[Palaeontology, Vol. 48, Part 1, 2005, pp. 171–183]

INTRA-TREE VARIABILITY IN WOOD ANATOMY AND ITS IMPLICATIONS FOR FOSSIL WOOD SYSTEMATICS AND PALAEOCLIMATIC STUDIES by HOWARD J. FALCON-LANG Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK; e-mail [email protected] Typescript received 30 April 2003; accepted in revised form 17 October 2003

Abstract: The validity of using quantitative analyses of

wood anatomical characters as systematic tools and as palaeoenvironmental proxies has been questioned on the basis that natural variability, and in particular intra-tree variability, tends to drown out the signal being sought. A detailed quantitative description of the wood anatomy of a balsam fir tree was undertaken along root–stump–trunk–branch transects to ascertain intra-tree variability, and to assess noise-to-signal ratio. Results demonstrate significant ontogenetic trends for anatomical parameters such as tracheid pit distribution, cross-field pit frequency, ray dimensions, ray spacing, tracheid diameter, mean ring width and mean sensitivity.

Fossil wood is very common in post-Devonian successions worldwide and, if calcified, silicified or charred, may exhibit a high degree of anatomical preservation (Creber and Chaloner 1984). This material represents an important archive of data for palaeoenvironmental reconstructions for two reasons (Creber 1977). First, studies of fossil wood systematics improve our knowledge of the composition of ancient arborescent vegetation (e.g. Francis 1983; Ash and Creber 2000; Poole 2000a; Falcon-Lang et al. 2001). As fossil woods represent the remains of canopy-forming framework organisms, these plants have a particularly important place in palaeoecological reconstructions. Second, studies of growth ring patterns may provide key information regarding the regional palaeoclimate (e.g. Francis 1984; Creber and Chaloner 1984) and also aspects of the tree’s palaeoecology and phenology (e.g. Falcon-Lang 2000a; Falcon-Lang and Cantrill 2002). Both these strands of research are inherently quantitative in their approach. For example, by virtue of the fact that wood anatomy has remained highly conservative throughout its evolutionary history (Savidge 1996), many fossil wood taxa may only be distinguished from one another in a rigorous, repeatable manner by quantifying numerous anatomical characters in addition to qualitative description (Falcon-Lang and Cantrill 2000, 2001a; Poole and Cantrill 2001). Furthermore, the correlation of

ª The Palaeontological Association

However, although intra-tree variability is great, results suggest that fossil taxa nevertheless may be distinguished from one another on the basis of standard qualitative and quantitative procedures. With regard to palaeoenvironmental studies, results indicate that a significant, but not unrealistic, increase in sample size and an improved knowledge of specimen ontogeny is needed in the future if signals are to be distinguished from background intra-tree variability for some parameters. Key words: fossil wood, growth ring, anatomy, systematics, palaeoclimate.

various aspects of a tree’s phenology, ecology and environment to growth ring anatomy has resulted in a proliferation of quantifiable parameters such as mean ring width, mean sensitivity and percentage of latewood (Creber 1977; Creber and Chaloner 1984; Francis 1986; Morgans 1999; Falcon-Lang 2000a, b). However, a flaw with these quantitative approaches, as presently applied, is that the natural variability of measured parameters is usually very poorly known (Poole 2000b). One of the most important sources of variability in wood anatomy is related to the position and ontogenetic age of the wood, i.e. whether it forms part of a branch, root, stump or the trunk (Jane 1962). According to Larson (1967, p. 145), ‘more variability in wood characteristics exists within a single tree than among [average values for] trees growing on the same … or different sites’. In palaeoclimatic and systematic studies of recent and subfossil woods, this kind of sample variability is minimized by sampling wood at a constant locus in the tree; in most cases mature wood from the trunk at breast height is sampled (Fritts 1976). Unfortunately, it is often difficult to determine the ontogenetic age and original position of a fossil wood fragment (Creber and Chaloner 1984), except through a detailed analysis based on exceptionally well-preserved material (e.g. Falcon-Lang and Cantrill 2002). The problem is exacerbated by the fact

171

172

PALAEONTOLOGY, VOLUME 48

that in most papers on fossil wood, especially in the older literature, no attempt is made to determine the ontogenetic age of the material described. It is clear that if fossil wood anatomical data are to be interpreted with greater precision, detailed knowledge of the magnitude of intra-tree variability of living plants is essential. However, with a few important exceptions (e.g. Bailey and Faull 1934; Chapman 1994), most studies of intra-tree wood variability are too generalized to be readily applied to fossil woods, or focus on parameters that cannot be measured in fossil material, such as specific gravity (Bamber and Burley 1983; Zobel and van Buijtenen 1989). In this paper, a qualitative and quantitative description of intra-tree variations in wood anatomy is presented for a single specimen of balsam fir, Abies balsamea (Linnaeus) Miller, using parameters directly applicable to fossil material. This taxon was chosen for analysis because pinaceous conifer woods are common in the post-Triassic fossil record (e.g. Bannan and Fry 1957; Roy and Stewart 1971; Creber 1972; Roy and Hills 1972; Poole 1996; Figueiral et al. 1999; Morgans 1999; Meijer 2000). The data presented allow the robustness of some systematic, palaeoenvironmental and palaeoecological inferences based on fossil woods to be assessed. They also highlight the problems arising from the fragmentary nature of many fossil wood specimens and indicate aspects of methodology that may require modification if fossil descriptions are to enhance our scientific understanding in the future.

MATERIAL AND METHODS The balsam fir specimen studied was a 12Æ8-m-high monopodial tree with a diameter at breast height of 20 cm. It was growing in a thin (< 30 cm) stony soil overlying Lower Palaeozoic leucogranitic bedrock, at a site near Halifax, Nova Scotia, Canada (Lat. 44Æ65
N, Long. 63Æ60
W), and was uprooted in a storm in April 2002. Climatic parameters from the meteorological station closest to this site are as follows: mean annual temperature 6Æ6
C, cold month mean temperature )4Æ9
C, warm month mean temperature 18Æ1
C, and mean annual rainfall 1394 mm (http://www.worldclimate.com). Four-centimetre-thick transversely orientated discs were cut from the specimen at 1-m intervals. One disc was taken from the roots, 12 discs were taken along the length of the trunk and three discs were taken from a lateral branch attached to the trunk at a point 8 m above ground level (Text-fig. 1). Each disc was sanded down to improve the clarity of the growth rings and ring width sequences were measured to the nearest 0Æ1 mm with the aid of a binocular microscope. As the trunk cross-sectional shape deviated from a circle, ring sequences were measured along both the widest and the narrowest files (Text-fig. 1B). Two standard parameters were then calculated using these data, mean ring width and mean sensitivity. This latter parameter is given by the equation:

T E X T - F I G . 1 . Diagrammatic representation of the studied specimen showing the position and accession number of each disc specimen. A crosssection through the stump disc (SO) is given as an inset, illustrating how the narrowest (A–B) and widest (B–C) growth ring sequences were measured for each disc.

FALCON-LANG: IMPLICATIONS OF VARIABILITY IN WOOD ANATOMY



X
2ðxtþ1
xt Þ
1 t¼n
1

MS ¼ n
1 t¼1
xtþ1 þ xt
where x is ring width, n is the number of rings in the sequence analysed and t is the year number of each ring. Cubic 1–2-cm-diameter blocks were then cut from the discs. Blocks were taken from the most mature wood in each of the discs and given the accession numbers R1 (for root wood positioned 1 m from the stump), S0–S12 (for trunk wood positioned 0–12 m from trunk base) and B0–B2 (for branch wood positioned 0–2 m from the point of branch attachment). In addition, in order to investigate ontogenetic trends within the stump wood (the organ most commonly encountered in fossil forests; Chapman 1994) a sequence of eight blocks was taken from the basal trunk disc (S0) from the centre to its outer edge. These were labelled S01 (for the most mature stump wood) to S08 (for the most juvenile stump wood). Standard 20-lm-thick sections were then taken along radial longitudinal (RLS), tangential longitudinal (TLS) and transverse (TS) orientations for each sample using a block sliding microtome housed at the Jodrell Laboratory, Royal Botanic Gardens, Kew. Sections were stained using Safranin (lignin) and Alicerin Blue (cellulose) and permanently mounted on glass slides. Using a binocular transmitted light microscope, the woods were described using a quantitative technique specifically devised for fossil woods by Chapman and Smellie (1992), and subsequently further developed by FalconLang and Cantrill (2000, 2001a). In addition to general qualitative observations, the following parameters were quantified. (1) The proportion of uniseriate vs. multiseriate bordered pitting was ascertained for tracheids in RLS. (2) The degree of bordered pit contiguity was measured by counting the length of contiguous pit sequences along tracheids in RLS. (3) The proportion of blank tracheids (lacking pits) was measured in RLS. (4) The number of pits per cross-field region was measured in RLS. (5) Ray height expressed in number of cells was measured in TLS. (6) Ray spacing in micrometres was measured in TS along a tangential orientation (parallel to the growth ring boundary). For each of these six parameters up to 50 measurements were taken in each sample. However, as several authors have previously noted (e.g. Bailey and Faull 1934), there is considerable heterogeneity in the abundance and distribution of most of these features along the length of individual tracheids and also across growth increments. Consequently, in order for the data to be collected in a repeatable and meaningful way, measurements were standardized in the following manner. First, when measuring parameters 1–6, only earlywood regions were targeted because (1) earlywood tracheids are easiest to observe and (2) they exhibit much greater anatomical

173

variability than latewood cells. Targetting earlywood regions has an additional advantage in that anatomical preservation is usually best in the earlywood of fossil woods, thereby enhancing the applicability of the methodology to the fossil realm. Second, when measuring parameters 1–3, variations along the length of tracheids were homogenized by ascertaining mean values for each tracheid; this was achieved by systematically measuring parameters along the entire length of selected individual tracheids. Although this method obscures the finest scale of anatomical heterogeneity (intratracheid variability), it is advantageous in that fossil wood preservation is typically insufficient to measure this level of variability satisfactorily across wide areas. In addition to these parameters, a contiguous sequence of tracheid diameters was measured across a single file comprising one growth increment for each sample in TS. From these data percentage latewood was calculated using the method of Creber and Chaloner (1984). The cumulative algebraic sum of each tracheid cell’s deviation from the mean (CSDM curve) was calculated using the method of Falcon-Lang (2000a). Mean whole-ring tracheid diameter was also ascertained. Mean tracheid length, a key characteristic often studied in ontogenetic analyses of modern trees (e.g. Gartner 1995; Lee and Wang 1996), was not ascertained in this study because this feature is very hard to measure in fossil woods owing to the difficulty of producing accurate longitudinal sections for fossil material and to the structural buckling commonly encountered in fossil woods. Similarly, specific gravity, another key ontogenetic parameter (Bamber and Burley 1983; Zobel and van Buijtenen 1989) was also not measured, as this parameter is strongly affected by taphonomic processes in fossil material.

RESULTS All the wood thin sections analysed in this study are stored in the Gymnosperm Collection in the Anatomy Section of the Jodrell Laboratory, Royal Botanic Gardens, Kew, UK. The general anatomical features of Abies balsamea are illustrated for mature trunk wood at breast height (disc S01) in Text-figure 2. A detailed description of ontogenetic wood anatomical trends is given below, and raw data are presented in Tables 1–2 and plotted in Text-figures 3–4. A more general description of the wood of A. balsamea may be obtained in the classic studies of Wiesehuegel (1932) and Greguss (1955, 1972).

Tracheid characteristics Mean tracheid diameter shows a marked decrease with decreasing ontogenetic age both from the outer edge to the centre of the stump (R2 ¼ 0Æ808) and along the

174

PALAEONTOLOGY, VOLUME 48

Mature wood anatomy of the Abies balsamea specimen at breast height (disc S1). A, wide tracheids exhibiting uniseriate bordered pitting with moderate contiguity, RLS; · 300. B, wide tracheids exhibiting uniseriate bordered pitting with low contiguity, RLS; · 300. C, ray composed of long, narrow cells with nodular end walls, and 2–3 taxodioid cross-field pits, RLS; · 300. D, tall uniseriate rays, TLS; · 50. E, small, irregularly distributed pits on tangential tracheid walls, TLS; · 300. F, medium-sized ring increment showing widely spaced rays, TS; · 30. TEXT-FIG. 2.

length of the trunk (R2 ¼ 0Æ649) (Text-fig. 3). In both cases mean tracheid diameter decreases from close to 40 lm to < 30 lm. Mean tracheid diameter for juvenile branch woods is 27 lm. For stump, trunk and branch woods, maximum earlywood tracheid diameter ranges from 45 to 58 lm, and minimum latewood tracheid diameter ranges from 7 to 10 lm. For root woods, tracheid dimensions are generally much greater, with mean tracheid diameter being 54 lm, maximum earlywood diameter ranging from 58 to 73 lm and minimum latewood diameter ranging from 17 to 20 lm. Tracheids exhibit bordered pits on tangential and radial walls. Tangential pits have small circular borders (7–8 lm diameter), small circular apertures (2–3 lm diameter) and are very irregularly distributed. They increase in frequency with decreasing ontogenetic age; they are also especially common near zones of traumatic tissue. In contrast, pits on radial tracheid walls have large circular borders (18–25 lm diameter) with circular apertures (5–6 lm diameter). Their dimensions decrease to 13–15 lm and 3–5 lm,

respectively, in mature latewood, and latewood apertures are distinctly oval and obliquely orientated. All these pit dimensions are about 20 per cent smaller in juvenile woods. Radial tracheid pits dominantly (97–100%) have a uniseriate arrangement; however, in root wood where tracheid diameter is significantly greater, biseriate, opposite pitting characterized by distinct bars of Sanio occurs on 17 per cent of tracheids (Table 1). In the earlywood, the spacing of tracheid pits also varies with ontogenetic age along both stump (R2 ¼ 0Æ540) and trunk (R2 ¼ 0Æ758) transects, with the mean contiguity of bordered pitting increasing from values of 2–3 in the most mature wood to more than 7 in juvenile wood (Text-fig. 3). Earlywood in both branches and roots has moderate to high contiguity values (3–5). Latewood tracheids, irrespective of ontogenetic age, exhibit uniformly spaced bordered pits with contiguity values approaching 1. This trend towards more contiguous pitting in juvenile woods is mirrored, perhaps unsurprisingly, by a decrease

FALCON-LANG: IMPLICATIONS OF VARIABILITY IN WOOD ANATOMY TABLE 1.

175

Quantitative data concerning the ontogenetic variability for several wood anatomical characteristics (see text for further

details). Sample no.

R1 S01 S02 S03 S04 S05 S06 S07 S08 Mean S0 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 Mean B0 B1 B2 Mean Overall mean SD

Tracheid characteristics

Ray characteristics

Growth ring characteristics

Uniseriate bordered pitting (%)

Blank tracheids (%)

Pit contiguity value

No. of pits per ray cross-field

Ray height (no. of cells)

Ray spacing (lm)

No. of tracheids in ring

Mean tracheid diameter (lm)

Latewood (%)

CSDM skew (%)

83 100 99 100 100 100 100 100 100 99Æ87 100 97 100 100 99 100 99 100 99 100 100 100 100 99Æ56 100 100 100 100 99Æ04

1 17 15 14 8 10 11 2 1 9Æ75 17 21 5 12 12 21 7 10 6 3 16 7 0 10Æ53 23 8 18 16Æ33 10Æ60

5Æ56 2Æ55 2Æ89 2Æ66 3Æ46 3Æ62 4Æ94 3Æ14 4Æ43 3Æ46 2Æ55 2Æ07 3Æ75 3Æ44 3Æ26 3Æ21 4Æ21 3Æ91 5Æ89 5Æ97 5Æ00 7Æ04 5Æ41 4Æ29 5Æ21 2Æ95 4Æ52 4Æ22 4Æ06

2Æ59 1Æ82 1Æ95 1Æ98 2Æ03 2Æ01 2Æ03 2Æ08 2Æ31 2Æ02 1Æ82 1Æ53 1Æ91 1Æ85 1Æ73 1Æ90 1Æ84 1Æ99 2Æ08 2Æ23 2Æ25 2Æ48 2Æ88 2Æ03 2Æ30 2Æ39 2Æ80 2Æ49 2Æ11

6Æ32 12Æ82 10Æ06 10Æ62 8Æ78 9Æ68 7Æ46 6Æ36 5Æ68 8Æ93 12Æ82 11Æ10 10Æ66 10Æ3 9Æ48 8Æ22 9Æ52 8Æ74 8Æ56 8Æ72 7Æ94 8Æ20 6Æ30 9Æ27 6Æ56 5Æ90 4Æ88 5Æ78 8Æ62

353 258 295 214 304 233 222 217 121 233 258 230 253 246 260 212 245 214 217 220 140 149 129 213 140 252 137 176 220

7 36 36 35 37 41 44 33 52 39Æ25 36 46 46 37 43 40 44 30 43 46 44 77 46 44 39 38 45 41 41

53Æ92 39Æ93 39Æ30 39Æ85 36Æ01 35Æ67 31Æ76 26Æ66 31Æ34 35Æ06 39Æ93 34Æ89 36Æ63 31Æ08 33Æ02 33Æ93 32Æ96 33Æ33 35Æ17 32Æ10 29Æ71 30Æ29 26Æ84 33Æ08 26Æ41 34Æ61 21Æ39 27Æ47 33Æ86

28Æ57 33Æ33 33Æ33 28Æ57 29Æ72 48Æ78 29Æ54 33Æ33 26Æ92 32Æ94 33Æ33 30Æ43 30Æ43 48Æ64 41Æ86 50Æ00 31Æ81 33Æ33 32Æ55 34Æ21 25Æ00 23Æ37 26Æ08 33Æ93 23Æ08 42Æ10 11Æ11 25Æ43 32Æ38

42Æ85 33Æ33 5Æ55 42Æ85 40Æ54 2Æ43 22Æ72 33Æ33 46Æ15 28Æ36 33Æ33 26Æ08 17Æ39 2Æ70 16Æ27 0Æ00 22Æ73 33Æ33 25Æ55 31Æ57 50Æ00 53Æ24 43Æ47 27Æ35 7Æ69 15Æ78 51Æ11 24Æ86 27Æ99

3Æ41

6Æ74

1Æ28

0Æ33

2Æ13

58

6Æ21

8Æ73

16Æ34

in the number of blank tracheids with decreasing ontogenetic age across the stump transect (R2 ¼ 0Æ841) and, very weakly, along the trunk transect (R2 ¼ 0Æ358) (Textfig. 3). An exception to this trend is in the branch tissue, where pits are relatively uncommon owing to the high frequency of spiral thickening (reaction wood), which ornaments up to 70 per cent of tracheids in some ring increments. This kind of spiral tertiary thickening is rare elsewhere in the tree, occurring with background levels of only 1–5 per cent.

Ray characteristics Ray length was measured in TS. Although this technique may result in minimal ray lengths being recorded because rays may ‘drop out’ of cross-sectional view prior to their

11

ends, it is the only way of approximating ray length in a manner applicable to both modern and fossil woods. In fossil woods, rays are usually radially buckled so that length cannot be measured along the radial plane. Measured in TS, rays range in length from 1Æ75 to 4Æ58 mm (rarely up to 7Æ32 mm) in mature basal trunk woods, decreasing to 0Æ65–2Æ75 mm in juvenile trunk and branch woods. Ray height is very closely related to ontogenetic age across both stump (R2 ¼ 0Æ902) and trunk (R2 ¼ 0Æ849) transects (Text-fig. 3). The most mature woods possess rays with mean heights of 10–13 cells in TLS (range 1–28), whereas rays in juvenile woods are only 6–7 cells high (range 1–21). Similarly, juvenile root and branch wood possesses equally short rays with mean heights on the order of 5–7 cells (range 1–19). In addition, ontogenetic age weakly influences ray spacing with an evident relationship between these two parameters across both

176

PALAEONTOLOGY, VOLUME 48

TABLE 2.

Quantitative data concerning the ontogenetic variability of growth ring width and mean sensitivity (see text for further

details). Sample no.

R1 S0 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 Mean B0 B1 B2 Mean Overall mean SD

Trunk diameter (mm)

No. of tree rings (mm)

Mean ring width Widest file (mm)

Narrowest file (mm)

Mean (mm)

Widest file (lm)

Narrowest file (lm)

Mean (lm)

49 256 209 197 186 182 177 164 157 149 126 103 79 42

46 88 78 70 66 63 55 47 41 38 34 29 19 11

58 31 14

26 12 4

0Æ609 1Æ475 1Æ138 1Æ294 1Æ157 1Æ335 1Æ360 1Æ572 1Æ595 1Æ753 1Æ491 1Æ717 1Æ753 1Æ391 1Æ464 1Æ042 1Æ125 0Æ975 1Æ047 1Æ340

0Æ148 0Æ987 0Æ949 0Æ980 1Æ121 1Æ035 1Æ240 1Æ206 1Æ302 1Æ305 1Æ238 1Æ107 1Æ316 1Æ245 1Æ156 0Æ465 0Æ942 0Æ850 0Æ752 1Æ026

0Æ378 1Æ231 1Æ043 1Æ137 1Æ139 1Æ185 1Æ300 1Æ389 1Æ448 1Æ529 1Æ364 1Æ412 1Æ534 1Æ318 1Æ309 0Æ753 1Æ033 0Æ912 0Æ899 1Æ183

0Æ520 0Æ256 0Æ296 0Æ243 0Æ216 0Æ241 0Æ252 0Æ275 0Æ219 0Æ347 0Æ328 0Æ401 0Æ434 0Æ305 0Æ293 0Æ386 0Æ321 0Æ309 0Æ339 0Æ315

0Æ334 0Æ285 0Æ259 0Æ257 0Æ231 0Æ305 0Æ276 0Æ277 0Æ292 0Æ318 0Æ299 0Æ328 0Æ421 0Æ381 0Æ302 0Æ277 0Æ528 0Æ039 0Æ281 0Æ300

0Æ427 0Æ270 0Æ277 0Æ250 0Æ223 0Æ273 0Æ264 0Æ276 0Æ255 0Æ332 0Æ313 0Æ364 0Æ427 0Æ343 0Æ297 0Æ331 0Æ429 0Æ174 0Æ311 0Æ308

0Æ301

0Æ304

0Æ291

0Æ080

0Æ095

0Æ071

stump (R2 ¼ 0Æ509) and trunk (R2 ¼ 0Æ706) transects. Rays are relatively widely spaced in mature stump and trunk woods (250–300 lm apart, i.e. 3–4 rays per linear millimetre), but more closely packed in juvenile stump, trunk and branch woods (120–150 lm apart, i.e. 7–8 rays per linear millimetre). Exceptionally widely spaced rays in root wood (350 lm apart) probably reflect the very large-diameter intervening tracheids in this organ (Text-fig. 3). Rays are almost exclusively uniseriate in TLS, although rare examples in the branch wood or in juvenile trunk wood may possess short (2–5 cells high) biseriate portions. They are composed of both ray tracheids and parenchyma. Ray tracheids are rare, and are commonly only found associated with traumatic tissue. Ray parenchyma cells are ubiquitous and their dimensions change with ontogenetic age. Long and thin ray parenchyma cells occur in mature wood (70–275 lm long, 17–28 lm high, 10–15 lm wide), with relatively short and fat parenchyma cells found in juvenile wood (35–115 lm long, 20–37 lm high, 13–20 lm wide). All ray parenchyma have distinctly nodular tangential end walls, and may possess irregular indentures and calcium oxalate crystals, features that are found throughout the tree.

Mean sensitivity

Ray parenchyma exhibit pitting on all walls, transverse, tangential and especially radial. The diagnostic cross-field pits on the radial walls are small (5–7 lm diameter), oval and either lack borders completely or possess very narrow, subdued borders defining a large oval aperture. The former cross-field pit type would be classified as podocarpoid, whereas the latter are taxodioid in character. The number of pits per cross-field increases with decreasing ontogenetic age across both stump (R2 ¼ 0Æ786) and trunk (R2 ¼ 0Æ766) transects (Text-fig. 3). In both cases mature woods typically possess 1Æ5–1Æ9 pits per field (range 1–4), whereas juvenile wood in the upper trunk, branches and roots possesses a mean cross-field pit number of 2Æ2–2Æ9 (range 1–6).

Growth ring characteristics Growth rings have marked boundaries continuous around the circumference of all organs. Mean ring width is weakly related to ontogenetic age along the trunk transect (R2 ¼ 0Æ571) with basal trunk ring widths on the order of 1Æ0–1Æ2 mm, increasing to 1Æ3–1Æ5 mm in the upper trunk. Lateral branches have lower mean ring widths

FALCON-LANG: IMPLICATIONS OF VARIABILITY IN WOOD ANATOMY

177

T E X T - F I G . 3 . Scatter graphs showing ontogenetic relationship between different wood anatomical parameters. Symbols: closed diamonds, trunk wood; open squares, stump wood; open triangles, branch wood; crosses, root wood. Ontogenetic age is plotted on the x-axis; the y-axis is indicated by the title of each graph.

(0Æ8–1Æ0 mm) with roots exhibiting exceptionally narrow rings (0Æ4 mm). Ring width is asymmetric around the circuit of all organs, but asymmetry is greatest in root

wood where the widest file is 4Æ11 times wider than the narrowest, in contrast to asymmetry values for branch and trunk wood of 1Æ39 and 1Æ26, respectively. Like ring

178

PALAEONTOLOGY, VOLUME 48

TEXT-FIG. 4.

Scatter graphs showing ontogenetic relationship of different growth ring parameters. Symbols and axis labels as for

Text-figure 3.

width, mean sensitivity increases with decreasing ontogenetic age (R2 ¼ 0Æ556), with lower trunk discs having values around 0Æ22–0Æ27 compared with upper trunk values ranging up to 0Æ43 (Text-fig. 4). Branch wood exhibits the most highly variable mean sensitivity values of any organ (0Æ17–0Æ43), although this variability may be related to the relatively short nature of the ring sequences (4–26 rings long) that do not permit statistically significant data to be collected. The single root specimen has very high mean sensitivity values (0Æ43). Neither percentage latewood nor percentage CSDM skew values exhibit any obvious relationship to ontogenetic age (R2 ¼ 0Æ066–0Æ351 and 0Æ052–0Æ352, respectively), although latewood values are highest and CSDM values lowest in mature trunk wood (Text-fig. 4). Both parameters exhibit considerable variability, with whole-tree values ranging from 11 to 50 per cent (mean 32Æ38 per cent; standard deviation ± 8Æ73 per cent) and

0–53 per cent (mean 27Æ99 per cent; standard deviation ± 16Æ34 per cent), respectively. Despite this variability, standard deviation values stabilize within 10 per cent of the whole-tree mean for percentage latewood after nine readings are averaged and within 5 per cent of the mean after averaging 16 readings. Percentage CSDM skew values are more variable, with standard deviation values stabilizing with 15 per cent of the final mean after five readings are averaged. No relationship was observed between percentage latewood or percentage CSDM and ring width, expressed either in terms of number of cells (R2 ¼ 0Æ002) or in millimetres (R2 ¼ 0Æ026). Frost rings only occur within the juvenile wood of the upper trunk (S10–S12) and branches (B0–B2) where they are defined by impersistent rings of large-diameter traumatic parenchyma (up to 120 lm diameter) dominantly distributed at the start of the earlywood and the end of the latewood (i.e. beginning and end of the

FALCON-LANG: IMPLICATIONS OF VARIABILITY IN WOOD ANATOMY

growing season). In one sample containing only six ring increments, eight frost rings were found. Locally frost rings may grade around the stem circuit into concentric zones of smaller than normal diameter tracheids (i.e. false rings) with common interspersed, smaller and more ordered axial parenchyma cells. These latter axial parenchyma cells may either be restricted to a narrow concentric zone or be scattered widely throughout traumatized ring increment. Axial parenchyma cells are vertically elongate (20–25 lm diameter; 100–120 lm high) with nodular end walls and profusely pitted longitudinal walls. They are never encountered in mature untraumatized woods. Their observed increase in frequency with decreasing ontogenetic age therefore presumably reflects the greater susceptibility of younger stems to frost damage.

DISCUSSION All quantified parameters show a significant (R2 > 0Æ5) relationship with ontogenetic age with the exceptions of percentage latewood, percentage CSDM skew and to some extent the percentage of blank tracheids in trunk wood. The limited number of published studies of conifer wood ontogeny (in a form applicable to fossil studies) suggest that the trends and magnitudes of variation encountered here for Abies balsamea may be true for conifers in general, with the most mature conifer woods having narrower ring widths, proportionally more latewood, longer tracheids (not measured here), less spiral thickening (blank tracheids) and less numerous rays (see Gartner 1995 for review; Bamber and Burley 1983; de Kort 1993; Lee and Wang 1996). This large intra-tree anatomical variability has several important implications for palaeontological studies as follows.

179

would plot as single continuous entities in ‘morphology space’. The results of this paper are generally supportive of the approach of Falcon-Lang and Cantrill (2000, 2001a). A ‘morphology space’ diagram using the two parameters found to be most discriminating for some fossil wood studies (tracheid pit contiguity and ray height) is shown in Text-figure 5, and plots Cretaceous fossil wood data (Falcon-Lang and Cantrill 2001a) together with data on Abies balsamea (this paper). As predicted for a true biological species, the A. balsamea data plot as a continuous field in morphology space. This holds true for all the characters, although these are not illustrated. A partial exception to this observation is root wood, which often possesses characters with quantitatively very different values from other parts of the same tree. Consequently, there would seem to be an especially great danger of erroneously erecting new taxa based on fossil root wood. One example of this is the way that anatomically preserved root wood of a type of Upper Carboniferous cordaite tree has been named Ameylon whereas attached trunks have been referred to the genus Pennsylvanioxylon (Cridland 1964; Costanza 1985). In addition to this problem with root wood, the general differences between intergrading juvenile and mature wood characters seen in A. balsamea bring in to question some specific taxonomic determinations, such as the distinction between Falcon-Lang and Cantrill’s (2001a) two Araucarioxylon species. The main defining feature of the

Implications for systematic studies As noted in the introduction, qualitative descriptive analyses alone have, in many cases, proved inadequate in distinguishing fossil conifer wood species owing to the conservative nature of this group. This has resulted in a proliferation of species names. For example, up to the time of their writing, Vaudois and Prive´ (1971) had recognized over 44 species within the genus Cupressinoxylon, and Schultze-Motel (1966) had recognized 45 species within the genus Araucarioxylon. Many of these probably represent ontogenetic variants of the same species. Developing the work of Chapman and Smellie (1992), FalconLang and Cantrill (2000, 2001a) introduced quantitative multivariant analyses to fossil wood taxonomy to correct for this problem, reasoning that true biological species

Plot of ray height vs. tracheid pit contiguity showing the distribution of some fossil wood taxa (data from Falcon-Lang and Cantrill 2001a) and modern Abies balsamea in ‘morphology space’. Juvenile Abies data points, closed squares; mature data points, open circles (root wood not plotted).

TEXT-FIG. 5.

180

PALAEONTOLOGY, VOLUME 48

second Araucarioxylon species (relative to the first) is its shorter rays composed of shorter and fatter ray parenchyma, the occurrence of resinous matter in the tracheids, and as recorded in Falcon-Lang and Cantrill (2002), its narrower ring width and greater mean sensitivity. In light of the A. balsamea data, it is possible that instead of representing a second biological species, these differences may simply represent the ontogenetically youngest wood of A. sp. 1. It is noteworthy that the magnitude of morphological variability for A. balsamea is substantially greater than that exhibited by the fossil taxa (Text-fig. 5), and if the two Araucarioxylon species were combined into one, the resultant species would occupy a more closely similar morphological range to the modern taxon. Perhaps an even stronger case for the mistaken demarcation of fossil wood species on the basis of quantitative grounds is found in Poole and Cantrill (2001). Their Podocarpoxylon communis and P. chapmanae are very closely similar except for the greater frequency of uniseriate pitting, the smaller diameter of tracheid pit borders and the lower ray height in the former species. These quantitative differences could be interpreted as suggesting that P. communis represents the ontogenetically youngest wood of P. chapmanae. Although it is difficult to prove this assertion without detailed reanalysis of the original material, the results of this paper strongly emphasize the importance of rigorously considering ontogenetic variability before splitting fossil wood species.

Implications for palaeoenvironmental analyses More significant problems are raised by the ontogenetic data for palaeoenvironmental studies. Mean ring width has been used as an important indicator of the favourability of growing conditions (especially climate), and mean sensitivity has been used as a proxy for year-to-year environmental variability and hence for ecological stress (e.g. Francis 1984; Morgans 1999; Falcon-Lang et al. 2001). Although significant ontogenetic variability in these parameters has long been known (e.g. Bailey and Faull 1934), quantitative intra-tree variations have not been explicitly published in a form applicable to fossil wood studies. It is notable that growth ring sequences for the mature trunk of Abies balsamea are narrow (1Æ0–1Æ2 mm) and complacent (< 0Æ3), and yet are generally wider (1Æ3–1Æ5 mm) and sensitive (> 0Æ3) for juvenile organs. Should these organs be preserved separately in the fossil record, wood remains from different parts of the same tree may be interpreted differently, the mature assemblage being indicative of uniform, relatively unfavourable growing conditions (based on the conventional wisdom of Creber and Chaloner 1984), with the juvenile assemblage indicating variable but relatively favourable growing con-

ditions. In fact the taphonomic separation of mature and juvenile organs is very likely, especially, for example, in a situation where autochthonous stump assemblages are compared with allochthonous assemblages (which are mostly sourced from crown regions). Two recommendations are, therefore, made for future analyses of growth rings in fossil woods, which if adhered to will help to circumvent some of the problems raised above. First, it is important that every effort is made to understand the systematics and ontogenetic age of the wood being analysed, as exemplified by the studies of Chapman and Smellie (1992), Falcon-Lang et al. (2001) and Falcon-Lang and Cantrill (2002). In addition to the quantitative data of this paper, Chapman (1994) provided an invaluable summary of the wood characteristics useful in determining the position and ontogenetic age of a fossil wood specimen. Second, it is important that only large datasets are utilized, such as those collated by Francis and Poole (2002). The application of simple statistical measures of sample variability (i.e. standard deviation) is essential if intra-tree variability is to be distinguished from genuine palaeoenvironmental signals. A second type of palaeoenvironmental indicator commonly analysed in fossil woods is the presence or absence of traumatic rings, a phenomenon dominantly caused by sharp frosts during the growing season (e.g. Roy and Hills 1972; Chapman 1994; Falcon-Lang et al. 2001), although other factors such as fire or insect-mediated defoliation may result in similar features (Dechamps 1984). It is significant that, although frosts are extremely common during the beginning and end of the growing season in Nova Scotia (http://www.worldclimate.com), the occurrence of frost rings is highly localized in the specimen studied, being restricted to the juvenile wood of the upper trunk and lateral branches. Significantly, Chapman (1994) presented nearly identical results in her influential review of intra-tree wood variability. Juvenile woods are most susceptible to frost ring formation because, by virtue of being proximal to the crown, the vascular cambium of these regions is active much earlier and later in the growing season compared with the lower trunk (Bannan 1954; Glock et al. 1960; Glerum and Farrar 1966). The results presented here suggest that frost frequency may be greatly underestimated by palaeoclimatic studies based on mature wood specimens, and that the occurrence of frost rings in mature fossil trunk wood may indicate exceptionally harsh frost events.

Implications for phenological analyses Falcon-Lang (2000a) recently proposed percentage CSDM skew as a proxy for fossil conifer leaf longevity, warning

FALCON-LANG: IMPLICATIONS OF VARIABILITY IN WOOD ANATOMY

that the new technique should be applied with caution because of the small sample size used and the great variability exhibited by the dataset. That concern has, to some extent, been borne out by the present paper, which indicates considerable intra-tree variability in the CSDM value. Other recent studies have reached similar conclusions (Harland 2002). Nevertheless comparison of the mean whole-tree CSDM skew value obtained here for Abies balsamea (27Æ99 ± 16Æ33 per cent) with the original data set in Falcon-Lang (2000a) suggests a leaf retention time of c. 3–6 years, a value very similar to the leaf longevity value for the studied specimen (3–7 years) calculated using leaf trace persistence data (methodology of Falcon-Lang and Cantrill 2001b). Given that the standard deviation of the skew value for A. balsamea stabilized within 15 per cent of the final mean after five readings were averaged, it may be possible to obtain fossil leaf longevity estimates from CSDM data, despite significant intra-tree variability, provided that a sufficiently large data set is collected. As few as five measurements may be necessary merely to distinguish deciduous from evergreen taxa. However, a much safer approach to ascertaining fossil leaf longevity, and the one emphasized by Falcon-Lang (2000a), would be to use as many different techniques as possible in conjunction with one another. This multiple technique approach was used with great success by Falcon-Lang and Cantrill (2001b), who analysed the leaf phenology of an Early Cretaceous fossil forest on Alexander Island, Antarctica. They applied techniques that utilized leaf physiognomy, leaf trace persistence, leaf taphonomy and nearest living relative comparison, in addition to the CSDM methodology. The strength of this study was the way in which different techniques consistently gave similar leaf longevity values. In summary, whereas CSDM skew values may be used alone to determine fossil leaf longevity provided that the data set is sufficiently large, much more secure phenological interpretations necessary for regional-scale climate modelling (Osborne and Beerling 2002) may require a combined multiple technique analysis.

CONCLUSIONS Poole (2000b) described ontogenetic variability as ‘nature’s spanner’, a phenomenon with the potential to bring the entire task of using growth rings (and other features) as quantitative palaeoenvironmental proxies to a grinding halt. The results of this study are not quite so pessimistic. Although ontogenetic variability raises some genuine major problems for a variety of palaeontological techniques using fossil wood (i.e. systematics and palaeoenvironmental proxies), it is possible that many of

181

these can be satisfactorily circumvented by increasing sample size and controlling the sample locus through ontogenetic analyses. Thus, the more detailed quantitative knowledge of the ontogenetic variability of conifers presented here, rather than raising insurmountable problems for the various palaeontological techniques discussed, actually lays the foundations for them to be refined in the future. Acknowledgements. I thank Dr John Calder for helping to saw up the studied specimen on his property, and Dr Peter Gasson at the Jodrell Laboratory, Royal Botanic Gardens, Kew (UK), for teaching me how to prepare the wood sections used in this study. This research was jointly funded through a Killam Fellowship held at Dalhousie University (Canada) and a NERC Fellowship (NER ⁄ 2001 ⁄ 738) held at the University of Bristol (UK). The careful and critical reviews of Dr Jenny Chapman (University of Cambridge, UK) and Dr Elisabeth Wheeler (University of North Carolina, USA) improved this paper greatly.

REFERENCES A S H , S. R. and C R E B E R , G. T. 2000. The Late Triassic Araucarioxylon arizonicum trees of the Petrified Forest National Park, Arizona, USA. Palaeontology, 43, 15–28. B A I L E Y , I. W. and F A U L L , A. F. 1934. The cambium and its derivative tissues, IX. Structural variability in the redwood, Sequoia sempervirens, and its significance in the identification of fossil woods. Journal of the Arnold Arboretum, 15, 99–106. B A M B E R , R. K. and B U R L E Y , J. 1983. The wood properties of Radiata Pine. Commonwealth Agricultural Bureau, Slough, 84 pp. B A N N A N , M. W. 1954. The wood structure of some Arizonan and Californian species of Cupressus. Canadian Journal of Botany, 32, 285–307. —— and F R Y , W. L. 1957. Three Cretaceous woods from the Canadian Arctic. Canadian Journal of Botany, 35, 327–337. C H A P M A N , J. L. 1994. Distinguishing internal developmental characteristics from external palaeoenvironmental effects in fossil wood. Review of Palaeobotany and Palynology, 81, 19–32. —— and S M E L L I E , J. L. 1992. Cretaceous fossil wood and palynomorphs from Williams Point, Livingston Island, Antarctic Peninsula. Review of Palaeobotany and Palynology, 74, 164– 192. C O S T A N Z A , S. H. 1985. Pennsylvanioxylon of Middle and Upper Pennsylvanian coals from the Illinois Basin and its comparison with Mesoxylon. Palaeontographica B, 197, 81–121. C R E B E R , G. T. 1972. Gymnospermous wood from the Kimmeridgian of East Sutherland and from the Sandringham Sands of Norfolk. Palaeontology, 15, 655–661. —— 1977. Tree rings: a natural data-storage system. Biological Reviews, 52, 349–383. —— and C H A L O N E R , W. G. 1984. Influence of environmental factors on the wood structure of living and fossil trees. The Botanical Review, 50, 357–448.

182

PALAEONTOLOGY, VOLUME 48

C R I D L A N D , A. A. 1964. Amyelon in American coal balls. Palaeontology, 7, 189–209. D E C H A M P S , R. 1984. Evidence of bush fires during the PlioPliestocene in Africa (Omo and Sahabi) with the aid of fossil woods. 231–239. In C O E T Z E E , J. A. and V A N Z I N D E R E N B A K K E R , E. M. (eds). The palaeoecology of Africa and surrounding islands, 16. Balkema, Cape Town, 431 pp. F A L C O N - L A N G , H. J. 2000a. A method to distinguish between woods produced by evergreen and deciduous coniferopsids on the basis of growth ring anatomy: a new palaeoecological tool. Palaeontology, 43, 775–783. —— 2000b. The relationship between leaf longevity and growth ring markedness in modern conifer woods and its implications for palaeoclimatic studies. Palaeogeography, Palaeoclimatology, Palaeoecology, 160, 317–328. —— and C A N T R I L L , D. J. 2000. Cretaceous (Late Albian) Coniferales of Alexander Island, Antarctica. Part 1: Wood taxonomy: a quantitative approach. Review of Palaeobotany and Palynology, 111, 1–7. —— —— 2001a. Gymnosperm woods from the Cretaceous (mid-Aptian) Cerro Negro Formation, Byers Peninsula, Livingston Island, Antarctica: the arborescent vegetation of a volcanic arc. Cretaceous Research, 22, 277–293. —— —— 2001b. Leaf phenology of some mid-Cretaceous polar forests, Alexander Island, Antarctica. Geological Magazine, 138, 39–52. —— —— 2002. Terrestrial ecology of an Early Cretaceous highlatitude, volcanic archipelago, Byers Peninsula and President Head, South Shetlands, Antarctica. Palaios, 17, 535–549. —— —— and N I C H O L S , G. J. 2001. Biodiversity and terrestrial ecology of a mid-Cretaceous, high-latitude floodplain, Alexander Island, Antarctica. Journal of the Geological Society, London, 158, 709–725. F I G U E I R A L , I., M O S B R U G G E R , V., R O W E , N. P., A S H R A F , A. R., U T E S C H E R , T. and J O N E S , T. P. 1999. The Miocene peat-forming vegetation of northwestern Germany: an analysis of wood remains and comparison with previous palynological interpretations. Review of Palaeobotany and Palynology, 104, 239–266. F R A N C I S , J. E. 1983. The dominant conifer of the Jurassic Purbeck Formation, England. Palaeontology, 26, 277–294. —— 1984. The seasonal environment of the Purbeck (Upper Jurassic) fossil forests. Palaeogeography, Palaeoclimatology, Palaeoecology, 48, 285–307. —— 1986. Growth rings in Cretaceous and Tertiary wood from Antarctica and their palaeoclimatic implications. Palaeontology, 29, 665–684. —— and P O O L E , I. 2002. Cretaceous and early Tertiary climates of Antarctica: evidence from fossil wood. Palaeogeography, Palaeoclimatology, Palaeoecology, 182, 47–64. F R I T T S , H. C. 1976. Tree rings and climate. Academic Press, London, 576 pp. G A R T N E R , B. L. 1995. Patterns of xylem variation within a tree and their hydraulic and mechanical consequences. 125–149. In G A R T N E R , B. L. (ed.). Plant stems: physiology and functional morphology. Academic Press, London, 432 pp.

G L E R U M , C. and F A R R A R , J. L. 1966. Frost ring formation in the stems of some coniferous species. Canadian Journal of Botany, 44, 879–886. G L O C K , W. S., S T U D H A L T E R , R. A. and A G E R T E R , S. R. 1960. Classification and multiplicity of growth layers in the branches of trees. Smithsonian Miscellaneous Collections, 140. Publication 4421, Washington, DC, 184 pp. G R E G U S S , P. 1955. Identification of living gymnosperms on the basis of xylotomy. Akademia Kiado, Budapest, 263 pp. —— 1972. Xylotomy of the living conifers. Akademia Kiado, Budapest, 172 pp. H A R L A N D , M. 2002. Reconstructing polar-forest dynamics from fossil wood and computer models. Palaeontological Association Annual Meeting, Cambridge, Abstracts with Programs, p. 128. J A N E , F. W. 1962. The structure of wood. A & C Black, London, 427 pp. K O R T , I. de 1993. Relationships between sapwood amount, latewood percentage, moisture content and crown vitality of douglas fir, Pseudotsuga menziesii. IAWA Journal, 14, 413–427. L A R S O N , R. R. 1967. Silvicultural control on the characteristics of wood used for furnish. 143–150. In A N O N ( ed.). Proceedings of the 4th TAPPI Forest Biology Conference. Pulp and Paper Research Institute of Canada, Quebec. L E E , C. H. and W A N G , S. Y. 1996. A new technique for the demarcation between juvenile and mature wood in Cryptomeria japonica. IAWA Journal, 17, 125–131. M E I J E R , J. I. F. 2000. Fossil woods from the Late Cretaceous Aachen Formation. Review of Palaeobotany and Palynology, 112, 297–336. M O R G A N S , H. S. 1999. Lower and Middle Jurassic woods of the Cleveland Basin (North Yorkshire), England. Palaeontology, 42, 303–328. O S B O R N E , C. P. and B E E R L I N G , D. J. 2002. A processbased model of conifer forest structure and function with special emphasis on leaf life span. Global Biogeochemical Cycles, 16, 1097, doi: 10.1029/2001GB001467. P O O L E , I. 1996. Conifer twigs from the London Clay (Eocene) of southeast England. Review of Palaeobotany and Palynology, 94, 25–37. —— 2000a. Fossil angiosperm wood: its role in the reconstruction of biodiversity and palaeoenvironment. Botanical Journal of the Linnean Society, 134, 361–381. —— 2000b. Variation: nature’s spanner or an unrecognized tool? Palaios, 15, 371–372. —— and C A N T R I L L , D. 2001. Fossil woods from Williams Point Beds, Livingston Island, Antarctica: a Late Cretaceous southern high latitude flora. Palaeontology, 44, 1081–1112. R O Y , S. K. and H I L L S , L. V. 1972. Fossil woods from the Beaufort Formation (Tertiary), northwestern Banks Island, Canada. Canadian Journal of Botany, 50, 2637–2648. —— and S T E W A R T , W. N. 1971. Oligocene woods from the Cypress Hills Formation in Saskatchewan, Canada. Canadian Journal of Botany, 49, 1867–1877. S A V I D G E , R. A. 1996. Xylogenesis, genetic and environmental regulation: a review. IAWA Journal, 17, 269–310.

FALCON-LANG: IMPLICATIONS OF VARIABILITY IN WOOD ANATOMY

S C H U L T Z E - M O T E L , J. 1966. Literatur u¨ber die Gattung Dadoxylon Endlicher (Araucarioxylon Kraus). Geologie, 11, 716–731. V A U D O I S , N. and P R I V E´ , C. 1971. Revision des bois fossiles de Cupressaceae. Palaeontographica B, 134, 61–86.

183

W I E S E H U E G E L , E. G. 1932. Diagnostic characteristics of xylem of the North American Abies. Botanical Gazette, 93, 55–70. Z O B E L , B. J. and B U I J T E N E N , J. P. van 1989. Wood variation: its causes and controls. Springer-Verlag, Berlin, 363 pp.