Thompson, R.J. 1997. Fatty acids as trophic markers of phytoplankton blooms in the Bahía Blanca estuary (Buenos Aires, Argentina) and in Trinity Bay (Newfoundland, Canada). Biochemical Systematics and Ecology 25, 739-755. Nimer, N.A., Iglesias-Rodriguez, M.D. and Merrett, M.J. 1997. Bicarbonate utilization by marine phytoplankton species. Journal of Phycology 33, 625-631. Opute, F.I. 1974. Lipid and fatty-acid composition of diatoms. Journal of Experimental Botany 25, 823835. Pancost, R.D., Freeman, K.H., Wakeham, S.G. and Robertson, C.Y. 1997. Controls on carbon isotope fractionation by diatoms in the Peru upwelling region. Geochimica et Cosmochimica Acta 61, 4983-4991. Pancost, R.D., Freeman, K.H. and Wakeham, S.G. 1999. Controls on the carbon-isotope composition of compounds in Peru surface waters. Organic Geochemistry 30, 319-340. Parsons, T., Harrison, P.J. and Waters, R. 1978. An experimental simulation of changes in diatoms and flagellate blooms. Journal of Experimental Marine Biology and Ecology 32, 285-294. Perry, G.J., Volkman, J.K., Johns, R.B. and Bavor, H.J. 1979. Fatty acids of bacterial origin in contemporary marine sediments. Geochimica et Cosmochimica Acta 43, 1715-1725. Perry, R.I., Thompson, P.A., Mackas, D.L., Harrison, P.J. and Yelland, D. 1999. Stable carbon isotopes as pelagic food web tracers in adjacent shelf and slope regions off British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 56, 2477-2486. Peterson B.J. and Fry B. 1987. Stable isotopes in ecosystem studies. Annual review of ecology and systematics 18, 293-320. Piveteau, F., Gandemer, G., Baud, J.-P. and Demaimay, M. 1999. Changes in lipid and fatty acid compositions of European oysters fattened with Skeletonema costatum diatom for six weeks in ponds. Aquaculture International 7, 341-355.

122 Pohl, P. and Zuhrheide, F. 1979. Fatty acids and lipids of marine algae and the control of their biosynthesis by environmental factors. In: Hoppe, H.A., Levring, T. and Tanaka, Y. (eds) Marine algae in pharmaceutical science. Walter de Gruyer, Berlin, 473-523. Pollingher, U. and Zemel, E. 1981. In situ and experimental evidence of the influence of turbulence on cell division processes of Peridinium cinctum forma westii (Lemm.) Lefevre. British Phycological Journal 16, 281-287. Popp, B.N., Laws, E.A., Bidigare R.R., Dore, J.E., Hanson, K.L. and Wakeham, S.G. 1998. Effect of phytoplankton cell geometry on carbon isotopic fractionation Geochimica et Cosmochimica Acta 62, 69-77. Prahl, F.G., Eglinton, G., Corner, E.D.S., O’Hara, S.C.M. and Forsberg, T.E.V. 1984. Changes in plant lipids during passage through the gut of Calanus. Journal of the Marine Biological Association of the United Kingdom 64, 317-334. Rau, G.H., Takahashi, T. and Des Marais, D.J. 1989. Latitudinal variations in plankton δ13C: implications for CO2 and productivity in past oceans. Nature 341, 516-518. Rau, G.H., Teyssie, J.-L., Rassoulzadegan, F. and Fowler, S.W. 1990. 13C/12C and 15N/14N variations among size-fractionated marine particles: implications for their origin and trophic relationships. Marine Ecology Progress Series 59, 33-38. Rau, G.H., Riebesell, U. and Wolf-Gadrow, D. 1997. CO2aq-dependent photosynthetic 13C fractionation in the ocean: A model versus measurements. Global Biogeochemical Cycles 11, 267-278. Rau, G.H., Chavez, F.P. and Friederich, G.E. 2001. Plankton 13C/12C variations in Monterey Bay, California: evidence of non-diffusive inorganic carbon uptake by phytoplankton in an upwelling environment. Deep-Sea Research I 48, 79-94. Reemtsma, T. and Ittekkot, V. 1992. Determination of factors controlling the fatty acid composition of settling particles in the water column by principal-component analysis and their quantitative assessment by multiple regression. Organic geochemistry 18, 121-129. Reuss, N. and Poulsen, L.K. 2002. Evaluation of fatty acids as biomarkers for a natural plankton community. A field study of a spring bloom and a post-bloom period off west Greenland. Marine Biology 141, 423-434. Rhead, M.M., Eglinton, G., Draffan, G.H. and England, P.G. 1971. Conversion of oleic acid to saturated fatty acids in Severn estuary sediments. Nature 232, 327-330. Riebesell, U. and Wolf-Gadrow, D. 1995. Growth limits on phytoplankton. Nature 373, 28.

123 Riegman, R., Kuipers, B.R., Noordeloos, A.A.M. and Witte, H.J. 1993. Size-differential control of phytoplankton and the structure of plankton communities. Netherlands Journal of Sea Research 31, 255265. Roeske, C.A. and O’Leary, M.H. 1984. Carbon isotope effects on the enzyme catalyzed carboxylation of ribulose bisphosphate. Biochemistry 23, 6275-6284. Sargent, J. R., Parkes, R. J., Mueller-Harvey, I, and Henderson, R. J. 1988. Lipid biomarkers in marine ecology. In: M. A. Sleigh (ed.), Microbes in the sea. Ellis Horwood, Ltd. Chichester, U.K., p.119-138. Sciandra, A. 1991. Coupling and uncoupling between nitrate uptake and growth rate in Prorocentrum minimum (Dinophyceae) under different frequencies of pulsed nitrate supply. Marine Ecology Progress Series 72, 261-269. Sharkey, T.D. and Berry, J.A. 1985. Carbon isotope fractionation in algae as influenced by inducible CO2 concentrating mechanism. In: W.J. Lucas and J.A. Berry (eds.), Carbon uptake by aquatic photosynthetic organisms. American Society of Plant Physiology, Rockville, p. 389-401. Smayda, T.J. 1997. Harmful algal blooms: Their ecophysiology and general relevance to phytoplankton blooms in the sea. Limnology and Oceanography 42, 1137-1153. St. John, M.A. and Lund, T. 1996. Lipid biomarkers: linking the utilization of frontal plankton biomass to enhanced condition of juvenile North Sea cod. Marine Ecology Progress Series 131, 75-85. Stolte, W. and Riegman, R. 1995. Effect of phytoplankton cell-size on transient-state nitrate and ammonium uptake kinetics. Microbiology – U.K. 141, 1221-1229. Teece, M.A., Fogel, M.L., Dollhopf, M.E. and Nealson, K.H. 1999. Isotopic fractionation associated with biosynthesis of fatty acids by a marine bacterium under oxic and anoxic conditions. Organic Geochemistry 30, 1571-1579. Thomas, W.H., Dodson, A.N. and Reid, F.M.H. 1978. Diatom productivity compared to other algae in natural marine phytoplankton assemblages. Journal of Phycology 14, 250-253. Thompson, P.A., Guo, M., Harrison, P.J. and Whyte, J.N.C. 1992. Effects of variation in temperature. II. On the fatty acid composition of eight species of marine phytoplankton. Journal of phycology 28, 488-497. Thomson, R.E., Hickey, B.M. and Leblond, P.H. 1989. The Vancouver Island coastal current: fisheries barrier and conduit. In: R.J. Beamish and G.A. McFarlane (eds.) Effects of ocean variability on recruitment and an evaluation of parameters used in stock assessment models. Canadian Special Publication of Fisheries and Aquatic Sciences 108, p. 265-296. Thomson, R.E. and Ware, D.M. 1996. A current velocity index of ocean variability. Journal of geophysical research 101, 14297-14310.

124

Tortell, P.D., Rau, G.H. and Morel, F.M.M. 2000. Inorganic carbon acquisition in coastal Pacific phytoplankton communities. Limnology and Oceanography 40, 1485-1500. Turpin, D.H. and Harrison, P.J. 1980. Cell size manipulation in natural marine, planktonic, diatom communities. Canadian Journal of Fisheries and Aquatic Sciences 37, 1193-1195. Utermöhl, H. 1958. Zur Verollkommnung der quantitativen Phytoplankton-Methodik. Mitteilungen der Internationalen Vereinigung für theoretische und angewandte Limnologie 9: 1-38. Villinski, J., Dunbar, R.B. and Mucciarone, D.A. 2000. Carbon 13/carbon 12 ratios of sedimentary organic matter from the Ross Sea, Antarctica: A record of phytoplankton bloom dynamics. Journal of Geophysical Research 105, 14163-14172. Viso, A.-C. and Marty, J.-C. 1993. Fatty acids from 28 marine microalgae. Phytochemistry 34, 1521-1533. Volkman, J.K., Johns, R.B., Gillan, F.T., Perry, G.J. and Bavor, Jr., H.J. 1980. Microbial lipids of an intertidal sediment. I. Fatty acids and hydrocarbons. Geochimica et Cosmochimica Acta 44, 1133-1143. Volkman, J.K., Smith, D.J., Eglinton, G., Forsberg, T.E.V. and Corner, E.D.S. 1981. Sterol and fatty acid composition of four marine Haptophycean algae. Journal of the Marine Biological Association of the United Kingdom 61, 509-527. Volkman, J.K., Jeffrey, S.W., Nichols, P.D., Rogers, G.I. and Garland, C.D. 1989. Fatty acid and lipid composition of 10 species of microalgae used in mariculture. Journal of Experimental Marine Biology and Ecology 128, 219-240. Wood, B.J.B. 1974. Fatty acids and saponifiable lipids. In: W.D.P. Stewart (ed.), Algal physiology and biochemistry. Botanical Monographs 10, Blackwell Scientific Publications, Edinburgh, p. 236-265. Wong, W.W. and Sackett, W.M. 1978. Fractionation of stable carbon isotopes by marine phytoplankton. Geochimica et Cosmochimica Acta 42, 1809-1815. Young, E., Beardall, J. and Giordano, M. 2001. Inorganic carbon acquisition by Dunaliella tertiolecta (Chlorophyta) involves external carbonic anhydrase and direct HCO3- utilization insensitive to the anion exchange inhibitor DIDS. European Journal of Phycology 36, 81-88. Zhang, J.A., Quay, P.D. and Wilber, D.O. 1995. Carbon-isotope fractionation during gas-water exchange and dissolution of CO2. Geochimica et Cosmochimica Acta 59, 107-114.

125

Appendix

H0: Correlation coef.=0

Salinity

-0.65

0.63

1

-0.37 -0.02 -0.18

1

0.26

1

0.45

0.29

-0.42

0.24

0.19

0.55

1

-0.10

0.21

0.06

0.01

-0.02

0.05

0.67

0.26

1

-0.87

0.66

0.53

0.20

0.43

-0.43 -0.03

0.19

0.15

1

0.64

-0.60 -0.40 -0.42 -0.33

0.58

0.02

0.02

-0.20 -0.62

0.03

0.24

0.04

0.14

0.38

0.34

0.47

-0.53 -0.33 -0.04 -0.25

0.29 -0.07 -0.13 -0.22 -0.61

0.84

-0.21

1

0.88

-0.69 -0.54 -0.50 -0.48

0.78

0.23

0.07

-0.01 -0.74

0.81

0.03

0.57

1

0.76

-0.60 -0.58 -0.43 -0.48

0.93

0.31

0.21

0.01

-0.56

0.69

0.01

0.39

0.86

1

-0.81

0.61

0.54

0.18

0.50

-0.57 0.10

0.08

0.31

0.82

-0.72

0.24 -0.70 -0.77 -0.67

1

-0.85

0.58

0.47

0.38

0.32

-0.45 0.07

0.12

0.18

0.83

-0.67 -0.05 -0.61 -0.71 -0.55

0.80

1

-0.82

0.53

0.41

0.37

0.34

-0.47 -0.11

0.08

0.06

0.84

-0.61 -0.04 -0.59 -0.69 -0.58

0.77

0.90

1

-0.15

0.17

0.12 -0.12 -0.24

0.17

0.37

0.35

0.08

0.30

-0.08 -0.15 -0.35

0.07

0.15

0.13

0.38

0.43

1

20:5n3

0.73

-0.62 -0.65 -0.22 -0.55

0.66

0.19

0.04

-0.08 -0.67

0.68

-0.33

0.49

0.79

0.78

-0.75 -0.65 -0.58

0.20

1

22:6n3

0.20

-0.44 -0.33 0.21

0.27 -0.11 -0.17 -0.18 -0.15

0.27

-0.43

0.13

0.27

0.27

-0.25 -0.11 0.11

0.46

0.52

1

0.27

-0.31

18:4n3

1

-0.52

18:1n9

-0.03

0.32

16:2n4

1

0.07

18:3n3

Si(OH)4 Discr. score 16:1n7/16:0 iso 15:0 16:1n7 16:2n4 16:4n1 18:1n9 18:2n6 18:3n3 18:4n3 20:5n3 22:6n3

18:2n6

-0.15

16:4n1

0.09

16:1n7

0.08

HPO42-

iso 15:0

-0.57 -0.55 -0.39 -0.45

16:1n7/16:0

-0.05 -0.21 -0.15

0.69

NO3-

Si(OH)4

1

Discr. score

0.55 -0.05

HPO42-

0.52

NO3-

-0.57

1

Chl a (0m)

1

MLD

-0.76

P<0.001 Temp.

1

Dis. to shore

Bulk δ13C Bulk δ13C Dis. to shore Salinity Temp. MLD Chl a (0m)

P<0.01

Appendix 1. Correlation matrix showing the correlation coefficients (R) for each pair of variables, with negative correlations showing up as negative numbers. For the construction of the matrix 30 POM samples (for which all variables were measured) were used. For the R-values highlighted with dark grey there is 99.9% confidence that the coefficient is in fact smaller or higher than zero. The values highlighted with lighter grey have a confidence level of 99% (F-test).

126

4. Stable Carbon Isotope Ratios of Individual Fatty Acids in Marine Pelagic Organisms

4.1 Abstract The stable carbon isotope ratios of individual fatty acids can offer more detailed, time integrated information on the dietary intake of an organism than the

13

C/12C ratio of

whole samples. To test the applicability in ecological studies, and to advance the understanding of the variation in

13

C/12C ratios between individual fatty acids, stable

carbon isotope ratios of fatty acids from almost 200 samples of marine pelagic organisms and organic matter were measured. These samples consisted of particulate organic matter (POM), zooplankton, larval fish and juvenile salmon from off the west coast of Vancouver Island, Canada. It was anticipated that the δ13C of essential fatty acids in organisms feeding at the higher trophic positions would best reflect the dietary signal. However, this was not found to be necessarily the case. For example, the δ13C of the C14 saturated fatty acid in zooplankton and larval fish often showed closer agreement with its counterpart in POM. A large range (typically about 7‰) in the δ13C values of fatty acids was observed within single samples. The variation in the δ13C between the individual fatty acids was found to be reproducible. The same pattern was also discovered in values reported in other studies. This suggests the presence of common underlying mechanisms, most likely biosynthetic effects, producing the semi-predictable offsets between the δ13C of fatty acids. The same reproducibility observed in the δ13C differences between fatty acids in POM is also observed in the zooplankton, larval fish and juvenile salmon. Therefore, the recurring pattern in the δ13C differences among individual fatty acids in the primary producers is largely passed on to higher organisms through feeding. Mechanisms that could produce the large stable carbon isotope fractionations between fatty acids are evaluated. The factors that can potentially produce the largest carbon isotope fractionation seem to be desaturation of the fatty acid, different timing of lipid class

127 synthesis during the diurnal growth cycle of autotrophs, and perhaps also the proportion of polyunsaturated fatty acids (PUFAs) synthesized via an alternative (polyketide synthase-catalyzed) pathway. The extra information provided by the δ13C values of fatty acids is believed to be especially beneficial in identifying whether an organism has had a recent shift in its diet. Fatty acid δ13C values can be of great value in well-outlined source identification problems. Finally, 13C/12C ratios of fatty acids can give information on the importance of various biosynthetic pathways (e.g., the use of the polyketide synthase-pathway to produce PUFAs).

4.2 Introduction The main objective of this study was to examine the potential of the δ13C of essential fatty acids as a food web tracer. When a heterotroph is incapable of synthesizing a fatty acid at a sufficient rate to meet its physiological requirements this fatty acid is an essential fatty acid for this particular organism (Arts et al., 2001). Vertebrates, including fish, are known to lack the ability to introduce double bonds between the C9 position and the terminal methyl group of a fatty acid (Henderson and Tocher, 1987; Gurr and Harwood, 1991; Sargent et al., 1995). Therefore, the n-6 and n-3 (also termed ω-6 and ω3) fatty acids cannot be synthesized de novo (i.e., “from scratch”) by these animals, and have to be supplied by dietary intake. Limited synthesis of the long-chain polyunsaturated n-3 and n-6 fatty acids can occur by elongation and desaturation of dietary n-3 and n-6 fatty acid precursors. This limited ability to produce polyunsaturated fatty acids (PUFAs), such as the 20:4n-6, 20:5n-3 and 22:6n-3, from the 18:2n-6 and 18:3n-3 precursors has been shown to be sufficient in maintaining the structural integrity of cell membranes in freshwater fish, but not in marine fish (Henderson and Tocher, 1987; Buzzi et al., 1997 and references therein). In invertebrates, such as slugs, snails and insects, a ∆12 desaturase (i.e., the enzyme responsible for inserting a double bond at the C12 position) has been found (Weinert et al., 1993; Batcabe et al., 2000). This makes it possible for them to produce linoleic acid

128 (18:2n-6) de novo. Very few studies, though, have directly investigated the capability of marine zooplankton animals to synthesize n-6 and n-3 PUFAs. Tracer studies by Moreno et al. (1979a, b) showed that the wild copepod Paracalanus parvus could not produce the n-6 and n-3 PUFAs de novo. But, the copepod did show a capacity for elongation and desaturation of dietary n-6 and n-3 fatty acids. The lack of substantial de novo n-3, and n-6 fatty acid synthesis in the higher trophic levels of marine food webs is more than compensated for by the high abundance of these fatty acids in marine phytoplankton. It is believed that most marine fish have not developed an adequate efficiency to convert C18 n-6 and n-3 fatty acids to C20/C22 PUFAs because the latter are so abundant in their diet (Buzzi et al., 1997). Because of the unaltered transfer, or only slight modification of the n-3 and n-6 PUFAs by organisms from higher trophic levels, the δ13C of these essential fatty acids can be expected to remain unchanged throughout the marine food web. Indeed, Stott et al. (1997) found in a controlled animal feeding study that a particularly close similarity existed between the δ13C of linoleic acid in the diet and bones of pigs. These animals are known to be incapable of de novo synthesis of this fatty acid. The stable carbon isotope composition of fatty acids in organisms from hydrothermal vents has been studied extensively (shrimp: Pond et al., 1997a, b, 2000; Rieley et al., 1999; mussels and worms: Pond et al., 1998, 2002, respectively). Also, several researchers have measured the δ13C of fatty acids from organisms in hydrocarbon seep communities (mytilids and vestimentiferans: Fang et al., 1993; Abrajano et al., 1994; MacAvoy et al., 2002). These studies showed that the δ13C of individual fatty acids can be useful in detecting the contribution of carbon derived from symbionts (see also Uhle et al., 1997). Other applications of this technique include estimating the importance of seagrass as a food source (ducks: Hammer et al., 1998; bacteria: Boschker et al., 1999, 2000). Additionally, researchers have been using the δ13C of fatty acids to differentiate between marine and terrestrial- (foxes: Pond et al., 1995; Gilmour et al., 1995; particulate organic matter: Canuel et al., 1995, 1997), and also C3- and C4 plant sources (vegetable oil: Woodbury et al., 1995).

129 Little attention has been paid to the stable carbon isotope composition of fatty acids in marine pelagic food webs. Parker (1962, 1964) was the first to measure δ13C values for a few individual fatty acids and reported some values for marine plankton mostly comprising euphausiids. Pond et al. (2000) reported δ13C values for microplankton (<200µm) collected from different depths (i.e., 4 and 48-75m, respectively). They found that the δ13C of fatty acids typically has a range of about 7‰ in the microplankton, and no large differences between the samples from the various depths were found. A similar variation in the δ13C of fatty acids from settling particles was observed by Ramos et al. (2003). For this study particulate organic matter, zooplankton, larval fish and juvenile salmon were collected on several cruises off the west coast of Vancouver Island (Canada) during May 1998, May 1999 and July 1999. The results of measurements of the δ13C of fatty acids from almost 200 samples are discussed. A detailed evaluation of the potential metabolic- and biosynthetic effects on the δ13C of fatty acids is included in the discussion.

4.3 Methods 4.3.1 Sample collection and preparation The procedures followed during sample collection have been described in detail in Chapter 2. In Figure 2.1 the locations of sampling are indicated. Most of the juvenile salmon were collected with the CSS W.E. Ricker in waters on the continental shelf off Vancouver Island in May 1998 and 1999 (cruises R98-8155 and HS9913). However, four of the samples reported here were sampled on a line close to the Dixon Entrance (off Northern British Columbia, Canada; between 54º34.56’N 132º20.88’W and 54º27.84’N 132º48.96’W). Additionally, six samples were collected in the proximity of Baranof Island (off Alaska’s “panhandle”, between 56º15.3’N 135º01.5’W and 56º08.16’N 135º17.58’W). Location-maps for the juvenile salmon collected off Vancouver Island can be found in Chapter 7.

130 The juvenile salmon were collected using a trawl net. The nets used had mouth openings that varied from 12 to 16 m deep, and generally the headrope of the net was towed at the surface. This means that, at the moment of capture, the salmon were residing between the surface and approximately 16 metres depth. After determining the species and measuring their size, the juvenile salmon were kept frozen at approximately -20ºC onboard the ship. The salmon from the 1999 cruise were stored at -80ºC upon arrival at the Pacific Biological Station in Nanaimo. However, the 1998 samples were stored at -20ºC for over a year. Muscle tissue from below the dorsal fin was sampled for further analyses. The skin and 1-2mm of sub-dermal muscle tissue was removed from the samples. To test whether any substantial oxidation of the 1998 samples had taken place, the fatty acid profiles of the 1998 and 1999 salmon were compared. The 1998 samples did on average not have a lower PUFA contents than the 1999 samples. Therefore, it was assumed that the storage at -20ºC had not significantly altered the composition of the sampled muscle tissue.

4.3.2 Fatty acid extraction, methyl ester preparation and isotope ratio measurements For the methods followed for the extraction and fatty acid methyl ester preparation, and also the stable carbon isotope ratio measurements, please refer to the methods section in Chapter 2.

4.4 Results and discussion 4.4.1 Trophic transfer As mentioned in the introduction it was anticipated that the δ13C of essential fatty acids (EFAs) would be relatively constant throughout the food web. Due to the limited ability of organisms other than plants to produce EFAs such as 20:5n-3 and 22:6n-3, most of the carbon in these fatty acids can be expected to be derived from the primary producers. In Figure 4.1 the δ13C of fatty acids (δ13CFA) from zooplankton and larval fish samples are

131 compared with the δ13C of identical fatty acids from POM collected at the same stations during the same cruise. Though the median differences between the δ13CFA of POM and the δ13CFA of zooplankton and larval fish (δ13CFA-POM) generally stays within 2‰, large variability exists (see standard deviations shown in Figure 4.1). Also, EFAs like 20:5n-3 and 22:6n-3 in zooplankton and larval fish do not show more similarity in δ13C of POM from the same station than many other fatty acids.

13

13

Median (δ CFA - POM δ CFA) [‰]

5 4

Zoo < 425µm (n=12), small error bar cap Zoo > 425µm (n=22), med. error bar cap Larval fish (n=25), large error bar cap

3 2 1 0 -1 -2 -3

Bulk

Tot FAs*

14:0 15:0 16:0 16:1n-7 16:2n-4 17:0 16:4n-1 18:0 18:1n-9 18:1n-7 18:2n-6 18:3n-3 18:4n-3 20:4n-6 20:5n-3 22:6n-3

-4

Fatty acid Figure 4.1. Differences in δ13C (∆δ13C) of fatty acids from zooplankton and larval fish with those in POM from the same location. The organisms that are suspected to have moved (Table 2.4) were left out. Here the median of those differences is shown ± 1 standard deviation. The error bars with smallest, medium, and largest cap width represent the error for small zooplankton, large zooplankton and larval fish, respectively. The ∆δ13C values of the bulk (i.e., the unextracted whole sample) are also shown for comparison. * The δ13C of the total fatty acids was calculated by computing the weighted average δ13C (using the relative abundances of the fatty acids as weighting factors).

The large variability observed in Figure 4.1 can be explained by several mechanisms. One of the factors that may account for a large part of the variability is fluctuation of the δ13C of POM over time due to varying growth conditions. These

132 fluctuations will be averaged over different time-scales for different animals. Also, different fatty acids will turnover at varying time-scales (Chapter 2). Another factor causing an offset with the δ13C of POM fatty acids is movement of the sampled organisms from waters with POM exhibiting different 13C/12C ratios. For the construction of Figure 4.1 the samples of organisms that were suspected to have moved from the shelf to the off shelf and vice versa (Table 2.4) were left out. However, within each of those regions the δ13C of seston can still vary considerably (e.g., see Figures 2.5 and 3.9).

Alternatively, selectivity in food intake may cause some differences between zooplankton and POM δ13C when variation in the δ13C of the food items exist. Some large phytoplankton cells and colonies, for example, are not available to the zooplankton smaller than 425µm. Additionally, certain zooplankton species (>425µm), such as Calanus finmarchicus, have been found to preferentially select diatoms and dinoflagellates (Meyer-Harms et al., 1999). Therefore, the δ13C of the total POM may not always represent the δ13C of the diet of these zooplankton animals. Finally, some of the variability in Figure 4.1 may be explained by measurement error. For example, for the 18:4n-3 fatty acid, the average of the standard deviations of replicate δ13C measurements amounted to almost 0.5‰ (Table 2.1). Since two measurements were subtracted, the precision estimate for the δ13CFA-POM becomes approximately ±0.7‰, due to the propagation of errors. It is interesting that the δ13C of the saturated C14 fatty acid (14:0) shows the closest similarity between POM, zooplankton and larval fish. Partly, this may be due to the good precision attained for the δ13C measurement on this fatty acid (Table 2.1). Moreover, Sargent and Whittle (1981) state that “while significant catabolism of 14:0 acid certainly occurs at various trophic levels, this acid may be transmitted through the food chain in very significant quantities with little or no additional biosynthetic input beyond phytoplankton”. If this is true, the δ13C of the 14:0 fatty acid would be an ideal trophic marker when sufficient variation in the

13

C/12C ratios of this fatty acid exists between

different algae or algae from different regions.

133 -18 -20

13

δ C of fatty acid (‰)

-22

POM Zooplankton < 425µm Zooplankton > 425µm Sablefish (32-40mm) Sablefish (40-60mm) Sandlance (27-35mm)

LG3, May 1998

-24 -26 -28 -30

Bulk

Tot FAs*

14:0 15:0 16:0 16:1n-7 16:2n-4 17:0 16:4n-1 18:0 18:1n-9 18:1n-7 18:2n-6 18:3n-3 18:4n-3 20:4n-6 20:5n-3 22:5n-3 22:6n3

-32

Fatty acid Figure 4.2. Stable carbon isotope composition of fatty acids and bulk from organisms all collected at station LG3 in May 1998 (for location, see Fig. 2.1). Precision of the measurements of each of the fatty acids is shown in Table 2.1, and the error bars of the bulk δ13C measurements are the size of the symbol. * The δ13C of the total fatty acids was calculated by computing the weighted average δ13C (using the relative abundances of the fatty acids as weighting factors).

Figure 4.2 illustrates the variation observed for the δ13C of fatty acids from various organisms collected at the same station (LG3) and the same time (May 1998). Despite this variability, the difference in 13C/12C ratios frequently observed between shelf and off shelf-food webs is large enough to distinguish between organisms originating from those two environments (Figure 4.3).

4.4.2 Bulk – fatty acid δ13C difference The median differences between the δ13C of the whole sample (bulk) and the various fatty acids are shown in Figure 4.4. The median values were computed for measurements on POM, zooplankton, larval fish and juvenile salmon samples combined from several

134 -14 -16

LBP2 Zoopl. < 425µ m LBP2 Zoopl. > 425µ m LBP2 Sablefish (20mm) LBP7 Zoopl. < 425µ m LBP7 Zoopl. > 425µ m LBP7 Sablefish (10-25mm)

Line BP, May 1999

-18

-22 -24 -26 -28

13

δ C of fatty acid (‰)

-20

-30 -32

Bulk

Tot FAs*

22:6n-3

20:5n-3 22:5n-3

18:4n-3 20:4n-6

18:2n-6 18:3n-3

18:1n-7

18:0 18:1n-9

17:0 16:4n-1

16:2n-4

16:0 16:1n-7

14:0 15:0

-34

-14 -16

Line G, May 1998

LG3 Zoopl. < 425µ m LG3 Zoopl. > 425µ m LG7 Sablefish (32-40mm) LG7 Zoopl. < 425µ m LG7 Zoopl. > 425µ m LG7 Un-id. larval fish

-18

-22 -24 -26 -28

13

δ C of fatty acid (‰)

-20

-30 -32

Bulk

Tot FAs*

22:6n-3

22:5n-3

20:5n-3

20:4n-6

18:4n-3

18:3n-3

18:2n-6

18:1n-7

18:0

18:1n-9

16:4n-1

17:0

16:1n-7

16:0

15:0

14:0

-34

Fatty acid

Figure 4.3. Comparison of the δ13C values of fatty acids (and bulk) from zooplankton and larval fish collected at stations in continental shelf waters (LBP2 and LG3) with those collected off shelf (LBP7 and LG7). Precision of the measurements of each of the fatty acids is shown in Table 2.1, and the precision of the bulk δ13C measurements is the size of the symbol. For location of the stations, please refer to Figure 2.1. * The δ13C of the total fatty acids was calculated by computing the weighted average δ13C (using the relative abundances of the fatty acids as weighting factors).

cruises. As can be seen in Figure 4.4, a large variation exists among the fatty acids in the deviation from the bulk δ13C. Mechanisms that may explain this variability are discussed

135 in detail in later sections.

0 Min. 2 SD

Max. 2 SD

Median ∆δ13CFA - Bulk [‰]

-2 -4 -6 -8 -10 -12

POM (n=69) Zooplankton (n=46) Larval fish (n=25) Juv. salmon (n=34) Tot FAs*

14:0 iso 15:0 15:0 16:0 16:1n-7 16:2n-4 16:4n-1 17:0 18:0 18:1n-9 18:1n-7 18:2n-6 18:3n-3 18:4n-3 20:2n-6 20:4n-6 20:4n-3 20:5n-3 21:5n-3 22:5n-3 22:6n-3

-14

Fatty acid Figure 4.4. Differences between the δ13C of fatty acids and the bulk of the sample for each of the trophic groups. The minimum and maximum double standard deviations are indicated at the top (see also Fig. 4.7 for a better indication of standard deviation of ∆δ13C values of specific fatty acids). Organisms that were suspected to have switched their diet recently (Table 2.4) were left out. * The δ13C of the total fatty acids was calculated by computing the weighted average δ13C (using the relative abundances of the fatty acids as weighting factors).

As expected the δ13C of the fatty acids are, in general, considerably lower than the bulk (Figure 4.4). In the early 1960s, Parker (1962, 1964) determined that the δ13C of lipids and individual fatty acids derived from marine organisms, were consistently depleted in 13C compared to the total organism. In addition, Parker (1964) noted a great variability (0.5 – 15‰) in the difference between the lipid fraction and the bulk organism. Park and Epstein (1961) also found that the lipids in the plants measured were about 2 to 9‰ depleted in 13C relative to the whole plant. DeNiro and Epstein (1977) and Monson and Hayes (1982a, b) determined that the

13

C depletion in lipids can be explained by

isotope fractionation occurring during the formation of acetyl-coenzyme A (CoA). The carbon of the acetyl-group of acetyl-CoA is used in the synthesis of fatty acids, and

136 contributes to other lipids (e.g., isoprenoids) and some amino acids. The pyruvate dehydrogenase (PDH) complex discriminates against

13

C during the decarboxylation of

pyruvate. The fractionation is such that the carbon localized at the carboxyl position in the acetyl-CoA product is on average more depleted in

13

C than the carbon atom at the

methyl position (DeNiro and Epstein, 1977; Monson and Hayes, 1982a). The isotope effect during the pyruvate dehydrogenase reaction was confirmed by Melzer and Schmidt (1987) for PDH from both the prokaryote Escherichia coli and the eukaryote Saccharomyces cerevisiae. The isotopic order in the acetyl sub-unit is carried through the biosynthesis, such that carbon is depleted in 13C at alternate positions in the carbon chain of the fatty acid (Monson and Hayes, 1982a). For the estimation of the isotope effect of the PDH reaction, DeNiro and Epstein (1977) and Monson and Hayes (1982a) assumed that the hexose used for the synthesis of pyruvate is isotopically homogeneous. On the contrary, Rossmann et al. (1991) found that natural glucose is, in fact, isotopically heterogeneous. These authors reported δ13C values between 4 to 6‰ higher for carbon at position 4 in glucose, and carbon at position 6 was on average 5‰ depleted in

13

C with respect to the other carbon atoms. Since

carbon from positions 3 and 4 are lost as CO2 when producing acetyl-CoA, lipids can be expected to be depleted in

13

C with respect to glucose (Rossman et al., 1991). Even an

intra-molecular alternation of 13C depleted and enriched carbons in the fatty acid carbon chain can be predicted due to the fact that carbon of position 6 in glucose will be systematically transferred to the methyl position in acetyl-CoA (Rossman et al., 1991; Van Dongen et al., 2002). However, it can be calculated that the intra-molecular δ13C differences observed in glucose can only explain a 13C depletion of 1-2‰ of total lipids with respect to glucose. Additionally, the δ13C differences between neighbouring carbon atoms in the fatty acid hydrocarbon chain would only be about 1.5 to 2‰. Moreover, since the methyl group, and not the carbonyl group of acetyl-CoA is depleted in 13C, the intra-molecular δ13C differences would be in a reverse order compared to observations by Monson and Hayes for E. coli and soybeans (Monson and Hayes, 1982a, b). Therefore, it seems that the bulk – lipid δ13C difference can mostly be attributed to the isotope effect during oxidation of pyruvate by PDH, which was unequivocally determined by Melzer and Schmidt (1987). Van Dongen et al. (2002), however, stress that their δ13C

137 measurements on carbohydrates leave room for more intra-molecular variability in the carbon isotope composition of hexoses than observed by Rossmann et al. (1991).

Another apparent trend that can be observed in Figure 4.4 is that the difference in δ13C between fatty acids and the bulk (∆δ13CFA-Bulk) tends to be largest for larval fish and salmon, and lowest for POM and zooplankton. In fact, the ∆δ13CFA-Bulk often follows the following pattern: POM < zooplankton < larval fish < juvenile salmon (muscle). This is also the expected order of the respective trophic levels for these organisms. One possible explanation of the higher ∆δ13CFA-Bulk per trophic level is that lipid content and production decreases going from seston to juvenile salmon muscle. When a higher proportion of the bulk is composed of lipids the ∆δ13CFA-Bulk automatically decreases. Moreover, when the production of lipids is relatively higher a greater proportion of pyruvate will have to be oxidized to form acetyl-CoA. Because a greater fraction of the pyruvate is used the isotope effect associated with the pyruvate dehydrogenase reactions will result in a lower carbon isotope fractionation between the acetyl-group and pyruvate (see DeNiro and Epstein, 1977; Monson and Hayes, 1982a). Assume that the lipid content of POM was unusually high (Wakeham et al., 1997) at 20% of the dry weight (e.g., when POM consisted only of growing algae; Shifrin and Chisholm, 1981), and lipids comprised only 5% of dried juvenile salmon muscle (below values reported for salmon by Henderson and Tocher, 1987). Furthermore, assume that lipids have on average a δ13C that is 6‰ lower than the δ13CBulk. Then, the difference in lipid content can result in a δ13CBulk that is almost 1‰ lower for POM than for the salmon muscle. However, next to being an unrealistic difference in lipid content, such a difference would only partly explain the observed differences in ∆δ13CFA-Bulk. In a feeding experiment Focken and Becker (1998) found a “trophic shift” of about 1.3‰ between the δ13C of fat-free matter of carp and its diet. However, the carbon isotope composition of lipids extracted from the fish was virtually indistinguishable from lipids from its diet. Grice et al. (1998) and Klein-Breteler et al. (2002) also did not find an enrichment in 13C in lipids during the transfer to a higher trophic level. A trophic shift

138 in the bulk δ13C, but not in the δ13C of fatty acids would, to some extent, explain the pattern observed in Figure 4.4. The reasons for this effect are not clear. It may be that the proportion of fatty acids synthesized de novo in animals is much lower than that for fatty acids assimilated directly from the diet. Goulden and Place (1990) estimated from the results of their

14

C-labelling experiment that only 2% of the fatty acids accumulated by

daphniids were synthesized by these zooplankton animals themselves. However, this is probably a low estimate due to the short incubation time during the experiment and may not hold for fish. In Figure 4.4 it can be observed that it is especially the essential fatty acids (n-6 and n-3 PUFAs) that show the largest differences ∆δ13CFA-Bulk between fish and POM. This pattern could be explained by a trophic shift in the δ13C of compound classes, such as proteins and carbohydrates, but not in lipids. The de novo synthesis of (non-essential) fatty acids uses carbon of the degradation products of (the slightly

13

C enriched)

carbohydrates and (some) of proteins. Therefore, the (non-essential) fatty acids readily produced by the fish, such as the 16:0, 18:0 and 18:1n-9, will be a mixture of both dietary fatty acids and (13C enriched) fatty acids synthesized by fish. In contrast, the essential fatty acids, mostly n-6 and n-3 PUFAs, will almost solely consist of fatty acids directly derived from the diet. Hence, the essential fatty acids are then expected to show a larger ∆δ13CFA-Bulk in higher trophic level animals because the bulk δ13C has shifted, but not the δ13C of the fatty acid. This effect should also have affected the pattern of Figure 4.1. However, it is not that clearly observed that the non-essential fatty acids have higher δ13CFA – δ13CPOM differences. The fact that the zooplankton and larval fish were not continuously feeding at the same location, and possible variation of the POM δ13C over time, may have masked the effect. Additionally, juvenile salmon, for which the biggest trophic shift in bulk δ13C is expected, could not be shown here because these animals would certainly have moved a considerable distance prior to sampling.

4.4.3 Differences in δ13C among individual fatty acids The stable carbon isotope composition of the various fatty acids was normalized by comparing the δ13C values with the δ13C of the C14 saturated fatty acid (Figure 4.5). This

139 fatty acid was chosen because 1) it was always prevalent in high enough amounts, 2) good measurement precision could be attained for it (see Table 2.1 and Figure 2.2), and 3) the fatty acid is a direct product of the fatty acid synthase reactions and has not undergone any extra elongation or desaturation. A range of 5-8‰ in the δ13C of fatty acids commonly occurs within a single sample (Figures 4.2 and 4.3). The same pattern of relative δ13C values for individual fatty acids can be observed for POM, zooplankton, larval fish and muscle tissue of juvenile salmon (Figure 4.5). The curves shown in Figures 4.4 and 4.5 are constructed using the median of values obtained from samples collected on 3 different cruises. This pattern is also quite similar in plots of values from each individual cruise (not shown). Fatty acid δ13C values obtained by Murphy and Abrajano (1994) and Pond et al. (2000) were normalized against the δ13C value reported for the 14:0 fatty acid and compared to values found in this study (Figure 4.6). Though the absolute values sometimes deviate considerably from values reported here, the pattern observed by these authors for marine organisms from different parts of the world is remarkably similar. This suggests the presence of common underlying mechanisms producing semi-predictable offsets between the δ13C of fatty acids.

Mechanisms causing the correspondence between the offsets between the δ13C of the various fatty acids from organisms from all trophic levels should already be operating at the base of the food web. Schouten et al. (1998) have measured the δ13C of fatty acids in 13 different cultured algae. However, they only reported the δ13C values of groups of fatty acids lumped together. For example, they reported one δ13C value of the C18 fatty acids. In this study, however, the δ13C of the individual fatty acids part of this group often showed a range of 7‰. Tolosa et al. (1999) also measured the δ13C of fatty acids in algae. The range of the δ13C values measured for different fatty acids in a dinoflagellate was about 3‰, whereas in a diatom this range was found to be about 5‰. The variation in δ13C values of fatty acids did not correspond well with the pattern observed here and in other studies (Murphy and Abrajano, 1994; Pond et al., 2000; Bec, 2003). Unfortunately,

140

6

Median ∆δ13CFA - 14:0 (‰)

4 2 0 -2 -4

Min. 2 SD

Max. 2 SD

14:0 iso 15:0 15:0 16:0 16:1n-7 16:2n-4 16:4n-1 17:0 18:0 18:1n-9 18:1n-7 18:2n-6 18:3n-3 18:4n-3 20:2n-6 20:4n-6 20:4n-3 20:5n-3 21:5n-3 22:5n-3 22:6n-3

-6

POM (n=73) Zooplankton (n=56) Larval fish (n=29) Juv. salmon (n=37)

Median ∆δ13CFA - Tot FAs (‰)

4 2 0 -2 -4

Min. 2 SD Max. 2 SD

14:0 iso 15:0 15:0 16:0 16:1n-7 16:2n-4 17:0 16:4n-1 18:0 18:1n-9 18:1n-7 18:2n-6 18:3n-3 18:4n-3 20:2n-6 20:4n-6 20:4n-3 20:5n-3 21:5n-3 22:5n-3 22:6n-3

-6

POM (n=70) Zooplankton (n=56) Larval fish (n=29) Juv. salmon (n=34)

Fatty acid Figure 4.5. The median δ13C of fatty acids normalized against the 14:0 fatty acid (top) and the weighted average, or total fatty acids (bottom), plotted for each of the trophic groups. The minimum and maximum double standard deviations are indicated at the lower right. Organisms that were suspected to have shifted diet recently (Table 2.4) were left out. The δ13C of the total fatty acids was calculated by computing the weighted average δ13C (using the relative abundances of the fatty acids as weighting factors).

141 the exact conditions under which the algae of the study by Tolosa et al. (1999) were grown, and the species used were not reported. Unpublished work by Bec (2003) does show a correspondence to the variation in δ13C values of fatty acids in POM reported here. In the next sections several potential mechanisms to explain the differences between the δ13C of individual fatty acids will be discussed.

6

Murphy and Abrajano (1994): Mussels (n=5) Pond et al. (2000): "Microplankton" (n=10) This study: Zooplankton (n=56)

5 4 3

1 0 -1

13

∆δ CFA - 14:0 (‰)

2

-2 -3 -4

22:6n-3

20:5n-3

18:5n-3

18:4n-3

18:3n-3

18:2n-6

18:1n-7

18:1n-9

18:0

16:1n-7

16:0

15:0

14:0

-5

Fatty acid Figure 4.6. Comparison with δ13C values obtained by other authors. The reported δ13C values of fatty acids were normalized against the δ13C of the 14:0 fatty acid in the same samples. Values reported by Pond et al. (2000) are an average of 10 samples of microplankton (<200µm) from 4m depth above the Mid-Atlantic Ridge. The average from 5 mussels collected in estuarine environments around Newfoundland was reconstructed from Murphy and Abrajano (1994), Figure 3.

4.4.4 Source effect Different sources with a different stable carbon isotope composition may contribute to the POM. When the various sources have different fatty acid distributions the δ13C of the individual fatty acids in the POM will vary. For example Murphy and Abrajano (1994)

142 and Abrajano et al. (1994) suggested a “unique input” for the 18:4n-3 fatty acid due to its low δ13C value. Murphy and Abrajano (1994) proposed a bacterial source for this fatty acid. To test whether the δ13C pattern observed in the present study was predominantly caused by a source effect, the ∆δ13CFA-Bulk values of POM with different fatty acid profile characteristics were plotted (Figure 4.7). Out of 70 samples 15 with the highest- and 15 with the lowest abundance of the 16:2n-4 fatty acid were selected. As shown in the previous chapter the 16:2n-4 fatty acid abundance can be used as a semi-quantitative measure for diatom abundance. Similarly, the iso-15:0 fatty acid abundance was used as a qualitative measure for bacterial input, and the 15 samples with highest relative amount of iso-15:0 were selected as samples that are richest in bacterial matter. The differences between the ∆δ13CFA-Bulk trends of the “diatom-poor/rich”, and “bacteria-rich” POM do not seem large enough to state that the typical δ13C pattern for individual fatty acids can be explained by a source-effect (Figure 4.7). Moreover, it would be surprising if the mussel samples, “microplankton” and zooplankton from different environments (Figure 4.6) all had had a similar input of different sources to produce a comparable δ13C pattern. There is no doubt that different sources can have an effect on the δ13C of individual fatty acids (see e.g., Boschker et al., 2000). However, the presented evidence does not support the hypothesis that the observed reproducible pattern in the δ13C variation for individual fatty acids (Figures 4.2-4.7) is caused by input from different sources. Neither do the presented results point to a “unique input” or bacterial source for the 18:4n-3 fatty acid, as suggested by Murphy and Abrajano (1994) and Abrajano et al. (1994).

4.4.5 Effect of preferential catabolism Different lipid classes, such as phospholipids and glycolipids (polar lipids), and triacylglycerols and wax esters (neutral lipids) have distinct fatty acid compositions. Triacylglycerols (TAGs) contain proportionally more saturated fatty acids (SFAs) and monounsaturated fatty acids (MUFAs) than phospholipids and glycolipids (e.g., Henderson and Sargent, 1989; Hodgson et al., 1991; Sukenik et al., 1993; Albers et al.,

143 1996). Wax esters, especially in copepods, are typically rich in MUFAs (Sargent and Falk-Petersen, 1988; Albers et al., 1996).

0

13

Median ∆δ CFA - Bulk (‰)

-2 -4 -6 -8 "Diatom-poor" POM (n=15) "Diatom-rich" POM (n=15) "Bacteria-rich" POM (n=15)

-10

22:6n-3

20:5n-3

20:4n-6

18:4n-3

18:3n-3

18:2n-6

18:1n-7

18:1n-9

18:0

17:0

16:2n-4

16:0

16:1n-7

15:0

iso 15:0

14:0

-12

Fatty acid Figure 4.7. The median δ13C of fatty acids normalized to the δ13C of the bulk of 15 POM samples from three pre-selected groups: POM believed to be richest in diatoms, poorest in diatoms, and with the most bacterial matter (15 samples in each group). See text for more details. The vertical bars represent ±1 standard deviation.

TAGs and wax esters predominantly function as energy storage molecules. The catabolism of fatty acids takes place mainly through the β-oxidation pathway (Gurr and Harwood, 1991). Before the fatty acids can be catabolized they are released from the glycerol backbone by hydrolytic enzymes called lipases. Due to their role as fuel molecules a greater proportion of TAGs and wax esters will get catabolized. Thus, when an isotope effect is associated with the hydrolysis by lipases (lipolysis), it can be expected that SFAs and MUFAs will be relatively enriched in

13

C due to their

preponderance in TAG and wax esters. The isotope fractionation will have to take place at the carboxyl (C-1) position in the fatty acid.

144 Vogler and Hayes (1980) indeed found that carbon at the C-1 positions of the 16:0 fatty acid in corn and soybeans was more depleted in 13C than in the 18:1, which was, in turn, more depleted in 13C than the carboxyl group of the 18:2 fatty acid. These authors already stated that “it is conceivable, e.g., that the catabolism of fats is accompanied by an isotope effect leading to the accumulation of carbon-13 at the carboxyl carbon positions in the unutilized residual fats” (Vogler and Hayes, 1980). Their results indicate that a fractionation of as much as 12-14‰ can occur on the carboxyl carbon. For a whole C18-fatty acid this would, however, only result in a δ13C value about 0.7-0.8‰ higher than without the lipolysis effect. It should also be borne in mind that the observed differences in the δ13C of the carboxyl carbons, observed by Vogler and Hayes (1980), may have been the result of more than one isotope effect. When catabolism has an isotope effect, enriching the residual fatty acids in

13

C, it

13

can be expected that fatty acids will have consistently higher δ C values when going up through the food web. However, such an effect is not clearly observed in Figure 4.1. Therefore, preferential catabolism is apparently not large enough to produce the patterns observed in Figures 4.2-4.6.

4.4.6 Effect of different timing of lipid class synthesis Many factors can influence the relative production rates of neutral versus polar lipids in algae. These factors include timing within the growth cycle, culture age, nutrient depletion, light and temperature (Shifrin and Chisholm, 1981; Henderson and Sargent, 1989; Roessler, 1990; Sukenik and Carmeli, 1990; Hodgson et al., 1991). For example, during aging of a culture and nutrient deplete conditions the lipid content of algae can more than double (Roessler, 1990; Hodgson et al., 1991). Typically a shift towards higher lipid content in the cell is associated with a shift towards a greater proportion of neutral lipids (see e.g. Roessler, 1990; Sukenik and Carmeli, 1990). Concomitantly, the production of proteins declines, with C:N ratios often increasing several fold (Shifrin and Chisholm, 1981). Therefore, during times of relatively high lipid production, it can be expected that a higher proportion of pyruvate will be oxidized by the PDH complex to form acetyl-CoA. Due to the use of a higher proportion of the reactant, the isotope effect

145 associated with the reaction will result in less fractionation between the product and the reactant. Hence, less fractionation occurs between the acetyl groups used in the synthesis of fatty acids (acetyl-CoA) and the carbon in pyruvate. Since relatively more neutral lipids are produced, it is especially the SFAs and MUFAs that will be enriched in

13

C

relative to previous conditions. Hence, a disparity can arise in the 13C/12C ratios between the PUFAs that are more prevalent in the remaining polar lipids and the SFAs and MUFAs formed under the conditions favouring high (neutral) lipid production. Sukenik and Carmeli (1990) studied the chemical composition and lipid biosynthesis of a marine eustigmatophyte Nannochloropsis sp. grown under a 12:12 h light-dark regime. They found that [1-14C]acetate was incorporated mostly into lipids during the light period, whereas during the dark period the labelled acetate taken up by Nannochloropsis was incorporated more into non-lipid compounds. Additionally, it was observed that neutral lipids, such as TAG, were synthesized and accumulated in the light and showed rapid turnover in the dark. In the light this resulted in an increase in the percentage of the 16:0 and 16:1 fatty acids, which were associated with TAGs. In the dark, on the other hand, the relative abundance of the 20:5n-3 fatty acid (associated with glycolipids) increased (Sukenik and Carmeli, 1990). The results of Sukenik and Carmeli (1990), together with those of Otsuka and Morimura (1966) show that during a diurnal growth cycle substantial fluctuations in the routing of the carbon flow exist. As argued in section 4.4.2, this may have an effect on the δ13C composition of fatty acids, creating more variability. In order to get an idea of the magnitude of the effect of the changing carbon flow on the δ13C of fatty acids the following scenario is considered. It is assumed that: 1. During the growth phase in the light 50% of all pyruvate produced in the cells gets used to form acetyl-CoA. 2. During the course of the cell division in the dark only 25% of the pyruvate gets oxidized to form acetyl-CoA. 3. The isotope effect, α, on C-2 of pyruvate associated with the PDH reaction is 1.025, where α is the ratio of the reaction rate constants for the respective isotopes (k12/k13). The C-2 of pyruvate becomes the C-1 in the acetyl group of

146 acetyl-CoA, and it is assumed that the C-2 in acetyl-CoA did not undergo any isotope effect. The assumptions 1 and 2 are rough estimations based on the differences in acetate incorporation rates measured by Sukenik and Carmeli (1990) for the whole cell and lipids. Although the magnitudes (25% and 50%) may be on the small side, it is the difference (25%) that matters for the following calculation. Assumption 3 is based on the determination of the isotope effect of PDH in yeast by Melzer and Schmidt (1987). Because of the mass balance requirement, the following should be true: δ pyrC −2 = f Ac ⋅ δ AcC −1 + (1 − f Ac ) ⋅ δ unpyrC −2 , where

(4.1)

δpyrC-2 is the δ13C of the carbon at the C-2 position in the pyruvate supplied, fAc is the fraction of pyruvate being oxidized to form acetyl-CoA, δAcC-1 is the δ13C of the carbon at position 1 (C-1) in acetyl-CoA (derived from the C-2 in pyruvate), and δunpyrC-2 is the δ13C of the C-2 in the “unused pyruvate”. And it follows that δ AcC −1 = δ pyrC −2 − (1 − f Ac ) ⋅ ∆ PDH , where

∆ PDH = δ pyrC − 2 − δ AcC −1 ≈ 103 ⋅ ln α = 24.7‰

(4.2)

(Melzer and Schmidt, 1987)

(4.3)

When filling in fAc=0.5 (50%) in equation 4.2, for growth in the light, the value for δAcC-1 becomes 12.3‰ lower than δpyrC-2. For the dark period, with fAc=0.25 (25%), δAcC-1 will be 18.5‰ lower than δpyrC-2. These results are insensitive to the choice of δ13C for the pyruvate (see equation 4.3). Since both the fractionated- and unfractionated carbons of acetyl-CoA end up in the fatty acids, the difference between the δ13C of lipids and the supplied pyruvate will be 6.2‰ and 9.3‰, respectively. The exercise above indicates that hypothetically opposite timing of the synthesis of fatty acids can lead to a difference of as much as 3‰ (i.e., 9.3 – 6.2‰). However, it should be noted that none of the major fatty acids is found exclusively in one lipid class, and the lipid classes are not exclusively formed during one of the growth periods.

147 Therefore, when taking these considerations into account, this “timing-effect” cannot be the sole explanation for a range of 6-9‰ in the difference of the δ13C among fatty acids shown in Figures 4.4 - 4.7. Additionally, this scenario cannot explain why for example the 20:5n-3 and 22:6n-3 fatty acids are not as depleted in

13

C as the 18:4n-3. All are

typically major components of polar lipids and should thus have a comparable δ13C.

When studying the isotope composition of individual compounds in algal cultures it is important to recognize the carbon flow and biosynthesis timing-effect mentioned above. Sukenik and Livne (1991) showed that in chemostat cultures of Isochrysis galbana the fatty acid/protein ratio co-varied with growth rate. They found a fatty acid/protein ratio of 2.89 at a growth rate of 0.2d-1 and a ratio of 0.59 at a growth rate of 0.9d-1. Additionally, lipid mass and proportion of TAGs can increase substantially during the age of a batch culture (see e.g., Hodgson et al., 1991). Furthermore, during the shift in pH in batch cultures, the carbon fixed by the algae will get increasingly enriched in 13C over time (Van Dongen et al., 2002). Since the various compound classes and different fatty acids have different turnover times, larger differences in their δ13C can be expected to arise over time in batch cultures. Hence, careful control and the availability of detailed reports on the growth conditions are necessary when comparing δ13C values of individual compounds in cultured algae.

4.4.7 Biosynthetic effects The last option explored here is that the variation in the δ13C of individual fatty acids can be explained by isotope fractionating steps after the fatty acid synthase (FAS) reactions. The majority of the carbon atoms in all fatty acids is assembled in the FAS reaction cycles. Therefore, any variation in the isotope composition should be caused by reactions after the release of the fatty acid from FAS, i.e., when it is assumed that differences in timing of the biosynthesis play no role in causing this variation. To date, the pioneering research by Monson and Hayes (1980, 1982a, b) and Vogler and Hayes (1980) still provides the best source of information on the possibilities and magnitudes of isotope

148 effects for such reactions. In the following sections several mechanisms are explored. Here will be focussed on processes occurring in plants, as it seems clear that the typical δ13C pattern (Figure 4.5) is already present in POM samples (most of which are believed to be dominated by algal matter).

4.4.7a Hydrolysis of fatty acyl-ACP During the FAS reactions the fatty acid chain is elongated by sequential addition of the two-carbon units derived from acetyl-CoA. As its hydrocarbon chain grows, the fatty acyl group is linked to the acyl carrier protein (ACP). Typically, when the fatty acyl-group has reached a length of 16 or 18 carbons this process is terminated, this termination can also occur at shorter lengths. In plants the fatty acid synthesis happens in the plastids, and the typical chloroplast exports primarily 16:0 and 18:1 fatty acids (Schmid and Ohlrogge, 1996). The fatty acid gets released for transport out of the plastid due to the action of thioesterases. Acyltransferases can transfer the fatty acyl-group directly to membrane lipids. The work by Monson and Hayes (1982a, b) suggests that an isotope effect is associated with these type of catalyzed hydrolysis reactions in E. coli and yeast. The α values (k12/k13) for these reactions can be as large as 1.013 (Monson and Hayes, 1982a, b). This means that a maximum fractionation of about 13‰ (at the C-1) can occur at a branch point in the carbon flow instigated by the release from ACP. For, example a lipidincorporated 16:0 fatty acid will then be depleted by maximally 0.8‰ with respect to the total carbon of all the fatty acids produced downstream.

4.4.7b Desaturation In plants desaturation can take place by a soluble acyl-ACP desaturase, or the double bond can be introduced while the fatty acid is membrane-bound (Harwood, 1996; Schmid and Ohlrogge, 1996). It has been found that all membrane-bound desaturases examined to date feature a large primary deuterium kinetic isotope effect with α = kH/kD ≈ 5-8 (Buist and Behrouzian, 1996, 1998; Pinilla et al., 1999; Behrouzian et al., 2001b; Savile et al., 2001). On the other hand, for desaturation catalyzed by the soluble stearoyl-ACP ∆9

149 desaturase no notable kinetic isotope effect (KIE) was measured (Behrouzian et al., 2001a). An ω-3 type desaturase derived from a nematode that uses fatty acyl-CoA thioesters as a substrate, also caused a large deuterium KIE, with an α-value of ca. 7.8 (Meesapyodsuk et al., 2000). The aforementioned studies all found that the KIE occurred at the C-H bond closest to the C-1 position, and no KIE was observed for the C-H furthest away from the terminal carbonyl group. In Figure 4.8 the mechanistic scheme for fatty acid desaturation according to Behrouzian et al. (2001a) is shown. First, oxidation is initiated by an energetically difficult, and rate determining C-H abstraction step. This generates a very short-lived, carbon-centred radical intermediate, or its iron-bound equivalent (not shown), which collapses to result in the unsaturated product (Behrouzian et al., 2001a). Since the

Fe

(IV)

Fe O

H

H

H 10 9

Fe

(IV)

H

(IV)

Fe

OH

slow KIE

H

H

fast H

10 9

Fe

(III)

no KIE

(III)

Fe H

O

(III)

+

H

H H

10

9

Figure 4.8. Mechanistic scheme for the desaturation of a fatty acid. During the process two hydrogens get removed. The removal of the first hydrogen (closest to the C-1, carbonyl group) is the slowest, rate determining step. The radical intermediate quickly collapses to release the second hydrogen. A kinetic isotope effect (KIE) only occurs during the first reaction shown in this scheme. The example here depicts the desaturation by a ∆9-desaturase (note carbon numbers). Drawn after Behrouzian et al. (2001a).

first hydrogen abstraction step is rate determining and takes place at the C-H bond closest to C-1, the KIE is localized at that position.

As both C and H atoms influence the C-H bond strength, an isotope effect should also occur at the carbon position. To get an idea of how the KIE on the carbon atom relates to the observed hydrogen isotope effect, consider the influence of isotopic substitution on the C-H bond strength. The bond strength is closely related to the

150 vibrational frequency of the atoms participating in the bond. For an ideal diatomic harmonic oscillator:

v=

1 2π

f

µ

, where

(4.4)

v is the vibrational frequency, f the forcing constant, which remains constant when substituting different isotopes, and µ is the reduced mass, defined as:

µ=

m1m2 m1 + m2

, where

(4.5)

m1 and m2 are the masses of the two atoms participating in the bond. Considering the difference between C-H and C-D bonds, it can be noted that the reduced mass of a C-D bond is about 1.85 times higher than for C-H. The vibrational frequency for a C-D bond should thus be 1/√(1.85) = 0.73 times that of a C-H bond. For the situation where either a C-H or a C-D bond gets entirely broken at the transition state, it can be derived (see Carpenter, 1984) that the rate constants relate as follows:

α=

k H k1  hc( v1 − v2 )  = ≅ exp  , k D k2  2kT 

where

(4.6)

α is the isotope effect associated with the C-H/D bond, kH and kD are the reaction rate constants of the reactions involving breaking a C-H and C-D bond, respectively; h is Planck’s constant, c is the speed of light, v1 and v2 the vibration frequencies of C-H and C-D (in this case), respectively; k is Boltzmann’s constant, and T is the temperature. When using a typical C-H bond vibrational frequency of 2950 cm-1, and filling in (1 0.73) * 2950cm-1 for (v1 - v2) in equation 4.6, the calculated α is 6.6. The α values obtained for the examined desaturases are indeed in the range of this theoretical value (see above). To calculate the expected carbon isotope effect of desaturase catalyzed reactions the reduced masses (µ) of 12C-H with 13C-H are compared. Subsequently, in equation 4.6 the

151 difference in vibrational frequencies can be filled in (i.e., [1 – 0.997] * 2950cm-1), and an α value of 1.022 is obtained. This is in exact agreement with the calculation made by Monson and Hayes (1982b) who also report about the existence of a hydrogen isotope effect during desaturation. Their calculations were based on a diatomic molecular dissociation model as well. The estimate implies that an unsaturated C18 fatty acid, for example, can have a δ13C of about 1.2‰ (maximum) lower than its precursor due to the desaturation. Though, it is important to note that some of this difference can be partially masked when this fatty acid is a reactant in subsequent reactions with associated isotope effects. In Figure 4.10a the δ13C of fatty acids from the C18-series is plotted against the degree of unsaturation. The δ13C indeed declines with an increasing number of double bonds. However, when just regarding the POM, the difference in δ13C between the 18:1n9 and 18:2n-6 fatty acids is larger than expected when desaturation would be the sole mechanism for depletion in

13

C. Additionally, the relatively high δ13C value for the

20:5n-3 and especially the 22:6n-3 values will have to be explained by other factors.

4.4.7c Elongation The data presented here suggests that the elongation of fatty acids is associated with enrichment in

13

C, rather than depletion (Figure 4.9b). For the elongation of fatty acids

with C18 and longer carbon chains, the fatty acid will need to be released from the glycerolipid. This hydrolysis or transacylation reaction may have a small KIE associated with it. However, the KIE will cause a small isotope fractionation favouring 12-carbon at the C-1 position (e.g., Monson and Hayes, 1982b), which would cause the δ13C value to decrease with greater chain-length. This is opposite to what is observed in Figure 4.9b. Schneider and Roessler (1994) found in a radiolabeling study of lipids in Nannochloropsis that different metabolic pools of malonyl-CoA exist for de novo synthesis and fatty acid elongation. Schwender and Ohlrogge (2002) studied the metabolic fluxes in developing embryos of canola (Brassica napus). They found that amino acids can contribute substantially to the carbon in the terminal C2 units of C20 and

152

13

Median ∆δ CFA - 14:0 (‰)

6 POM (n=73) Zooplankton (n=56) Larval fish (n=29) Juv. salmon (n=37)

4 2 0 -2 -4 -6

C18-fatty acid series 0

6 4

1 2 3 Number of double bonds

4

b

POM (n=73) Zooplankton (n=56) Larval fish (n=29) Juv. salmon (n=37)

2 0

13

Median ∆δ CFA - 14:0 (‰ )

a

-2 -4 -6

(n-3)-fatty acid series 18

19 20 21 Number of carbon atoms

22

Figure 4.9. The median δ13C of fatty acids, normalized against the 14:0 fatty acid δ13C, plotted for the C18-series against the number of double bonds present in the respective fatty acids (a). Below, the median δ13C of fatty acids, normalized against the 14:0 fatty acid δ13C, were plotted for the (n-3)-series against the number of carbons present in the respective fatty acids (b). Values for each of the trophic groups are shown. The vertical bars represent ±1 standard deviation.

C22 fatty acids. It can therefore be concluded that cytosolic acetyl-CoA, which is used for the elongation of long-chain fatty acids, has a more complex biogenic origin than plastidic acetyl-CoA, involved in de novo fatty acid synthesis. To produce a 3‰ difference between a C18 and a C20 fatty acid (Figure 4.5), the added two carbon atoms should each have a δ13C that is 30‰ higher than the C18 fatty acid. Amino acids can have δ13C values that are as much as 7‰ higher than the whole cell (Macko et al., 1987). However, such a difference is still not enough to explain the difference observed, even if

153 all of the added C2 units were derived from amino acid carbon. Additionally, the amino acid degradation products will have to be converted by the PDH complex (with associated isotope effect) to produce acetyl-CoA. Therefore, it remains unclear why fatty acids such as the 20:5n-3 and 22:6n-3 are so enriched in 13C (3-6‰) with respect to the 18:2n-6 and 18:3n-3/4, which are likely precursors.

4.4.7d The polyketide synthase pathway The mechanisms outlined in the previous discussion cannot account for the relatively high δ13C values measured for some of the longer chain fatty acids, such as the 20:5n-3 and 22:6n-3. The C20 and C22 PUFAs are expected be even more depleted in

13

C than

their C18-PUFA precursors when they are formed via elongation- and desaturation steps, as a result of the KIEs associated with these steps. Recently, Metz et al. (2001) reported an alternative pathway to form fatty acids such as 20:5n-3 and 22:6n-3 (see also Wallis et al., 2002). They found that polyketide synthases (PKSs), distinct from previously examined PKSs, catalyze PUFA synthesis in the investigated marine bacterium (Shewanella) and a marine protist (Schizochytrium). PKSs carry out some of the same reactions as FAS and also use ACP as a covalent attachment site for the growing carbon chain (Metz et al., 2001). However, some of the steps in the formation of the fatty acid are abbreviated. Additionally, PUFA synthesis catalyzed by PKS does not require desaturation and elongation of saturated fatty acids (Metz et al., 2001). Therefore, it is conceivable that PUFAs synthesized via the PKScatalyzed pathway may have experienced less isotope fractionation, and are enriched in 13

C with respect to PUFAs formed via the “traditional” pathway. Hence, the relatively

high δ13C values for the long chain PUFAs may perhaps be explained by a high proportion of those fatty acids formed via an alternative biosynthetic route, catalyzed by PKS. If indeed a high amount of, for example, the 22:6n-3 fatty acids in the sampled marine POM was synthesized by the action of PKS, this would suggest that this pathway has to be quite common in the marine environment. The marine protist Schizochytrium, in

154 which the particular PKS was identified by Metz et al. (2001), is a member of the Heterokont phylum (López-García et al., 2001). Among others, algae such as Chrysophytes and diatoms belong to the same phylum. Metz et al. (2001) raised the possibility that the PUFA PKS has undergone lateral gene transfer. Therefore, species that are related to Schizochytrium may not necessarily have a higher chance that they will possess a similar PKS. Berry et al. (2002) found PKS-encoding genes in the dinoflagellate Pfiesteria schumwayae, and suggested that those particular genes “may be involved in PUFA biosynthesis rather than toxin production”. The use of the PKS pathway by (some) dinoflagellates may also be in line with the observations made by Henderson and Mackinlay (1991). They investigated the incorporation of

14

C labelled

18:0, 18:1n-9 and acetate by the dinoflagellate Crypthecodinium cohnii. It was found that after 3 days growth in the presence of labelled acetate, 48% of the radioactivity incorporated into polar lipids was located in 22:6n-3. However, the corresponding values for when

14

C-18:0 and

14

C-18:1n-9 were administered, were 2% and 4%. This could be

interpreted to indicate that 22:6n-3 was mostly synthesized de novo (catalyzed by PKS), and only little amounts of 22:6n-3 was formed by elongation and desaturation of 18:0 and 18:1n-9. Hence, potentially the PKS catalyzed synthesis of PUFAs can play an important role in dinoflagellates and perhaps also other algae. Certainly more research is necessary to investigate whether the PUFA PKS enzymes are ubiquitous in the marine environment or not. Also, the implications on the stable carbon isotope ratio of PUFAs will have to be investigated more directly. For now, the synthesis of long chain PUFAs catalyzed by PKS seems to be the only plausible explanation of the relatively high δ13C of these fatty acids.

4.4.8 Practical implications Due to the variation caused by some of the aforementioned effects, and the lower precision of the δ13C measurements on individual fatty acids than on bulk sample, bulk δ13C may reflect the δ13C of the diet just as well. However, it should be borne in mind that when using bulk δ13C a slight correction should be made for the (fairly predictable) trophic fractionation (DeNiro and Epstein, 1978; Peterson and Fry, 1987; Post, 2002).

155 Additionally, bulk δ13C values are an average of diet intake over a longer time period, compared to that reflected by the fatty acid δ13C signal. Therefore, in most food web studies bulk IRMS may be sufficient and certainly more cost efficient. However, there are cases where the extra information of the δ13C of fatty acids is quite useful. As demonstrated in Chapter 2, the fatty acid δ13C data can add information about the time dimension. This is because the different turnover times of the various fatty acids can cause abnormal differences in δ13C between the fatty acids and also the bulk sample after a shift in the diet (see Chapter 2). δ13C values of fatty acids are also useful in confirming whether the diet of a heterotroph consisted of a mix of two known endmembers, or of a third intermediate source. In addition, the δ13C of fatty acids can provide insight into biosynthetic pathways. For example, when indeed the PKS catalyzed pathway produces PUFAs with higher

13

C/12C ratios, this technique may be able to

confirm the use of PKS for the synthesis of PUFAs.

4.5 Conclusions The δ13C of essential fatty acids in zooplankton and larval fish was not found to be more similar to those in POM (from the same location) than the δ13C of other fatty acids. For example, the δ13C of the 14:0 fatty acid often showed closer agreement with its counterpart in POM. A large range (typically about 7‰) in the δ13C values of fatty acids is observed within single samples. Such a range in δ13C values is in accordance with other studies that measured the stable carbon isotope ratios of fatty acids from mussels and marine plankton (Fang et al., 1993; Abrajano et al., 1994; Murphy and Abrajano, 1994; Pond et al., 2000). It was shown here that the observed variation in δ13C was more affected by biosynthetic effects than by different contributions of fatty acids from sources with distinct 13C/12C ratios. Some of the variation in δ13C between individual fatty acids was found to reproducible, not only in samples measured for this research, but also when comparing values found in other studies. This reproducible, average pattern is most likely the end-result of several mechanisms and isotope fractionations caused by the action of a

156 range of different enzymes. The factors that can potentially produce the largest carbon isotope fractionation seem to be desaturation, different timing of lipid class synthesis during the diurnal growth cycle of autotrophs, and perhaps also the proportion of PUFAs synthesized via an alternative (PKS-catalyzed) pathway. De novo synthesis via the PKScatalyzed pathway was the only hypothesis that could account for the relative enrichment in 13C of some of the C20 and C22 PUFAs. The reproducible pattern in the δ13C differences between individual fatty acids, produced by the primary producers, is subsequently passed on to higher organisms through feeding. Some deviations from the POM fatty acid δ13C pattern in higher organisms may occur due to desaturation, hydrolysis and re-esterification of dietary fatty acids. Additionally, when the δ13C of dietary fatty acids has shifted over time, a larger variation can be expected due to different turnover rates of the various fatty acids. Stable carbon isotope ratios of individual fatty acids are especially useful in ecological studies to investigate recent diet shifts. Stable carbon isotope measurements of individual fatty acids can also confirm that a diet consists of a mix of two end-members when these two sources have a distinct δ13C and fatty acid composition. Finally, the δ13C of fatty acids may provide new insights into biosynthetic pathways.

4.6 References Abrajano, Jr, T.A., Murphy, D.E., Fang, J., Comet, P. and Brooks, J.M. 1994. 13C/12C ratios in individual fatty acids of marine mytilids with and without bacterial symbionts. Organic Geochemistry 21, 611-617. Albers, C.S., Kattner, G. and Hagen, W. 1996. The compositions of wax esters, triacylglycerols and phospholipids in Arctic and Antarctic copepods: evidence of energetic adaptations. Marine Chemistry 55, 347-358. Arts, M.T., Ackman, R.G. and Holub, B.J. 2001. “Essential fatty acids” in aquatic ecosystems: a crucial link between diet and human health and evolution. Canadian Journal of Fisheries and Aquatic Sciences 58, 122-137.

157 Batcabe, J.P., Howell, J.D., Blomquist, G.J., Borgeson, C.E. 2000. Effects of developmental age, ambient temperature, and dietary alterations on ∆12 desaturase activity in the house cricket, Acheta domesticus. Archives of Insect Biochemistry and Physiology 44, 112-119. Bec, A. 2003. Ph.D. thesis. Université Blaise Pascal, Clermont Ferrand, France. Behrouzian, B., Buist, P.H. and Shanklin, J. 2001a. Application of KIE and thia approaches in the mechanistic study of a plant stearoyl-ACP ∆9 desaturase. Chemical Communications 2001, 401-402. Behrouzian, B., Fauconnot, L., Daligault, F., Nugier-Chauvin, C., Patin, H. and Buist, P.H. 2001b. Mechanism of fatty acid desaturation in the green alga Chlorella vulgaris. European Journal of Biochemistry 268, 3545-3549. Berry, J.P., Reece, K.S., Rein, K.S., Baden, D.G., Haas, L.W., Ribeiro, W.L., Shields, J.D., Snyder, R.V., Vogelbein, W.K. and Gawley, R.E. 2002. Are Pfiesteria species toxicogenic? Evidence against production of ichthyotoxins by Pfiesteria shumwayae. Proceedings of the National Academy of Science, U. S. A. 99, 10970-10975. Boschker, H.T.S., de Brouwer, J.F.C. and Cappenberg, T.E. 1999. The contribution of macrophyte-derived organic matter to microbial biomass in salt-marsh sediments: Stable carbon isotope analysis of microbial biomarkers. Limnology and Oceanography 44, 309-319. Boschker, H.T.S., Wielemaker, A., Schaub, B.E.M. and Holmer, M. 2000. Limited coupling of macrophyte production and bacterial carbon cycling in the sediments of Zostera spp. meadows. Marine Ecology Progress Series 203, 181-189. Buist, P.H. and Behrouzian, B. 1996. Use of deuterium kinetic isotope effects to probe the cryptoregiochemistry of ∆9 desaturation. Journal of the American Chemical Society 118, 6295-6296. Buist, P.H. and Behrouzian, B. 1998. Deciphering the cryptoregiochemistry of oleate ∆12 desaturase: a kinetic isotope effect study. Journal of the American Chemical Society 120, 871-876. Buzzi, M., Henderson, R.J. and Sargent, J.R. 1997. The biosynthesis of docosahexaenoic acid [22:6(n-3)] from linolenic acid in primary hepatocytes isolated from wild northern pike. Journal of Fish Biology 51, 1197-1208. Canuel, E.A., Cloern, J.E., Ringelberg, D.B., Guckert, J.B. and Rau, G.H. 1995. Molecular and isotopic tracers used to examine sources of organic matter and its incorporation into the food webs of San Francisco Bay. Limnology and Oceanography 40, 67-81. Canuel, E.A., Freeman, K.H. and Wakeham, S.G. 1997. Isotopic compositions of lipid biomarker compounds in estuarine plants and surface sediments. Limnology and Oceanography 42, 1570-1583.

158 Carpenter, B.K. 1984. Determination of organic reaction mechanisms. John Wiley & Sons, New York, pp. 247. Colman, B., Huertas, I.E., Bhatti, S. and Dason, J.S. 2002. The diversity of inorganic carbon acquisition mechanisms in eukaryotic microalgae. Functional Plant Biology 29, 261-270. DeNiro, M.J. and Epstein, S. 1977. Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197, 261-263. DeNiro, M.J. and Epstein, S. 1978. Influence of diet on the distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta 42, 495-506. Fang, J., Abrajano, T.A., Comet, P.A., Brooks, J.M., Sassen, R. and MacDonald, I.R. 1993. Gulf of Mexico hydrocarbon seep communities XI. Carbon isotopic fractionation during fatty acid biosynthesis of seep organisms and its implications for chemosynthetic processes. Chemical geology 109, 271-279. Focken, U. and Becker, K. 1998. Metabolic fractionation of stable carbon isotopes: implications of different proximate compositions for studies of the aquatic food webs using δ13C data. Oecologia 115, 337343. Gilmour, I., Johnston, M.A., Pillinger, C.T., Pond C.M., Mattacks, C.A. and Prestrud, P. 1995. The carbon isotopic composition of individual fatty-acids as indicators of dietary history in arctic foxes on Svalbard. Philosophical Transactions of the Royal Society of London Series B 349, 135-142 Goulden, C.E. and Place, A.R. 1990. Fatty acid synthesis and accumulation rates in daphniids. Journal of Experimental Zoology 256, 168-178. Gurr, M.I. and Harwood, J.L. Lipid Biochemistry. Fourth edition, Chapman and Hall, London, pp. 406. Hammer, B.T., Fogel, M.L. and Hoering, T.C. 1998. Stable carbon isotope ratios of fatty acids in seagrass and redhead ducks. Chemical Geology 152, 29-41. Harrison, P.J., Waters, R.E. and Taylor, F.J.R. 1980. A broad spectrum artificial seawater medium for coastal and open ocean phytoplankton. Journal of Phycology 16, 28-35. Harwood, J.L. 1996. Recent advances in the biosynthesis of plant fatty acids. Biochimica et Biophysica Acta 1301, 7-56. Henderson, R.J. and Tocher, D.R. 1987. The lipid composition and biochemistry of freshwater fish. Progress in Lipid Research 26, 281-347. Henderson, R.J. and Sargent, J.R. 1989. Lipid composition and biosynthesis in ageing cultures of the marine cryptomonad, Chroomonas Salina. Phytochemistry 28, 1355-1361.

159

Henderson, R.J. and Mackinlay, E.E. 1991. Polyunsaturated fatty acid metabolism in the marine dinoflagellate Crypthecodinium cohnii. Phytochemistry 30, 1781-1787. Hodgson, P.A., Henderson, R.J., Sargent, J.R. and Leftley, J.W. 1991. Patterns of variation in the lipid class and fatty acid composition of Nannochloropsis oculata (Eustigmatophyceae) during batch culture I. The growth cycle. Journal of Applied Phycology 3, 169-181. López-García, P., Rodríguez-Valera, F. Pedrós-Alió, C. and Moreira, D. 2001. Unexpected diversity of small eukaryotes in deep-sea Antarctic plankton. Nature 409, 603-607. MacAvoy, S.E., Macko, S.A., Joye, S.B. 2002. Fatty acid carbon isotope signatures in chemosynthetic mussels and tube worms from gulf of Mexico hydrocarbon seep communities. Chemical Geology 185, 1-8. Macko, S.A., Fogel, M.L., Hare, P.E. and Hoering, T.C. 1987. Isotopic fractionation of nitrogen and carbon in the synthesis of amino acids by microorganisms. Chemical Geology 65, 79-92. Meesapyodsuk, D., Reed, D.W., Saville, C.K., Buist, P.H. and Covello, P.S. 2000. Characterization of the regiochemistry and cryptoregiochemistry of a Caenorhabditis elegans fatty acid desaturase (FAT-1) expressed in Saccharomyces cerevisiae. Biochemistry 39, 11948-11954. Melzer, E. and Schmidt, H.-L. 1987. Carbon isotope effects on the pyruvate dehydrogenase reaction and their importance for relative carbon-13 depletion in lipids. Journal of Biological Chemistry 262, 81598164. Metz, J.G., Roessler, P., Facciotti, D., Levering, C., Dittrich, F., Lassner, M., Valentine, R., Lardizabal, K., Domergue, F., Yamada, A., Yazawa, K., Knauf, V. and Browse, J. 2001. Production of polyunsaturated fatty acids by polyketide synthases in both prokaryotes and eukaryotes. Science 293, 290-293. Meyer-Harms, B., Irigoien, X., Head, R. and Harris, R. 1999. Selective feeding on natural phytoplankton by Calanus finmarchicus before, during, and after the 1997 spring bloom in the Norwegian Sea. Limnology and Oceanography 44, 154-165. Monson, K.D. and Hayes, J.M. 1980. Biosynthetic control of the natural abundance of carbon-13 at specific positions within fatty acids in Escherichia coli. Journal of biological chemistry 255, 11435-11441. Monson, K.D. and Hayes, J.M. 1982a. Carbon isotopic fractionation in the biosynthesis of bacterial fatty acids. Ozonolysis of unsaturated fatty acids as a means of determining the intramolecular distribution of carbon isotopes. Geochimica et Cosmochimica Acta 46, 139-149. Monson, K.D. and Hayes, J.M. 1982b. Biosynthetic control of the natural abundance of carbon 13 at specific positions within fatty acids in Saccharomyces cerevisiae. Journal of Biological Chemistry 257, 5568-5575.

160

Moreno, V.J., De Moreno, J.E.A. and Brenner, R.R. 1979a. Fatty acid metabolism in the calanoid copepod Paracalanus parvus: 1. Polyunsaturated fatty acids. Lipids 14, 313-317. Moreno, V.J., De Moreno, J.E.A. and Brenner, R.R. 1979b. Fatty acid metabolism in the calanoid copepod Paracalanus parvus: 2. Palmitate, stearate, oleate and acetate. Lipids 14, 318-322. Murphy, D.E. and Abrajano, T.A. Jr. 1994. Carbon isotope compositions of fatty acids in mussels from Newfoundland estuaries. Estuarine, coastal and shelf science 39, 261-272. Otsuka, H. and Morimura, Y. 1966. Changes in fatty acid composition of Chlorella ellipsoidea during its cell cycle. Plant Cell Physiology 7, 663-670. Park R. and Epstein, S. 1961. Metabolic fractionation of 13C and 12C in plants. Plant Physiology 36, 133138. Parker, P.L. 1962. The biogeochemistry of the carbon of fatty acids. Carnegie Institution of Washington Yearbook 61, 187-190. Parker, P.L. 1964. The biogeochemistry of the stable isotopes of carbon in a marine bay. Geochimica et Cosmochimica Acta 28, 1155-1164. Peterson B.J. and Fry B. 1987. Stable isotopes in ecosystem studies. Annual review of ecology and systematics 18, 293-320. Pinilla, A., Camps, F. and Fabrias, G. 1999. Cryptoregiochemistry of the ∆11 myristoyl CoA desaturase involved in the biosynthesis of Spodoptera littoralis sex pheromone. Biochemistry 38, 15272-15277. Pond, C.M., Mattacks, C.A., Gilmour, I., Johnston, M.A., Pillinger, C.T. and Prestrud, P. 1995. Chemical and carbon isotopic composition of fatty acids in adipose tissue as indicators of dietary history in wild arctic foxes (Alopex lagopus) on Svalbard. Journal of Zoology, London 236, 611-623. Pond, D.W., Dixon, D.R., Bell, M.V., Fallick, A.E. and Sargent, J.R. 1997a. Occurrence of 16:2(n-4) and 18:2(n-4) fatty acids in the lipids of the hydrothermal vent shrimps Rimicaris exoculata and Alvinocaris markensis: nutritional and trophic implications. Marine Ecology Progress Series 156, 167-174. Pond, D.W., Segonzac, M., Bell, M.V., Dixon, D.R., Fallick, A.E. and Sargent, J.R. 1997b. Lipid and lipid carbon stable isotope composition of the hydrothermal vent shrimp Mirocaris fortunata: evidence for nutritional dependence on photosynthetically fixed carbon. Marine Ecology Progress Series 157, 221-231. Pond, D.W., Bell, M.V., Dixon, D.R., Fallick, A.E., Segonzac, M. and Sargent, J.R. 1998. Stable-carbonisotope composition of fatty acids in hydrothermal vent mussels containing methanotrophic and thiotrophic bacterial endosymbionts. Applied and Environmental Microbiology 64, 370-375.

161

Pond, D.W., Sargent, J.R., Fallick, A.E., Allen, C., Bell, M.V. and Dixon, D.R. 2000. δ13C values of lipids from phototrophic zone microplankton and bathypelagic shrimps at the Azores sector of the Mid-Atlantic Ridge. Deep-Sea Research I 47, 121-136. Pond, D.W., Allen, C.E., Bell, M.V., Van Dover, C.L., Fallick, A.E., Dixon, D.R. and Sargent, J.R. 2002. Origins of long-chain polyunsaturated fatty acids in the hydrothermal vent worms Ridgea piscesae and Protis hydrothermica. Marine Ecology Progress Series 225, 219-226. Post, D.M. 2002. Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83, 703-718. Ramos, C.S., Parrish, C.C., Quibuyen, T.A.O. and Abrajano, T.A. 2003. Molecular and carbon isotopic variations in lipids in rapidly settling particles during a spring phytoplankton bloom. Organic Geochemistry 34, 195-207. Rieley, G., Van Dover, C.L., Hedrick, D.B. and Eglinton, G. 1999. Trophic ecology of Rimicaris exoculata: a combined lipid abundance/stable isotope approach. Marine Biology 133, 495-499. Roessler, P.G. 1990. Environmental control of glycerolipid metabolism in microalgae: commercial implications and future research directions. Journal of Phycology 26, 393-399. Rossmann, A., Butzenlechner, M. and Schmidt, H.-L. 1991. Evidence for a nonstatistical carbon isotope distribution in natural glucose. Plant Physiology 96, 609-614. Sargent, J.R. and Falk-Petersen, S. 1988. The lipid biochemistry of calanoid copepods. Hydrobiologia 167/168, 101-114. Sargent, J.R., Bell, J.G., Bell, M.V., Henderson, R.J. and Tocher, D.R. 1995. Requirement criteria for essential fatty acids. Journal of Applied Ichtyology 11, 183-198. Sargent, J. R. and Whittle, K.J. 1981. Lipids and hydrocarbons in the marine food web. In: A.R. Longhurst (ed.), Analysis of Marine Ecosystems. Academic Press, London, U.K., p. 491-533. Savile, C.K., Reed, D.W., Meesapyodsuk, D., Covello, P.S. and Buist, P.H. 2001. Cryptoregiochemistry of a Brassica napus fatty acid desaturase (FAD3): a kinetic isotope effect study. Journal of the Chemical Society – Perkin Transactions 1 2001, 1116-1121. Schmid, K.M. and Ohlrogge, J.B. 1996. Lipid metabolism in plants. In: D.E. Vance and J.E. Vance (Eds.) Biochemistry of Lipids, Lipoproteins and Membranes, New Comprehensive Biochemistry 31, Elsevier Science, Amsterdam, p. 363-389. Schneider, J.C. and Roessler, P. 1994. Radiolabeling studies of lipids and fatty acids in Nannochloropsis (Eustigmatophyceae), an oleaginous marine alga. Journal of Phycology 30, 594-598.

162

Schouten, S., Klein-Breteler, W.C.M., Blokker, P., Schogt, N., Rijpstra, W.I.C., Grice, K., Baas, M. and Sinninghe-Damsté, J.S. 1998. Biosynthetic effects on the stable carbon isotopic compositions of algal lipids: Implications for deciphering the carbon isotopic biomarker record. Geochimica et Cosmochimica Acta 62, 1397-1406. Schwender, J. and Ohlrogge, J.B. 2002. Probing in vivo metabolism by stable isotope labeling of storage lipids and proteins in developing Brassica napus embryos. Plant Physiology 130, 347-361. Shifrin, N.S. and Chisholm, S.W. 1981. Phytoplankton lipids: interspecific differences and effects of nitrate, silicate and light-dark cycles. Journal of Phycology 17, 374-384. Stott, A.W., Davies, E. and Evershed, R.P. 1997. Monitoring the routing of dietary and biosynthesised lipids through compound- specific stable isotope (δ13C) measurements at natural abundance. Naturwissenschaften 84, 82-86. Sukenik, A. and Carmeli, Y. 1990. Lipid synthesis and fatty acid composition in Nannochloropsis sp. (Eustigmatophyceae) grown in a light-dark cycle. Journal of Phycology 26, 463-469. Sukenik, A. and Livne, A. 1991. Variations in lipid and fatty acid content in relation to acetyl CoA carboxylase in the marine prymnesiophyte Isochrysis galbana. Plant Cell Physiology 32, 371-378. Sukenik, A., Yamaguchi, Y. and Livne, A. 1993. Alterations in lipid molecular species of the marine eustigmatophyte Nannochloropsis sp.. Journal of Phycology 29, 620-626. Tolosa, I., Lopez, J.F., Bentaleb, I., Fontugne, M. and Grimalt, J.O. 1999. Carbon isotope ratio monitoringgas chromatography mass spectrometric measurements in the marine environment: biomarker sources and paleoclimate applications. Science of the Total Environment 237/238, 473-481. Uhle, M.E., Macko, S.A., Spero, H.J., Engel, M.H. and Lea D.W. 1997. Sources of carbon and nitrogen in modern planktonic foraminifera: the role of algal symbionts as determined by bulk and compound specific stable isotopic analyses. Organic Geochemistry 27, 103-113. van Dongen, B.E., Schouten, S., Sinninghe Damsté, J.S. 2002. Carbon isotope variability in monosaccharides and lipids of aquatic algae and terrestrial plants. Marine Ecology Progress Series 232, 8392. Vogler, E.A. and Hayes, J.M. 1980. Carbon isotopic compositions of carboxyl groups of biosynthesized fatty acids. In: A.G. Douglas and J.R. Maxwell (eds.) Advances in Organic Geochemistry, 1979, Pergamon Press, 697-704. Wakeham, S.G., Hedges, J.I., Lee, C., Peterson, M.L. and Hernes, P.J. 1997. Compositions and transport of lipid biomarkers through the water column and surficial sediments of the equatorial Pacific Ocean. DeepSea Research II 44, 2131-2162.

163

Wallis, J.G., Watts, J.L. and Browse, J. 2002. Polyunsaturated fatty acid synthesis: what will they think of next? TRENDS in Biochemical Sciences 27, 467-473. Weinert, T.J., Blomquist, G.J. and Borgeson, C.E. 1993. De novo biosynthesis of linoleic-acid in 2 noninsect invertebrates – The land slug and the garden snail. Experientia 49, 919-921. Woodbury, S.E., Evershed, R.P., Rossell, J.B., Griffith, R.E. and Farnell, P. 1995. Detection of vegetable oil adulteration using gas chromatography combustion/isotope ratio mass spectrometry. Analytical Chemistry 67, 2685-2690.

164

5. The Fatty Acid Composition of Marine Pelagic Organisms off Vancouver Island, Canada; Nature and Nurture

5.1 Abstract The fatty acid composition of particulate organic matter (POM), zooplankton, larval fish and juvenile salmon (muscle), collected off Vancouver Island, Canada, was measured and analyzed with multivariate statistical techniques. As reported in Chapter 3, POM was found to vary spatially, such that waters on the continental shelf were more diatom-rich than POM off shelf. The resulting differences in fatty acid composition of the POM also caused a spatial disparity in the fatty acid composition of higher trophic level organisms at least up to larval fish. In spite of the differences at the base of the food webs, the largest part of the variation in the fatty acid compositions of all analyzed organisms can be explained by differences between the trophic groups (POM, zooplankton, larval fish). The proportion of C20 and C22 polyunsaturated fatty acids (PUFAs) was found to increase with trophic level. When plotting the abundance of docosahexaenoic acid (22:6n-3; DHA) versus that of 14:0 or 22:5n-3, good separation between the trophic groups was achieved. Furthermore, it was found that the DHA/20:5n-3 ratio increased, with POM < zooplankton < larval fish < juvenile salmon. A positive correlation was found between the δ15N and the abundance of DHA, suggesting that biomagnification of DHA occurs. However, data from previous studies indicate that the DHA abundance versus trophic level relationship does not hold beyond larval/juvenile fish. In land mammals and birds a decrease in the proportion of DHA with body size has been reported, suggesting physiological constraints on the abundance of this fatty acid. Fatty acids that are not synthesized by the animal, but are not essential for the organism, seemed to be the best trophic markers. Previous studies suggest that adipose tissue may be the tissue that reflects the dietary fatty acid compositions most faithfully.

165

5.2 Introduction In Chapter 3 factors influencing the fatty acid composition of particulate organic matter (POM) were discussed. This chapter will focus on trends observed in the fatty acid profiles of organisms when going up in the marine pelagic food web off Vancouver Island. As shown by many studies, different diets with dissimilar fatty acid profiles will result in distinct fatty acid composition of an organism (e.g., Graeve et al., 1994; St. John and Lund, 1996; Kirsch et al., 1998). Therefore, fatty acid profiles can harbour valuable information on the dietary preferences of heterotrophs. The advantage of the fatty acid data over gut content analysis data is that they give temporally, and spatially integrated information. Hence, the fatty acid composition can help in elucidating more general feeding habits, i.e., the average food consumption, rather than the most recent food ingested. After food intake, dietary lipids are made soluble and are hydrolysed in the gut. Once absorbed, the fatty acids and the digestive products are activated and re-esterified. The various lipids, such as the triacylglycerols and phosphoglycerides, are then packaged into particles called chylomicrons. These plasma lipoproteins collect in the lymphatic system and ultimately enter the blood vascular system. The chylomicrons are degraded in tissues such as adipose tissue and muscle. Here, they can be hydrolyzed by lipases, absorbed by the tissue, and become re-esterified again. Also, smaller lipoproteins can be formed as breakdown products. These can then be carried to other organs, such as the liver, for further metabolism or degradation (for reviews of absorption, transportation, deposition and mobilization of lipids, see Sheridan, 1988; Gurr and Harwood, 1991). Naturally, due to selective incorporation and metabolic processes the fatty acid composition of an organism will not be the same as its diet. In copepods, for example, it has been shown that polyunsaturated fatty acids (PUFAs) are preferentialy removed as food is passing through the gut (Prahl et al., 1984; Neal et al., 1986; Harvey et al., 1987).

166 Fatty acids derived from the diet can be elongated in mitochondria and the endoplasmic reticulum (Cook, 1996). Fatty acids can also be further desaturated. Most animals lack desaturases that can introduce double bonds beyond the C9-position and therefore are not able to synthesize n-6 and n-3 fatty acids de novo. However, ∆12 desaturases have been found in some terrestrial invertebrates, allowing them to synthesize n-6 fatty acids de novo (Weinert et al., 1993; Batcabe et al., 2000). Although, it is unclear whether marine invertebrates are also able to do so. The marine fish studied so far show a general inability to desaturate and elongate C18 PUFAs, making preformed C20 and C22 PUFAs essential fatty acids in their diet (Henderson and Tocher, 1987; Buzzi et al., 1997 and references therein). Polar lipids, such as the phosphoglycerides, are particularly important in maintaining the structural integrity of cell membranes. Therefore, it can be expected that these lipids have larger constraints on the variability of their fatty acid composition than triacylglycerols (TAGs), which serve mostly as energy storage molecules. Also, some tissues have stricter requirements, and thus, display less variable fatty acid compositions. Crawford et al. (1976) reported that the fatty acid profiles of phospholipids from brains of various animals, with very different diets and body sizes, are remarkably similar. Despite the physiological constraints and modifications, described above, dietary signatures can be quite evident in the fatty acid profiles of higher organisms (see e.g., Jeffries, 1970; Desvilettes et al., 1994; Graeve et al., 1994; St. John and Lund, 1996; Napolitano et al., 1997; Kirsch et al., 1998; Oltra et al., 2000). Most studies that researched the trophic transfer of fatty acids focussed on a single predator-prey step. Only a small number of investigations has examined the fatty acid transfer across multiple trophic levels (e.g., Fraser et al., 1989; St. John and Lund, 1996). Here, data of the fatty acid composition of POM, zooplankton, larval fish and juvenile salmon (muscle), collected off Vancouver Island, Canada, are presented and discussed. The applicability of fatty acids as trophic markers is briefly examined. Additionally, multivariate techniques have been employed, and apparent trends are compared with literature data.

167

5.3 Methods 5.3.1 Sample collection Sample collection details are described in Chapter 2. The locations of sampling are shown in Figure 2.1. The sampling procedure for the juvenile salmon is described in Chapter 4. However, in addition to the samples described in Chapter 4, juvenile salmon from a cruise in June 1999 were analyzed for fatty acid composition. These samples were taken from the CSS W.E. Ricker (Cruise No. HS9914) in coastal waters on the continental shelf off Vancouver Island. Similar procedures as outlined in Chapter 4 were followed, and on land the samples were stored at -80ºC.

5.3.2 Fatty acid methyl ester preparation and analyses Please refer to the methods section in Chapter 2 for the fatty acid methyl ester preparation, separation, quantification and identification methods.

5.3.3 δ15N measurements Approximately 0.4 mg of freeze-dried and powdered sample was weighed and placed into a tin capsule. In the case of POM samples, a sub-sample of the glass fibre filter was cut into 0.2 cm2 pieces and placed into the tin capsule. The stable nitrogen isotope ratio was measured using a Carlo Erba elemental analyzer interfaced with the Finnigan MAT 252 isotope ratio mass spectrometer. N2 gas with a known δ15N (vs. air) was used as a reference during each run, and an acetanilide standard (δ15N = -2.19‰) was run 5 times during a 50-sample session. The standard deviation of the repeated acetanilide runs was typically below 0.1‰ (range: 0.04 – 0.12‰). However, δ15N values of actual samples were less reproducible due to inhomogeneities in the sample material, and standard deviations of 0.2 and 0.3‰ were commonly observed. It was found that no reliable δ15N value could be obtained for most zooplankton samples. This is due to contamination of un-evaporated (recrystallized) ammonium

168 formate that was used to rinse sea-salt off the zooplankton. Only seven zooplankton samples that were not contaminated with the ammonium formate were used here.

5.3.4 Multivariate statistics A matrix with the relative amounts of 32 fatty acids for 197 samples of POM, zooplankton and larval fish combined (from all 3 cruises) was used. As described in Chapter 2, linear discriminant analysis was used to separate between samples collected in waters on the continental shelf (stations with water depth <200m) and samples from off the shelf (water depth >200m). In order to find a smaller set of variables that optimized the separation, stepwise discriminant analysis was used as a variable selection technique prior to the discriminant analysis. The same procedures were followed to find optimal separation between samples collected north of Line G (see Figure 2.1) and samples from Line G and to the south of it. Table 5.1 shows the fatty acids used, and the coefficients of the discriminant functions. As can be observed from the various estimates of the classification error rates (see Chapter 2 for explanation), no large north–south distinction was observed.

Table 5.1. Results of discriminant analysis on combinations of fatty acids selected by stepwise discriminant analysis. The shelf-off shelf discriminant function (top) best separates samples taken from waters above the continental shelf (stations with a water depth <200m) and samples collected away from the shelf (stations with >200m water depth). The north-south discriminant function (bottom) best separates samples taken from northern stations (north of Line G) and southern stations (Line G and more south). The coefficients of the standardized discriminant function and the partial F-test results are a measure of separation contributed by the individual fatty acids (in the presence of the others). *: df: 1, n-p-1.

Shelf – Off shelf discriminant analysis (n=197, p=8) Fatty acid

14:0 16:1n5 16:2n4 18:1n9 18:1n7

20:0 20:2n6 20:4n3

Discr. Funct.

0.238

1.729

-1.479

0.298

-0.616

-2.063

2.539

1.170

Standardized Discr. Funct.

1.378

0.536

-0.764

0.904

-0.791

-0.584

0.691

0.622

Part.F-values*

18.97

4.54

8.51

15.66

10.57

6.75

9.24

5.41

Re-substitution Leave-1-out Error rate (%)

14.7

15.7

Leave-5-out Mean-1/3-subset 15.9

16.7

169 North - South discriminant analysis (n=197, p=5) Fatty acid

16:1n7

16:2n4

18:1n5

20:0

20:5n3

Discr. Funct.

0.109

1.352

0.956

-1.454

0.115

Standardized Discr. Funct.

0.562

0.697

0.432

-0.410

0.737

2.22

4.62

5.02

3.59

11.18

Part.F-values*

Error rate (%)

Re-substitution

Leave-1-out

Leave-5-out

Mean-1/3-subset

25.3

27.8

27.4

28.0

For the principal component analysis the correlation matrix was used in order to standardize the data set (see also Meglen, 1992). All matrix computations and statistical tests were performed by using the software package Matlab (version 4.2b, The Mathworks Inc., Natick, MA, U.S.A.).

5.4 Results and Discussion 5.4.1 Spatial variation in fatty acid composition To illustrate the variation of the fatty acid composition of different marine algal taxa, data from the literature were collected and principal component analysis was performed. All the reported fatty acid compositions that were used for the compilation were from single species lab cultures. As often a limited number of fatty acids are reported, only 12 fatty acids were used for the principal component analysis. Literature data were plotted on the axes of the first two principal components in Figure 5.1. As can be observed, algae from similar taxa often plot together, indicating that the fatty acid composition can be taxon specific. Therefore, fatty acids have previously been proposed as dietary indicators or trophic markers since it has been found that differences in the algal species consumed are reflected in the fatty acid composition of zooplankton (e.g., Fraser et al., 1989; Desvillettes et al., 1994; Graeve et al., 1994; St. John and Lund, 1996; Napolitano et al., 1997; Virtue et al., 2000). It should be noted, however, that large variation occurs between single species, and overlap exists between the fatty acid composition of algae from different taxa. López Alonso et al. (1992) even found considerable variation in the

170 fatty acid composition of different isolates of a single strain of Isochrysis galbana grown under identical conditions. Hence, only qualitative and, perhaps, semi-quantitative assessments of the dietary intake can be made when measuring the fatty acid composition of a heterotroph captured in the field.

0.6 18:3 (all isom.)

PC-2 (15.4% of variance)

0.4 18:2n-6 0.2

0.0

-0.2

16:3 (all isom.) 16:2 (all isom.)

18:1 (all isom.) 18:4n-3

20:5n-3 16:1 (all isom.) 14:0 18:0 16:0 22:6n-3

-0.4

-0.6 -0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

PC-1 (32.4% of variance) 30

PC-2 (15.4% of variance)

20 Centric diatoms Pennate diatoms Dinoflagellates Prymnesiophytes Cryptophytes Chrysophytes Chlorophytes Xantophytes Eustigmatophytes Cyanobacteria POM this study

10

0

-10

-20 -30

-20

-10

0

10

20

30

PC-1 (32.4% of variance)

Figure 5.1. Loading plot (top) and score plot (bottom) of principal component analysis of fatty acid abundance data from cultured algae (derived from the literature). The references used are mentioned in the text. POM measured for this study is also plotted, and the dashed box is enlarged in Fig. 5.2. For the 16;1, 16:2, 18:1 and 18:3 fatty acids, all isomers were grouped, as in several reports no distinction was made between the various isomers.

171

10

Shelf / South Shelf / North Off shelf / South Off shelf / North

8 6 4

PC-2

2 0 -2 -4 -6 -8 -10 -10

-8

-6

-4

-2

0

2

4

6

8

10

PC-1 Figure 5.2. POM samples plotted on the first and second principal component axes that were produced by the principal component analysis on the algal culture data (Fig. 5.1). The shelf samples, and also some northern off shelf samples, plot mostly in the lower left quadrant. In this quadrant the majority of the cultured diatoms plot as well, suggesting a higher abundance of diatoms in the POM samples occupying that part of the plot.

To compare the fatty acid composition of the POM sampled for this study with that of the algal cultures from the literature the POM samples were plotted on the same principal component axes (Figures 5.1 and 5.2). First of all it is apparent that the POM samples are clustered tightly close to the middle, in between the algal groups. This is partly due to the fact that the POM consists of a mixture of different algal classes and dead organic matter. In Figure 5.2 it can be observed that on average the samples collected in shelf waters plot more towards the diatom cluster than POM away from the shelf. Many samples from the off shelf region plot in proximity of the dinoflagellate cluster. However, no great abundance of dinoflagellates has been found in water from the off shelf stations sampled for phytoplankton counts (see Chapter 3). In the literature data set the dinoflagellates were found to be more enriched in the C16 and C18 saturated fatty acids (see lower right corner in Figure 5.1). In contrast to their labile unsaturated

172 counterparts, these fatty acids are known to be preferentially preserved in sinking dead organic matter (Wakeham et al., 1984, 1997). Thus, the off-shelf samples plotting near the dinoflagellate cluster are not necessarily rich in dinoflagellates, instead they may be richer in degrading organic matter.

Shelf / South Shelf / North Off shelf / South Off shelf / North

North - south discriminant score

8

6

4

2

0

-2

0

2

4

6

8

10

Shelf - off shelf discriminant score Figure 5.3. Score-plot of two discriminant functions that optimally separate shelf- and off shelf samples, and samples taken from northern and southern stations. The sample set includes POM, zooplankton and larval fish. See Table 5.1 for the coefficients of the two discriminant functions, and results of the classification analysis.

Discriminant analysis and a variable selection technique were employed on fatty acid composition data of POM, zooplankton and larval fish samples from three separate cruises off the west coast of Vancouver Island. Typically, it was found to be possible to distinguish between organisms that were collected in either shelf- or off shelf waters with the use of fatty acid data (see Figures 5.3 and 2.3). This distinction was more evident off Estevan point (line G, see Figure 2.1) and at the more southern stations (Figure 5.3). As discussed in Chapter 3, the shelf - off shelf difference in POM can mostly be attributed to a higher abundance of diatoms in shelf waters.

173 It was tested which fatty acids could individually distinguish shelf- from off shelf food web organisms best (by using only one variable in the discriminant analysis). The best four fatty acids were found to be the 16:2n-4, 16:4n-1, 18:1n-7 and 16:3n-4 fatty acids. The C16 unsaturated fatty acids are characteristically found in higher amounts in diatoms (Chuecas and Riley, 1969; Pohl and Zuhrheide, 1979; Sargent et al., 1988). Therefore, the differences in fatty acid composition of the POM are also expressed in zooplankton and larval fish, and seem to persist through at least the lower end of the food web.

In conclusion, the variation in fatty acid composition of the sampled POM was found to be small in comparison to the variation observed among cultured algae from different taxonomic groups. Still, the differences were large enough to allow an approximate differentiation between zooplankton and larval fish feeding in diatom-rich waters versus those feeding in diatom-poor waters.

5.4.2 Differences between trophic groups Principal component analysis was performed on the same data set as used in Figure 5.3. However, the 32 most abundant fatty acids were used and no variable selection method was employed before the analysis, as was done prior to the discriminant analysis. As can be seen in Figure 5.4, most of the variance in the data set is caused by differences in the fatty acid composition between the different trophic groups (i.e., POM, zooplankton and larval fish). As indicated by the loading plot (Fig. 5.4a), these differences are mostly caused by higher abundance of the C20 and C22 PUFAs in the zooplankton and particularly in larval fish. Fatty acids with shorter carbon chains tend to be more concentrated in POM (i.e., they plot on the left hand side in Fig. 5.4a). To illustrate the observations from the principal component analysis the relative abundances of docosahexaenoic acid (DHA: 22:6n-3) and myristic acid (14:0) were plotted against each other (Figure 5.5a). Samples of crab larvae collected on the same cruises and samples of juvenile salmon muscle tissue were added to the plot. Figure 5.5a

174

PC-2 score (13.2% of variance)

0.9 18:4n3

0.6

16:3n4 16:4n1 16:2n4

0.3

0.0

16:1n9 14:0 16:1n7 iso 15:0

-0.3

22:1 20:1

20:3n3 iso 17:0

18:3n3 16:1n5 18:2n6

20:4n3 20:4n6

18:1n5 16:4n3

22:5n3

17:0

18:1n7

18:1n9

16:0

-0.3

22:6n3

20:2n6

15:0

-0.6

20:5n3

22:5n6

20:0

-0.6 -0.9

21:5n3

0.0

0.3

18:0

0.6

0.9

PC-1 score (30.8% of variance)

PC-2 score (13.2% of variance)

40

POM / off shelf POM / shelf Zoopl. / off shelf Zoopl. / shelf L.Fish / off shelf L.Fish / shelf

30 20 10 0 -10 -20 -30 -40

-30

-20

-10

0

10

20

30

40

PC-1 score (30.8% of variance)

Figure 5.4. Loading plot (top) and score plot (bottom) of principal component analysis of fatty acid abundance data from POM, zooplankton and larval fish collected during three separate cruises. The 32 most abundant fatty acids were used in the analysis. By comparing the score- and loading plots it can be observed that from POM to zooplankton, to larval fish the proportion of C20 and C22 PUFAs increases, and fatty acids with shorter carbon chains decrease.

shows an apparent trend with the abundance of DHA increasing in the following order: POM < zooplankton < larval fish < juvenile salmon muscle. The abundance of the 14:0 fatty acid shows the opposite trend, with the highest relative amounts present in POM.

175 Additionally, the two C22 PUFAs, DHA and docosapentaenoic acid (DPA), were plotted against each other. As expected (both have high loadings for PC-1; Fig. 5.4a), the abundance for both fatty acids increases, with POM < zooplankton < larval fish < juvenile salmon muscle (Figure 5.5b). Especially when plotting these two fatty acids a

22:6n-3 abundance (% of total fatty acids)

very good separation of the different trophic groups was achieved.

45

POM Zooplankton < 425µm Zooplankton > 425µm Crab larvae Jellyfish Larval fish Squid larvae Juvenile salmon (muscle)

40 35 30 25

a

20 15 10 5 0 0

2

4

6

8

10 12 14 16 18 20 22 24

22:6n-3 abundance (% of total fatty acids)

14:0 abundance (% of total fatty acids)

b

45 40 35 30 25 20 15 10 5 0 0.2

0.4

0.6 0.8 1

2

4

22:5n-3 abundance (% of total fatty acids)

Figure 5.5. The relative abundance of the 14:0 (a) and 22:5n-3 (b) fatty acids plotted against that of 22:6n-3 (DHA). Note that the relative abundance of 22:5n-3 has been plotted on a logarithmic scale.

176

Table 5.2. Ratios of the abundances (% of total) of unsaturated over saturated fatty acids, and PUFAs over saturated and monounsaturated fatty acids. The abundance of the fatty acids from the n-3 and n-6 series, and their ratio. The abundance of 22:6n-3 (DHA). The ratios of 22:5n-3 (DPA) over 14:0, DHA over 14:0 and DHA over 20:5n-3 (EPA). All values are reported as an average and median for each trophic group, and the standard deviations are indicated. * Next to samples from off the west coast of Vancouver Island, both zooplankton and juvenile salmon groups include also samples taken near Baranof Island (off south Alaska) and Dixon Entrance (off British Columbia, Canada). Category

Statistic

Unsat/Sat Poly/(Sat+Mono)

POM

Average

1.2

(n=70)

Median ± 1 SD

DHA/14:0 DHA/EPA

n-3

n-6

n-3 / n-6

DHA

DPA/14:0

0.4

18

3.0

6.5

4.3

0.02

0.3

0.8

1.2

0.3

18

3.0

6.0

3.8

0.02

0.3

0.7

±0.3

±0.2

±6.8

±0.9

±2.9

±2.3

±0.01

±0.3

±0.4

Zooplankton* Average

2.4

0.9

39

3.1

16

15

0.15

3.6

1.0

(n=118)

Median

2.6

0.9

39

2.9

13

14

0.13

3.4

1.0

± 1 SD

±0.9

±0.4

±10

±1.6

±11

±5.5

±0.12

±2.7

±0.3

Larval Fish

Average

2.4

1.0

45

3.0

16

24

0.60

12

1.8

(n=52)

Median

2.4

1.1

46

3.0

16

25

0.50

10

1.8

± 1 SD

±0.6

±0.2

±7.4

±0.7

±5.1

±5.8

±0.40

±8.1

±0.4

Juv. Salmon* Average

2.8

1.6

56

3.6

19

36

3.4

37

2.7

(n=121)

Median

2.8

1.7

57

2.7

20

37

3.2

35

2.7

± 1 SD

±0.2

±0.3

±5.1

±2.4

±7.0

±5.7

±1.5

±16

±0.6

177 Similarly, the DHA/eicosapentaenoic acid (EPA: 20:5n-3) ratio increases, with: POM< zooplankton< larval fish< juvenile salmon muscle (Figure 5.6 and Table 5.2). Another trend is an increasing proportion of unsaturated fatty acids, coinciding with a greater share of n-3 unsaturated fatty acids in the same order (Table 5.2). However, this trend is mostly driven by an increase in DHA (Table 5.2). Therefore, the presented data seem to indicate that a significant biomagnification of DHA occurs, with DHA getting more concentrated in marine animals feeding at higher trophic levels (up to juvenile salmon, at least).

POM Zooplankton < 425µm Zooplankton > 425µm Crab larvae Jellyfish Larval fish Squid larvae Juv.salmon (muscle)

2:1

4:1

22:6n-3 abundance (% of total fatty acids)

45 40 35 30

1 1:

25 20

1:2

15 10 5 0 0

5

10

15

20

25

30

35

40

45

20:5n-3 abundance (% of total fatty acids)

Figure 5.6. The relative abundance of the 20:5n-3 fatty acid (EPA) plotted against that of 22:6n-3 (DHA). Lines with constant DHA/EPA ratios are indicated.

5.4.3 Fatty acid proxy for trophic level? Due to urinary loss of

15

N depleted ammonium and urea the

15

N/14N isotope ratio of

nitrogen found in organisms increases with trophic position (Peterson and Fry, 1987, and

178 references therein). To further test the notion that DHA tends to get more concentrated in organisms feeding at higher trophic levels, the abundance of DHA was compared with δ15N measurements on the same organisms (Figure 5.7a). Although a large variation in the δ15N of POM was observed, a good correlation (R2=0.69) between the relative amount of DHA and the δ15N was found (Figure 5.7a). This correlation confirms that in this dataset DHA increases with trophic level. An equally good correlation was found between (the logarithm of) the DHA/14:0 ratio and δ15N (Figure 5.7b). The DHA/EPA ratio was less successful in predicting the δ15N (Figure 5.7c). Natural variation in diet, physiology and physical parameters such as temperature can play an important role in determining the abundance of DHA (see e.g., Henderson and Tocher, 1987; March, 1993). However, superimposed on this, there seems to be a general trend of increasing levels of DHA with increasing tropic position (Figure 5.7a). Unfortunately, the proportions of the various lipid classes were not measured for this study. There is a distinct possibility that the relative abundance of TAGs decreases with POM > zooplankton > larval fish > juvenile salmon muscle. If that were indeed the case, patterns observed in Figures 5.5-5.7 can be expected, since TAGs are generally richer in saturated and mono-unsaturated fatty acids than polar lipids.

5.4.4 Literature data To further investigate whether biomagnification of DHA is a common and useful effect, data from the literature were compiled. Data were included in the dataset only when the whole organism or muscle tissue was analyzed, and when the relative abundance of the total fatty acids (from all lipid classes) was reported. Except for the algae, only fatty acid compositions of wild marine organisms were included in the dataset. No distinction was made between organisms which were part of benthic or pelagic food webs (or both). Table 5.3 lists the references used for the fatty acid composition of the (wild) marine organisms. The algae included in the data set are all lab-cultured algae (same as shown in Fig. 5.1). For the fatty acid data on algal cultures the following references were used:

179 Pohl and Zuhrheide (1979) and references therein, Volkman et al. (1981, 1989), BenAmotz et al. (1987), Kattner and Brockmann (1990), Thompson et al. (1992), Viso and Marty (1993), Graeve et al. (1994) and Mansour et al. (1999).

a

14 12

8

15

δ N (‰)

10

POM Zooplankton Crab larvae Larval Fish Juv. Salmon

6 4 2 0

10

20

30

40

50

DHA (% of total fatty acids) 16

15

δ N (‰)

14

16

b

12

12

10

10

8

8

6

6

4

4

2

0.1

c

14

1

DHA/14:0

10

2

0

1

2

3

4

DHA/EPA

Figure 5.7. The relative abundance of DHA (22:6n-3) plotted against the (bulk) δ15N of the same samples (a), showing an increase in DHA (%) per trophic level. The lower plots (b and c) show an increase of the DHA/14:0 and DHA/EPA fatty acid ratios with δ15N, respectively.

180 Table 5.3. References used for Figures 5.8a and b (references for algae are listed in the text). Only references were used that reported measurements on whole organisms or muscle tissue. Additionally, only total fatty acid weight percent values were used. All references report on analyses on wild marine organisms. Description

n

Location

Reference

5 1 5 10 8 5 2

SW Finland, Baltic Sea Bahía Blanca, Argentina Antarctic Ocean South Georgia, Antarctica Weddell Sea South Georgia, Antarctica ?

Linko et al. (1985) Napolitano et al. (1997) Nelson et al. (2001) Cripps and Atkinson (2000) Kattner and Hagen (1998) Cripps et al. (1999) Sargent and Falk-Petersen (1988)

6 22

Off Antarctic Peninsula ?

Phleger et al. (1997) Joseph (1982)

2 1 12

Seto Inland Sea, Japan NE Atlantic Off Antarctic Peninsula

Fukuda and Naganuma (2001) Pond and Sargent (1998) Nelson et al. (2000)

19 5 5 2 2 6 13 28

Black Sea and Marmara Sea SW Finland, Baltic Sea off SW Alaska Fraser River/estuary, BC NW Atlantic Balsfjorden, N Norway E Bass Strait, Australia Pr. William Sound, Alaska

Tanakol et al. (1999) Linko et al. (1985) Ozawa et al. (1993) Mjaavatten et al. (1998) Kirsch et al. (1998) Henderson et al. (1984) Dunstan et al. (1988) Iverson et al. (2002)

9

E Bass Strait, Australia

Dunstan et al. (1988)

3

E Bass Strait, Australia

Dunstan et al. (1988)

8 1 6

? NW Atlantic Pr. William Sound, Alaska

Joseph (1982) Kirsch et al. (1998) Iverson et al. (2002)

Zooplankton Mixed zooplankton >150um Zooplankton>200um Amphipods Various zooplankton Euphausiids (incl. eggs) Krill Copepod eggs & nauplii

Molluscs Pteropods (planktonic) Various molluscs

Gelatinous animals Jellyfish (Aurelia aurita) Doliolids (Tunicate) Cnidaria and Ctenophora

Teleost fish Various fish (dorsal muscle) Herring fillet Salmon adult (dorsal muscle) Salmon smolt (muscle) Mackerel and cod Capelin (muscle) Various fish (flesh excl. skin) Various fish

Cartilaginous fish Sharks, skates and dogfish (flesh excl. skin)

Cephalopods Octopus, cuttlefish and squid (flesh excl. skin) Planktonic squid Squid Squid and Octopus

181 In Figure 5.8a and b the relative abundance of DHA is plotted against that of the 14:0 and the 20:5n-3 fatty acids, respectively. As can be seen, when comparing with the Figures 5.5a and 5.6, there is a rough agreement between the patterns observed in the data of this study and the data collected from the literature. Most fish (together with squid) show relatively high levels of DHA, moderate levels are found in invertebrates, and most algae (except dinoflagellates) contain low amounts of DHA (Figures 5.5a, 5.6 and 5.8a, b). This pattern is less clear in the literature data. Perhaps, this can be ascribed to a wider variety of primary producers and a varying importance of benthic food organisms in the respective food webs. Benthic algae and macro-algae tend to contain higher quantities of (n-6) fatty acids (e.g. Sargent and Whittle, 1981; Pohl and Zuhrheide, 1979). It was expected that the cartilaginous fish, such as sharks, would contain the highest abundance in DHA, since many are top-predators. The DHA/EPA ratios found for cartilaginous fish are, indeed, some of the highest (Figure 5.8b). However, shark flesh (analyzed by Dunstan et al., 1988) did not exhibit higher proportions of DHA than the juvenile salmon analyzed for this study (compare Figs. 5.8 and 5.5). This may in part be explained by qualitative differences in the base of the food web, or by the fact that not all are top predators. For example, at least one of the sharks (the Port Jackson shark), was probably feeding on macro-algae-consuming sea urchins and sea snails (Dunstan et al., 1988).

The recent study by Iverson et al. (2002) measured the fatty acid composition of a very large number (1153) of fish and invertebrates from Prince William Sound, Alaska. All fish for which individuals of different sizes or age were available in this study showed a decrease of DHA concentration with age or size. This trend coincided with a decrease in PUFAs with respect to monounsaturated fatty acids (MUFAs). The proportion of saturated fatty acids (SFAs) remained roughly the same (Iverson et al., 2002). For example, Pink salmon smolts had a DHA abundance of, on average, 36% of the total fatty acid content, which is not unlike values found for juvenile salmon reported here. However, the adult Pink salmon displayed a DHA abundance of just under 20% (Iverson et al., 2002). Analogously, DHA abundances decreased from juveniles to adults in Pacific sandlance (from 23% to 16%), in Walleye pollock (from 26% to 18%), and in Pacific tomcod (from 30% to 18%). In the same study the fatty acid composition of 360

182 Pacific herring was measured. The smallest (~6cm) herring had DHA abundances of above 30%, which decreased with size to, on average, 11% in the older and larger herring (Iverson et al., 2002). Likewise, Budge et al. (2002) found a decrease in DHA abundance

22:6n-3 abundance (% of total fatty acids)

with size for capelin from the north-western Atlantic.

a

50

40

30

20

10

0 0

5

10

15

20

25

30

35

22:6n-3 abundance (% of total fatty acids)

14:0 abundance (% of total fatty acids)

b

50 Algae Dinoflagellates Zooplankton Molluscs Gelatinous animals Teleost fish Cephalopods Cartilagenous fish

40

30

20

10

0 0

10

20

30

40

50

20:5n-3 abundance (% of total fatty acids)

Figure 5.8. Data derived from the literature (see Table 5.3 for references). The relative abundance of the 14:0 (a) and 20:5n-3 (b) fatty acids are plotted against that of 22:6n-3.

183 Therefore, from the literature it seems clear that a decline of the DHA concentration with age and size is common. The fish species from the aforementioned studies increasingly feed on organisms from higher trophic levels as they grow (Grosse and Hay, 1988). Hence, the trophic level versus DHA abundance relationship, observed in the presented study, does not seem to hold beyond larval and juvenile fish. In fact, the trend may even reverse.

5.4.5 Terrestrial animals and size To further investigate whether in other food webs DHA becomes more concentrated in organisms feeding at higher trophic positions, the available literature on fatty acids in land animals was consulted. In 1976 Crawford, Casperd and Sinclair reported on the abundance of PUFAs in phosphoglycerides derived from a suite of different animals. At first glance, the fatty acid composition of liver ethanolamine phosphoglycerides (EPG) reported for the various animals seemed to corroborate the notion of biomagnification of DHA occurring in food webs. The lowest levels of DHA were found in large ruminantand non-ruminant herbivores. The often omnivorous primates (including humans) showed higher concentrations, and the carnivorous species (cat, civet, leopard and a lion) displayed the highest abundance of DHA in liver EPG (Crawford et al., 1976). Earlier, Rivers et al. (1975, 1976) already found that cats and lions are not able to desaturate linoleic- and linolenic acid to form the essential PUFAs. Therefore, selective retention and accumulation of DHA must be occurring in these carnivores. Not all of the observations by Crawford et al. (1976), however, fit the pattern of biomagnification of DHA. Notably, liver tissue of the analyzed small herbivorous mammals had the same amount of DHA as the carnivores. Other studies also found higher amounts of DHA in various tissues of smaller animals, compared to larger ones (Gudbjarnason et al., 1978; Couture and Hulbert, 1995; Pamplona et al., 2000; PorteroOtín et al., 2001; Hulbert et al., 2002a, b). In fact, the aforementioned studies found significant negative correlations between the concentration of DHA in various tissues of animals and their body size, basal metabolic rate or maximum lifespan. These relationships were described for mammals, as well as for birds (Hulbert et al., 2002a, b).

184 Several reasons for the relationship between the DHA content and, for example, body size are mentioned in the literature. In addition to DHA, the abundance of other essential long chain PUFAs, such as 20:4n-6 was shown to decrease with body size and maximum lifespan (e.g., Portero-Otín et al., 2001). In contrast, the precursor to this fatty acid, 18:2n-6, was found to increase with body size. It has been argued by Crawford et al. (1976) that the chain elongation and desaturation process does not reach completion in the liver and muscle tissue of the fastest growing animals, and in animals that attain large body weights. A supporting argument for this is the trend that fast growing, large mammals have the smallest brains in proportion to weight (Spector, 1956; Crawford et al., 1976). The brain seems to have a specific requirement for DHA (see e.g. Crawford et al., 1999). This organ shows a remarkable constancy in its fatty acid composition, regardless of species or size of the animal (Crawford et al., 1976; Hulbert et al., 2002a). It could thus be argued that a very slow rate of conversion of shorter, less unsaturated fatty acids to DHA may have been rate-limiting in the evolution of the brain in large mammals. For these reasons Crawford et al. (1999) and Broadhurst et al. (2002) propose that during the critical time of the evolution of Homo sapiens our ancestors must have had access to lacustrine (or coastal marine) food sources, which are relatively rich in DHA and also 20:4n-6. For land mammals, humans have an anomalously high brain volume in relation to their body size (Crawford et al., 1999). Another explanation of the inverse size versus membrane polyunsaturation relationship lies in the metabolic restrictions dictated by a large body size. Hulbert et al. (2002a) argue that the concentration of PUFAs (especially DHA) and size are inversely correlated because the metabolic intensity and size of a species correlate as well. Hulbert and Else (1989) found a higher metabolic rate in a (endothermic) mammal compared to a (ectothermic) reptile of the same body size and temperature. The same authors showed that the higher level of metabolism in the mammal was associated with a higher DHA content in liver and kidney phospholipids. It is known from a number of studies that a higher proportion of PUFAs (with respect to saturated and monounsaturated fatty acids) in the membrane enhances the activity of membrane proteins (see Hulbert and Else, 1999, and references therein). Phospholipids containing DHA specifically have been proposed to be essential for the assembly of membrane proteins such as rhodopsin, ion pumps and

185 the various complexes of the mitochondrial electron transport chain (see Infante et al., 2001). The lateral diffusion coefficients of membrane lipids are two orders of magnitude greater than those of proteins (Storch and Kleinfeld, 1985). This implies that membrane proteins are constantly bombarded by the membrane lipids. Hulbert and Else (1999) propose that energy transfer during intermolecular collisions within the membrane is facilitated by a higher number of double bonds. DHA has both the greatest number of double bonds and the most even distribution of these units throughout the bilayer depth. Therefore, this fatty acid is proposed to be optimal for facilitating energy transfer, and thus protein activity (Hulbert and Else, 1999). Thus, according to Hulbert and Else (1999), the reason why larger animals have a lower degree of polyunsaturation and less DHA is because the metabolic rate would become too high. It has been calculated that if a rhinoceros had a metabolic rate similar to a mouse it would require a surface temperature of 100ºC to relieve itself from its basal heat production (Hulbert and Else, 1999). Therefore, a decline in metabolic intensity, attenuated by a lower concentration of PUFAs, became necessary during the evolution of large animals.

In conclusion, land animals show some biomagnification of DHA within the terrestrial food web (see Crawford et al., 1976). However, instead of the diet composition, physiological differences seem to be the overruling factor causing lower concentrations of PUFAs (especially DHA) in large animals.

5.4.6 Marine mammals; evolutionary control? Seals eat predominantly fish and also squid (e.g., Olesiuk, 1993). Therefore, they feed at a higher trophic level than any of the organisms analyzed in this study. If indeed the concentration of DHA is a viable trophic level indicator in marine food webs, as suggested by Figure 5.7a, the relative abundance of DHA in seals should be very high (at least 40% of total fatty acids; Figure 5.7a). Budge et al. (2002) measured over 900

186 samples of 28 species of fish (between 10 and 40cm) and invertebrates that could be prey items of seals in eastern Canada. The median abundance of DHA in those species was about 19% (of total fatty acids), with a range of 8-31%. Durnford and Shahidi (2002) measured the fatty acid composition of various tissues from all six species of eastern Canadian phocid seals. They found that the muscle lipids of the analyzed seal species contained on average between 7 and 10% DHA. Other tissues, except brain tissue, did not exhibit higher levels of DHA (Durnford and Shahidi, 2002). Processed meat from harp seals hunted in the coastal regions of Newfoundland only had a DHA abundance of 5.5% (Shahidi and Synowiecki, 1991). Therefore, it seems clear that seals incorporate DHA in their tissues at a lower relative abundance than is available in their diet. This, again, goes against the trend observed for organisms feeding at lower trophic positions analyzed for this study (Fig. 5.7). Another disparity between the diet and seal tissue compositions is the ratio of n-3 over n-6 fatty acids. The average n-3/n-6 ratio in the diet organisms, approximated by calculation from the data of Budge et al. (2002), is just above 10. This is much higher than the average n-3/n-6 ratio of 2.8 in seal muscle, calculated from Durnford and Shahidi (2002), and the n-3/n-6 ratios of 2 and 2.4 for seal meat, reported by Shahidi and Synowiecki (1991). This difference is not only caused by the decrease of DHA in seal tissues, but also by a marked increase of arachidonic acid (AA; 20:4n-6). Hence, it appears that seals actively retain the n-6 fatty acids (especially AA) and/or selectively catabolize the n-3 fatty acids (since vertebrates are unable to synthesize n-6 fatty acids [Cook, 1996]). The n-3/n-6 ratio difference between the diet and various tissues of seals has been noted before (Engelhardt and Walker, 1974; Ackman and Hooper, 1974), and has also been observed for dolphins (Williams et al., 1977, 1987). Large land mammals and marsupials have a requirement for n-6 fatty acids, and are also known to incorporate fatty acids at lower n-3/n-6 ratios than their diet (Williams et al., 1987). The relatively high proportion of AA found in the marine mammals is of particular interest as it is an important precursor for the n-6 series of eicosanoids synthesized by land mammals. Therefore, Williams et al. (1987) noted that it may well be that the requirement for the n6 fatty acids is a remnant of the evolutionary past of marine mammals.

187 The blubber of the seals, on the other hand, seems to reflect the diet much better. For example, blubber from the seals analyzed by Durnford and Shahidi (2002) displayed n-3/n-6 ratios values of on average close to 10. This is not far off from values that can be expected for their diet (Budge et al., 2002) The fatty acid composition of adipose tissue of the dolphins was also in better agreement with the diet (Williams et al., 1977, 1987). This congruence between the fatty acid compositions of diet and blubber has been successfully used in foraging ecology studies on seals (Iverson, et al., 1997; Walton et al., 2000). Thus, it appears that the fatty acid composition of metabolically active tissues of marine mammals is strongly influenced by physiological constraints. Fortunately, blubber generally records past diet compositions more faithfully.

5.4.7 Fatty acids as trophic markers In the marine pelagic food web off the west coast of Vancouver Island the relative abundance of DHA was found to increase substantially from POM up to juvenile salmon (Figures. 5.5-5.7; Table 5.1). Preferential uptake of PUFAs has been observed in several studies. For example, Harvey et al. (1987) found that a marine copepod assimilated proportionally more PUFAs from an algal diet. In a fatty acid turnover study in rat tissues it was determined that 20:5n-3 was lost more rapidly than 22:5n-3 and DHA (Herzberg and Skinner, 1997). Human plasma lipids showed the same tendency (Hodge et al., 1993). A similar predisposition in the marine food web would explain the rising DHA/20:5n-3 ratios and coinciding increases in DHA and 22:5n-3 with trophic level (Figures 5.6 and 5.7). Crawford et al. (1976) found that guinea pigs and rats that were fed an identical diet displayed large differences in the PUFA composition of liver phospholipids. Additionally, the size versus PUFA content relationship found in mammals and birds (Hulbert et al., 2002a, b) seems to indicate that physiological constraints can strongly affect the fatty acid composition of animal tisues. Some indication exists that a similar negative correlation between size and DHA content also exists during growth of fish

188 (Iverson et al., 2002; Budge et al., 2002). Furthermore, the discrepancy between n-3/n-6 fatty acid ratios between the diet and various tissues of marine mammals underscores the importance of physiological requirements. Without a doubt, processes like selective assimilation or retention, de novo synthesis, elongation and desaturation of particular fatty acids obscure the dietary signature in animal tissues. Such effects limit the use of fatty acid profile analysis in foraging ecology studies. Analyzing tissues, lipid classes and fatty acids that are least affected by the aforementioned processes has shown the greatest potential to make semi-quantitative estimates of the diet composition. For example, Williams et al. (1987) and Durnford and Shahidi (2002) showed that, of all tissues analyzed, blubber or adipose tissue of marine mammals corresponded best with the fatty acid composition of the diet. It seems that the majority of the dietary fatty acids are absorbed and transported directly in chylomicrons to the blubber without modification (Durnford and Shahidi, 2002). This allows successful employment of fatty acid signature analysis for studying the foraging habits of seals (Iverson et al., 1997). The biosynthesis of certain fatty acids is very limited or non-existent beyond phytoplankton. Very little biosynthesis of the 14:0 fatty acid occurs in higher organisms (Sargent and Whittle, 1981). Also, a wide range of usually minor fatty acids, for example from the n-4 and n-1 series are almost certainly strictly produced by phytoplankton. In some invertebrates a ∆12 desaturase has been found, allowing the formation of n-4 and n6 fatty acids (Weinert et al., 1993; Batcabe et al., 2000). Though, to my knowledge no synthesis of n-4 fatty acids by invertebrates has been observed. Vertebrates, including fish, are known to lack the ability to introduce double bonds beyond the C9 position (Cook, 1996). Therefore, when spatial variation exists, fatty acids such as the 14:0 and fatty acids strictly produced by phytoplankton can be good markers that can differentiate food webs from different areas. Indeed, as mentioned in section 5.4.1, the abundance of fatty acids such as the 16:2n-4, 16:4n-1 and 16:3n-4 could best distinguish between the shelf and off shelf food webs off Vancouver Island. As illustrated by Figure 5.4, spatial differences were found to be much smaller than differences in fatty acid compositions between different groups of organisms. Budge et

189 al. (2002) also found that within-species variation, due to foraging at different locations, was substantially less than among-species variation. Nevertheless, with multivariate classification techniques some differences of spatially separated food webs are still discernable. Such differences can aid in distinguishing stocks that have a different feeding history. On the other hand, Joensen et al. (2000) showed that two stocks of identically reared cod (fed the same diet) could still be distinguished with the fatty acid compositions of heart tissue. They suggested, therefore, that genetic control of the fatty acid composition of certain tissues allow classification below the species level.

5.5 Conclusions Spatial differences in POM were found to result in consistent differences in the fatty acid composition of higher trophic level organisms (at least) up to larval fish. However, the largest part of the variation in the fatty acid compositions of all analyzed organisms is explained by differences between the trophic groups (POM, zooplankton, larval fish). From POM up to juvenile salmon (muscle) the proportion of C20 and C22 PUFAs increases. A correlation between the δ15N and abundance of DHA suggests that biomagnification of DHA occurs. However, data from previous studies indicate that the DHA content versus trophic level relationship breaks down beyond larval/juvenile fish. Despite the physiological constraints of the fatty acid composition of the various tissues, dietary differences are still discernable. As reported by other researchers, the least metabolically active tissues (such as adipose tissues) offer most potential to record dietary signatures. Fatty acids that are not essential and cannot be synthesized by the animal seem to be the best trophic markers.

5.6 References Ackman, R.G. and Hooper, S.N. 1974. Long chain monoethylenic and other fatty acids in heart, liver and blubber lipids of two harbor seals (Phoca vitulina) and one grey seal (Halichoerus grypus). Journal of the Fisheries Research Board of Canada 31, 333–341.

190

López Alonso, D., Molina Grima, E., Sánchez Pérez, J.A., Garcia Sánchez and García Camacho, F. 1992. Fatty acid variation among different isolates of a single strain of Isochrysis Galbana. Phytochemistry 31, 3901-3904. Batcabe, J.P., Howell, J.D., Blomquist, G.J., Borgeson, C.E. 2000. Effects of developmental age, ambient temperature, and dietary alterations on 12 desaturase activity in the house cricket, Acheta domesticus. Archives of Insect Biochemistry and Physiology 44, 112-119. Ben-Amotz, A., Fishler, R. and Schneller, A. 1987. Chemical composition of dietary species of marine unicellular algae and rotifers with emphasis on fatty acids. Marine Biology 95, 31-36. Broadhurst, C.L., Wang, Y.Q., Crawford, M.A., Cunnane, S.C., Parkington, J.E. and Schmidt, W.F. 2002. Brain-specific lipids from marine, lacustrine, or terrestrial food resources: potential impact on early African Homo sapiens. Comparative Biochemistry and Physiology 131B, 653-673 Budge, S.M., Iverson, S.J., Bowen, W.D., Ackman, R.G. 2002. Among- and within-species variability in fatty acid signatures of marine fish and invertebrates on the Scotian Shelf, Georges Bank, and southern Gulf of St. Lawrence. Canadian Journal of Fisheries and aquatic Sciences 59, 886-898. Buzzi, M., Henderson, R.J. and Sargent, J.R. 1997. The biosynthesis of docosahexaenoic acid [22:6(n-3)] from linolenic acid in primary hepatocytes isolated from wild northern pike. Journal of Fish Biology 51, 1197-1208. Chuecas, L. and Riley, J.P. 1969. Component fatty acids of the total lipids of some marine phytoplankton. Journal of the Marine Biological Association of the United Kingdom 49, 97-116. Cook, H.W. 1996. Fatty acid desaturation and chain elongation in eukaryotes. In: D.E. Vance and J.E. Vance (Eds.) Biochemistry of Lipids, Lipoproteins and Membranes, New Comprehensive Biochemistry 31, Elsevier Science, Amsterdam, p. 363-389. Couture, P. and Hulbert, A.J. 1995. Membrane fatty acid composition is related to body mass of mammals. Journal of Membrane Biology 148, 27-39. Crawford, M.A., Casperd, N.M. and Sinclair, A.J. 1976. The long-chain metabolites of linoleic and linolenic acids in liver and brain in herbivores and carnivores. Comparative Biochemistry and Physiology 54B, 395–401. Crawford, M.A., Bloom, M., Broadhurst, C.L., Schmidt, W.F., Cunnane, S.C., Galli, C., Gehbremeskel, K., Linseisen, F., Lloyd-Smith, J. and Parkington, J. 1999. Evidence for the unique function of docosahexaenoic acid during the evolution of the modern hominid brain. Lipids 34, Supplement S, S39S47.

191 Cripps, G.C., Watkins, J.L., Hill, H.J. and Atkinson, A. 1999. Fatty acid content of Antarctic krill Euphausia superba at South Georgia related to regional populations and variations in diet. Marine Ecology Progress Series 181, 177-188. Cripps GC, Atkinson A. 2000. Fatty acid composition as an indicator of carnivory in Antarctic krill, Euphausia superba. Canadian Journal of Fisheries and aquatic Sciences 57, Supplement 3, 31-37. Desvilettes, Ch., Boudier, G., Breton, J.C. and Combrouze, Ph. 1994. Fatty acids as organic markers for the sudy of trophic relationships in littoral cladoceran communities of a pond. Journal of Plankton Research 16, 643-659. Dunstan, G.A., Sinclair, A.J., O’Dea, K. and Naughton, J.M. 1988. The lipid content and fatty acid composition of various marine species from southern Australian coastal waters. Comparative Biochemistry and Physiology 91B, 165-169. Durnford, E. and Shahidi, F. 2002. Comparison of FA compositions of selected tissues of phocid seals of eastern Canada using one-way and multivariate techniques. Journal of the American Oil Chemists Society 79, 1095-1102. Engelhardt, F.R. and Walker, B.L. 1974. Fatty acid composition of the harp seal, Pagophilus groenlandicus (Phoca groenlandica). Comparative Biochemistry and Physiology 47B, 169–179. Fraser, A.J., Sargent, J.R., Gamble, J.C. and Seaton, D.D. 1989. Formation and transfer of fatty acids in an enclosed marine chain comprising phytoplankton, zooplankton and herring (Clupea harengus L.) larvae. Marine Chemistry 27, 1-18. Fukuda, Y. and Naganuma, T. 2001. Potential dietary effects on the fatty acid composition of the common jellyfish Aurelia aurita. Marine Biology 138, 1029-1035. Graeve, M., Kattner, G. and Hagen, W. 1994. Diet-induced changes in the fatty acid composition of Arctic herbivorous copepods: experimental evidence of trophic markers. Journal of experimental marine biology and ecology 182, 97-110. Grosse, D.J. and Hay, D.E. 1988. Pacific Herring, Clupea harengus pallasi, in the northeast Pacific and Bering Sea. In: N.J. Wilimovsky, L.S. Incze and S.J. Westrheim (eds.), Species synopses : life histories of selected fish and shellfish of the northeast Pacific and Bering Sea. Washington Sea Grant Program and Fisheries Research Institute, University of Washington, Seattle, U.S.A., p. 34-54. Gudbjarnason, S., Doell, B., Oskardottir, G. and Hallgrimsson, J. 1978. Modification of cardiac phospholipids and catecholamine stress tolerance. In: C. deDuve and O. Hayaishi (eds.), Tocopherol, Oxygen and Biomembranes. Elsevier, Amsterdam, p. 297-310. Gurr, M.I. and Harwood, J.L. Lipid Biochemistry. Fourth edition, Chapman and Hall, London, pp. 406.

192 Harvey, H.R., Eglinton, G., O’Hara, S.C.M. and Corner, E.D.S. 1987. Biotransformation and assimilation of dietary lipids by Calanus feeding on a dinoflagellate. Geochimica et Cosmochimica Acta 51, 3031-3040. Henderson, R.J., Sargent, J.R. and Hopkins, C.C.E. 1984. Canges in the content and fatty acid composition of lipid in an isolated population of the capelin Mallotus villosus during sexual maturation and spawning. Marine Biology 78, 255-263. Henderson, R.J. and Tocher, D.R. 1987. The lipid composition and biochemistry of freshwater fish. Progress in Lipid Research 26, 281-347. Herzberg, G.R. and Skinner, C. 1997. Differential accumulation and release of long-chain n-3 fatty acids from liver, muscle, and adipose tissue triacylglycerols. Canadian Journal of Physiology and Pharmacology 75, 945-951. Hodge, J., Sanders, K. and Sinclair, A.J. 1993. Differential utilization of eicosapentaenoic acid and docosahexaenoic acid in human plasma. Lipids 28, 525-531. Hulbert, A.J. and Else, P.L. 1989. The evolution of mammalian endothermic metabolism: mitochondrial activity and changes in cellular composition. American Journal of Physiology 256, Part 2, R63-R69. Hulbert, A.J. and Else, P.L. 1999. Membranes as possible pacemakers of metabolism. Journal of Theoretical Biology 199, 257-274. Hulbert, A.J., Rana, T. and Couture, P. 2002a. The acyl composition of mammalian phospholipids: an allometric analysis. Comparative Biochemistry and Physiology 132B, 515-527. Hulbert, A.J., Faulks, S., Buttemer, W.A. and Else, P.L. 2002b. Acyl composition of muscle membranes varies with body size in birds. Journal of Experimental Biology 205, 3561-3569. Infante, J.P., Kirwan, R.C. and Brenna, J.T. 2001. High levels of docosahexaenoic acid (22 : 6n-3)containing phospholipids in high-frequency contraction muscles of hummingbirds and rattlesnakes. Comparative Biochemistry and Physiology 130B, 291-298. Iverson, S.J., Frost, K.J. and Lowry, L.F. 1997. Fatty acid signatures reveal fine scale structure of foraging distribution of harbor seals and their prey in Prince William Sound, Alaska. Marine Ecology Progress Series 151, 255-271. Iverson, S.J., Frost, K.J. and Lang, S.L.C. 2002. Fat content and fatty acid composition of forage fish and invertebrates in Prince William Sound, Alaska: factors contributing to among and within species variability. Marine Ecology Progress Series 241, 161-181. Jeffries, H.P. 1970. Seasonal composition of temperate plankton communities: fatty acids. Limnology and Oceanography 15, 419-426.

193

Joensen, H., Steingrund, P., Fjallstein, I. and Grahl-Nielsen, O. 2000. Discrimination between two reared stocks of cod (Gadus morhua) from the Faroe Islands by chemometry of the fatty acid composition in the heart tissue. Marine Biology 136, 573-580. Joseph, J.D. 1982. Lipid composition of marine and estuarine invertebrates. Part II: Mollusca. Progress in Lipid Research 21, 109-153. Kattner, G. and Hagen, W. 1998. Lipid metabolism of the Antarctic euphausiid Euphausia crystallorophias and its ecological implications. Marine Ecology Progress Series 170, 203-213. Kattner, G. and Brockmann, U.H. 1990. Particulate and dissolved fatty acids in an enclosure containing a unialgal Skeletonema costatum (Greve.) Cleve culture. Journal of Experimental Marine Biology and Ecology 141, 1-13. Kirsch, P.E., Iverson, S.J., Bowen, W.D., Kerr, S.R. and Ackman, R.G. 1998. Dietary effects on the fatty acid signature of whole Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 55, 1378-1386. Linko, R.R., Kaitaranta, J.K. and Vuorela, R. 1985. Comparison of the fatty acids in the Baltic herring and available plankton feed. Comparative Biochemistry and Physiology 82B, 699-705. Mansour, M.P., Volkman, J.K., Jackson, A.E. and Blackburn, S.I. 1999. The fatty acid and sterol composition of five marine dinoflagellates. Journal of Phycology 35, 710-720. March, B.E. 1993. Essential fatty acids in fish physiology. Canadian Journal of Physiology and Pharmacology 71, 684-689. Meglen, R.R. 1992. Examining large databases: a chemometric approach using principal component analysis. Marine Chemistry 39, 217-237. Mjaavatten, O., Levings, C.D. and Poon, P. 1998. Variation in the fatty acid composition of juvenile chinook and coho salmon from Fraser river estuary determined by multivariate analysis; role of environment, and genetic origin. Comparative Biochemistry and Physiology 120B, 291-309. Napolitano, G.E., Pollero, R.J., Gayoso, A.M., MacDonald, B.A. and Thompson, R.J. 1997. Fatty acids as trophic markers of phytoplankton blooms in the Bahía Blanca estuary (Buenos Aires, Argentina) and in Trinity Bay (Newfoundland, Canada). Biochemical Systematics and Ecology 25, 739-755. Neal, A.C., Prahl, F.G., Eglinton, G., O’Hara, S.C.M. and Corner, E.D.S. 1986. Lipid changes during a planktonic feeding sequence involving unicellular algae, Elminius nauplii and adult Calanus. Journal of the Marine Biological Association of the United Kingdom 66, 1-13.

194 Nelson, M.M., Phleger, C.F., Mooney, B.D. and Nichols, P.D. 2000. Lipids of gelatinous Antarctic zooplankton: Cnidaria and Ctenophora. Lipids 35, 551-559. Nelson, M.M., Mooney, B.D., Nichols, P.D., Phleger, C.F. 2001. Lipids of Antarctic Ocean amphipods: food chain interactions and the occurrence of novel biomarkers. Marine Chemistry 73, 53-64. Olesiuk, P.F. 1993. Annual prey consumption by harbour seals (Phoca vitulina) in the Strait of Georgia, British Columbia. Fisheries Bulletin (Dublin) 91, 491-515. Oltra, R., Todolí, R., Bosque, T., Lubián, L.M. and Navarro, J.C. 2000. Life history and fatty acid composition of the marine rotifer Synchaeta Cecilia valentina fed different algae. Marine Ecology Progress Series 193, 125-133. Ozawa, A., Satake, M. and Fujita, T. 1993. Comparison of muscle lipid composition between marine and landlocked forms of sockeye salmon (Oncorhynchus nerka). Comparative Biochemistry and Physiology 106B, 513-516. Pamplona, R., Portero-Otín, M., Ruiz, C., Gredilla, R., Herrero, A. and Barja, G. 2000. Double bond content of phospholipids and lipid peroxidation negatively correlate with maximum longevity in the heart of mammals. Mechanisms of Ageing and Development 112, 169-183. Peterson B.J. and Fry B. 1987. Stable isotopes in ecosystem studies. Annual review of ecology and systematics 18, 293-320. Phleger, C.F., Nichols, P.D. and Virtue, P. 1997. Lipids and buoyancy in Southern Ocean pteropods. Lipids 32, 1093-1100. Pohl, P. and Zuhrheide, F. 1979. Fatty acids and lipids of marine algae and the control of their biosynthesis by environmental factors. In: Hoppe, H.A., Levring, T. and Tanaka, Y. (eds) Marine algae in pharmaceutical science. Walter de Gruyer, Berlin, 473-523. Pond, D.W. and Sargent, J.R. 1998. Lipid composition of the pelagic tunicate Dolioletta gegenbauri (Tunicata, Thaliacea). Journal of Plankton Research 20, 169-174. Portero-Otín, M., Bellmunt, M.J., Ruiz, M.C., Barja, G. and Pamplona, R. 2001 Correlation of fatty acid unsaturation of the major liver mitochondrial phospholipid classes in mammals to their maximum life span potential. Lipids 36, 491-498. Prahl, F.G., Eglinton, G., Corner, E.D.S., O’Hara, S.C.M. and Forsberg, T.E.V. 1984. Changes in plant lipids during passage through the gut of calanus. Journal of the Marine Biological Association of the United Kingdom 64, 317-334. Rivers, J.P.W., Sinclair, A.J. and Crawford, M.A. 1975. Inability of the cat to desaturate essential fatty acids. Nature 258, 171-173.

195

Rivers, J.P.W., Hassam, A.G., Crawford, M.A. and Brambell, M.R. 1976. Inability of the lion, Panthera leo L. to desaturate linoleic acid. FEBS Letters 67, 269-270. Sargent, J. R. and Whittle, K.J. 1981. Lipids and hydrocarbons in the marine food web. In: A.R. Longhurst (ed.), Analysis of Marine Ecosystems. Academic Press, London, U.K., p. 491-533. Sargent, J. R. and Falk-Petersen, S. 1988. The lipid biochemistry of calanoid copepods. Hydrobiologia 167/168, 101-114. Sargent, J. R., Parkes, R. J., Mueller-Harvey, I, and Henderson, R. J. 1988. Lipid biomarkers in marine ecology. In: M. A. Sleigh (ed.), Microbes in the sea. Ellis Horwood, Ltd. Chichester, U.K., p.119-138. Shahidi, F. and Synowiecki, J. 1991. Cholestrol content and lipid fatty acid composition of processed seal meat. Canadian Institute of Food Science and Technology Journal 24, 269-272. Sheridan, M.A. 1988 Lipid dynamics in fish: aspects of absorption, transportation, deposition and mobilization. Comparative Biochemistry and Physiology 90B, 679–690. Spector, W. S. (ed.) 1956. Handbook of Biological Data, W. B. Saunders Co., Philadelphia, U.S.A., pp. 584. St. John, M.A. and Lund, T. 1996. Lipid biomarkers: linking the utilization of frontal plankton biomass to enhanced condition of juvenile North Sea cod. Marine Ecology Progress Series 131, 75-85. Storch, J. and Kleinfeld, A.M. 1985. The lipid structure of biological membranes. Trends in Biochemical Sciences 10, 418-421. Tanakol, R., Yazici, Z., Sener, E. and Sencer, E. 1999. Fatty acid composition of 19 species of fish from the Black Sea and the Marmara Sea. Lipids 34, 291-297. Thompson, P.A., Guo, M., Harrison, P.J. and Whyte, J.N.C. 1992. Effects of variation in temperature. II. On the fatty acid composition of eight species of marine phytoplankton. Journal of phycology 28, 488-497. Virtue, P., Mayzaud, P., Albessard, E. and Nichols, P. 2000. Use of fatty acids as dietary indicators in northern krill, Meganyctiphanes norvegica, from northeastern Atlantic, Kattegat, and Mediterranean waters. Canadian Journal of Fisheries and aquatic Sciences 57 (Suppl. 3), 104-114. Viso, A.-C. and Marty, J.-C. 1993. Fatty acids from 28 marine microalgae. Phytochemistry 34, 1521-1533.

196 Volkman, J.K., Smith, D.J., Eglinton, G., Forsberg, T.E.V. and Corner, E.D.S. 1981. Sterol and fatty acid composition of four marine Haptophycean algae. Journal of the Marine Biological Association of the United Kingdom 61, 509-527. Volkman, J.K., Jeffrey, S.W., Nichols, P.D., Rogers, G.I. and Garland, C.D. 1989. Fatty acid and lipid composition of 10 species of microalgae used in mariculture. Journal of Experimental Marine Biology and Ecology 128, 219-240. Wakeham, S.G., Lee, C., Farrington, J.W. and Gagosian, R.B. 1984. Biogeochemistry of particulate organic matter in the oceans: results from sediment trap experiments. Deep-Sea Research 31, 509-528. Wakeham, S.G., Hedges, J.I., Lee, C., Peterson, M.L. and Hernes, P.J. 1997. Compositions and transport of lipid biomarkers through the water column and surficial sediments of the equatorial Pacific Ocean. DeepSea Research II 44, 2131-2162. Walton, M.J., Henderson, R.J. and Pomeroy, P.P. 2000. Use of blubber fatty acid profiles to distinguish dietary differences between grey seals Halichoerus grypus from two UK breeding colonies. Marine Ecology Progress Series 193, 201-208. Weinert, T.J., Blomquist, G.J. and Borgeson, C.E. 1993. De novo biosynthesis of linoleic-acid in 2 noninsect invertebrates – The land slug and the garden snail. Experientia 49, 919-921. Williams, G., Davidson, D.C., Stevens, P.A. and Crawford, M.A. 1977. Comparative fatty acids of the dolphin and herring. Journal of the American Oil Chemists Society 54, 328-330. Williams, G., Crawford, M.A., Perrin, W.F. 1987. Comparison of the fatty acid component in structural lipids from dolphins, zebra and giraffe: possible evolutionary implications. Journal of Zoology 213, 673684.

197

6. The Effect of Starvation on the Relative Abundance and δ13C of Fatty Acids in Rotifers (Brachionus plicatilis)

6.1 Abstract A large portion of fatty acids is lost via catabolism during the transfer between trophic levels. In order to study the effect of catabolism on the δ13C of individual fatty acids, rotifers (Brachionus plicatilis) were subjected to starvation, and a time-series of measurements of the concentration, and δ13C of the fatty acids was obtained. Fatty acids with shorter carbon chain lengths and a higher degree of unsaturation were typically found to have been preferentially lost during fasting. This is in accordance with previously reported observations of selective mobilization of fatty acids from triacylglycerols in rodents. The concentration of most fatty acids decreased exponentially, but the δ13C values of the total fatty acids showed a linear increase over the course of the starvation period. It is shown that the δ13C of the food of the rotifers, Tahitian Isochrysis galbana, must have shifted substantially during the time leading up to the starvation experiment. The neutral, and polar lipid fatty acid pools have different turnover rates. Therefore, the shift in the δ13C of the food must have resulted in a large disparity between the 13C/12C ratios of fatty acids from neutral, and polar lipids. Consequently, the rise in the δ13C of the various fatty acids in the rotifers was not the result of a kinetic isotope effect during catabolism, but due to a decrease in the proportion of (13C depleted) neutral lipid fatty acids. This is illustrated here with a model. Because of the unusually large difference between the 13

C/12C ratios of fatty acids from neutral, and polar lipids at the onset of the starvation,

the isotope results from this experiment are not representative for natural conditions. Most likely, some retroconversion of 18:4n-3 to 16:4n-3, and probably also of 22:6n-3 to 20:5n-3 occurred during starvation.

198

6.2 Introduction One of the requirements for the δ13C of fatty acids to be a good trophic marker, is that the fatty acids should be transferred through the food web without any significant alterations. Because several kinetic isotope effects are known to occur during the synthesis of a fatty acid (Monson and Hayes, 1982a, b; Chapter 4 this thesis), the fatty acids that are strictly produced by primary producers are the best candidates as trophic markers. These may be either essential fatty acids, or non-essential fatty acids that cannot be produced by the heterotrophs. Fractionation during the assimilation or catabolism of fatty acids, could still be able to distort their original δ13C signature that was fixed at the bottom of the food web. Evidence exists that no substantial stable carbon isotope fractionation takes place between the lipids of an animal and those of its diet. In a feeding experiment Focken and Becker (1998) found that fat-free matter of tilapia and carp was enriched in 13C by about 3 and 1.3‰, respectively, relative to fat-free matter of their diet. However, lipids from the fish and their diet showed on average only a few tenths of a per mil difference in δ13C. Grice et al. (1998) and Klein Breteler et al. (2002) showed that the δ13C values of sterols and alkenones in a copepod were similar to that of its diet. This also suggests that lipids are not subjected to any substantial trophic fractionation. Marine rotifers (Brachionus plicatilis), which are small (few tenths of a millimetre) zooplankton animals, were starved for three days to further investigate the effect of catabolism on the δ13C of fatty acids. At several time-intervals, sub-samples of the rotifers were taken in order to obtain a time-series of δ13C measurements of their fatty acids during the starvation. Frolov and Pankov (1992) have already studied the effect of starvation on the biochemical composition, including the fatty acid profile, of Brachionus plicatilis. Additionally, Olsen et al. (1993) tested the same rotifers for the dependence of the loss rate of lipids and n-3 fatty acids on temperature during starvation. However, the corresponding stable carbon isotope ratio was not measured in these investigations.

199 Few studies have investigated the effect of starvation on the

13

C/12C ratio of the

whole body, and mixed results were obtained by the investigations. For example, Oelbermann and Scheu (2002) found that spiders (Pardosa lugubris) were enriched in 13

C after starvation, whereas Gorokhova and Hansson (1999) did not find any significant

change in the δ13C of mysids during 5 weeks of starvation. Several researchers have shown how a shift from the use of one type of metabolic fuel to another can change the δ13C of the expired CO2 (e.g., Miller et al., 1985; Gautier et al., 1996). Hence, a change in proportion of the various biochemical fractions due to differential catabolism may cause a shift in whole-body δ13C during starvation. This is the first study to report measurements of δ13C values of individual fatty acids during starvation.

6.3 Methods 6.3.1 Experimental conditions and procedure Marine rotifers (Brachionus plicatilis; purely asexual culture of L-type amictic females), initially isolated from the Seto Inland Sea, Japan (Nagata and Whyte, 1992), were cultivated from resting eggs in 100L plastic bags with filtered (1µm) and UV-sterilized seawater, at approximately 24°C. The rotifers were daily supplied with Tahitian Isochrysis galbana (T-Iso) to maintain satiation feed levels, and continuous irradiation was provided with warm-white fluorescent lights. A total of approximately 4 million rotifers were rinsed and transferred into two receptacles with 0.2µm-filtered seawater (23°C). The rotifers in each receptacle came from a different bag. From one bag 1.5 million, and from another 2.5 million rotifers were withdrawn. To avoid including damaged or dead rotifers in the experiment both batches were carefully decanted into new receptacles, leaving behind about 100,000 rotifers that had sunk to the bottom. A “t=0” sample, containing approximately 200,000 rotifers, was collected from each batch. This was approximately 30 minutes after the transfer of the first batch, and 10 minutes after the second batch was placed into filtered seawater. Then the rotifers from the two batches were mixed and transferred to four 4L glass beakers that were filled with 3 liters of 0.2µm-filtered seawater. The concentration of the rotifers in the beakers was

200 approximately 325 individuals/mL. The water was constantly bubbled with air, and the temperature of the water in the beakers was about 21°C throughout the experiment. Samples were taken after 2, 4, 8, 12, 24, 48 and 72 hours after the t=0 time. The sampled rotifers were collected on a 44µm screen and rinsed with an isotonic solution of 3.2% ammonium formate in distilled water. Subsequently, they were transferred into a vial and kept frozen at –80°C until further analysis several days later. During the experiment few rotifers settled to the bottom of the beakers. No count was done as to determine how many, however it is estimated that no more than 10 to 20% had dropped out throughout the experiment. The first visible amount of individuals that settled down was observed at t=8 h. Five times the rotifers at the bottom were sampled for analysis with a large pipette from the fourth beaker. However, no large differences between the fatty acid composition of these and the other sampled rotifers were found. The results shown in this chapter are from the analysis of samples taken from the first three beakers. Before taking a sample of the rotifers in these beakers, the water was gently stirred as to get a representative sample of the rotifers from the whole water column. At the same time that samples were obtained from each beaker an aliquot was preserved in formalin for a fecundity assessment. To measure the fecundity, the number of eggs per 50 rotifers was counted for each sample.

6.3.3 T-iso samples and culture conditions Two (150mL) sub-samples of a 250L batch culture of Tahitian Isochrysis galbana (T-Iso) were taken for further analysis. A count of the T-Iso cells was performed to determine that the sampled culture water contained approximately 1.5 million cells/L, and it was confirmed that the culture consisted only of T-Iso cells. The samples were filtered through a pre-combusted glass-fibre filter, and the filters were then treated exactly as described for POM samples in chapter 2. The T-Iso batch culture, grown at the algal lab at the Pacific Biological Station (Department of Fisheries and Oceans, Nanaimo, B.C., Canada) had been inoculated in

201 previously filtered (1µm) and pasteurized seawater. 1mL vitamin solution/ 20L seawater and 1mL nutrients/ L seawater had been added to the seawater (modified from Harrison et al. (1980)’s saltwater enrichment solution). The pH was controlled by bubbling CO2 into the culture on demand from a pH probe that activates a solenoid valve. This ensured that the culture was kept within a pH range of 7.4-7.8. The algae were grown in continuous light at a light intensity of 250 µmol m-2 s-1 using full-spectrum fluorescent lights. The lab temperature was 18oC ± 1oC. A sample of the T-Iso culture used for the cultivation of the rotifers was only taken two days after the initiation of the starvation experiment.

6.3.4 Fatty acid methyl ester preparation and δ13C measurements Samples were freeze-dried and the fatty acid methyl esters were prepared according to the method described by Whyte (1988). Please refer to the methods section in Chapter 2 for descriptions of this procedure and also the stable carbon isotope ratio measurements.

6.4 Results 6.4.1 Fatty acid composition of T-Iso and rotifers The fatty acid composition of the T-Iso culture, sampled two days after the commencement of the experiment, is shown in Figure 6.1. The most abundant fatty acids are the 14:0 and 18:4n-3, each comprising just over 20% of the total weight of the fatty acids. The relative amounts of 18:1n-9, 16:0, 18:3n-3 and the 22:6n-3 were determined at 11.1, 9.6, 8.3 and 7.9 weight percent of the total fatty acids, respectively. Rotifers exhibited, next to the fatty acids derived from T-Iso, a wide range of fatty acids not found at appreciable levels in T-Iso. A good example is 20:4n-3, which amounted to about 5% of the total fatty acids in the rotifers, but was not detected in the algae. Also notable is that linoleic acid (18:2n-6) is the principal fatty acid in the rotifers, whereas in T-iso it is only moderately abundant (5% of total). The ratio of n-6 over n-3

202 fatty acids in T-Iso was found to be nearly 0.19. The rotifers, on the other hand, displayed n-6/n-3 ratios of about 0.50.

T-Iso culture

20

Fatty acid abundance (% of total)

15 10 5 0 12

Rotifers at t=0 h Rotifers at t=72 h

10 8 6 4 2

14:0 15:0 16:0 16:1n-7 16:2n-4 16:4n-3 18:0 18:1n-11? 18:1n-9 18:1n-7 18:2n-6 18:3n-4? 18:3n-3 18:4n-3 20:1n-11 20:1n-9 20:1n-7 20:2n-6 20:4n-6 20:3n-3 20:4n-3 20:5n-3 22:1n-9 21:5n-3 22:5n-6 22:5n-3 22:6n3

0

Fatty acid Figure 6.1. Fatty acid profiles of Tahitian Isochrysis galbana (top) and of the rotifers (Brachionus plicatilis) before and after 72 hours of starvation.

6.4.2 Effect of starvation on fatty acid profile Table 6.1 lists the relative abundances of the fatty acids found in the rotifers at several times during the starvation experiment. The concentration of the total fatty acids (in mg/g of sample), also shown in Table 6.1, decreased by almost 50% in 72 hours. The largest decrease was found in the first time interval between 0 and 2 hours. As can be observed in Figure 6.1, the general characteristics of the fatty acid profile do not change considerably. However, among the fatty acids a great variety exists in the relative amount lost during starvation (Figure 6.2). Two fatty acids, 16:4n-3 and 22:5n-3, were actually found to increase in concentration (Figure 6.2). The highest loss was

203

Table 6.1. Relative abundance of fatty acids (in weight percent of total fatty acids) in rotifers before and during starvation. The top line shows the concentration of total fatty acids in the (freeze-dried) rotifers. The numbers are given ± 1 standard deviation, calculated from the analyses on 3 different batches of rotifers. Starvation time mg FA/g sample Total FA

0h

2h

4h

8h

12 h

24 h

48 h

72 h

77.3

63.3 ± 0.5

60.6 ± 2.6

56.6 ± 4.2

52.7 ± 1.3

48.7 ± 3.5

37.9 ± 2.1

40.7 ± 3.4

% of total FA wt. 14:0 iso 15:0 15:0 16:0 16:1n-7 16:1n-5 iso 17:0 16:2n-7? 16:2n-4 17:0 16:3n-4 16:4n-3 18:0 18:1n-11? 18:1n-9 18:1n-7 18:1n-5 18:2n-7? 18:2n-6 18:3n-4? 18:3n-3 18:4n-3 20:1n-11 20:1n-9 20:1n-7 20:2n-6

10.48 0.36 0.45 11.62 5.00 0.46 tr 0.34 0.54 tr 0.40 0.81 2.61 1.30 6.07 3.02 tr 0.39 11.69 0.60 8.99 6.30 0.91 2.33 0.55 0.60

10.25 ± 0.13 0.34 ± 0.02 0.45 ± 0.00 11.44 ± 0.23 4.80 ± 0.07 0.36 ± 0.02 0.23 ± 0.01 0.36 ± 0.00 0.56 ± 0.01 0.23 ± 0.01 0.36 ± 0.01 1.08 ± 0.04 2.64 ± 0.02 1.32 ± 0.03 5.86 ± 0.05 2.93 ± 0.04 tr 0.39 ± 0.01 11.40 ± 0.11 0.61 ± 0.01 8.78 ± 0.09 6.00 ± 0.14 0.92 ± 0.03 2.41 ± 0.01 0.57 ± 0.01 0.59 ± 0.01

9.96 ± 0.22 0.33 ± 0.00 0.45 ± 0.01 11.44 ± 0.07 4.65 ± 0.05 0.33 ± 0.03 0.23 ± 0.01 0.35 ± 0.01 0.51 ± 0.01 0.23 ± 0.00 0.34 ± 0.01 1.21 ± 0.03 2.76 ± 0.02 1.38 ± 0.00 5.82 ± 0.01 2.96 ± 0.02 tr 0.40 ± 0.01 11.42 ± 0.09 0.60 ± 0.01 8.61 ± 0.13 5.63 ± 0.05 0.91 ± 0.02 2.51 ± 0.01 0.59 ± 0.01 0.58 ± 0.01

9.55 ± 0.29 0.33 ± 0.01 0.46 ± 0.01 11.50 ± 0.19 4.44 ± 0.12 0.33 ± 0.02 0.23 ± 0.02 0.34 0.47 ± 0.02 0.25 ± 0.00 0.30 ± 0.01 1.47 ± 0.08 2.96 ± 0.01 1.50 ± 0.01 5.72 ± 0.09 3.02 ± 0.03 tr 0.39 ± 0.02 11.59 ± 0.07 0.57 ± 0.01 8.40 ± 0.03 4.98 ± 0.06 0.97 ± 0.03 2.65 ± 0.04 0.63 ± 0.01 0.60 ± 0.01

9.06 ± 0.19 0.32 ± 0.01 0.47 ± 0.01 11.62 ± 0.18 4.24 ± 0.10 0.29 ± 0.02 0.23 ± 0.01 0.35 ± 0.01 0.41 ± 0.00 0.26 ± 0.00 0.27 ± 0.01 1.58 ± 0.06 3.19 ± 0.02 1.63 ± 0.01 5.65 ± 0.10 3.15 ± 0.04 0.28 ± 0.09 0.43 ± 0.00 11.86 ± 0.09 0.55 ± 0.02 8.16 ± 0.05 4.33 ± 0.02 1.04 ± 0.02 2.77 ± 0.01 0.65 ± 0.01 0.60 ± 0.01

7.94 ± 0.29 0.37 ± 0.03 0.49 ± 0.01 11.33 ± 0.20 4.10 ± 0.16 0.29 ± 0.02 0.25 ± 0.01 0.37 ± 0.01 0.34 ± 0.01 0.28 ± 0.01 tr 2.03 ± 0.07 3.50 ± 0.04 1.88 ± 0.01 5.14 ± 0.08 3.76 ± 0.12 0.32 ± 0.03 tr 12.02 ± 0.11 0.47 ± 0.02 7.50 ± 0.08 3.09 ± 0.06 1.22 ± 0.03 2.82 ± 0.02 0.73 ± 0.01 0.65 ± 0.01

7.28 ± 0.02 0.69 ± 0.07 0.56 ± 0.01 10.87 ± 0.05 4.22 ± 0.15 0.30 ± 0.07 0.45 ± 0.01 0.42 ± 0.00 0.29 ± 0.01 0.31 ± 0.00 tr 2.55 ± 0.18 3.78 ± 0.06 2.20 ± 0.06 4.74 ± 0.03 4.38 ± 0.17 0.55 ± 0.05 tr 11.94 ± 0.08 0.40 ± 0.02 6.72 ± 0.11 2.13 ± 0.07 1.36 ± 0.01 2.76 ± 0.03 0.78 ± 0.01 0.59 ± 0.00

6.95 ± 0.19 0.99 ± 0.11 0.31 ± 0.04 10.69 ± 0.13 4.56 ± 0.22 0.26 0.59 ± 0.04 0.45 ± 0.02 0.25 0.32 ± 0.03 tr 2.62 ± 0.07 3.88 ± 0.06 2.39 ± 0.05 4.56 ± 0.04 5.00 ± 0.39 0.78 ± 0.07 tr 11.74 ± 0.09 0.35 ± 0.04 6.10 ± 0.17 1.70 ± 0.10 1.48 ± 0.08 2.68 ± 0.08 0.87 ± 0.05 0.60 ± 0.04

204

20:4n-6 20:3n-3 20:4n-3 20:5n-3 22:1n-11 22:1n-9 21:5n-3 22:4n-6 22:5n-6 22:4n-3 22:5n-3 22:6n-3

2.25 0.76 5.01 2.74 0.46 0.59 0.85 0.36 1.45 tr 0.54 6.14

2.26 ± 0.03 0.76 ± 0.01 5.17 ± 0.11 2.84 ± 0.02 0.59 ± 0.02 0.66 ± 0.01 0.87 ± 0.06 0.38 ± 0.04 1.47 ± 0.02 tr 0.63 ± 0.02 6.34 ± 0.05

2.30 ± 0.03 0.76 ± 0.01 5.29 ± 0.12 2.87 ± 0.05 0.53 ± 0.01 0.70 ± 0.02 0.88 ± 0.03 0.41 ± 0.06 1.45 ± 0.07 0.25 0.84 ± 0.11 6.15 ± 0.23

2.46 ± 0.06 0.79 ± 0.00 5.69 ± 0.13 3.04 ± 0.09 0.48 ± 0.02 0.70 ± 0.02 0.89 ± 0.02 0.42 ± 0.02 1.40 ± 0.09 tr 1.02 ± 0.01 5.65 ± 0.39

2.58 ± 0.05 0.79 ± 0.01 5.93 ± 0.18 3.04 ± 0.11 0.43 ± 0.03 0.72 ± 0.01 0.85 ± 0.05 0.43 ± 0.03 1.31 ± 0.07 tr 1.20 ± 0.06 5.36 ± 0.08

2.79 ± 0.07 0.77 ± 0.01 6.34 ± 0.10 2.99 ± 0.03 0.56 ± 0.04 0.71 ± 0.01 0.86 ± 0.04 0.50 ± 0.03 1.21 ± 0.05 0.31 1.45 ± 0.12 4.94 ± 0.25

3.02 ± 0.06 0.74 ± 0.01 6.25 ± 0.08 3.08 ± 0.10 0.52 ± 0.01 0.67 ± 0.01 0.77 ± 0.01 0.55 ± 0.02 1.00 ± 0.02 0.27 ± 0.05 1.80 ± 0.10 4.25 ± 0.20

3.04 ± 0.01 0.72 ± 0.01 6.02 ± 0.19 3.09 ± 0.17 0.57 ± 0.03 0.65 ± 0.04 0.80 ± 0.07 0.60 ± 0.03 0.86 ± 0.04 0.28 ± 0.05 1.77 ± 0.11 3.85 ± 0.11

Tot un-ID

2.60

3.16 ± 0.04

3.54 ± 0.19

4.14 ± 0.05

4.30 ± 0.27

5.99 ± 0.28

7.07 ± 0.31

8.07 ± 0.41

Σ Saturated Σ Mono-unsat. Σ Di-unsat. Σ 3-6 Unsats.

25.8 20.8 13.6 37.2

25.6 ± 0.2 20.4 ± 0.2 13.3 ± 0.1 37.5 ± 0.5

25.4 ± 0.3 20.4 ± 0.1 13.3 ± 0.1 37.4 ± 0.3

25.2 ± 0.6 20.4 ± 0.2 13.2 ± 0.1 37.1 ± 0.8

25.1 ± 0.5 20.8 ± 0.2 13.4 ± 0.3 36.4 ± 0.7

24.1 ± 0.6 21.5 ± 0.3 13.4 ± 0.1 35.0 ± 0.8

23.8 ± 0.3 22.5 ± 0.2 13.2 ± 0.1 33.4 ± 0.7

23.7 ± 0.2 23.6 ± 0.8 12.9 ± 0.1 31.7 ± 0.7

Σ n-3 Σ n-6 (n-6)/(n-3)

32.1 16.3 0.51

32.5 ± 0.4 16.1 ± 0.1 0.50 ± 0.01

32.3 ± 0.2 16.2 ± 0.1 0.50 ± 0.00

31.9 ± 0.6 16.5 ± 0.1 0.52 ± 0.01

31.3 ± 0.6 16.8 ± 0.1 0.54 ± 0.01

30.1 ± 0.6 17.2 ± 0.1 0.57 ± 0.01

28.5 ± 0.7 17.1 ± 0.1 0.60 ± 0.01

26.9 ± 0.7 16.8 ± 0.1 0.63 ± 0.01

205

% loss of FA (% of mg FA/g of sample)

100 80 60 40 20 0 -20 -40 -60

14:0 16:0 16:1n-7 16:2n-4 16:4n-3 18:0 18:1n-11? 18:1n-9 18:1n-7 18:2n-6 18:3n-4? 18:3n-3 18:4n-3 20:1n-11 20:1n-9 20:1n-7 20:2n-6 20:3n-3 20:4n-6 20:4n-3 20:5n-3 21:5n-3 22:1n-9 22:5n-6 22:5n-3 22:6n-3

-80

Fatty acid Figure 6.2. Percent loss of concentration of the various fatty acids from 72 hours of starvation. The error bars represent ± 1 standard deviation, calculated from measurements on 3 batches of starved rotifers. Negative numbers indicate a rise in concentration.

observed for 18:4n-3, which lost about 86% of the initial amount per sample weight unit. The fatty acids in Figure 6.2 are ordered such that from left to right the chain length increases, and in the same direction fatty acids with similar chain lengths increase in unsaturation. Although many exceptions exist, a weak trend of increased loss with increased unsaturation can be observed. Also, when fatty acids with a similar degree of unsaturation are compared, the fatty acids with shorter chain lengths have often lost a higher proportion during the starvation than their longer counterparts (Figure 6.2). In Figure 6.3 the concentration of several fatty acids during the starvation has been plotted. As can be seen in the figure, the fatty acid concentrations tend to follow an exponential decay curve. However, as noted before, not all fatty acids show a decrease in concentration at similar rates. The decrease in total fatty acid concentration is shown in Figure 6.4.

206

Fatty acid concentration (mg FA/g sample)

8

14:0 20:5n-3 22:5n-3 22:6n-3

7 6 5 4 3 2 1 0 0

10

20

30

40

50

60

70

Time of starvation (h)

80

-52

70

-53 13

δ C of total FAs (‰)

Tot. FA concentration (mg FA/g sample)

Figure 6.3. Concentration measurements (triplicates) of four fatty acids plotted against time of starvation.

-54

60

-55 50 -56 40 -57 0

10

20

30

40

50

60

70

Time of starvation (h) Figure 6.4. Concentration and the δ13C of the total fatty acids (weighted average, with fatty acid abundance as weighting factor) plotted against time starved.

207 -36 -38 -40 -42 -44

13

δ C (‰)

-46 -48 -50 -52 -54 -56 -58 -60 -64

14:0 15:0 16:0 16:1n-7 16:2n-4 17:0 16:4n-3 18:0 18:1n-9 18:1n-7 18:2n-6 18:3n-3 18:4n-3 20:2n-6 20:4n-6 20:4n-3 20:5n-3 21:5n-3 22:5n-6 22:5n-3 22:6n-3

-66

Rotifers at t=0 h Rotifers at t=2 h T-iso 2 days after start expt. Bulk

-62

Fatty acid Figure 6.5. The δ13C of individual fatty acids in rotifers before and 2 hours into starvation compared to δ13C values of the same fatty acids in T-Iso sampled two days after the commencement of starvation study. Bulk δ13C measurements are also shown. Error bars of ± 1 S.D. are based on triplicate measurements.

6.4.3 Unusual variation in stable carbon isotope ratios The δ13C of the fatty acids derived from rotifers prior to, and 2 hours into starvation is shown in Figure 6.5. The same graph shows the stable carbon isotope composition of fatty acids from T-Iso used for the rotifer culture, but sampled two days later. First of all, a remarkably large range of 15‰ was found in the δ13C values measured for fatty acids in T-Iso. Additionally, the bulk of the algae was found to be more depleted in 13C than some of the fatty acids. On top of these unusual results an even higher range of 17‰ was observed in the δ13C values of fatty acids in rotifers. In a minimum of 2 days enormous differences had developed between the δ13C of fatty acids from the rotifers and the T-Iso culture (up to 17‰, for 14:0; see Figure 6.5). Later in this chapter it will be argued that a change of the CO2 tank, used for bubbling of the T-Iso culture, must have taken place within a few days before the experiment was set up.

208

13

δ C of bulk sample (‰)

-44.5

-45.0

-45.5

-46.0

-46.5 0

10

20

30

40

50

60

70

Starvation time (h)

Figure 6.6. Bulk δ13C measurements of rotifers plot against time of starvation.

6.4.4 Effect of starvation on stable carbon isotope composition Whereas the fatty acid concentrations in the rotifers showed an exponential decay curve, the δ13C of the total fatty acids shows a linear increase during 72 hours of starvation (Figure 6.4). Similarly, the

13

C/12C ratio of bulk sample increased in a linear fashion

(Figure 6.6). The δ13C of the bulk increased about 1.5‰, and that of the total fatty acids approximately 3‰. In Figure 6.7 all the δ13C measurements of the time series for each fatty acid are plotted, with symbol size increasing with time of starvation. Examples of trends in the δ13C over time are shown for several fatty acids in Figure 6.8. Fatty acids such as the 16:0, 22:6n-3 and 18:4n-3 all show a linear increase of 2-3‰. The 14:0 did not change much until the last 24 hours, and the 13C/12C ratio of 20:5n-3 actually decreased. In fact, this decrease is not linear, but actually follows a decay curve (Figure 6.8). The relative changes in the δ13C of the various fatty acids between t=2 and t=72 hours are shown in Figure 6.9. Over this 70 hour time period the δ13C of most fatty acids

209 increased, but also a few fatty acids showed a modest decrease in the 13C/12C ratio (see Figure 6.9). The highest increase was measured for 18:1n-7, however this may be an artefact due to a decrease in the peak size of the shouldering 18:1n-9 at the front-end of 18:1n-7 (see discussion).

-44 -46

δ13C of fatty acid (‰)

-48

t=0 h

-50

t=2 h

-52

t=4 h

-54

t=8 h

-56 -58

t=12 h t=24 h

-60

t=48 h

-62

t=72 h

-64

14:0 iso 15:0 15:0 16:0 16:1n-7 16:2n-4 17:0 16:4n-3 18:0 18:1n-9 18:1n-7 18:2n-6 18:2n-4 18:3n-3 18:4n-3 20:2n-6 20:4n-6 20:4n-3 20:5n-3 21:5n-3 22:5n-6 22:5n-3 22:6n-3

-66

Fatty acid Figure 6.7. δ13C measurements during the 72 hour time series plotted for each fatty acid, with symbol size increasing with time of starvation (see legend). The open circle indicates the δ13C of the fatty acid before starvation.

6.4.5 Fecundity The average number of eggs per rotifer decreased from about 0.1, in the initial stages of starvation, to almost 0 after 48 hours of starvation (see Figure 6.10). The observation of a decrease in fecundity during starvation is in accordance with previously reported findings (Yoshinaga et al., 2000).

210 -50

-52

-56

16:0 20:5n-3 14:0 22:6n-3 18:4n-3

-58

13

δ C of fatty acid (‰)

-54

-60

-62

-64 0

10

20

30

40

50

60

70

Time of starvation (h)

Figure 6.8. Examples of the variation in δ13C of fatty acids during 72 hours of starvation.

6.5 Discussion 6.5.1 Decrease in total fatty acid concentration The various energy reserves (protein, lipids, carbohydrates) are not used in similar order and are not of similar importance for all invertebrate species (Whyte et al., 1986, and references therein). Frolov and Pankov (1992) found that during starvation of the same rotifer as used here (Brachionus plicatilis) the proportion of lipids and carbohydrates decreased and that of total proteins increased. The proportion of lipids decreased most rapidly in the initial 20 hours, the rate of decrease in the glycogen level was highest between 20 and 48 hours, and the fraction of other carbohydrates was depleted most rapidly between 48 and 72 hours of starvation (Frolov and Pankov, 1992). Here, the

211 fraction of fatty acids per unit sample weight decreased to half its initial value within 48 hours (Fig. 6.4). Therefore, also during this experiment lipids seem to be preferentially used as a source of energy during the initial stages of starvation.

6

4

13

δ C at t=72h - δ C at t=2h (‰)

8

2

13

0

22:6n-3

22:5n-3

22:5n-6

20:5n-3

20:4n-3

20:4n-6

20:2n-6

18:4n-3

18:3n-3

18:2n-6

18:1n-7

18:1n-9

18:0

16:4n-3

17:0

16:2n-4

16:0

16:1n-7

15:0

14:0

iso 15:0

-2

Fatty acid Figure 6.9. Change in δ13C of fatty acids from 2 hours after the commencement (t=2 h) to 72 hours of starvation (t=72 h). Positive numbers indicate enrichment in 13C from t=2 h to t=72 h, negative numbers indicate the fatty acids became more depleted in 13C during that time period. Error bars represent ± 1 S.D. determined from measurements on the three batches of rotifers starved.

About 55% of the total decrease in fatty acid concentration already occurred in the first 8 hours of starvation, with the largest drop in the first two hours (Fig. 6.4). This is significantly more than observed by Frolov and Pankov (1992), who found that only 30% of the 72-hour decrease of the total lipid fraction occurred in the first 8 hours. The temperatures at which both experiments were conducted were approximately the same (18-22°C vs about 21°C here). However, there are also differences between the experiments. A different alga was fed to the rotifers (Monochrysis lutheri vs T-iso here), and Frolov and Pankov (1992) deprived the rotifers of food for a 40-minute period prior to the start of the time series, in order to allow gut clearance. In the experiment reported

212 here some time (10-30 min.) elapsed between placing the rotifers in filtered seawater and taking the t=0 sample. However, this may possibly not have resulted in total gut clearance prior to t=0. Therefore, the question arises whether the initial rapid decrease in total fatty acid concentration was due to gut clearance.

0.16

Fecundity (eggs per rotifer)

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0

10

20

30

40

50

60

70

Time of starvation (hours)

Figure 6.10. Fecundity (average number of eggs per rotifer) during the 72 hours of starvation. Error bars represent ± 1 S.D. determined from counts on the three batches of rotifers starved.

The initial drop in total fatty acid concentration, between 0 and 2 hours of starvation, was from 77 to 63mg fatty acid/g of sample. For this drop to be the result of gut clearance the excreted material would have to have been richer in fatty acids than the rotifer itself. The concentration of fatty acids in T-Iso has been determined here, and amounted to about 65mg fatty acids/g of sample. This is lower than that of the rotifer at t=0. However, it should be borne in mind that probably the concentration of lipids could have been somewhat higher in the excreted material than in the T-Iso. Roman (1991) found in a study on a copepod fed labelled algae that the highest ratio of incorporation to ingestion was found for proteins (0.81), with lower ratios for polysaccharides and lipids (0.27 and 0.25, respectively). Still, even with this difference in mind, the gut clearance of

213 partly digested T-Iso material cannot account for the initial drop in fatty acid concentration. Hence, it seems that most of the decrease in fatty acid content should be ascribed to catabolism. It may be possible that this heightened catabolic activity was still part of the specific dynamic action, which is the increase in respiration rate associated with feeding (Grisolia and Kennedy, 1966). The main contributors to specific dynamic action are energetic costs associated with biomass formation, absorption and assimilation (Kiørboe et al., 1985).

6.5.2 Selective mobilization of fatty acids In Figure 6.2 a weak trend can be observed with higher relative losses for the more highly unsaturated fatty acids. Raclot and others, repeatedly observed selective mobilization of fatty acids from triacylglycerols (TAGs) (Raclot and Groscolas, 1993, 1995; Conner et al., 1996; Raclot, 1997; Perona et al., 2000). These authors found that fatty acids that are shorter, more unsaturated, and have their double bonds closer to the methyl-end are more readily mobilized than their counterparts during lipolysis. The reasons for this effect are not entirely clear, but it seems that the different physicochemical properties will cause heterogenous distribution of TAGs in oil droplets. The polarity of the fatty acids increases with decreasing chain length and increasing unsaturation. The more polar TAGs may be more concentrated on the outside of the oil droplet, and are therefore more available for lipolysis (Raclot, 1997). Additionally, more affinity to polar fatty acids by lipases may result in the observed selective mobilization. Although different animals (rats and rabbits) were used in the aforementioned studies, the observed pattern seems to fit the data obtained during this experiment (Figure 6.4). The trends with unsaturation and chain length are probably somewhat obscured here by the fact that the fatty acids are not only derived from TAGs but also from polar lipids.

6.5.3 Retroconversion? Two fatty acids, the 16:4n-3 and 22:5n-3, were found to show an increase in concentration (mg FA/g of sample) during the 72 hours of starvation (Figure 6.2). It is

214 important to note that this does not necessarily imply that an increase in the absolute amount of these fatty acids occurred, since the total weight per individual also must have decreased. On the basis of the findings of Raclot and others (see above), 16:4n-3 and 22:5n-3 would be expected to have been mobilized quite readily. Instead, their relative abundance was found to increase. Two mechanisms can potentially explain this phenomenon. First, both fatty acids may have been present strictly in the polar lipids, which were spared the most from catabolism. A second mechanism could involve the formation of 16:4n-3 and 22:5n-3 during starvation. A labelling study by Williard et al. (1998) showed that 16:4n-3 was the most abundant product of (incomplete) peroxisomal β-oxidation of 20:5n-3 in human skin fibroblasts. Hence, it is conceivable that also here some 16:4n-3 was formed by two cycles of β-oxidation of 20:5n-3, and/or by β-oxidation of 18:4n-3. The 22:5n-3 may have been formed by elongation of 20:5n-3. In fact, Grønn et al. (1991) found that, even during fasting, rat hepatocytes would retroconvert 22:6n-3 into 20:5n-3, which subsequently was partly elongated to 22:5n-3. During the starvation of rotifers, Frolov and Pankov (1992) also noted an increase of 22:5n-3 in the polar lipids, but the proportion of the same fatty acid decreased in the neutral lipids.

6.5.4 Shift in T-Iso δ13C before and after start of rotifer starvation From Figure 6.5 it is clear that the δ13C of the T-Iso culture must have undergone a large shift after the start of the starvation experiment. Fatty acids from the rotifers at the start of the experiment were found to be depleted in 13C by about 4‰ to as much as 20‰ than the T-Iso culture sampled 2 days later. A large shift in the δ13C values of a magnitude of 20‰ has to be explained by a shift in the δ13C of the carbon source for the algae. The TIso culture was bubbled with CO2. Hence, it is likely that the CO2 tank was changed, possibly with an air-tank, just after the starvation of the rotifers. Unfortunately, no records were kept to verify a tank switch. The remarkably large range of 15‰ found in the δ13C values of T-Iso supports the hypothesis of a recent shift in carbon source. Saoudi-Helis et al. (1994) have measured

215 the fatty acid composition of various lipid classes from T-iso. They found that the 18:1n9, followed by the 16:0 and 14:0 fatty acids are the most abundant fatty acids in TAGs of T-iso. Particularly, the 18:1n-9 was proportionally more concentrated in the TAGs, with respect to the polar lipids. Conversely, the 18:4n-3 was particularly prevalent in the phospholipids. Hence, the difference in turnover times for these lipid classes, coupled with a shift to a 13C enriched carbon source, may explain the high δ13C of 18:1n-9, 16:0 and 14:0 and still low values for 18:4n-3 and 22:6n-3. Moreover, it was found that the TIso bulk sample has δ13C values that are up to 5‰ lower than some of the fatty acids (see Figure 6.5). From a biosynthetic point of view, fatty acids should be

13

C depleted with

respect to the bulk (DeNiro and Epstein, 1977; Monson and Hayes, 1982; Melzer and Schmidt, 1987). Fatty acids in natural particular organic matter are typically between 4 and 8‰ depleted in 13C compared to the bulk (see Chapter 4). The lower turnover time of other biochemical fractions, such as proteins, will cause a slower “response time” to a new carbon source than lipids (see also Chapter 2). Therefore, a bulk more depleted in 13

C relative to some of the fatty acids is in accordance with a shift to a different carbon

source that is richer in 13C. Obviously, the shift in δ13C of the T-iso culture after commencement of the starvation experiment would not affect the stable carbon isotope composition of the rotifers used in the experiment. However, there is also reason to believe that the δ13C of T-Iso was shifting also before the rotifers were placed under food deprivation. The unusually large range of 17‰ in the δ13C values measured for fatty acids of the rotifers at t=0 attests to this (Figure 6.5). Additionally, the bulk δ13C would normally be expected to adjust slower than most of the fatty acids to the isotope composition of the new diet. The δ13C of 18:4n-3 is about 17‰ lower than the bulk. This would be very hard to explain when a shift toward more 13C enriched food took place, since this a fatty acid that is very abundant in T-Iso (Figure 6.1) and, therefore, can be expected to turnover relatively fast. Hence, it is believed that before the commencement of the starvation experiment the δ13C of T-Iso shifted to an isotope composition that was much more depleted in 13C.

216

6.5.5 Difficulties for interpretation One of the aims of this experiment was to study the effect of catabolism on the δ13C of individual fatty acids. So far, two problems have been identified that will prevent reaching any firm conclusions on this matter. First of all, there are at least two pools of fatty acids, i.e., constituents of neutral, and polar lipids, that undergo different rates of catabolism. For example, Frolov and Pankov (1992) found that the abundance of TAGs decreased from about 55% to 4% of the total lipids, whereas the proportion of polar lipids increased during 120 hours of starvation. Thus, TAGs are primarily catabolized by the rotifers during starvation. Due to their different turnover rates, together with the recent shift in δ13C of T-Iso, identical fatty acids from the two pools will exhibit dissimilar 13

C/12C ratios. Hence, when fatty acids from one pool are catabolized more than those

from the other pool, the δ13C of the total fatty acids will shift. Without knowing the relative loss rates for the different pools (e.g., TAGs and polar lipids), it remains unclear how much of this shift (if any) is due to an isotope effect due to catabolism. Secondly, due to the shift in δ13C of the diet of the rotifers (T-Iso) over time, newly derived carbon has a different 13C/12C than earlier assimilated carbon. When there is any differential catabolism of “younger” and “older” fatty acids, a shift in δ13C will occur in the remaining pool of fatty acids. This could be misinterpreted as being an effect of catabolism on the δ13C of fatty acids in general.

6.5.6 Observed changes in δ13C As outlined above, fluctuations in the δ13C of T-Iso while being fed to the rotifers were not monitored. Additionally, no measurements were made on the various pools of fatty acids separately. In spite of these shortcomings, below an attempt will be made to elucidate some of the processes operating during the starvation. Due to the recent shift in diet δ13C, it is anticipated that the TAGs and polar lipids have dissimilar

13

C/12C ratios as a result of their different turnover times. Since TAGs

have a higher turnover rate, they will contain a higher proportion of newly assimilated fatty acids. Furthermore, it is expected, and found by Frolov and Pankov (1992), that

217 TAGs are lost more rapidly than polar lipids during starvation. As a result, the fatty acids can be expected to shift toward isotope compositions reflecting more of the older diet signal. If indeed all of the assumptions stated above are correct, the results would confirm that the rotifers were fed T-Iso that shifted to more 13C depletion in the final days prior to the experiment (as suggested in section 6.5.4). This is because it was found that most of the fatty acids shifted towards more

13

C enriched values during starvation (Figures 6.4

and 6.7-6.9), which is proposed to be a shift to the “older diet signal”. Before starvation, the δ13C of the total fatty acids was about 10‰ lower than the bulk (Figures 6.4 and 6.5). The concentration and the δ13C of the total lipids were not measured throughout the experiment. Frolov and Pankov (1992) found that the proportion of lipids decreased from about 20% to 10% as a result of 72 hours of starvation. Perhaps, a somewhat larger decrease occurred during this experiment due to the extra large decrease between t=0 and t=2 h. (see Fig. 6.4) not observed by Frolov and Pankov (1992). Typically, not all lipids are as depleted in

13

C as fatty acids (Schouten et al.,

1998). Still, the rise in bulk δ13C (about 1.5‰) lies within the range of being entirely accounted for by the decrease in proportion of (13C-depleted) lipids. This is because the fatty acids, and perhaps also other lipids, that were catabolized had lower 13C/12C ratios than the average, as is revealed by the increase in δ13C of the total fatty acids (Fig. 6.4).

6.5.7 Conceptual model A conceptual model was constructed in order to test whether the assumptions outlined in the previous section can, in fact, explain the observed changes. The equations used for the model are described in the Appendix of this chapter. In the model the fatty acids of the neutral- and polar lipids of the rotifers are treated as two separate pools of fatty acids that are catabolized at different rates. At the start of the experiment the two pools of fatty acids are assumed to have a dissimilar δ13C due to a recent shift in the 13C/12C ratio of the food and different turnover rates of the two pools.

218 For simplicity the fatty acids of the rotifers are regarded as a mixture of fatty acids from “new food” (with low δ13C) and “old food” (with high δ13C). For the “old- and new food” a δ13C value of -46 and -64‰ was chosen, respectively. These values correspond approximately with the maximum and minimum values measured for the fatty acids in the rotifers at the start of the experiment (Figure 6.5). To attain a δ13C close to what was measured for the total fatty acids (see Fig. 6.4), it was assumed that 80% of the neutral lipid fatty acids, and 30% of the polar lipid fatty acids were derived from the “new food”. This is not unreasonable considering the different turnover times of the two fatty acid pools. The different proportions of newly assimilated fatty acids led to a δ13C difference of 9‰ between the fatty acids of the two lipid fractions. For the model it was assumed that catabolic processes are causing an exponential decrease in the amount of fatty acids during starvation. The neutral lipid fatty acids were allowed to decrease at a rate of 3% per hour, whereas the polar lipid fatty acids were given a catabolic rate of 0.3% per hour. These values were chosen so that they approximately correspond to values calculated from the study by Frolov and Pankov (1992). For the fraction of fatty acids derived from neutral lipids a value of 0.6 was given. This number correspondents with results from Fernández-Reiriz and Labarta (1996), who measured the fatty acid composition of Brachionus plicatilis fed Isochrysis galbana. For the amount of fatty acids at t=2 h 65mg was chosen, in analogy to the concentration of 65 mg/g of sample measured for the rotifers at that time. To start with, it was assumed that no isotope effect is associated with catabolism. In Figure 6.11 the output of the model is compared with the measured data. As can be seen in Figure 6.11, the basic features observed in the real data are described by the model output. Most importantly, the model approximates the linear increase in δ13C of the fatty acids, while the total amount of fatty acids decreases exponentially. It was not attempted here to model the concentration of the total fatty acids, as had been measured. The faster decrease in the modeled total amount of fatty acids, as compared to the measured mg/g concentration (Figure 6.11) can be expected. This is due to the concomitant decrease in the body weight of the rotifers as a result of the catabolism of carbohydrates (especially glycogen) and perhaps some proteins. On the

219 other hand, the slight curve in the modeled δ13C of the fatty acids may be an indication of

Concentration of total fatty acids (mg/g sample)

Amount of total fatty acids (mg)

inaccuracy of the model or the starting values.

80

Measured Model

70

60

50

40

30

20 0

10

20

30

40

50

60

70

-52

-54

-55

-56

13

δ C of total fatty acids (‰)

-53

Measured Model

-57

0

10

20

30

40

50

60

70

Time of starvation (hours)

Figure 6.11. Measured values of total fatty acid concentration compared to modeled amount of total fatty acid (top). Modeled δ13C of total fatty acids plot with the measured δ13C of total fatty acids during 72 hours of starvation of rotifers (bottom).

In Figure 6.12a the effect of an isotope effect during the catabolism of fatty acids can be observed. Shown here is that even an unrealistically large isotope effect of 1.030 would only produce an extra increase of about one per mil during the 72 hours of starvation. In Figure 6.12b the effect of an isotope effect is shown for the case when the

220 starting values of the two fatty acid pools are similar. Clearly, the larger carbon isotope fractionation within the faster catabolized pool (neutral lipids) is partly masked by the fatty acids of the pool that is catabolized at a slower rate (polar lipids). Additionally, the isotope effect results in a more curved increase in the δ13C of the total fatty acids, which is not according to what has been observed (Figures 6.4 and 6.9). It is possible to produce a linear increase with an isotope effect associated with catabolism when only one pool of fatty acids exists. Beside the fact that this would not be realistic from a physiological point of view, with a similar loss rate of fatty acids an isotope effect with an α-value of about 1.060 would be needed to produce the observed shift in δ13C. This is even more unrealistic, and not at all in accordance with observations by Vogler and Hayes (1980), Monson and Hayes (1982a,b) and the results of Chapter 4. Therefore, the linear increase in the δ13C of the total fatty acids cannot be explained by a kinetic isotope effect operating during the catabolism of a single pool of fatty acids. The model is quite sensitive to the proportion of the neutral- and polar lipids (Figure 6.12c), and also the values entered for the δ13C of the new- and old diet fatty acids and their proportions in the two fatty acid pools (not shown). In Figure 6.12c it can be observed that when a higher proportion of the fatty acids is derived from the neutral lipids a slightly more linear increase in δ13C can be attained. It would therefore be possible to fit the observed data even better by changing this proportion, and concomitantly adjust the estimates of the δ13C of the new- and old diet fatty acids. In summary, the model seems to support the hypothesis that, at the time of sampling, the rotifer culture had just undergone a shift in diet δ13C. This resulted in a large difference in δ13C of the neutral- and polar lipids, with the neutral lipids being depleted in

13

C possibly by as much as 9‰ relative to the polar lipids. The differential

loss rates of these two fatty acid pools during starvation gave rise to a steady increase in δ13C of the total fatty acids.

221

δ C of fatty acids 13

Amount of fatty acids 70

Neutral- and polar lipids have different δ13C

1.030

-53

1.010 1

50

-54

40

-55

30

-56

a

20 0

-57 10

70

20

30

40

50

60

70

50 40 30

0

20

30

40

50

60

70

-53 -54 -55

α 1.030 1.020 1.010 1

-56 -57

13

b

20

10

-52

Neutral- and polar lipids 13 have similar δ C

60

0

δ C of fatty acids (‰)

Amount of total fatty acids (mg/g of initial sample weight)

60

α

-52

10

70

20

30

40

50

60

70

10

20

30

40

50

60

70 %Ntrl/%Pol

-52

Different proportions of neutral- and polar lipids

60

0

50/50 60/40

-53

70/30

-54

50

-55

40

%Ntrl/%Pol

30

50/50 60/40

c

20 0

70/30

10

20

30

40

50

60

70

-56 -57 0

10

20

30

40

50

60

70

Time of starvation (hours) Figure 6.12. Sensitivity of model to different parameters. On the left-hand side the model total amount of fatty acids, and on the right-hand side the modeled δ13C of the total of fatty acids is shown as a function of starvation time. In the Appendix the starting values of the model are given. At the top (a) the influence of introducing a kinetic isotope effect during catabolism of fatty acids is shown (α=1 means no isotope effect). In the middle (b) the effect of introducing an isotope effect is shown when the neutral- and polar lipids do not exhibit different δ13C values at the start of the experiment. At the bottom (c) the effect of changing the proportion of neutral- and polar lipids before starvation are shown.

222

6.5.8 Differential 13C-enrichment The majority of the fatty acids shows an increase in δ13C during the starvation (Figures 6.7-6.9). However, not all of them increased in δ13C to the same extent, and some even showed a decrease (Figures 6.7-6.9). The extremely large enrichment in 13C of 18:1n-7, however, should be regarded with some caution. As the 18:1n-9 and 18:1n-7 are not completely separated down to the base-line in the chromatogram, some of the

13

C-

depleted tail-end of the 18:1n-9 fatty acid is bound to be partly integrated with the 18:1n7. Because the amount of 18:1n-9 decreased more rapidly during the 72 hours of starvation, the “contamination” from its tail-end may have become less. Hence, δ13C increase measured for 18:1n-7 may be artificially high. On the other hand, the 18:1n-9 fatty acid has perhaps undergone a larger shift in its δ13C than recorded here. Some of the other differences in increase of δ13C of the various fatty acids may be explained by differences in turnover time and dissimilar distributions over the neutral- and polar lipids. As argued earlier in section 6.5.3, perhaps some retroconversion due to incomplete β-oxidation has taken place during starvation. This hypothesis can now be tested again with the help of the δ13C values for the individual fatty acids. The δ13C of 18:4n-3 was about 12‰ lower than that of 16:4n-3 at the onset of the experiment (Figure 6.5). Therefore, it would be expected that if 18:4n-3 was retroconverted to 16:4n-3, the latter should not become enriched in

13

C (as opposed to most fatty acids). Indeed, a slight

decrease in δ13C was observed for 16:4n-3 (Figures 6.7 and 6.9), and hence, the hypothesis that retroconversion of 18:4n-3 to 16:4n-3 took place is not refuted by the isotope data. Likewise, retroconversion of 22:6n-3 to 20:5n-3, and some of the 22:5n-6 to 20:4n-6 would be supported by the δ13C data. However, any substantial elongation of 20:5n-3 to 22:5n-3, as suggested in section 6.5.3, is contested because 22:5n-3 became more enriched in 13C. The retroconversion of 22:6n-3 to 20:5n-3 has been shown to occur in Artemia (Navarro et al., 1999), and has also been reported for humans (Conquer and Holub, 1997). Indeed, the exponential decrease in δ13C of 20:5n-3 (Figure 6.8), concurrent with the exponential decrease in the concentration of 22:6n-3 (Figure 6.3), would fit the hypothesis of retroconversion of 22:6n-3 to 20:5n-3 in starving rotifers.

223

6.5.9 Recommendations The insights from this starvation experiment, and its shortcomings, lead to several recommendations for carrying out a similar, but improved experiment in the future. In order to avoid large differences in δ13C between the various lipid fractions, it is very important to grow the rotifers for a sufficiently long period (2 weeks?) on a diet with a constant

13

C/12C ratio. This would not only be good for the study of the effects of

catabolism on the δ13C of fatty acids, but also for establishing whether any fractionation occurs in the transfer of fatty acids from the diet to a heterotroph. Ideal would probably be the feeding of a well-mixed batch of dried food, such as spray-dried Schizochytrium sp. (e.g., Barclay and Zeller, 1996). Another improvement would be to separate the lipids into a polar and neutral lipid fractions before preparing the fatty acid methyl esters for identification, quantification and δ13C measurements. Also, isolating the carbohydrates and proteins for either bulk- or compound specific stable carbon isotope analyses would lead to an even more complete understanding of the effects of catabolism on the δ13C of an organism. Such information can then be used to better understand the variation in trophic fractionation observed by researchers trying to elucidate trophic relationships. Next to carrying out a starvation study it would also be instructive to carry out a controlled diet switch experiment. Time series measurements on the quantity and 13C/12C ratio of the various biochemical fractions and individual compounds could lead to a detailed assessment of the various turnover times. This knowledge can be of use to confirm the switch in diet by an organism, or recent changes at the base of a food web.

6.6 Conclusions Fatty acids with shorter carbon chain lengths and a higher degree of unsaturation had some preference over their counterparts during catabolic processes in rotifers while fasting. This is in accordance with previously reported observations of selective mobilization of fatty acids from TAGs in rodents. Most likely, some retroconversion of 18:4n-3 to 16:4n-3, and probably also of 22:6n-3 to 20:5n-3 occurred during starvation.

224 The δ13C of the food of the rotifers (T-Iso) must have shifted during the time leading up to the starvation experiment. This resulted in a large disparity between the 13

C/12C ratios of fatty acids from the neutral- and polar lipids due to their different

turnover rates. Hence, the linear increase in δ13C of the total fatty acids during starvation was not the result of a kinetic isotope effect, but due to a decrease in the proportion of (13C depleted) neutral lipid fatty acids. Future research investigating the effect of catabolism should, prior to the starvation period, raise the animal on a diet with a constant δ13C for a sufficiently long time. Additionally, to get better insight in the stable carbon isotope dynamics during starvation, the different biochemical fractions and lipid classes should be analyzed separately.

6.7 References Barclay, W. and Zeller, S. 1996. Nutritional enhancement of n-3 and n-6 fatty acids in rotifers and Artemia nauplii by feeding spray-dried Schizochytrium sp.. Journal of the World Aquaculture Society 27, 314-322. Connor, W.E., Lin, D.S. and Colvis, C. 1996. Differential mobilization of fatty acids from adipose tissue. Journal of Lipid Research 37, 290-298. Conquer, J.A. and Holub, B.J. 1997. Dietary docosahexaenoic acid as a source of eicosapentaenoic acid in vegetarians and omnivores. Lipids 32, 341-345. DeNiro, M.J. and Epstein, S. 1977. Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197, 261-263. Fernández-Reiriz, M.J. and Labarta, U. 1996. Lipid classes and fatty acid composition of rotifers (Brachionus plicatilis) fed two algal diets. Hydrobiologia 330, 73-79. Focken, U. and Becker, K. 1998. Metabolic fractionation of stable carbon isotopes: implications of different proximate compositions for studies of the aquatic food webs using δ13C data. Oecologia 115, 337343. Frolov, A.V. and Pankov, S.L. 1992. The effect of starvation on the biochemical composition of the rotifer Brachionus plicatilis. Journal of the Marine Biological Association of the United Kingdom 72, 343-356.

225 Gautier, J.F., Pirnay, F., Lacroix, M., Mosora, F., Scheen, A.J., Cathelineau, G. and Lefebvre, P.J. 1996. Changes in breath (CO2)-13C/(CO2)-12C during exercise of different intensities. Journal of Applied Physiology 81, 1096-1102. Gorokhova, E. and Hansson, S. 1999. An experimental study on variations in stable carbon and nitrogen isotope fractionation during growth of Mysis mixta and Neomysis integer. Canadian Journal of Fisheries and Aquatic Sciences 56, 2203-2210. Grice, K., Klein Breteler, W.C.M., Schouten, S., Grossi, V., de Leeuw, J.W., Sinninghe Damsté J.S. 1998. Effects of zooplankton herbivory on biomarker proxy records. Paleoceanography 13, 686-693. Grisolia, S. and Kennedy, J. 1966. On specific dynamic action, turnover and protein synthesis. Perspectives in Biology and Medicine 9, 578-583. Grønn, M., Christensen, E., Hagve, T.-A. and Christophersen, B.O. 1991. Peroxisomal retroconversion of docosahexaenoic acid (22:6(n-3)) to eicosapentaenoic acid (20:5(n-3)) studied in isolated rat liver cells. Biochimica et Biophysica Acta 1081, 85-91. Harrison, P.J., Waters, R.E. and Taylor, F.J.R. 1980. A broad spectrum artificial seawater medium for coastal and open ocean phytoplankton. Journal of Phycology 16, 28-35. Kiørboe, T., Møhlenberg, F. and Hamburger, K. 1985. Bioenergetics of the planktonic copepod Acartia tonsa: relation between feeding, egg production an respiration, and composition of specific dynamic action. Marine Ecology-Progress Series 26, 85-97. Klein Breteler, W.C.M., Grice K., Schouten S., Kloosterhuis H.T., Sinninghe Damsté J.S. 2002. Stable carbon isotope fractionation in the marine copepod Temora longicornis: unexpectedly low δ13C value of faecal pellets. Marine Ecology-Progress Series 240, 195-204. Melzer, E. and Schmidt, H.-L. 1987. Carbon isotope effects on the pyruvate dehydrogenase reaction and their importance for relative carbon-13 depletion in lipids. Journal of Biological Chemistry 262, 81598164. Miller, R.F., Orr, G.L., Fritz, P., Downer, R.G.H. and Morgan, A.V. 1985. Stable carbon isotope ratios in Periplaneta americana L., the American cockroach. Canadian Journal of Zoology 63, 584-589. Monson, K.D. and Hayes, J.M. 1982a. Carbon isotopic fractionation in the biosynthesis of bacterial fatty acids. Ozonolysis of unsaturated fatty acids as a means of determining the intramolecular distribution of carbon isotopes. Geochimica et Cosmochimica Acta 46, 139-149. Monson, K.D. and Hayes, J.M. 1982b. Biosynthetic control of the natural abundance of carbon 13 at specific positions within fatty acids in Saccharomyces cerevisiae. Journal of Biological Chemistry 257, 5568-5575.

226 Nagata, W.D. and Whyte, J.N.C. 1992. Effects of yeast and algal diets on the growth and biochemical composition of the rotifer Brachionus plicatilis (Muller) in culture. Aquaculture and Fisheries Management 23, 13-21. Navarro, J.C., Henderson, R.J., McEvoy, L.A., Bell, M.V. and Amat, F. 1999. Lipid conversions during enrichment of Artemia. Aquaculture 174, 155-166. Oelbermann, K. and Scheu, S. 2002. Stable isotope enrichment (δ15N and δ13C) in a generalist predator (Pardosa lugubris, Araneae : Lycosidae): effects of prey quality. Oecologia 130, 337-344. Olsen, Y., Reitan, K I. and Vadstein, O. 1993. Dependence of temperature on loss rates of rotifers, lipids, and ω3 fatty acids in starved Brachionus plicatilis cultures. Hydrobiologia 255/256, 13-20. Perona, J.S., Portillo, M.P., Macarulla, M.T., Tueros, A.I., Ruiz-Gutiérrez V. 2000. Influence of different dietary fats on triacylglycerol deposition in rat adipose tissue. British Journal of Nutrition 84, 765-774. Raclot, T. and Groscolas, R. 1993. Differential mobilization of white adipose tissue fatty acids according to chain length, unsaturation, and positional isomerism. Journal of Lipid Research 34, 1515-1526. Raclot, T. and Groscolas, R. 1995. Selective mobilization of adipose tissue fatty acids during energy depletion in the rat. Journal of Lipid Research 36, 2164-2173. Raclot, T. 1997. Selective mobilization of fatty acids from white fat cells: Evidence for a relationship to the polarity of triacylglycerols. Biochemical Journal 322, 483-489. Roman, M.R. 1991. Pathways of carbon incorporation in marine copepods: Effects of developmental stage and food quantity. Limnology and Oceanography 36, 796-807. Saoudi-Helis, L., Dubacq, J.-P., Marty, Y., Samain, J.-F. and Gudin, C. 1994. Influence of growth rate on pigment and lipid composition of the microalga Isochrysis aff. galbana clone T.iso. Journal of Applied Phycology 6, 315-322. Vogler, E.A. and Hayes, J.M. 1980. Carbon isotopic compositions of carboxyl groups of biosynthesized fatty acids. In: A.G. Douglas and J.R. Maxwell (eds.) Advances in Organic Geochemistry, 1979, Pergamon Press, 697-704. Whyte, J.N.C., Englar, J.R., Carswell, B.L. and Medic, K.E. 1986. Influence of starvation and subsequent feeding on body composition and energy reserves in the prawn Pandalus platyceros. Canadian Journal of Fisheries and aquatic Sciences 43, 1142-1148. Whyte, J.N.C. 1988. Fatty acid profiles from direct methanolysis of lipids in tissue of cultured species. Aquaculture 75, 193-203.

227

Williard, D.E., Kaduce, T.L., Harmon, S.D. and Spector, A.A. 1998. Conversion of eicosapentaenoic acid to chain-shortened ω3 fatty acid metabolites by peroxisomal oxidation. Journal of Lipid Research 39, 978986. Yoshinaga, T., Hagiwara, A. and Tsukamoto, K. 2000. Effect of periodical starvation on the life history of Brachionus plicatilis O.F. Müller (Rotifera): a possible strategy for population stability. Journal of Experimental Marine Biology and Ecology 253, 253-260.

228

Appendix This appendix describes the equations used in the conceptual model used to further understanding of the response of the δ13C of (total) fatty acids in the rotifers to starvation. Because the diet has shifted in isotope composition, and the neutral- and polar lipid fatty acids have different turnover times, the δ13C of the total fatty acids is weighted average of the δ13C of the two fatty acid pools:

δ TotFA (t ) =

w N (t ) w P (t ) ⋅ δ N (t ) + ⋅ δ P (t ) , wTotFA (t ) wTotFA (t )

(6.1)

where δTotFA (t) is the δ13C of the total fatty acids (as a function of time), and the δN and δP are the δ13C of neutral- and polar lipid fatty acids, respectively. wTotFA is the amount (weight) of total fatty acids, wN is the amount of neutral lipid fatty acids, and wP the amount of polar lipid fatty acids, with: wTotFA (t ) = w N (t ) + w P (t ) ,

(6.2)

where

(

)

w N (t ) = f N (0) ⋅ (1 − catf N ) t ⋅ wTotFA (0) ,

(6.3)

and

(

)

wP (t ) = f P (0) ⋅ (1 − catf P )t ⋅ wTotFA (0)

(6.4)

where fN (0) and fP (0) are the fractions of the total fatty acids that consist of neutral lipidand polar lipid fatty acids, respectively, at t=0. The catfN and catfP are the fractions of neutral lipid- and polar lipid fatty acids that are being catabolized per time unit (here per hour).

229 The δ13C of the neutral lipid fatty acids (δN) and that of polar lipid fatty acids (δP) are calculated as follows (taking a Raleigh distillation effect into account): α −1  wN ( t )   f N (0) ⋅ wTotFA (0) 



δ N (t ) = δ N (0) ⋅ 

(6.5)

and



 w P (t )   f P (0) ⋅ wTotFA (0) 

α −1

δ P (t ) = δ P (0) ⋅ 

(6.6)

where α is the isotope effect (k12/k13) associated with catabolism of the fatty acids, and δN(0) and δP(0) are the δ13C values at t=0 of the neutral lipid- and polar lipid fatty acids, respectively, which are calculated as follows: δ N (0) = newf N ⋅ newδ FA + (1 − newf N ) ⋅ oldδ FA

(6.7)

δ P (0) = newf P ⋅ newδ FA + (1 − newf P ) ⋅ oldδ FA

(6.8)

where newfN and newfP are the fractions of neutral lipid- and polar lipid fatty acids that consist of fatty acids derived from the diet after a shift in δ13C of the diet. The oldδFA and newδFA are the δ13C values of the total fatty acids from the diet before and after the δ13C shift, respectively. As a best estimate, to correspond optimally with the rotifers used in the reported experiment the parameters were given the following values (see text for explanation for the choice of values; for model output see Figure 6.11): newδFA = -64‰ oldδFA = -46‰ newfN = 0.8 newfP = 0.3 fN(0) = 0.6 fP(0) = 0.4

230 catfN = 0.03 catfP = 0.003 wTotFA (0) = 65 α = 1.000

231

7. Juvenile Salmon; a natural diet switch experiment

7.1 Abstract The migration of juvenile salmon to the ocean offers an ideal opportunity to study the effects of a natural dietary shift, both on the fatty acid composition and on the

13

C/12C

ratio of the bulk and individual fatty acids. For this study juvenile Pacific salmon were collected along the west coast of Vancouver Island and further north, up to Alaska, during three cruises in May 1998, May and June 1999. Stable carbon- and nitrogen isotope ratios of bulk muscle tissue were measured. Also, the molecular- and stable carbon isotope composition of fatty acids from four different species of salmon were determined and compared. Bulk muscle tissue from the juvenile salmon collected in May 1998 display a wide range of δ13C and δ15N values (-28 to -16‰, +8 to +16‰, respectively). In May 1999 the bulk muscle δ13C and δ15N values show less variation (range: -22 to -16‰, +10 to +14‰, respectively). The variation in isotope ratios in both years reflects the varying degrees of adjustment from a freshwater composition to typical marine signatures. Further illustration for the disequilibrium between body composition and the diet is an 8‰ disparity in the δ13C, together with a 4‰ difference in δ15N between the muscle tissue and the stomach contents for one of the salmon. Typically, the salmon depleted in

13

C

(i.e., δ13C < -23‰) also exhibited higher n-6/n-3 fatty acid ratios, in accordance with a remnant freshwater signal. Next to a higher abundance of n-6 fatty acids, more of the 18:3n-3, 18:4n-3 and 20:4n-3 fatty acids were found in the salmon that putatively emigrated from their freshwater habitats more recently. Muscle tissue from a group of juvenile sockeye salmon (Oncorhynchus nerka) sampled in May 1998 showed aberrant δ13C and δ15N values. Their δ13C values (< -23‰) suggest recent ocean-entry, but the

15

N/14N ratios (δ15N of about +13‰) indicate that

these salmon had already spent a considerable amount of time in the ocean. It is argued here that rearing in lakes and subsequent feeding in Barkley Sound (SW Vancouver

232 Island) is the most plausible explanation for the values observed for these juvenile sockeye salmon. The differences between δ13C values measured for the bulk and individual fatty acids of muscle tissue were greatest for the group of salmon showing typical marine isotope signatures (δ13C > -18‰, δ15N > +12‰). The other groups of salmon that were still adjusting their body composition showed on average lower fatty acid-bulk δ13C differences. This is consistent with faster turnover rates for fatty acids than bulk tissue. However, the deviation in bulk-fatty acid ∆δ13C deviations from the reference may not be large enough to serve as a clear, definitive indicator for a recent diet switch. Discriminant analysis on fatty acid abundance data for the muscle tissue shows that the four analyzed species of salmon can be distinguished on the basis of their fatty acid composition.

7.2 Introduction Juvenile salmon do not only face a change in their physical environment, but also dietary changes as they migrate from the freshwater- to the marine environment. Due to the differences between primary producers and their carbon- and nitrogen sources in both environments, organisms from the respective food webs can be distinguished by their fatty acid composition and their stable isotope composition. Freshwater organisms tend to have higher n-6/n-3 fatty acid ratios (Ackman, 1967; Sargent, 1976; Henderson and Tocher, 1987; Smith et al., 1996), and are depleted in 13C and 15N relative to their marine counterparts (Chisholm et al., 1982; Schoeninger et al., 1983; Owens, 1987; Smith et al., 1996). Therefore, the migration of juvenile salmon to the ocean offers an opportunity to study the effects of a naturally occurring diet-switch, both on the fatty acid composition and on the 13C/12C ratio of the bulk and individual fatty acids. Among the Pacific salmon, pink (Oncorhynchus gorbuscha) and chum (Oncorhynchus keta) fry usually migrate into estuaries and coastal waters soon after emergence from the egg. Coho (Oncorhynchus kisutch) generally rear for one or more years before entering the ocean as fingerlings. Sockeye (Oncorhynchus nerka) typically

233 migrate rapidly from streams into lakes, where they rear up to several years. Both chinook (Oncorhynchus tschawytscha) and sockeye, however, may have extremely variable life histories. Some stocks migrate to the ocean at a small size soon after hatching, and others rear up to several years in freshwater (Sedgwick, 1982; Groot and Margolis, 1991). Juvenile salmon were collected during three cruises in May of 1998, May and June of 1999. Muscle tissue of the salmon was analyzed for bulk stable carbon- and nitrogen isotope ratios. Additionally, the molecular- and stable carbon isotope composition of fatty acids from the muscle tissue of four different species of salmon were measured and compared.

7.3 Methods 7.3.1 Sample collection and location of sampling Juvenile salmon were collected with the CSS W.E. Ricker in waters on the continental shelf (< 200m water depth) off Vancouver Island in May 1998, May 1999, and June 1999 (cruises R98-8155, HS9913, HS9914, respectively). Also, in May 1999 juvenile salmon were captured more north, close to Dixon Entrance and Baranof Island. In Figures 7.1 and 7.2 maps with the sampling locations are shown. The juvenile salmon analyzed in this study from cruise R98-8155 were collected between May 18 and 27, 1998. The salmon samples from 1999 were taken between May 18 and 28, and from June 23 to 26, 1999. The juvenile salmon were collected using a trawl net. The nets used had mouth openings that varied from 12 to 16 m deep, and generally the headrope of the net was towed at the surface. This means that, at the moment of capture, the salmon were residing between the surface and approximately 16 metres depth.

234

Figure 7.1 Map with sample locations from May 1998 cruise (R98-8155), with longitude (in °W) on the x-axis and latitude (in °N) on the y-axis. The black dots show the starting- and stopping locations of the net tows from which salmon were analyzed. The numbers indicate the tow number. The grey lines with negative numbers are isobaths with the respective water depths (in meters). The 200m-isobath is defined as the shelf-break and is indicated with a thick black line.

7.3.2 Handling and processing of samples After determining the species and measuring their size, the juvenile salmon were kept frozen at approximately -20ºC onboard the ship. The salmon from the 1999 cruise were stored at -80ºC upon arrival at the Pacific Biological Station in Nanaimo. However, the 1998 samples were stored at -20ºC for over a year till analysis. Muscle tissue from below the dorsal fin was sampled for further analyses. The skin and 1-2mm of sub-dermal muscle tissue was removed from the samples. To test whether any substantial oxidation of the 1998 samples had taken place, the fatty acid profiles of the 1998 and 1999 salmon were compared. The 1998 and 1999 samples have similar polyunsaturated fatty acid (PUFA) contents. Therefore, it was assumed that the storage at -20ºC had preserved the fatty acid composition of the sampled muscle tissue.

235

Figure 7.2. Map with sample locations from May 1999 and June 1999 cruises (HS9913 and HS9914, respectively), with longitude (in °W) on the x-axis and latitude (in °N) on the y-axis. Sample locations from the May 1999 cruise are indicated with stars, and for the June 1999 cruise with triangles. The labels are the names for the sampling lines: Baranof Island line, Dixon Entrance line, Triangle Island line (Tline), and the C-line is off Estevan point. The grey lines with negative numbers are isobaths with the respective water depths (in meters).

236

7.3.3 Fatty acid methyl ester preparation and isotope ratio measurements Please refer to the methods section in Chapter 2 for the procedures of the fatty acid methyl ester preparation, and stable carbon isotope ratio measurements.

7.3.4 Multivariate analyses For the principal component analysis the data set was standardized by using the correlation matrix instead of the variance-covariance matrix (Meglen, 1992). Three-group discriminant analysis was performed using the 18 most abundant fatty acids. Comparing the coefficients of the discriminant function can be an indication for the relative contribution of each variable to the group separation. However, such comparison is not helpful when the different variables do not have comparable variances. To assess the relative contribution of each fatty acid to the group separation, the discriminant function was standardized to produce standardized coefficients. For standardization the respective coefficients of the discriminant function are multiplied by the within-group standard deviation of the same variable (Rencher, 1995). All matrix computations for the principal component- and discriminant analysis were performed by using Matlab (v. 4.2b, The Mathworks Inc., Natick, MA, U.S.A.).

7.4 Results and discussion 7.4.1 Stable carbon- and nitrogen isotope composition The juvenile salmon collected in May 1998, in particular, display a wide range of δ13C values (-28 to -16‰) obtained from their bulk muscle tissue (Figure 7.3). In May 1999, the bulk muscle δ13C values showed less variation (range: -22 to -16.5‰; Figure 7.3). Figure 7.3 plots the stable carbon isotope compositions of the potential prey together with the δ13C values of the salmon muscle tissue. Since the juvenile salmon were found only in shelf waters, only zooplankton and larval fish that were collected inshore of the 200m

237

May 1999 Juv. salmon

Larval fish

Zooplankton

J.Salmon off Baranof Isl. J.Salmon off Dixon Entr. J.Salmon off Vanc. Isl. Larval fish off Vanc. Isl. Crab larvae off Vanc. Isl. Zoopl. off Baranof Isl. Zoopl. off Dixon Entr. Zoopl. off Vanc. Isl. -28

-26

-24

-22

-20

-18

-16

-22

-20

-18

-16

May 1998 Juv. salmon

Larval fish

Zooplankton

Juv. Salmon Larval fish Crab larvae Var. zooplankton -28

-26

-24

13

δ C (‰)

Figure 7.3. The δ13C of bulk muscle tissue of juvenile salmon and the δ13C of their potential marine prey collected in shelf waters (water depth < 200m). The zooplankton and larval fish in May 1998 were collected at approximately the same time (within a week). The zooplankton and larval fish collected off Vancouver Island in May 1999 were sampled about two weeks before the salmon were collected. Off Baranof Island and Dixon Entrance, however, the zooplankton was collected simultaneously with the juvenile salmon.

isobath (Figure 2.1 and 7.1) are shown. Figure 7.3 illustrates that a large proportion of the juvenile salmon collected in 1998 is more depleted in 13C than any of the zooplankton or larval fish collected at around the same period. In May 1999, only two out of 28 juvenile salmon had lower δ13C values than the zooplankton and larval fish that were sampled about two weeks earlier. The juvenile salmon of 1998 and May 1999 were sampled around the same time of the year (18 – 27 May vs. 18 – 28 May, respectively).

238

16 15

"conspicuous sockeye"

14

12 11

15

δ N (‰)

13

10

Chum 1998 Coho 1998 Sockeye 1998 Chinook 1998 Sockeye 1999 Chinook 1999

9 8 7 -28

-26

-24

-22

-20

-18

-16

13

δ C (‰) Figure 7.4. The δ13C plotted against the δ15N of bulk muscle tissue of juvenile salmon from the May 1998 and May 1999 cruises. The encircled group of sockeye will often in the text be referred to as “conspicuous sockeye”.

The δ13C and δ15N of bulk muscle tissue of salmon from both May 1998 and 1999 are plotted against each other in Figure 7.4. The δ13C and δ15N values show a positive correlation, however a group of juvenile sockeye salmon do not fit the relationship observed for the other salmon. These conspicuous sockeye were collected at wide variety of locations, i.e., at tows 4, 7, 10, 34, 36, 37, 39, 59, 67, 74 and 76 (cf. Figure 7.1). Figure 7.5 shows that both the 13C- and 15N depleted salmon are plotting at the lower end of the size spectrum, but are not the smallest. For example, the majority of the chum salmon are smaller than the

13

C-depleted sockeye (Figure 7.5). The chum, however, probably

migrated directly to the ocean after hatching, whereas the other salmon typically stay in the freshwater environment for at least a year (Sedgwick, 1982; Groot and Margolis, 1991).

239

16 -16

15

-18

14 13

δ N (‰)

15

-22

13

δ C (‰)

-20

-24

12 11 10

Sockeye Coho Chum Chinook

-26 -28 100

150

200

250

300

350

400

Sockeye Coho Chum Chinook

9 8 7 100

150

Fork length (mm)

200

250

300

350

400

Fork length (mm)

Figure 7.5. The δ13C (left) and the δ15N (right) of bulk muscle tissue of juvenile salmon from the May 1998 and May 1999 cruises plotted against their fork length.

Typically, organisms feeding in freshwater ecosystems are depleted in 13C and 15N with respect to marine species (Chisholm et al., 1982; Schoeninger et al., 1983; Owens, 1987; Smith et al., 1996). Therefore, it seems that the trend from low δ13C and δ15N values to more

13

C and

15

N enriched compositions is reflecting the fact that the various

individual had different times of ocean-entry. Because of the differences in ocean-entry time, the various salmon are at different stages in the adjustment of their body composition to a new equilibrium with the marine diet. Additionally, some differences in rates of adjustment to new δ13C and δ15N values can exist due to varying feeding rates. The fact that salmon collected in 1999 were not as depleted in

13

C and

15

N as in 1998

may indicate that the downstream migration of salmon took place earlier in 1999. However, some caution should be exercised in drawing this conclusion since in May 1999 more samples were collected further north (Figures 7.1 and 7.2).

7.4.2 δ13C and δ15N of gut contents

240 The δ13C and δ15N of the gut contents of four salmon caught in May 1998 were measured. Figure 7.6 indicates that the juvenile salmon (coho) with the lowest δ13C value (-24.8‰) had been feeding on prey items with a δ13C that was 8‰ higher than its muscle tissue. Furthermore, the δ15N of the gut contents was found to be about 4‰ higher than the salmon tissue (Fig. 7.6). The two salmon with δ13C values of –18.5 and –18.9‰ fed on a diet that had similar 13C/12C ratios as their muscle. However, the δ15N values do not show the usual trophic shift of about 3‰ (see Peterson and Fry, 1987; Michener and Schell, 1994). The coho for which a δ13C value of about -16‰ was measured does indeed show some enrichment in

15

N relative to its diet (δ15N of salmon is about 1.5‰ higher

than the stomach contents). These results are in accordance with the hypothesis that the various salmon were at different stages of adjustment to a new body composition as a result of the switch to a marine diet. The salmon with the lowest δ13C (approx. < -24‰) had recently migrated downstream to the ocean, whereas those with the highest δ13C values (around -16‰) had almost fully equilibrated with their marine diet.

16

-16

15 14

-20

13 12

15

δ N (‰)

13

δ C (‰)

-18

-22

11 10

-24

Salmon Stomach contents

Salmon Stomach contents

9

Tow 51 Chum 8

Tow 51 Chum 1

Tow19 Coho 16

Tow19 Coho 1

Tow 51 Chum 8

Tow 51 Chum 1

Tow19 Coho 16

Tow19 Coho 1

-26

Figure 7.6. Comparison of the δ13C and δ15N values of gut contents with values for muscle tissue from the same juvenile salmon. All four salmon were collected during the May 1998 cruise. The tow number (see Fig. 7.1), species and sample number are indicated on the x-axes.

241

7.4.3 Hypotheses for origin of conspicuous sockeye An explanation for the isotope composition of the conspicuous group of sockeye (δ13C <23‰, δ15N ca. +13‰; Figure 7.4), remains unresolved. On the one hand, their

13

C/12C

ratios seem to indicate that they only recently moved from the freshwater ecosystem. But on the other hand, the

15

N/14N ratios seem to suggest that they had already resided for

period in the marine environment. Three competing hypotheses were constructed to explain the stable isotope ratios of this group of sockeye salmon. Figure 7.7 schematically shows the change in stable isotope composition over time as proposed by all three hypotheses. One hypothesis (1, Figure 7.7) involves movement of the sockeye to waters beyond the shelf-break, after which they returned again to the coast to be captured. As shown in earlier chapters, in May 1998 the δ13C values of pelagic organisms off the shelf were lower by about 5‰ than organisms found in shelf waters (see Figure 2.5). Next to assimilating carbon that is more depleted in 13C, a more typically marine stable nitrogen isotope composition (δ15N of about 15‰, Fig. 7.4) would be attained off the shelf. The second hypothesis (2, Figure 7.7) requires the particular group of sockeye to have fed in a riverine or estuarine environment in which the nitrogen at the base of the food web has become enriched in 15N. High δ15N values (around +15‰ and higher) for suspended organic matter have regularly been observed in estuaries (Mariotti et al., 1984; Owens, 1987; Cifuentes et al., 1988; Middelburg and Nieuwenhuize, 1998). A third hypothesis (3, Figure 7.7) predicts that the sockeye were feeding in lakes shortly before ocean-entry. The dissolved inorganic carbon in lakes can become depleted in

13

C due to the input and reuse of carbon derived from decaying and respired organic

matter (Rau, 1978; Fry and Sherr, 1984). As a consequence phytoplankton in lakes can reach δ13C values as low as –35 to -45‰ (Rau, 1978, 1980; Fry and Sherr, 1984 and references therein). Further constraints can be placed on the origin of the conspicuous group of juvenile sockeye salmon using the fatty acid composition of the muscle tissue samples.

242 16

Marine end-member

15

Off-shelf end-member ?

14

15

δ N (‰)

13

"Conspicuous sockeye"

12 11 10 9 8 7

Freshwater end-member -34

-32

-30

1 -28

-26

-24

-22

-20

-18

-16

δ C (‰) 13

16

Marine end-member

Estuarine end-member ?

15 14

15

δ N (‰)

13

"Conspicuous sockeye"

12 11 10 9 8 7

Freshwater end-member -34

-32

-30

2 -28

-26

-24

-22

-20

-18

-16

δ C (‰) 13

16

14

15

δ N (‰)

13 12 11

Marine end-member

Chum 1998 Coho 1998 Sockeye 1998 Chinook 1998 Sockeye 1999 Chinook 1999

15

Lake end-member ?

"Conspicuous sockeye"

10 9 8 7

Freshwater end-member -34

-32

-30

3 -28

-26

-24

-22

-20

-18

-16

δ C (‰) 13

Figure 7.7. The change in stable carbon, and nitrogen isotope composition of the juvenile salmon over time as proposed by three hypotheses. The black arrows show the direction of the change over time. The circles represent isotopic end-members. See text for explanation of the three hypotheses.

243

7.4.4 Fatty acid signatures To gain further insight into the feeding history of the various groups of salmon, principal component analysis (PCA) was performed on the fatty acid composition data of all salmon (May 1998, May and June 1999). As can be observed in Figure 7.8, two groups of data points plot away from the majority. One group consists of salmon with more monounsaturated fatty acids such as 16:1n-7, 18:1n-9, 20:1 (all isomers) and 22:1n-11, and also the C14 saturated fatty acid. These fatty acids are known to occur in relatively high amounts in wax esters of copepods (Sargent and Falk-Petersen, 1988; Albers et al., 1996). The group of salmon that was found to be enriched in these “copepod markers” were all chinook and had relatively large body sizes (fork length >285mm; cf. Figure 7.5). However, there were also chinook salmon with fork lengths higher than 285mm that did not belong to this group. The other group of juvenile salmon that stands out in the PCA score-plot (Figure 7.8) are the salmon, mostly sockeye, that were found to be relatively depleted in

13

C

(δ13C < -23‰). This group includes also the salmon with low δ15N values (see Figure 7.4). What distinguishes the fatty acid composition of these salmon from the others is the higher levels of n-6 fatty acids, more 18:3n-3, 18:4n-3 and 20:4n-3 (Figure 7.8). Ozawa et al. (1993) compared the fatty acid compositions of marine- with land-locked sockeye salmon. They found that land-locked sockeye were enriched in the 18:2 (the isomer was not stated), 18:3n-3, 18:4n-3, 20:4n-3 and 20:4n-6 fatty acids. Also, it is known that freshwater organisms typically have a higher abundance of n-6 relative to n-3 fatty acids (Ackman, 1967; Sargent, 1976; Henderson and Tocher, 1987; Smith et al., 1996). A negative correlation between the n-6/n-3 fatty acid ratio and the δ13C is presented in Figure 7.9. This seems to confirm the adjustment from high n-6/n-3 ratios and low δ13C values, typical for freshwater organisms, to the low n-6/n-3 ratios and higher δ13C, more common for organisms from the marine environment. PCA was also performed on abundance data of 20 fatty acids from sockeye salmon alone to remove any species effect (Figure 7.10). Again, n-6 fatty acids, C18-PUFAs and 20:4n-3 were found to be higher in the salmon with relatively low δ13C values (< -23‰). These results confirm the similarity with the fatty acid composition of land-locked

244

0.4

14:0 16:1n-7

PC-2 (15.2% of variance)

0.3

18:1n-9 22:1n-11 16:4n-1

16:3n-4

0.2 21:5n-3

0.1

15:0 18:2n-4

18:4n-3 16:1n-5

0.0 18:2n-6

19:0

18:3n-3

-0.1

17:0 20:2n-6

20:1n-11 20:1n-9 20:1n-7 18:1n-5 18:1n-7

16:2n-4 20:0

20:4n-3 20:3n-3

18:0 16:0

20:4n-6

22:5n-6

20:5n-3

-0.2

22:6n-3

22:5n-3

-0.3 -0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

PC-1 (24.9% of variance) 12

Sockeye δ C < 23‰ 13 Chinook δ C < 23‰ Sockeye δ13C > 23‰* 13 Coho δ C > 23‰* 13 Chum δ C > 23‰* 13 Chinook δ C > 23‰* 13

PC-2 (15.2% of variance)

10 8 6 4 2 0

δ13C ?

-2 -4 -8

-6

-4

-2

0

2

4

PC-1 (24.9% of variance)

Figure 7.8. Loading plot (top) and score plot (bottom) of principal component analysis on the fatty acid abundance data from juvenile salmon from all three cruises. *The salmon that are indicated as having δ13C values higher than -23‰ also include salmon of which the δ13C was not measured, see, e.g., the sockeye marked with the arrow.

sockeye, as described by Ozawa et al. (1993). Therefore, the 13C-depleted sockeye seem to have recently migrated from a freshwater environment. Hence, the hypothesis that suggests recent feeding off the shelf and returning to coastal waters can be discounted.

245

0.55

n-6/n-3 fatty acid ratio

0.50

0.15

0.10

0.05

0.00 -30

-28

-26

-24

-22

-20

-18

-16

-14

13

δ C (‰)

Figure 7.9. The ratio of the abundance of n-6 over n-3 fatty acids in muscle tissue of juvenile salmon plot against the δ13C of the same tissue.

7.4.5 Origin of conspicuous sockeye The δ13C and δ15N values for the conspicuous group of sockeye form a fairly tight cluster in a scatter plot (Figure 7.4). Hence, there is reason to believe that these juvenile sockeye originated from the same freshwater region. The positive correlation between size and δ13C (Figure 7.11) seems to confirm a common origin, but suggests that differences in the amount of body mass acquired in the ocean exist. These variations in the proportion of carbon derived from marine sources can perhaps be explained by differences in the timing of ocean-entry. No correlation was found between the δ15N and body size. This observation may suggest that upon ocean-entry the δ15N of the individuals was already close to the +13‰ value measured.

Matthews and Mazumder (personal comm., 2003) found that the δ13C of zooplankton from 12 lakes on Vancouver Island varied from -29 to -35‰. Since sockeye typically rear in lakes for a year or more before entering the ocean, this may cause a

246 stronger depletion in

13

C relative to other salmon when leaving the freshwater

environment.

PC-2 (15.4% of variance)

2

1

0

-1

-2

Sockeye δ C < -23‰ Sockeye δ13C > -23‰ Sockeye δ13C n.a. 13

-5.5 -6.0

-8

-6

-4

-2

0

2

4

6

PC-1 (39.9% of variance) 0.6 18:0

PC-2 (15.4% of variance)

0.4

20:4n-6

0.2

16:0

22:5n-6

20:2n-6

17:0

18:2n-6 20:4n-3 16:2n-4 18:1n-9 0.0 18:3n-3 22:5n-3 20:5n-3 18:4n-3 21:5n-3 -0.2 16:1n-7

-0.4 -0.6 -0.4

14:0

-0.3

18:1n-7 22:6n-3

20:1

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

PC-1 (39.9% of variance)

Figure 7.10. Loading plot (top) and score plot (bottom) of principal component analysis on the fatty acid abundance data of sockeye salmon from all three cruises. The salmon that are indicated with an open circle had bulk muscle tissue δ13C values lower (i.e., more depleted in 13C) than -23‰; closed circles represent salmon displaying δ13C values higher than -23‰, and the triangles represent salmon for which the δ13C was not measured.

247 Next to the Fraser River and the Columbia River, Barkley Sound is an important source for juvenile sockeye entering the ocean at the west coast of Vancouver Island (Hyatt and Steer, 1987). Barkley Sound may, in fact, provide food items that are both more depleted in

13

C with respect to marine prey, and enriched in

15

N relative to

freshwater organisms. Input from terrestrial sources will supply carbon depleted in

13

C

(δ13C of about -28‰; Peterson and Fry, 1987) especially around the edges of Barkley Sound. Additionally, intrusions with marine organic matter, nitrate and ammonium provide a source for nitrogen more enriched in 15N. Hence, the Barkley Sound area as an origin for the conspicuous group of sockeye, together with their rearing in lakes prior to their residence in Barkley Sound, may offer an explanation that combines two of the hypotheses mentioned in section 7.4.3.

42 165

40 38 36

155

Weight (grams)

Fork length (mm)

160

150 145 140

34 32 30 28 26 24

135 130 -26.5

22 -26.0

-25.5 13

-25.0

-24.5

-24.0

δ C of muscle tissue (‰)

-23.5

-23.0

-26.5

-26.0

-25.5

-25.0

-24.5

-24.0

-23.5

-23.0

13

δ C of muscle tissue (‰)

Figure 7.11. The δ13C of muscle tissue of the “conspicuous sockeye salmon” (see Fig. 7.4) from the May 1998 plot against their fork length (left) and weight (right).

Wood et al. (1993) studied juvenile sockeye as they emigrated from the Great Central Lake and Sproat Lake and made their way through Alberni Inlet to Barkley Sound (see map, Figure 7.1). They found that in 1989 the downstream migration of the sockeye smolts began in mid-April, reached a peak at the beginning of May, and then declined gradually through May and June. The average weight reported by Wood et al. (1993) for the smolts leaving the lakes was 2.9 grams (average fork length of 71mm).

248 Hyatt et al. (1990) report a range of mean weights of 3.7 to 6.8 g for Great Central Lake smolts, and 3.5 to 4.7 g for Sproat Lake smolts. The average fork lengths of the sockeye reaching Barkley Sound was found to be about 90 mm (Wood et al., 1993), which corresponds to a weight of about 6-7 grams (inferred from weight – length data from salmon collected during cruise R98-8155, used in this study). At the Barkley Sound site closest to the open ocean Wood et al. (1993) measured an average fork length of 122 mm, corresponding to an inferred weight of about 18-19 grams. Therefore, the majority of the body mass acquired before entering the open ocean is accumulated in Barkley Sound. Two juvenile sockeye with weights of 23.1 and 26.2 grams, and δ13C values of – 26.1 and –26.0‰, respectively, were caught just offshore of Barkley Sound (tow 4, Figure 7.1). Given their weight and their location, it can be hypothesized that these salmon left Barkley Sound not too long ago. As mentioned earlier, the juvenile sockeye with distinctive stable isotope compositions (δ13C < -23‰, δ15N ≈ +13‰) were found all along the west coast of Vancouver Island. Close to the north tip of Vancouver Island (Figure 7.1), during tows 74 and 76, sockeye with a δ13C of –25.3 and –23.8‰, respectively, were collected. If indeed they originated from Barkley Sound they would have to have travelled at least 450 km along the coast. Assuming a starting δ13C of -27‰ at Barkley Sound, a marine diet with a δ13C of -18‰ (Figures 7.3 and 7.6), and taking an average daily growth rate of 1.5% body weight per day (Heintz et al., 2000; model by Perry et al., 1996), an average travelling speed can be calculated. By using the following mass-balance equation the fraction of carbon derived from the marine diet can be calculated: δ 13C = f mar ⋅ δ 13Cmar + (1 − f mar ) ⋅ δ 13C BS

(1)

where δ13C is the measured δ13C of the muscle tissue, δ13Cmar is the δ13C of the marine diet (-18‰), δ13CBS is the δ13C of the salmon muscle tissue before leaving Barkley Sound (-27‰), and fmar is the fraction of the carbon that had been assimilated from its marine diet (with a maximum of 1). When it is assumed that fmar is also the fraction of the body mass accumulated in the open marine environment, then at a growth rate of 1.5% per day, the following should hold true:

249 (1.015)t

=

1 (1 − f mar )

(2)

where t is the number of days spent in the marine environment after leaving Barkley Sound. Using these equations, the sockeye from tow 74 (with a δ13C of –25.3‰), is estimated to have left Barkley Sound 14 days before capture, and the sockeye from tow 76 (δ13C = –23.8‰) took 29 days according to these calculations. This would mean that these salmon travelled on average at least 1.3 km/h and 0.6 km/h, or 37 cm/s and 18 cm/s, respectively. The Vancouver Island Coastal Current has a poleward flow with typical speeds in excess of 10 cm/s (Thomson et al., 1989). Therefore, the swimming speed of the juvenile sockeye would be estimated at around 2 and 0.6 body lengths a second. This is less than the physiological maximum sustained swimming speed for sockeye salmon this size (Brett, 1965), and lower than the 1 km/h estimated swimming velocity during freshwater downstream migration observed by Wood et al. (1993). The above calculations are very sensitive to the δ13C estimated for juveniles leaving Barkley Sound. Also, higher growth rates than 1.5% body weight per day have been observed for juvenile salmon (LeBrasseur and Parker, 1964). Such adjustments could nudge the estimations of the swimming speed to implausible speeds. Additionally, the calculation does not take catabolic loss off carbon into account. However, due to the fast growth of the juvenile salmon, the effect of catabolic loss will only result in a marginal adjustment to a slightly higher estimated swimming speed (e.g., Fry and Arnold, 1982). The above calculations are meant to illustrate that a Barkley Sound origin is plausible. Indeed, with all data considered so far, a Barkley Sound origin for the conspicuous juvenile sockeye seems to be the most viable hypothesis.

Rankin and Hyatt (Pacific Biological Station, Dept. of Fisheries and Oceans, Nanaimo, B.C., Canada; K. Hyatt, personal comm.) studied the outmigration timing of sockeye juveniles at Great Central Lake and Sproat Lake in 1998 and 1999. At Great Central Lake the peak of the outmigration took place around April 30th in 1998 and at May 11th in 1999. The juveniles caught in 1998 for the presented study were collected between May 18th and 28th, and at May 18th in 1999. The sockeye embark on their

250 coastal migration route about 3 weeks after the peak of their migrations from their nursery lakes (K. Hyatt, personal comm.). When about 3 weeks are added to the outmigration peak times (April 30, 1998 and May 11, 1999), it can be concluded that in 1998 the sample collection took place while a large proportion of the sockeye smolts from the Great Central Lake were entering the ocean. Whereas in 1999 the sample collection took place before the majority of the sockeye were leaving Barkley Sound. This difference in match between the outmigration timing and sampling between the two years seems to be in accordance with the fact that more sockeye were caught in 1998. Furthermore, the timing difference perhaps explains the fact that the few sockeye that were collected in May 1999 do not exhibit the isotope signature ascribed to the Barkley Sound system.

A question that remains is where the other sockeye, especially those with the low δ15N values (δ15N < +12‰; Figure 7.4), originated from. As stated before, the Fraser River, the Columbia River and Barkley Sound are the important sources for juvenile sockeye found on the west coast of British Columbia. At the time of sample collection, the Columbia River sockeye are not expected to have reached the locations of sampling yet (K. Hyatt, personal comm.). Some evidence suggests that most Fraser River sockeye migrate north through the Georgia Strait rather than up the west coast of Vancouver Island (K. Hyatt, personal comm.). If they did migrate to the west coast of Vancouver Island their timing could have matched with the time of sample collection. However, these salmon would be expected to have a more typical marine stable isotope signature, and perhaps have a larger size due to the amount of distance covered, and time spent in marine waters. Therefore, if no other sources exist, all of the sockeye collected during the May cruises should have originated from the Barkley Sound system. The presented isotope data either suggests heterogeneity within the Barkley Sound system, with some sockeye rearing in an area that contains prey with a different isotope composition, or some of the collected sockeye did not originate from the Barkley Sound area.

251

7.4.6 δ13C difference between bulk and individual fatty acids As discussed in Chapter 2, the fatty acids are expected to turnover faster than the bulk of an organism. Therefore, because the fatty acids are depleted in

13

C relative to the bulk,

after a switch to a 13C- enriched (marine) diet the δ13C difference between the fatty acids and the bulk should become less. Subsequently, after the body composition has fully equilibrated with the new marine diet, the δ13C difference between bulk and fatty acids will be back to “normal” again. For plotting of the bulk-fatty acid δ13C difference the samples were split into four groups: 1) sockeye with a δ13C < -23‰, 2) salmon with both low δ13C and δ15N values (< -23‰ and < +10‰, respectively), 3) salmon with –18‰ > δ13C > -23‰, and 4) salmon with δ13C values > -18‰. The last group should be in approximate equilibrium with its diet, given the fact that their isotope ratios fully reflect marine values. Therefore, this group should be a good reference for the other juveniles that purportedly have switched to marine diets more recently. Figure 7.12 plots the differences between δ13C values measured for the bulk and individual fatty acids (∆δ13C) of muscle tissue. Indeed, the differences in δ13C are largest for the “equilibrated” reference group. In particular the two salmon that were both depleted in

13

C and

15

N showed lower fatty acid-bulk δ13C differences. The two other

groups do not always show significant differences (P<0.05) from the reference group. Therefore, for juvenile salmon the applicability of bulk-fatty acid ∆δ13C deviations from the reference pattern as a diet switch indicator may be limited.

7.4.7 Distinguishing species with their fatty acid signature It was attempted to distinguish between the various salmon species on the basis of the fatty acid composition of their muscle tissue. For the discriminant analysis, the 18 most abundant fatty acids and data for the sockeye, chinook and chum salmon were used. The coho were left out, but as a test they were subsequently plotted using the discriminant functions derived from the analysis using only the other three species. As can be seen in Figure 7.13 all four salmon species are, mostly, well separated. Only a few salmon plot in

252 the wrong species cluster. The fact that the coho, not used for the discriminant analysis, plotted as a separate group shows that classification of juvenile salmon using fatty acid data can be a viable method. Mjaavatten et al. (1998) showed success in distinguishing between species, organs and wild versus hatchery juvenile salmon using multivariate statistical techniques. Hence, when differences remain during adult life, multivariate analysis of fatty acid abundance data can potentially be useful for food inspection purposes.

-2 -4

-8 -10

13

∆δ CFA - Bulk (‰)

-6

-12 13

22:6n-3

22:5n-3

20:5n-3

20:4n-3

20:4n-6

18:4n-3

18:3n-3

18:2n-6

18:1n-7

18:1n-9

18:0

17:0

16:1n-7

14:0

-16

16:0

J. sockeye δ C < -23‰ (n=11) 15 J. salmon δ N < +10‰ (n=2) 13 J. salmon -18‰ > δ C > -23‰ (n=15) 13 J. salmon δ C > -18‰ (n=13)

-14

Fatty acid Figure 7.12. Differences between the δ13C of fatty acids and the bulk of muscle tissue of juvenile salmon. The various salmon were divided in groups with different stable isotope signatures (see legend), and the averages of each group (± 1 S.D.) were plotted.

As shown by PCA (Figure 7.8), factors other than differences in species have a larger impact on the fatty acid composition of the salmon. Nevertheless, the discriminant analysis shows that classifying the species with the use of their fatty profile is possible. The differences in fatty acid composition between the various species may be the result

Coeff. of standardized discriminant function 2

253 1.0 0.8

18:1n-9

22:6n-3

0.6

20:5n-3

0.4

20:1

18:2n-6

16:0

16:1n-7

0.2

18:4n-3

22:5n-6 18:0

0.0

21:5n-3

20:4n-3

16:2n-4 22:5n-3

-0.2

-0.8

18:1n-7

18:3n-3

-0.4 -0.6

-0.4

20:4n-6 14:0

-0.2

0.0

0.2

0.4

Coefficients of standardized discriminant function 1 3.6

Discriminant score 2

3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 -2.5

Sockeye Chinook Chum Coho -2.0

-1.5

-1.0

-0.5

0.0

Discriminant score 1

Figure 7.13. Results of discriminant analysis on fatty acid abundance data of muscle tissue of sockeye, chinook and coho salmon from all three cruises. As a test for robustness of the distinction made between the various species, the coho were plotted using the same discriminant function. The top plot shows the standardized coefficients for each fatty acid used for both discriminant functions. These coefficients are a measure of their relative contribution to the group separation (see methods section).

254 of different dietary preferences. On the other hand, some of the differences can also be due to genetic control on the fatty acid profiles (Pickova et al., 1997; Joensen et al., 2000). From the standardized coefficients of the discriminant functions it is not very clear what the differences in dietary preference would be. It can be speculated that chinook had a slightly larger proportion of their diet consisting of copepods, as indicated by somewhat higher levels of the 20:1 fatty acids. However, it should also be borne in mind that not all salmon were of the same age. Therefore, it may be that other salmon, if they become as large as the chinook (see Fig. 7.5), will also acquire more of the 20:1 fatty acids. For example, Iverson et al. (2002) found that adult pink salmon (Oncorhynchus gorbuscha) had higher levels of the 20:1 and 22:1 fatty acids than the smolts. This may be because the bigger salmon eat more copepods. However, it is also important to note that longchain fatty acids with a low degree of unsaturation, such as the 20:1 and 22:1, tend to be spared most from mobilization from triacylglycerols (Raclot and Groscolas, 1993, 1995; Conner et al., 1996; Raclot, 1997; Perona et al., 2000). Hence, because of lower losses due to catabolism for the 20:1 and 22:1 fatty acids, they may get more concentrated with age. Indeed, Iverson et al. (2002) found an increase in the abundance of these fatty acids with body size for all the fish in their study.

7.5 Conclusions Bulk muscle tissue from juvenile salmon showed a positive correlation between the δ13C and δ15N values, and the negative correlation between δ13C and n-6/n-3 fatty acid ratios. These observations are in accordance with the adjustment from freshwater to marine compositions. Bulk muscle tissue from juvenile salmon sampled at the same time of year had a wider range of δ13C and δ15N values (-28 to -16‰, +8 to +16‰, respectively) in May 1998 than in May 1999 (range: -22 to -16‰, 10 to 14‰, respectively). This was largely due to a group of juvenile sockeye salmon sampled in May 1998 that had low δ13C values (δ13C < -23‰), suggesting a recent entry into the ocean-entry by this group. However, their 15N/14N ratios (δ15N of ca. +13‰) indicate that a considerable amount of time had been spent in the ocean. It was argued that these salmon most likely reared in

255 lakes and subsequently fed in Barkley Sound to obtain the stable isotope signatures observed. The salmon that putatively had resided in the ocean the longest showed the largest differences between δ13C values of the bulk and individual fatty acids. Faster turnover rates for fatty acids than bulk tissue caused the salmon that made the switch to a marine diet more recently to have smaller fatty acid-bulk δ13C differences. The deviations observed from the bulk-fatty acid ∆δ13C reference pattern may not be large enough to serve as a general indicator for recent diet switch. It was shown that with the use of discriminant analysis on fatty acid abundance data the four analyzed species of salmon can be distinguished. This may potentially be useful for food inspection purposes.

7.6 References Ackman, R.G. 1967. Characteristics of the fatty acid composition and biochemistry of some freshwater fish oils and lipids in comparison with marine oils and lipids. Comparative Biochemistry and Physiology 22, 907–922. Albers, C.S., Kattner, G. and Hagen, W. 1996. The compositions of wax esters, triacylglycerols and phospholipids in Arctic and Antarctic copepods: Evidence of energetic adaptations. Marine Chemistry 55, 347-358. Brett, J.R. 1965. The relation of size to rate of oxygen consumption and sustained swimming speed of sockeye salmon (Oncorhynchus nerka). Journal of the Fisheries Research Board of Canada 22, 14911501. Chisholm, B.S., Nelson, D.E. and Schwarz, H.P. 1982. Stable-carbon isotope ratios as a measure of marine versus terrestrial protein in ancient diets. Science 216, 1131-1132. Cifuentes, L.A., Sharp, J.H. and Fogel, M.L. 1988. Stable carbon and nitrogen isotope biogeochemistry in the Delaware estuary. Limnology and Oceanography 33, 1102-1115. Connor, W.E., Lin, D.S. and Colvis, C. 1996. Differential mobilization of fatty acids from adipose tissue. Journal of Lipid Research 37, 290-298.

256 Fry, B. and Arnold, C. 1982. Rapid Oecologia 54, 200-204.

13

C/12C turnover during growth of brown shrimp (Penaeus aztecus).

Fry B. and Sherr, E.B. 1984. δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contributions in Marine Science 27, 13-47. Groot, C. and Margolis, L. (eds.). 1991. Pacific salmon life histories. UBC Press, Vancouver, B.C., Canada, pp. 564. Heintz, R.A., Rice, S.D., Wertheimer, A.C., Bradshaw, R.F., Thrower, F.P., Joyce, J.E. and Short, J.W. 2000. Delayed effects on growth and marine survival of pink salmon Oncorhynchus gorbuscha after exposure to crude oil during embryonic development. Marine Ecology Progress Series 208, 205-216. Henderson, R.J. and Tocher, D.R. 1987. The lipid composition and biochemistry of freshwater fish. Progress in Lipid Research 26, 281-347. Hyatt, K.D., and G.J. Steer. 1987. Barkley Sound sockeye salmon, (Oncorhynchus nerka): Evidence for over a century of successful stock development, fisheries management, research, and enhancement efforts. In: H.D. Smith, L. Margolis, and C.C.Wood (eds.), Sockeye salmon (Oncorhynchus nerka) Population Biology and Future Management. Canadian Special Publication of Fisheries and Aquatic Sciences 96, p. 435-457. Hyatt, K.D., Wright, M., Rankin, P., Miki, I. and Kolody, D. 1990. Sockeye salmon recruitment variations. In: T.D. Beacham (ed.), The marine survival of salmon program. Annual Progress Report, 1990. Department of Fisheries and Oceans, Pacific Biological Station, Nanaimo, B.C., Canada, p.13-25. Hyatt, K.D., Pacific biological Station, Department of Fisheries and Oceans, Nanaimo, B.C., Canada, V9T 6N7. Personal communication, April 2003. Iverson, S.J., Frost, K.J. and Lang, S.L.C. 2002. Fat content and fatty acid composition of forage fish and invertebrates in Prince William Sound, Alaska: factors contributing to among and within species variability. Marine Ecology Progress Series 241, 161-181. Joensen, H., Steingrund, P., Fjallstein, I. and Grahl-Nielsen, O. 2000. Discrimination between two reared stocks of cod (Gadus morhua) from the Faroe Islands by chemometry of the fatty acid composition in the heart tissue. Marine Biology 136, 573-580. LeBrasseur, R.J. and Parker, R.R. 1964. Growth rates of central British Columbia pink salmon. Journal of the Fisheries Research Board of Canada 21, 1101-1127. Mariotti, A., Lancelot, C. and Billen, G. 1984. Natural isotopic composition of nitrogen as a tracer of origin for suspended matter in the Scheldt estuary. Geochimica Cosmochimica Acta 48, 549-555.

257 Matthews, B. and Mazumder, A. 2003. Compositional and inter-lake variability of zooplankton affect baseline stable isotope signatures. Submitted to Limnology and Oceanography. Meglen, R.R. 1992. Examining large databases: a chemometric approach using principal component analysis. Marine Chemistry 39, 217-237. Michener, R.H. and Schell, D.M. 1994. Stable isotope ratios as tracers in marine aquatic food webs. In: K. Lajtha and R. H. Michener (eds.), Stable isotopes in ecology and environmental science. Blackwell Scientific Publications, Oxford, U.K., 138-157. Middelburg, J.J. and Nieuwenhuize, J. 1998. Carbon and nitrogen stable isotopes in suspended matter and sediments from the Schelde Estuary. Marine Chemistry 60, 217-225. Mjaavatten, O., Levings, C.D. and Poon, P. 1998. Variation in the fatty acid composition of juvenile chinook and coho salmon from Fraser river estuary determined by multivariate analysis; role of environment, and genetic origin. Comparative Biochemistry and Physiology 120B, 291-309. Owens, N.J.P. 1987. Natural variations in 15N in the marine environment. Advances in Marine Biology 24, 389-451. Ozawa, A., Satake, M. and Fujita, T. 1993. Comparison of muscle lipid composition between marine and landlocked forms of sockeye salmon (Oncorhynchus nerka). Comparative Biochemistry and Physiology 106B, 513-516. Perona, J.S., Portillo, M.P., Macarulla, M.T., Tueros, A.I., Ruiz-Gutiérrez V. 2000. Influence of different dietary fats on triacylglycerol deposition in rat adipose tissue. British Journal of Nutrition 84, 765-774. Perry, R.I., Hargreaves, N.B., Waddell, B.J. and Mackas, D.L. 1996. Spatial variations in feeding and condition of juvenile pink and chum salmon off Vancouver Island, British Columbia. Fisheries Oceanography 5, 73-88. Peterson B.J. and Fry B. 1987. Stable isotopes in ecosystem studies. Annual review of ecology and systematics 18, 293-320. Pickova, J., Dutta, P.C., Larsson, P.O. and Kiessling, A. 1997. Early embryonic cleavage pattern, hatching success, and egg-lipid fatty acid composition: comparison between two cod (Gadus morhua) stocks. Canadian Journal of Fisheries and Aquatic Sciences 54, 2410-2416. Raclot, T. and Groscolas, R. 1993. Differential mobilization of white adipose tissue fatty acids according to chain length, unsaturation, and positional isomerism. Journal of Lipid Research 34, 1515-1526. Raclot, T. and Groscolas, R. 1995. Selective mobilization of adipose tissue fatty acids during energy depletion in the rat. Journal of Lipid Research 36, 2164-2173.

258

Raclot, T. 1997. Selective mobilization of fatty acids from white fat cells: Evidence for a relationship to the polarity of triacylglycerols. Biochemical Journal 322, 483-489. Rau, G.H. 1978. Carbon-13 depletion in a subalpine lake: carbon flow implications. Science 201, 901-902. Rau, G.H. 1980. Carbon-13/carbon-12 variation in subalpine lake aquatic insects: food source implications. Canadian Journal of Fisheries and Aquatic Sciences 37, 742-746. Rencher, A.C. 1995. Methods of Multivariate Analysis. Wiley, New York, pp. 627. Sargent, J. R. 1976. The structure, metabolism and function of lipids in marine organisms. In: D. C. Malins and J.R. Sargent (eds.), Biochemical and biophysical perspectives in marine biology. Volume 3. Academic Press Inc. London, U.K., p.149-212. Sargent, J. R. and Falk-Petersen, S. 1988. The lipid biochemistry of calanoid copepods. Hydrobiologia 167/168, 101-114. Schoeninger, M.J., DeNiro, M.J. and Tauber, H. 1983. Stable nitrogen isotope ratios of bone collagen reflect marine and terrestrial components of prehistoric human diet. Science 220, 1381-1383. Sedgwick, S.D. 1982. The Salmon Handbook. The life and cultivation of fishes of the salmon family. Robert Hartnoll Ltd., Cornwall, Great Britain, pp. 242. Smith, R.J., Hobson, K.A., Koopman, H.N. and Lavigne, D.M. 1996. Distinguishing between populations of fresh- and salt-water harbour seals (Phoca vitulina) using stable-isotope ratios and fatty acid profiles. Canadian Journal of Fisheries and Aquatic Sciences 53, 272-279. Thomson, R.E., Hickey, B.M. and Leblond, P.H. 1989. The Vancouver Island coastal current: fisheries barrier and conduit. In: R.J. Beamish and G.A. McFarlane (eds.) Effects of ocean variability on recruitment and an evaluation of parameters used in stock assessment models. Canadian Special Publication of Fisheries and Aquatic Sciences 108, p. 265-296. Wood, C.C., Hargreaves, N.B., Rutherford, D.T. and Emmett, B.T. 1993. Downstream and early marine migratory behaviour of sockeye salmon (Oncorhynchus nerka) smolts entering Barkley Sound, Vancouver Island. Canadian Journal of Fisheries and Aquatic Sciences 50, 1329-1337.

259

8. Conclusions

8.1 Fatty acid composition Particulate organic matter (POM), zooplankton and larval fish from the continental shelf environment could typically be distinguished from samples collected off the shelf with the use of the fatty acid composition data. The differences between the adjoining food webs seem to be mostly driven by the different contributions of diatom-derived material to seston at the base of the food webs. When using a variable selection technique and subsequent discriminant analysis it was possible to distinguish between the organisms from the two regions. The largest part of the variation in the fatty acid compositions of all analyzed organisms, however, is caused by differences between the trophic groups (POM, zooplankton, larval fish). It was found that the proportion of C20 and C22 polyunsaturated fatty acids increases when increasing in trophic level from POM up to juvenile salmon (muscle). It was shown that with the use of discriminant analysis on fatty acid abundance data the four analyzed species of Pacific salmon can be distinguished.

8.2 Shelf - off shelf δ13C difference in POM The δ13C of POM collected in shelf waters was typically enriched in 13C with respect to waters further away from the coast. In May 1998 the difference in the δ13C of bulk POM was on average almost 5‰ (with a maximum of 9‰), but in May 1999 the differences were much smaller at an average of about 2‰ (maximum almost 6‰). With the aid of the δ13C of individual fatty acids it was concluded that the higher δ13C values of POM from the shelf waters could not be explained by a higher contribution of 13C-enriched diatom-derived carbon. Instead, it is suggested that diatoms seem to thrive better in “13C-enriching conditions”, which cause other algae also to be

260 enriched in

13

C. Important factors in creating such conditions are believed to be a high

daily average photon flux density, a high nutrient supply rate, but not necessarily high concentration. The average cell size and growth rate of phytoplankton are generally higher in coastal environments. Also, carbon concentrating mechanisms may be more employed by algae in this environment due to the higher nutrient flux and lower CO2(aq) concentration during times of high productivity. For these reasons a difference in δ13C of organisms feeding close to shore and individuals feeding further off shore are expected to be a common phenomenon.

8.3 δ13C of individual fatty acids The δ13C of essential fatty acids in zooplankton and larval fish was not found to be more similar to those in POM from the same location than the δ13C of other fatty acids. For example, the δ13C of the non-essential 14:0 fatty acid often showed closer agreement with its counterpart in POM. A large range (typically about 7‰) in the δ13C values of fatty acids is observed within single samples. The variation in δ13C among the individual fatty acids was found to be reproducible, not only in samples measured for this research, but also when comparing values found in other studies. This suggests the presence of common underlying mechanisms, most likely biosynthetic effects, producing the semi-predictable offsets between the δ13C of fatty acids. The factors identified here as having potentially the largest impact on the δ13C seem to be desaturation, different timing of lipid class synthesis during the growth cycle of autotrophs, and perhaps also the proportion of PUFAs synthesized via an alternative (polyketide synthase -catalyzed) pathway. However, more research is needed to fully explain the observed differences in δ13C among the various fatty acids.

261

8.4 Application and evaluation The δ13C of individual fatty acids in POM, zooplankton and larval fish could typically be used to distinguish organisms collected in shelf, and off-shelf waters. δ13C values of the bulk and/or the fatty acid composition are similarly succesful at making this distinction. It was found that particularly the combination of fatty acid composition data and stable carbon isotope data (either bulk or compound specific) can be very effective in movement studies and in determining the dietary history of an organism. In most ecological studies, bulk δ13C measurements may be as successful in solving the particular research problem as through employing the δ13C of individual fatty acids. However, in some cases δ13C values of individual fatty acids can have a great advantage over bulk isotope data. In particular situations where it is not possible to physically isolate the organisms to be investigated. Additionally, when fatty acids exhibit different turnover rates, an unusual difference between the δ13C of these fatty acids can be an indication for recent diet shift. The difference in δ13C between the bulk (or protein) and fatty acids may also be an indicator for a recent shift to a new diet. When the various turnover rates are well constrained an estimate of the timing of the diet switch may even be possible.

262

9 Recommendations and outlook

9.1 Polyketide Synthase pathway In order to explain the δ13C of some of the polyunsaturated fatty acids (PUFAs) in particulate organic matter it was suggested that the polyketide synthase (PKS) pathway for the synthesis of PUFAs is perhaps a common pathway in marine algae. Obviously, this needs to be tested. Berry et al. (2002) reported that there is some indication that PKS-encoding genes, involved in the biosynthesis of PUFAs, exist in the dinoflagellate Pfiesteria shumwayae. The genomics analysis of genes encoding enzymes of PUFA synthesis via the PKS pathway have been reported by Metz et al. (2001). The genomic sequence prepared from the marine bacterium Shewanella, Moritella marina is publicly available (Metz et al., 2001). Therefore, it would probably not be very difficult to test for the existence of the respective PKS-encoding genes in a wide range of marine algae. Combining a genomics study and measuring the δ13C of the fatty acids from the same algae could be very instructive. When indeed the PUFAs produced via the PKS pathway exhibit different δ13C signatures, the δ13C of fatty acids could be an indicator for whether or not a PKS is used in the synthesis of PUFAs by certain organisms.

9.2 n-3 PUFAs and human health A myriad of reports exist on the health benefits of n-3 PUFAs (Connor, 2000; Arts et al., 2001). These studies consist of epidemiological- and population studies, as well as clinical trials using supplementation of n-3 PUFAs. For example, a higher intake of n-3 PUFAs has been linked to a decrease in the mortality and morbidity of cardiovascular disease (Holub, 1988; Simon et al., 1995; Daviglus et al., 1997; Leaf, 1999; Harris, 2003), a reduction in thrombogenicity (Holub, 1988; Harris, 1997), reduced incidence

263 and/or severity of many cancers (Stillwell and Jenski, 2002), and may beneficial against inflammatory diseases, and perhaps even behavioural disorders (Connor, 2000). Some researchers have expressed their concerns about the future availability of n-3 PUFAs for human nutrition (Arts et al., 2001; Crawford, 2002). Marine fish and shellfish are the primary source of these fatty acids (Thomas and Holub, 1994). The world’s fish stocks have been in decline (Williams, 1998) and at the same time the human population is growing rapidly, with an expected increase of about 40% within the next 50 years (Lutz et al., 2001, and references therein). Hence, it is necessary to investigate the ability to acquire large amounts of n-3 PUFAs without depleting the global fish stocks. Batch cultures of the marine protist Schizochytrium, which uses the PKS-pathway for PUFA synthesis, are already used for the production of n-3 PUFA supplements (OmegaTech Inc., Boulder, CO, U.S.A.). Perhaps, more research investigating the possibility of inserting the respective PKS-encoding genes into easily cultured organisms can be rewarding. Additionally, nationally and globally a higher priority should be given to counter the threat of depleting the world’s fish stocks.

9.3 Evolutionary role and function of docosahexaenoic acid Particularly phospholipids containing docosahexaenoic acid (DHA; 22:6n-3) seem to play a specific structural cofactor role in ion pumps, receptors and other membrane proteins (Infante et al., 2001). DHA has been recognized to be a crucial fatty acid for nervous tissues, such as the brain and retina in which it is found in relatively high concentrations (Crawford et al., 1999; Cunnane et al., 2000). Additionally, high contraction frequency muscles were found to have higher DHA concentrations than that of other muscle tissue in the same animal (Infante et al., 2001). A question that could be asked is: Why do certain algae, such as most dinoflagellates, have a high abundance of DHA. Is DHA present in higher amounts in the flagellae? Flagellate algae exhibiting phototactic behaviour often possess a complex structure called the eyespot apparatus (EA). A large variety of eyespots exist, but the most conspicuous components of EAs are carotenoid-rich lipid globules which form the

264 eyespots sensu stricto (Kreimer, 1999). The location of the photoreceptors is thought to be overlying the eyespot or in a flagellar swelling, and the eyespot globules represent a light modifying structure (for reviews see: Dodge, 1973; Foster and Smyth, 1980; Melkonian and Robenek, 1984; Kreimer, 1994). Another question that can be asked is: Is DHA concentrated in the eyespots of algae? Obviously, making measurements of such small structures will be very challenging and may have to await further technological innovations. Perhaps as a first approach the measurements of the relative size of the eyespot apparatus and the flagellae can be compared to the concentration of DHA in samples of cultures from a whole suite of different algae. From an evolutionary point of view it would be interesting to know whether the capability of synthesizing DHA-rich phospholipids was a requirement for the evolution of organs like the brain or the eye. It seems that even missing only one double bond in the acyl group, such as in docosapentaenoic acid (22:5n-3), is highly disadvantageous for the lipid-protein interaction needed in the membranes of the brain (Crawford et al., 1999). The question of what is so special about DHA has not been fully answered yet.

9.4 δ13C of fatty acids in algae As shown in Chapter 4 the variation in δ13C values among the various fatty acids is reproducible, with semi-predictable offsets between the δ13C values of the individual fatty acids (Figure 4.5). This reproducible trend in the variation among the δ13C values of fatty acids is found in different organisms from different trophic levels and from both shelf- and off shelf waters. Furthermore, other researchers observed the same pattern for samples from different regions (Murphy and Abrajano, 1994; Pond et al., 2000). Moreover, work towards a Ph.D. by Alex Bec (Université Blaise Pascal, Clermont Ferrand, France) shows a remarkable similarity between the variation in the δ13C values of fatty acids from cultured algae (Rhodomonas sp. and Cyclidium sp.) and the POM of this study. Additionally, fatty acids from cultured Daphnia showed a similar trend in the δ13C values as found for zooplankton reported here (Bec, 2003).

265 To reveal which processes are responsible for the carbon isotope fractionations during the biosynthesis of the fatty acids more studies on the δ13C of fatty acids in cultured marine algae should be done. It would be instructive to determine the δ13C of fatty acids isolated from algae grown at different, well-controlled conditions. A difference in timing of synthesis during the growth cycle was mentioned in Chapter 4 as one of the possibilities for explaining some of the variation in δ13C values among fatty acids in algae. This can perhaps be tested by growing algae under different light regimes and taking samples for δ13C measurements of the fatty acids at different times during the growth cycle. For such an experiment the culture should have a phased growth cycle, such that the cells divide during each light-dark cycle and divisions take place during a restricted period of the cycle (Chisholm, 1981; Sukenik and Carmeli, 1990). Also, for a better understanding, the polar- and neutral lipids should be separated before measuring the δ13C of the fatty acids.

9.5 Diet switch experiment Several experiments have been performed in which an organism, reared on a diet with a specific

13

C/12C ratio, is switched to a diet with a different stable carbon isotope

composition (Fry and Arnold, 1982; Tieszen et al., 1983; Hobson and Clark, 1992; Hesslein et al., 1993; Frazer et al., 1997; Herzka and Holt, 2000). Some of these investigators measured the δ13C of different tissues of the animal used in the experiment (Tieszen et al., 1983; Hobson and Clark, 1992; Hesslein et al., 1993). Generally, carbon from the various tissues were found to have a different turnover time. Tieszen et al. (1983) found that the half-life for carbon ranged from 6.4 days in the liver and 47.5 days in hair of gerbils. Differences should also exist in the turnover time of carbon in different biochemical constituents (e.g., Roman, 1991). It would be interesting to do a diet switch experiment for which the δ13C of lipids (perhaps separated into a neutral and a polar fraction), proteins and carbohydrates are measured. Such an experiment can be useful for studies examining small animals in highly dynamic systems, such as estuaries, where the

266 interplay between inputs of freshwater and marine derived carbon can complicate the interpretation of stable isotope data.

9.6 Lipid – lipid-free matter δ13C difference as trophic level indicator There seems to be no indication that carbon in lipids gets enriched in

13

C during the

transfer to higher trophic levels (this study; Focken and Becker, 1998; Grice et al., 1998; Klein-Breteler et al., 2002). However, lipid-free matter does show enrichment in

13

C in

higher trophic level organisms (Focken and Becker, 1998; Klein-Breteler et al., 2002). Therefore, the difference in δ13C of lipids and lipid-free matter should increase with trophic level. It would perhaps be worthwhile to investigate whether this difference could be used as a trophic level indicator, analogous to the use of δ15N data. However, caution should be exercised when it is suspected that a recent diet shift has occurred. As explained in the previous section, due to different turnover times the difference in the δ13C of lipids and lipid-free matter will be affected by a diet switch.

9.7 Corroboration of chl a – δ13C relationship Only when enough contrast exists between POM from shelf and off-shelf waters, the fatty acid composition and stable carbon isotope ratio (of either the bulk or fatty acids) can be used to monitor exchanges of organisms between the two environments. In this study some correlation was found between chlorophyll a (chl a) levels and the δ13C of POM. If indeed differences in chl a levels are a good proxy for the δ13C differences between environments, it could be pre-determined with satellite data whether the approach could be successful or not. Further research is needed to find out if indeed satellite chl a concentration estimates can predict the extent of the contrast in the δ13C of POM between two regions.

267

9.8 References Arts, M.T., Ackman, R.G. and Holub, B.J. 2001. "Essential fatty acids" in aquatic ecosystems: a crucial link between diet and human health and evolution. Canadian Journal of Fisheries and Aquatic sciences 58, 122-137. Bec, A. 2003. Ph.D. thesis. Université Blaise Pascal, Clermont Ferrand, France. Berry, J.P., Reece, K.S., Rein, K.S., Baden, D.G., Haas, L.W., Ribeiro, W.L., Shields, J.D., Snyder, R.V., Vogelbein, W.K. and Gawley, R.E. 2002. Are Pfiesteria species toxicogenic? Evidence against production of ichthyotoxins by Pfiesteria shumwayae. Proceedings of the National Academy of Science, U. S. A. 99, 10970-10975. Chisholm, S.W. 1981. Temporal patterns of cell division in unicellular algae. Canadian Bulletin of Fisheries and Aquatic Sciences 210, 150-181. Connor, W.E. 2000. Importance of n-3 fatty acids in health and disease. American Journal of Clinical Nutrition 71, Supplement, 171-175. Crawford, M.A., Bloom, M., Broadhurst, C.L., Schmidt, W.F., Cunnane, S.C., Galli, C., Gehbremeskel, K., Linseisen, F., Lloyd-Smith, J. and Parkington, J. 1999. Evidence for the unique function of docosahexaenoic acid during the evolution of the modern hominid brain. Lipids 34, Supplement, 39-47. Crawford, M.A. 2002. Docosahexaenoic and arachidonic acid acids were essential to human cerebral evolution:implications of aquatic resources as diet for humans. Plenary lecture at American Society of Limnology and Oceanography 2002 Summer Meeting, Victoria, B.C., Canada. Cunnane, S.C., Francescutti, V., Brenna, J.T., Crawford and M.A. 2000. Breasts-fed infants achieve a higher rate of brain and whole body docosahexaenoate accumulation than formula-fed infants not consuming dietary docosahexaenoate. Lipids 35, 105-111. Daviglus, M.L., Stamler, J., Orencia, A.J., Dyer, A.R., Liu, K., Greenland, P., Walsh, M.K., Morris, D. and Shekelle, R.B. 1997. Fish consumption and the 30-year risk of fatal myocardial infarction. New England Journal of Medicine 336, 1046-1053. Dodge, J.D. 1973. The eyespot. In: J.D. Dodge (ed.) The fine structure of algal cells. Academic Press, New York, p. 125-137. Focken, U. and Becker, K. 1998. Metabolic fractionation of stable carbon isotopes: implications of different proximate compositions for studies of the aquatic food webs using δ13C data. Oecologia 115, 337343. Foster, K.W. and Smyth, R.D. 1980. Light antennas in phototactic algae. Microbiological Reviews 44, 572630.

268

Frazer, T.K., Ross, R.M., Quetin, L.B. and Montoya, J.P. 1997. Turnover of carbon and nitrogen during growth of larval krill, Euphausia superba Dana: a stable isotope approach. Journal of Experimental Marine Biology and Ecology 212, 259-275.

Fry, B. and Arnold, C. 1982. Rapid Oecologia 54, 200-204.

13

C/12C turnover during growth of brown shrimp (Penaeus aztecus).

Grice, K., Klein Breteler, W.C.M., Schouten, S., Grossi, V., de Leeuw, J.W., Sinninghe Damsté J.S. 1998. Effects of zooplankton herbivory on biomarker proxy records. Paleoceanography 13, 686-693. Harris, W.S. 1997. n-3 Fatty acids and serum lipoproteins: Human studies. American Journal of Clinical Nutrition 65, Supplement, 1645-1654. Harris, W.S., Park, Y. and Isley, W.L. 2003. Cardiovascular disease and long-chain omega-3 fatty acids. Current Opinion in Lipidology 14, 9-14. Herzka, S.Z. and Holt, G.J. 2000. Changes in isotopic composition of red drum (Sciaenops ocellatus) larvae in response to dietary shifts: potential applications to settlement studies. Canadian Journal of Fisheries and Aquatic Sciences 57, 137-147. Infante, J.P., Kirwan, R.C. and Brenna, J.T. 2001. High levels of docosahexaenoic acid (22:6n-3)containing phospholipids in high-frequency contraction muscles of hummingbirds and rattlesnakes. Comparative Biochemistry and Physiology 130B, 291-298. Hesslein, R.H., Hallard, K.A. and Ramlal, P. 1993. Replacement of sulfur, carbon, and nitrogen in tissue of growing broad whitefish (Coregonus nasus) in response to a change in diet traced by δ34S, δ13C, and δ15N. Canadian Journal of Fisheries and Aquatic Sciences 50, 2071-2076. Hobson, K.A. and Clark, R.G. 1992. Assessing avian diets using stable isotopes I: turnover of tissues. The Condor 94, 181-188.

13

C in

Holub, B.J. 1988. Dietary fish oils containing eicosapentaenoic acid and the prevention of atherosclerosis and thrombosis. Canadian Medical Association Journal 139, 377-381. Klein Breteler, W.C.M., Grice K., Schouten S., Kloosterhuis H.T., Sinninghe Damsté J.S. 2002. Stable carbon isotope fractionation in the marine copepod Temora longicornis: unexpectedly low δ13C value of faecal pellets. Marine Ecology-Progress Series 240, 195-204. Kreimer, G. 1994. Cell biology of phototaxis in flagellate algae. International Review of Cytology 148, 229-310.

269

Kreimer, G. 1999. Reflective properties of different eyespot types in dinoflagellates. Protist 150, 311-323. Leaf, A. 1999. Dietary prevention of coronary heart disease - The Lyon Diet Heart Study. Circulation 99, 733-735. Lutz, W., Sanderson, W. and Scherbov, S. 2001. The end of world population growth. Nature 412, 543545. Melkonian, M. and Robenek, H. 1984. The eyespot apparatus of flagellated green algae: A critical review. In: F.E. Round and D.J. Chapman (eds.) Progress in Phycological Research Volume 3, Biopress, Bristol, U.K., p. 193-268. Metz, J.G., Roessler, P., Facciotti, D., Levering, C., Dittrich, F., Lassner, M., Valentine, R., Lardizabal, K., Domergue, F., Yamada, A., Yazawa, K., Knauf, V. and Browse, J. 2001. Production of polyunsaturated fatty acids by polyketide synthases in both prokaryotes and eukaryotes. Science 293, 290-293. Murphy, D.E. and Abrajano, T.A. Jr. 1994. Carbon isotope compositions of fatty acids in mussels from Newfoundland estuaries. Estuarine, coastal and shelf science 39, 261-272. Pond, D.W., Sargent, J.R., Fallick, A.E., Allen, C., Bell, M.V. and Dixon, D.R. 2000. δ13C values of lipids from phototrophic zone microplankton and bathypelagic shrimps at the Azores sector of the Mid-Atlantic Ridge. Deep-Sea Research I 47, 121-136. Roman, M.R. 1991. Pathways of carbon incorporation in marine copepods: Effects of developmental stage and food quantity. Limnology and Oceanography 36, 796-807. Simon, H.A., Hodgkins, M.L., Browner, W.S., Neuhaus, J.M., Bernert, J.T. and Hulley, S.B. 1995. Serum fatty acids and the risk coronary heart disease. American Journal of Epidemiology 142, 469-476. Stillwell, W. and Jenski, L. 2002. International workshop on cellular and molecular aspects of ω-3 fatty acids and cancer. Journal of Lipid Research 43, 1579-1580. Sukenik, A. and Carmeli, Y. 1990. Lipid synthesis and fatty acid composition in Nannochloropsis sp. (Eustigmatophyceae) grown in a light-dark cycle. Journal of Phycology 26, 463-469. Thomas, L.M. and Holub, B.J. 1994. Nutritional aspects of fats and oils. In: B.S. Kamel and Y. Kakuda (eds.) Technological advances in improved and alternate sources of lipids. Blackie Academic and Professional, Glasgow, U.K., p.16-49. Tieszen, L.L., Boutton, T.W., Tesdahl, K.G. and Slade, N.A. 1983. Fractionation and turnover of stable carbon isotopes in animal tissues: Implications for δ13C analysis of diet. Oecologia 57, 32-37.

270 Williams, N. 1998. Overfishing disrupts entire ecosystems. Science 279, 809-809.

271

Vita Surname: Veefkind

Given Names: Ruben Jelmar

Place of Birth: Amsterdam, The Netherlands Educational Institutions Attended: University of Victoria

1997 to 2003

Utrecht University

1991 to 1993 and 1994 to 1997

Degrees Awarded: Doctorandus

Utrecht University

1997

Publications: Sephton, M.A., Looy, C.V., Veefkind, R.J., Visscher, H., Brinkhuis, H. and de Leeuw, J.W. 1999. Cyclic diaryl ethers in a Late Permian sediment. Organic Geochemistry 30, 267-273. Sephton, M.A., Veefkind, R.J., Looy, C.V., Visscher, H., Brinkhuis, H. and de Leeuw, J.W. 2001. Lateral variations in end-Permian organic matter. In: E. Buffetaut and C. Koeberl (eds.), Geological and Biological Effects of Impact Events, Springer, pp. 11-24. Sephton, M.A., Looy, C.V., Veefkind, R.J., Brinkhuis, H., de Leeuw, J.W. and Visscher, H. 2002. A synchronous record of δ13C shifts in the oceans and atmosphere at the end of the Permian. In: Catastrophic Events and Mass Extinctions: Impacts and Beyond; Boulder, Colorado (eds. K. MacLeod and C. Koeberl), Geological Society of America Special Paper 356, p. 455-462. Veefkind, R.J., Whiticar, M.J., Perry, R.I. and Whyte, J.N.C. 2003. The molecular, and stable carbon isotope composition of fatty acids and bulk sample as natural tags in marine pelagic organisms. In preparation for Limnology and Oceanography. Veefkind, R.J., Harris, S.L., Whiticar, M.J., Perry, R.I. and Whyte, J.N.C. 2003. Regional and temporal patterns in the fatty acid, and stable carbon isotope composition of seston off the west coast of Vancouver Island, Canada. In preparation for Journal of Plankton Research. Veefkind, R.J., Whiticar, M.J., Whyte, J.N.C. and Perry, R.I. 2003. Stable carbon isotope ratios of individual fatty acids in marine pelagic organisms. In preparation for Organic Geochemistry. Veefkind, R.J., Whiticar, M.J., Perry, R.I. and Whyte, J.N.C. 2003. The stable carbon isotope ratio of fatty acids and bulk organism: recording a natural diet switch. In preparation for Oecologia. Veefkind, R.J., Whiticar, M.J., Whyte, J.N.C. and Perry, R.I. 2003. Trends in the fatty acid composition of marine pelagic organisms with trophic level: implications for the use of fatty acids as trophic markers. In preparation for Marine Ecology Progress Series.

272

University of Victoria Partial Copyright License I hereby grant the right to lend my dissertation to users of the University of Victoria Library, and to make single copies only for such users or in response to a request from the Library of any other university, or similar institution, on its behalf or for one of its users. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by me or a member of the University designated by me. It is understood that copying or publication of this thesis for financial gain by the University of Victoria shall not be allowed without my written permission.

Title of Dissertation:

Carbon Isotope Ratios and Composition of Fatty Acids: Tags and Trophic Markers in Pelagic Organisms

Author ____________________

Ruben Jelmar Veefkind June 19, 2003