FISHERIES OCEANOGRAPHY
Fish. Oceanogr. 15:1, 67–79, 2006
Covariation between the average lengths of mature coho (Oncorhynchus kisutch) and Chinook salmon (O. tshawytscha) and the ocean environment
BRIAN K. WELLS,1,* CHURCHILL B. GRIMES,1 JOHN C. FIELD1 AND CHRISTIAN S. REISS2
Northern Oscillation Index, Oscillation, Pacific salmon
Pacific
Decadal
1
NOAA Fisheries, Santa Cruz Laboratory, 110 Shaffer Road, Santa Cruz, CA 95060, USA 2 NOAA Fisheries, La Jolla Laboratory, 8604 La Jolla Shores Dr., La Jolla, CA 92037, USA
INTRODUCTION
ABSTRACT We used the average fork length of age-3 returning coho (Oncorhynchus kisutch) and age-3 ocean-type and age-4 stream-type Chinook (Oncorhynchus tshawytscha) salmon along the northeast Pacific coast to assess the covariability between established oceanic environmental indices and growth. These indices included the Multivariate El Nin˜o-Southern Oscillation Index (MEI), Pacific Decadal Oscillation (PDO), Northern Oscillation Index, and Aleutian Low Pressure Index. Washington, Oregon, and California (WOC) salmon sizes were negatively correlated with the MEI values indicating that ultimate fish size was affected negatively by El Nin˜o-like events. Further, we show that the growth trajectory of WOC salmon was set following the first ocean winter. Returning oceantype British Columbia-Puget Sound Chinook salmon average fork length was positively correlated with the MEI values during the summer and autumn of return year, which was possibly a result of a shallower mixed layer and improved food-web productivity of subarctic Pacific waters. Size variation of coho salmon stocks south of Alaska was synchronous and negatively correlated with warm conditions (positive PDO) and weak North Pacific high pressure during ocean residence. Key words: Aleutian Low Pressure Index, El Nin˜o-Southern Oscillation, environment, growth, Multivariate El Nin˜o-Southern Oscillation Index,
*Correspondence. e-mail:
[email protected] Received 9 June 2004 Revised version accepted 11 January 2005 2006 Blackwell Publishing Ltd.
The role of regional- and basin-scale environmental variability in coho (Oncorhynchus kisutch) and Chinook (Oncorhynchus tshawytscha) salmon abundance (Hare et al., 1999) and distribution (Pearcy and Schoener, 1987) has been investigated. Yet, research on growth relative to a variable environment is lacking. It is intuitive that mortality and growth of Pacific salmon are related to environmental conditions. Indeed, recent work confirms such relationships for survival and abundance (Beamish et al., 1999; Beamish et al., 2000; Hobday and Boehlert, 2001; Mueter et al., 2002a,b; Magnusson and Hilborn, 2003). Fewer studies have examined variation in return size of salmon relative to the environment (Healey, 1982; Beamish and Mahnken, 2001; Hobday and Boehlert, 2001). The contemporary understanding of return size as it relates to basin-scale forces is based on a single El Nin˜o-Southern Oscillation (ENSO) event (Pearcy and Schoener, 1987; Johnson, 1988). No long-term studies have examined species- and population-specific spatial and temporal covariation between size at maturation and environmental time series. While local environmental factors early in life have some effect on the ultimate size of returning salmon adults (Hobday and Boehlert, 2001; Mueter et al., 2002b; Beamish et al., 2004), it is also important to quantify the effect of the environment at a spatial scale equal to the population distribution at each lifehistory stage (Levin, 1992). Indeed, adaptive management practices require that the relationships between varying environmental conditions and growth be quantified at multiple spatio-temporal scales. Specifically, by using established simplified regional- and basin-scale environmental indices (e.g. unidimensional representations of multivariate data), such as those examined here, inter-jurisdictional management models can be built. These examinations are of particular interest with respect to an apparent
doi:10.1111/j.1365-2419.2005.00361.x
67
68
B.K. Wells et al.
negative correlation between the abundance and survival of salmon species north and south of Vancouver Island, British Columbia, where there is a transition zone between the upwelling and downwelling regions of the northeast Pacific coast (Ware and McFarlane, 1989; Hare et al., 1999; Mueter et al., 2002a). Although coho and ocean-type and stream-type Chinook salmon have distinctly different life histories, they share some similarities. We use hatchery fish in this study, but hatcheries attempt to mimic wild fish early life-history patterns (Healey, 1991; Sandercock, 1991). Most coho and stream-type Chinook salmon are released from the hatchery in their second year, at which time they pass through the estuary and reside in near-shore marine waters until late summer or early autumn. By the end of autumn, most have migrated to the open ocean and into Alaskan gyre waters to feed. By contrast, ocean-type Chinook salmon are released in their first year, and are believed to migrate to estuarine waters where some stocks may reside for up to 6 months. Afterward, the majority are believed to grow to maturation in coastal waters and the remaining proportion in open ocean and Alaskan gyre waters (Healey, 1991). The similarities and differences across species and life-history types afford us the opportunity to evaluate the covariation between growth and the environment at different temporal and spatial scales. The objective of this study is to determine the degree of covariation between commonly referenced environmental indices and size at return for coho salmon and ocean-type and stream-type Chinook salmon populations. These relationships can help to elucidate the mechanisms driving size-related mortality, forecast potential escapement biomass, and model variation in average fecundity. As an illustration of its importance, Johnson (1988) showed that the impact on growth of a single event, the 1982–83 ENSO, resulted in a nearly 30% reduction in the fecundity of returning Oregon coho salmon. Otherwise, work has focused on the effects of the environment on populations before and during the first ocean winter (1OW; but see Quinn et al., 2004). After 1OW, Pacific salmon are distributed across a larger geographic range and, therefore, larger-scale indices must be used to evaluate how growth and environment covary. To this end, our examination evaluates the utility of established environmental indices to model growth. To accomplish our objectives, we used data on the size at return for different regional groups of salmon obtained from the coded wire-tagging (CWT) database available from the Pacific States Marine Fisheries Commission Regional Mark Processing Center (PSMFC RMPC). We related the average fork length
of the escapement hatchery populations to four environmental indices including the Multivariate El Nin˜oSouthern Oscillation Index (MEI; Wolter and Timlin, 1998), Pacific Decadal Oscillation (PDO; Mantua et al., 1997), Northern Oscillation Index (NOI; Schwing et al., 2002), and Aleutian Low Pressure Index (ALPI; Beamish and Bouillon, 1993; Beamish et al., 1997). We strengthen our interpretations further by examining these populations across their entire northeast Pacific ocean ranges. MATERIALS AND METHODS Average escapement fork length We extracted the average fork length of escapement collections from CWT coho and Chinook salmon from brood years beginning as early as 1971 (PSMFC RMPC; http://www.rmis.org/index.html last accessed February 2004). We restricted our samples to age-3 coho returns and two-ocean-winter (2OW) Chinook salmon returns (age-3 ocean-type Chinook and age-4 stream-type Chinook salmon). We used age-1+ coho and stream-type Chinook salmon and age-0+ oceantype Chinook salmon release ages. To minimize spatio-temporal biases associated with patterns of release, we only used hatchery releases from regions that had a long-term record of releases with no more than one break in the time series of no more than 2 yr (Table 1). These release region groups, and their subsequent returning fish, were grouped into eco-regions (Table 1). These eco-regions included Alaska (AK), British Columbia-Puget Sound (BC-Puget), Columbia River (Col; to separate the interior group), and Washington, Oregon, and California (WOC) following coherent patterns of ocean survival as reported by Magnusson (2002). Because of inconsistent return composition, often associated with short gaps in tagged releases or small release batches, we could not use principal component analysis (or similar statistical approaches) to consolidate release region groups. For example, note the limited representation of Western Vancouver Island recoveries (Table 1). Previous work has demonstrated that the a priori arrangement that we chose consolidates coho salmon populations with synchronous recruitment dynamics (Hobday and Boehlert, 2001) and also reflects patterns in larger-scale environmental forces such as oceanic circulation and terrestrial dynamics (e.g. river length, flow, and estuarine habitat; Ware and McFarlane, 1989). However, catch numbers for WOC Chinook salmon stocks are not synchronous across ports of landing (Botsford and Lawrence, 2002). In this study, spatial and temporal 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
Covariation between salmon size and environment
69
Table 1. Data summary for coded wire-tagged age-3 coho and two ocean-winter ocean-type and stream-type Chinook salmon collected along their range. The columns include species, life-history type (ocean or stream), the eco-region, the years for which we had returns to the eco-region, the total number of fish used within the eco-region, the individual release regions within each eco-region from which hatchery fish were released, the brood years represented for each release region, and the proportion of the total return to the eco-region made up by fish from each release region. Species
Type
Eco-region
Return years
N
Release regions used
Brood years
Proportion of return
Coho
– –
AK BC-Puget
1981–2002 1974–2002
23 679 158 737
–
Col
1975–2001
155 792
–
WOC
1974–75/1977–2002
186 904
Ocean
BC-Puget
1974–78/1981–2001
21 008
Ocean
Columbia
1974–76/1978–2001
18 575
Ocean
WOC
1979–2002
13 878
Stream Stream
AK Col
1982–2002 1975–2001
7003 63 390
Southeast Alaska Western Vancouver Island Johnstone Strait Georgia Strait Strait of Juan de Fuca Mid Puget Sound Stillaguamish R.-Snohomish R. Nooksack R. Hood Canal Upper Columbia R. Central Columbia R. Lower Columbia R. North Coastal Washington Grays Harbor North Coastal Oregon South Coastal Oregon Western Vancouver Island Johnstone Strait Georgia Strait Strait of Juan de Fuca Mid Puget Sound South Puget Sound Nooksack R. Hood Canal Upper Columbia R. Central Columbia R. Lower Columbia R. North Coastal Washington North Coast Oregon South Coastal Oregon Sacramento R. San Joaquin R. Southeast Alaska Snake R. Upper Columbia R. Central Columbia R. Lower Columbia R.
1978–99 1972–99 1974–99 1971–99 1971–99 1971–99 1971–99 1973–99 1971–99 1976–98 1972–99 1972–98 1971–99 1971–99 1973–99 1975–99 1971–98 1971–98 1971–98 1971–98 1971–98 1971–98 1971–98 1971–98 1971–98 1972–98 1971–98 1976–99 1976–99 1976–99 1976–99 1976–99 1978–98 1974–97 1974–97 1973–97 1971–97
1.00 <0.01 <0.01 <0.01 0.05 0.36 0.28 0.09 0.22 <0.01 0.04 0.96 0.19 0.20 0.20 0.41 <0.01 <0.01 <0.01 0.02 0.59 0.15 0.10 0.13 0.14 0.38 0.48 0.26 0.05 0.42 0.19 0.08 1.00 0.12 0.15 0.35 0.38
Chinook
Chinook
patterns of sample sizes did not accommodate a finer geographic resolution of analysis (e.g. separating returns north and south of Cape Blanco; Brodeur et al., 2004). However by combining populations along the California Current System, we followed the largerscale oceanographic conditions (Ware and McFarlane, 1989) and we include the region managed by the Pacific Fishery Management Council, a United States regional marine fishery management system 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
established by the Magnuson Fishery Conservation and Management Act 1976 (public law 94-265). To mimic natural populations, releases of juveniles occurred March to June (mean of May with a 1-month standard deviation) for all groups except for oceantype WOC and ocean-type Col Chinook salmon, which were released March to November (mean of June with a 2-month standard deviation). California juvenile releases were also included in the WOC, yet
70
B.K. Wells et al.
only one population, ocean-type Chinook salmon, had consistent releases of juveniles across the brood years examined. Alaska and BC-Puget are represented by fish from only the southern regions of each eco-region. We restricted our analysis to the escapement samples (i.e. no fisheries landing data were included) to reduce the effect of variation in fishing regulations such as size limits and quotas on age-specific forklength estimates. In doing this, we assume any effect on variation in size distribution driven by regulations and selectivity in the fishery would result in noise in the data because the fishery is unlikely to vary at the same temporal and geographic scale as the indices we examined (linear trends were removed from the data before analysis). In our analyses we used the average fork length for each population, except Alaska fish, for which only mid-eye fork length was reported. The average return length for each release region was calculated and then averaged within eco-regions (Table 1). Thus, our analysis used means of means. In doing this we represent, as best as possible, the average return length of fish across the entire eco-region and adjust for unequal representation across regions through the years. By taking this approach, as opposed to using the average of individuals within an ecoregion, we reduced the variation across the time series. A preliminary examination of our analysis using the averages of individuals within eco-regions resulted in slightly greater variation and greater correlations with environmental indices, indicating that using a mean of means is the more conservative approach. We also evaluated the effect of length at release on return length using least-squares regression on the subset of the data for which the average release length was reported. The average release size by batch was the independent variable used to predict average return size of each batch (coho AK-N ¼ 230, coho BC-Puget-N ¼ 163, coho Col-N ¼ 217, coho WOC-N ¼ 137, Chinook ocean-type BC-Puget-N ¼ 115, Chinook ocean-type Col-N ¼ 159, Chinook ocean-type WOC-N ¼ 119, Chinook stream-type AK-N ¼ 130, and Chinook stream-type Col-N ¼ 386). A significant predictive relationship between release size and return size could complicate a simple correlative analysis like that presented here. We evaluated the effect of variation in the sex ratio (M/F) of returning fish on the relationship between average length at return and the environment. Yearly estimates of sex ratio were as complete as return length data for all groups except AK for which 1987 through 1991 sex ratios for Coho salmon and values before 1991 for Chinook salmon were unavailable. First, we evaluated the relationship between average return
length and sex ratio for each group. For all groups that had a significant correlation between sex ratio and average length we determined if the sex ratio was correlated with the same environmental indices as average length at return. Regardless of the relationship between average return length and sex ratio, if the sex ratio did not relate to the same indices as average length then any variation in sex ratio across the time series would only add variation and not bias. We used times-series analysis for all groups except AK, and because of the limited time series available for AK, we ignored any non-independence between years. Environmental indices We determined the degree of shared variation between MEI, NOI, PDO, and ALPI with growth. While these indices are correlated and there are causative relationships between the environmental factors that the indices report, each represents a different suite of environmental factors at different spatial and temporal scales. The MEI is primarily an equatorial index that is based on six variables (sea-level pressure, zonal and meridional components of surface wind, sea-surface temperature, surface air temperature, and cloudiness). It represents near-shore conditions as they relate to mixed-layer depth, upwelling conditions, freshwater input, and primary production (Wolter and Timlin, 1998). Because ENSO events (represented here by MEI) are defined by equatorial conditions, their impact is minimized northward from the equator. For example, while ENSO events have a dramatic effect along the California Current System they do not have an effect in the Gulf of Alaska (Hollowed et al., 2001). The NOI is calculated as the difference in sea-level pressures between the North Pacific and near Darwin, Australia and represents the strength of the North Pacific high pressure cell. The NOI characterizes the North Pacific manifestation of tropical basin-scale variability, and values are related proximately to coastal conditions south of Vancouver Island (Schwing et al., 2002). The PDO index is derived from the leading principal component of sea-surface temperatures north of the 20N latitude and, therefore, it reveals longer-term patterns in the open ocean (although any monotonic trend is removed). In contrast to the MEI, PDO values represent conditions in northern waters with decreasing representation southward (Hollowed et al., 2001). Finally, the ALPI represents the strength of the Aleutian low-pressure system, which in turn, has a strong influence on subarctic wind stress, mixed-layer depth, and production of Alaskan salmon (Beamish and Bouillon, 1993). Generally, positive MEI values indicate ENSO-like 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
Covariation between salmon size and environment
conditions, positive NOI values indicate stronger North Pacific high conditions, positive PDO values indicate warm-water conditions along the entire northeast Pacific coast, and positive ALPI conditions indicate a stronger Aleutian low-pressure system, warmer conditions, and stronger Alaskan gyre. Statistical analysis We used cross-correlation analysis to examine covariation between environmental indices and the mean size of the individual groups in the escapement sample. We examined MEI, PDO, NOI, and ALPI for the years 1974–2002. To reduce auto-correlation and remove trends related to fishing alterations, all biological and environmental time series were fitted to a linear least-squares regression with years as the independent variable, and the residuals were used for the analysis. This detrending resulted in removing monotonic ocean warming trends (Levitus et al., 2001) but was required not only to reduce auto-correlation but also for the calculation of degrees of freedom. Following methods in Pyper and Peterman (1998) we determined there was no need for a reduction in the degrees of freedom for any of the bivariate analyses. We used an a ¼ 0.05 to judge significance of crosscorrelations. Note, in three cases there was a short break in the return fork-length time series: WOC coho salmon, BC-Puget ocean-type Chinook salmon, and ocean-type Col Chinook salmon. For each break we used the detrended mean value to fill the gap before conducting any analyses. When month-specific values were available we grouped the index values into three-month seasons, e.g. the winter period included December–February, etc. These seasons represent different growth stanzas in early life. For instance, in spring, some fish move out of freshwater and arrive in coastal waters where they may grow as much as fourfold before the onset of summer (Godfrey et al., 1975). During summer, growth in these near-shore waters can be >1 mm day)1 (Godfrey et al., 1975). However, for fish that remain in estuarine waters through summer, as occurs in some oceantype Chinook salmon, growth may be retarded because of warm and crowded conditions (Reimers, 1973; Marine and Cech, 2004). These fish enter the ocean environment in autumn. Finally, in the 1OW sizedependent mortality sets recruitment (Beamish and Mahnken, 2001; Beamish et al., 2004). After 1OW, these seasons represent growth and migration cycles. Theoretically, cross-correlation can detect the previous effects of the physical environment by resolving lagged correlations. For instance, significant events that occurred during age 2 can be observed 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
71
because, in the analysis, one series is lagged over the other. If the environmental series must be pushed forward 1 yr to match the fork-length series for an escapement population we can infer that the effect of that environmental forcing factor was set during the second, and not the third, year of life (return year of age-3 fish). We lagged the environmental indices over the length series and considered lags )3 to 3. RESULTS Adult size at return was significantly related to release size for all coho salmon groups (P < 0.05) and for the ocean-type Col Chinook salmon, but the amount of variation in the model accounted for by release size was low (r2 ¼ 0.22, 0.17, 0.09, and 0.07 for Col, AK, WOC, and BC-Puget coho salmon and r2 ¼ 0.07 for ocean-type Col Chinook salmon). Residual analysis confirmed that the linear model was appropriate. Importantly, release size was not correlated (timeseries approach) with any of the environmental variables examined here for any eco-region group. Therefore, the effect of release size to return size resulted in increased variation rather than bias. The correlation between average return length and sex ratio (M/F) was significant for a few eco-region groups but did not affect our interpretation of the relationship between average return length and the environment. Specifically, average return length and sex ratio were correlated for Col Coho (r ¼ )0.59), WOC Coho (r ¼ )0.59), ocean-type Col Chinook (r ¼ )0.58), AK coho (r ¼ 0.52), and AK stream-type Chinook salmon (r ¼ )0.61). Of these eco-regions the only significant correlation that was shared by average return length to the environment and sex ratio at return to the environment was the relationship between Col coho salmon and ALPI. Therefore, only our results as they relate to return size of Col coho salmon to ALPI may be biased by the relationship with sex ratio. The average size of returning age-3 coho and 2OW Chinook salmon varied considerably over the years (Fig. 1). The average fork length of returning coho salmon populations south of Alaska was positively correlated between populations, suggesting that growth of these covarying groups was driven by similar mechanisms (Table 2). Chinook salmon did not show any patterns of covariation between populations. Through the years 1974–2002 most of the environmental series were correlated with one another (Table 2). Because all indices are dependent on basinscale atmospheric and ocean conditions, this was expected. However, examination of their inter-correlations allows inference about the relation to
72
B.K. Wells et al.
Figure 1. Average fork lengths (FL; mid-eye fork length for Alaska Chinook) of escapement samples of (a) age-3 coho and 2 ocean winter (b) ocean-type and (c) stream-type Chinook from Alaska (blue), British Columbia-Puget Sound (red), Columbia River (green), and Washington, Oregon, and California (black). 800
(a)
700 600 500 400 1970
FL (mm)
800
1975
1980
1985
1990
1995
2000
2005
1975
1980
1985
1990
1995
2000
2005
1975
1980
1985
1990
1995
2000
2005
(b)
700
600 1970 800
(c)
700
600 1970
Return year
fish-length variability. The NOI was negatively correlated to the PDO, MEI, and ALPI. The MEI and PDO positively covaried, however, the MEI series preceded winter and spring PDO values (lag ¼ )1 yr). Because ENSO events may trigger switches in the PDO values (Newman et al., 2003), it is reasonable that the MEI will lead the PDO. It is important that this offset is recognized, as it is possible that fish-length series could statistically lead environmental forces (negative lags) as a result of mismatched and interrelated environmental series. When it is logical and supported by understood biological and life-history events, we assume the lag of environment to return size is a direct result of previous year/season effects – those effects acting upon fish before the final year prior to maturation. If negative lags represented the maximum correlation we report the occurrence but ignore
them in our interpretations. Therefore, we assume that negative lags are caused by fish relating more strongly to an environmental time series that lags another environmental series or unexamined force. The lengths of age-3 coho and 2OW Chinook salmon covaried with the PDO, MEI, and NOI values (Table 2). In all cases where there was significant cross-correlation, the maximum correlation exhibited a distinct peak. In the rare cases where more than one significant lag value was observed (three of 43), there were no more than two over the length of any time series. Excepting Alaska coho salmon and stream-type Col Chinook salmon (Table 2), no lags >1 yr or <)1 yr were significant for comparisons of length at return to environmental indices. Together, these results indicate, for the most part, the correlations we noted are not spurious. Generally, the strength of the correlations of the PDO, MEI, and NOI to average return length were similar across populations (r |0.40 to 0.70|). Length variation of returning coho salmon was significantly correlated to regional- and basin-scale environmental indices (Table 2). The length of age-3 BC-Puget coho salmon was negatively correlated with the PDO during spring, summer, and autumn of the return year (following 1OW; Fig. 2a) and positively to summer NOI values during return year. BC-Puget coho age-3 escapement length was also negatively correlated to the MEI and positively to winter and spring NOI values close to the period of release. Taken together, these analyses suggests that growth trajectories were established after release (as indicated by lag ¼ 1 yr), but modified within the open ocean and gyre waters (as indicated by the lag ¼ 0 yr). Col coho salmon growth during the final year at sea was negatively correlated with spring MEI values spring, summer, and autumn PDO values (Fig. 2b), and ALPI values, but was positively correlated with spring NOI values (lags all ¼ 0 yr). The WOC coho salmon size variation was negatively correlated to winter MEI and summer and autumn PDO values although there were multiple significant lags (Table 2). It seems the relationship between growth and the examined environmental indices may not be direct. However, WOC coho salmon length variation was positively and strongly correlated with spring NOI index values at a lag of 0 yr (Fig. 2c). Alaskan coho salmon correlated only with ALPI but at an inexplicable lag (lag ¼ )2). Variation in the return length of 2OW Chinook salmon was correlated with both regional- and basinscale environmental indices (Table 2). Ocean-type BC-Puget Chinook salmon length variation in age-3 length was positively correlated with summer and 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
Coho BC-Puget
Coho WOC
Chinook Ocean BC-Puget
Chinook Ocean Col
–43 (1)
–57 (0)
Chinook Ocean WOC
46 (0)
Chinook Stream AK
–45 (1)
–41 (0)
–49 (1)
MEI (Dec–Feb)
MEI (Mar–May)
–1
MEI (Jun–Aug)
–40 (1)
–1
50 (1)
–1
2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79. –45 (0) –66 (0) –57 (0) –47 (0) –50 (0) –57 (0) –47 (0)
53 (1) 76 (0)
58 (1) 61 (0) PDO (Jun–Aug)
PDO (Sep–Nov)
PDO (Mar–May)
–72 (0)
48 (1)
NOI (Sep–Nov)
59 (0)
63 (1)
–42 (1) –46 (1)
NOI (Jun–Aug)
NOI (Mar–May)
75 (0)
53 (0)
–42 (1)
–78 (–1) –61 (–1) –59 (0) –73 (0)
NOI (Dec–Feb)
72 (1)
56 (0)
–46 (–1)
–54 (0)
52 (0)
39 (0)
55 (–1)
53 (–1)
48 (0)
52 (0)
–51 (0)
–2
ALPI
In the top panels black cells represent negative correlations, gray cells represent positive correlations (P < 0.05), and empty cells indicate a lack of significant correlation. The values in parentheses represent the amount the column variable was lagged against the row variable. If more than one lag was significant it is shown in the parentheses with the one corresponding to the highest correlation listed first. For biological correlations if the only significant lag was negative, it is shown without its corresponding correlation values. The bottom panel represents correlations between environmental variables.
41 (0)
–65 (–1) –53 (0) –66 (0) –79 (0)
–69 (0) –81 (0) –39 (0) –45 (0)
72 (0)
68 (0)
55 (0)
63 (–1)
61 (0)
70 (0)
48 (–1)
61 (–1)
59 (0)
PDO (Dec–Feb)
43 (–1)
MEI (Sep–Nov)
–81 (0) –75 (0) –49 (1) –64 (1)
55 (1)
–51 (0) –43 (0)
–1
41 (0)
61 (–1)
43 (1)
–1
59 (0)
46 (0)
47 (1,0)
91 (0)
47 (0)
–48 (0)
–44 (1)
43 (1)
MEI (Jun–Aug)
62 (0)
–57 (0)
–52(0,–1) –41(0,–1)
–55 (0)
–66 (0)
72 (1)
45 (0)
–39 (0)
–1
–1
–52 (0)
PDO (Jun–Aug)
–70 (0)
PDO (Sep–Nov)
77 (1)
89 (1)
–48 (2) –48 (1)
46 (0)
PDO (Mar–May)
–44 (0)
NOI (Dec–Feb)
66 (0)
–43 (1)
51 (0)
PDO (Dec–Feb)
–1
NOI (Mar–May)
86 (0)
–57 (1)
–51(–1,0)
–50 (0)
MEI (Sep–Nov)
–44 (2)
NOI (Jun–Aug)
MEI (Mar–May)
MEI (Dec–Feb)
–39 (1)
Chinook Stream Col
Chinook Stream Col
Chinook Stream AK
Chinook Ocean WOC
Chinook Ocean Col
Chinook Ocean BC-Puget
Coho WOC
56 (0)
40 (0) 53(1,0)
Coho Col
Coho Col
Coho BC-Puget
Coho AK
NOI (Sep–Nov)
Table 2. Maximum cross-correlations (·100) for the fork lengths of age-3 coho and 2 ocean winter Chinook returns to each other and environmental indices.
Covariation between salmon size and environment 73
74
B.K. Wells et al.
Figure 2. Standardized fork length (FL) of coho salmon returns for 1974–2002 to (a) southern British Columbia and Puget Sound, (b) Columbia River, and (c) Washington, Oregon, and California shown as the black time series. The gray time series represent the standardized index values that most highly correlated to the return-length time series. These indices were (a) negative Pacific Decadal Oscillation (PDO) for summer, (b) negative PDO for autumn, and (c) Northern Oscillation Index for spring. 3
(a)
2
Figure 3. Standardized fork length (FL) of ocean-type Chinook salmon returns for 1974–2002 to (a) southern British Columbia and Puget Sound, (b) Columbia River, and (c) Washington, Oregon, and California shown as the black time series. The gray time series represent the standardized index values that most highly correlated to the return-length time series. These indices were (a) Multivariate El Nin˜oSouthern Oscillation Index (MEI) for summer (solid gray) and negative Northern Oscillation Index for summer (dashed gray), (b) negative Pacific Decadal Oscillation for winter lagged forward 1 yr, and (c) negative MEI for spring lagged forward 1 yr.
1
3
0
2
–1
1
–2
0
–3 1970
1975
1980
1985
1990
1995
2000
–1
2005
–2 3
(b)
2 1 0 –1 –2 –3 1970 3
1975
1980
1985
1990
1995
2000
2005
(c)
2 1
Standardized FL and index value
Standardized FL and index value
(a)
–3 1970 3
1980
1985
1990
1995
2000
2005
1975
1980
1985
1990
1995
2000
2005
1975
1980
1985
1990
1995
2000
2005
(b)
2 1 0 –1 –2 –3 1970 3
0
1975
(c)
2
–1
1 –2
0 –3 1970
1975
1980
1985
1990
1995
2000
2005
Return year
–1 –2 –3 1970
Return year
autumn MEI and negatively correlated with summer and autumn NOI values at a lag of 0 yr, suggesting that it was during the return year that size at age 3 was determined (Fig. 3a). Ocean-type Col Chinook salmon were weakly correlated to the PDO in winter at a lag of 1 yr and spring at a lag of 0 yr, suggesting that the indices are poor indicators of return length of ocean-type Col Chinook salmon (Fig. 3b). Oceantype WOC Chinook salmon size at age 3 was negatively correlated with winter, spring, and summer MEI (Fig. 3c) and summer PDO values, but positively correlated with winter and spring NOI values, all at a lag of 1 yr. These results indicate that events occurring following 1OW set the size at return. Stream-type AK
Chinook salmon were negatively correlated with PDO values during the final summer (Fig. 4a). Stream-type Col Chinook salmon were related to fall MEI values at a lag of 2 yr and winter PDO (Fig. 4b) values at a lag of 1 yr, indicating the relationship of environment to return size was not straight forward. DISCUSSION The results of this work can be used to describe the variation in growth of salmon in the oceanic environment (Fig. 5). Size at return of all WOC salmon 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
Covariation between salmon size and environment
Figure 4. Standardized fork length (FL) of stream-type Chinook salmon returns for 1975–2002 to (a) southeastern Alaska and (b) Columbia River as the black time series. The gray time series represent the standardized index values that most highly correlated to the return-length time series. These indices were (a) negative Pacific Decadal Oscillation (PDO) for summer and (b) negative PDO for winter lagged forward 1 yr. 3
(a)
2
Standardized FL and index value
1 0 –1 –2 –3 1970 3
1975
1980
1985
1990
1995
2000
2005
1975
1980
1985
1990
1995
2000
2005
(b)
2 1 0 –1 –2 –3 1970
Return year
groups was negatively correlated with MEI values following 1OW indicating that fish along the United States Pacific coast were affected negatively by ENSOlike conditions. This was most apparent for WOC Chinook salmon from the 1984 run year and WOC coho salmon from the 1983 run year (Fig. 1). This finding suggests that the growth trajectory of the WOC salmon was set after 1OW, which is further supported by the result that only one lag was significant. In contrast to WOC salmon, BC-Puget Chinook salmon average return size was positively correlated to MEI and negatively to NOI values in the final year at sea, which is likely the result of residence in the more productive Alaskan coastal and gyre waters (Healey, 1991; Polovina et al., 1995). Average size at return of AK stream-type Chinook salmon was negatively correlated to PDO values, that is, they were correlated to environmental indices in the same fashion as WOC Chinook salmon. Growth variation of coho salmon stocks south of Alaska was negatively correlated with warm conditions (positive PDO) during ocean residence. We can demonstrate some general patterns regarding the interaction between environmental factors and the life history of coho salmon from 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
75
upwelling and downwelling domains of the northeast Pacific (Fig. 5). For each coho salmon population south of Alaska, generally the year of return was the period during which size at return was set. This suggests that once size-related mortality occurred at ocean entry (Beamish et al., 2004), the ultimate size at return of coho salmon was determined after 1OW. The PDO covaried with coho growth for each population south of Alaska, and the MEI was well correlated with WOC coho salmon and modestly correlated with Col and BC-Puget coho salmon. These findings are well matched with the geographic scale of the indices. While we cannot discount entirely the impact of shorter time series (e.g. less than three PDO cycles), we argue that the ultimate size of returning AK coho salmon was not related to the environmental indices examined here. This result supports the findings of Kaeriyama et al. (2004) suggesting that salmon in high-seas Alaskan waters adapted to environmental variation by switching diet as the composition of prey varied. Taking both Kaeriyama et al. (2004) and our findings together, we infer that it is not the conditions all coho salmon experience while in the open ocean and gyre waters that regulates return size, but rather, it is the conditions experienced during return from Alaskan waters. That WOC and Col coho salmon size at return were correlated to MEI and NOI values during the time of return supports this assertion. However, we discounted possible bias introduced into these relationships by variation in fishing pressure across the times series. This bias could have been manifested as patterns similar to those which we note for Alaskan fish returning to spawn. Specifically, if there was a large degree of interannual variation in the size selectivity of the fishery on returning Alaskan fish, the relationship between size at return and the environmental indices examined here may not be apparent. However, we did correct for time trends in size at return (detrending), which removes long-term linear trends in size selectivity of the fishery leaving only the variation in size at age (residuals) for the analysis. In the end, our results imply that management should incorporate the returnyear PDO values in models for all coho salmon groups south of Alaska and the return-year NOI and MEI values for WOC and Col coho salmon to improve estimates of size at return and its correlated variables (e.g. fecundity). As an example, the detrended size at return values for WOC coho salmon had a difference of 10 cm between the two return years 1977, with high NOI values, and 1983, with low NOI values (Figs 1a and 2c). General patterns about the effect of the ocean environment on size at return are less apparent for
76
B.K. Wells et al.
Chinook salmon than coho salmon because Chinook salmon have a more complex life history. The oceantype BC-Puget Chinook salmon return size was dependent on conditions in the final year at sea, while ocean-type WOC Chinook salmon return size was set approximately at the time immediately following 1OW. These findings suggest size at return of northern populations may result from growth during the final year at sea while resident in Alaska coastal and gyre waters. However, ocean-type WOC Chinook salmon which reside predominantly in CCS coastal waters (Healey, 1991) are affected by conditions earlier in life. Interestingly, ocean-type BC-Puget and oceantype WOC Chinook salmon were correlated oppositely to the same indices. This is similar to trends in abundance north and south of the transition zone (Hare et al., 1999). However, while the BC-Puget and WOC Chinook salmon groups have opposing correlation to the same indices, both ultimately respond to production regimes in the same way, but at different points in their life history (i.e. ocean entry as opposed to ocean feeding). Specifically, primary, secondary, and salmon production in Alaska is inversely related to that in CCS waters (Fig. 5). These results suggest that environmental models should be developed separately for the two groups. Alaska Chinook salmon size at return was negatively correlated to spring PDO values during the final year at sea, which is counter to what is understood about primary, secondary, and salmon production in Alaska waters (Fig. 5). It is possible our results are spurious for AK Chinook salmon, there was bias from fishing, or Alaskan fish size at return was negatively affected by positive primary and secondary production directly. Alternatively, we suggest that there may have been an impact from a density-dependent
relationship. Density-dependent relationships have been demonstrated for coho salmon (Hobday and Boehlert, 2001), pink salmon (Oncorhynchus gorbuscha; Ricker, 1995), sockeye salmon (Oncorhynchus nerka; Ruggerone et al., 2003), and possibly the whole Oncorhynchus complex (Bigler et al., 1996), whereby trophic resources were stripped during high survival years. This explanation may also elucidate why AK coho salmon were not correlated to any of the environmental indices examined. Importantly, size at return for Alaskan fish may be modeled better using biological parameters. We cannot be confident that the patterns we describe for growth are solely environmentally mediated, because size-mediated mortality or density dependence may exist. For example, Hobday and Boehlert (2001) noted an inverse relationship between survival and ultimate size at maturation for coho salmon, which may suggest there exists a density-dependent relationship (Beamish et al., 1995). Specifically, Hobday and Boehlert (2001) observed an inverse relationship between survival and mixed-layer depth and a direct relationship between return size and mixed-layer depth. Growth was slower in conditions conducive to high survival and abundance. The MEI and NOI values are loose surrogates for the mixed-layer depth along the northeast Pacific coast, and PDO values correlate with the mixed-layer depth in the subarctic (Polovina et al., 1995). However, by design, these indices reflect larger-scale environmental conditions relative to the mixed-layer values examined by Hobday and Boehlert (2001). As noted by Hobday and Boehlert (2001), at larger scales, as one confounds the ultimate effect of the environment on the individual, the patterns of synchrony between biological and environmental factors can change. Importantly, after
Figure 5. Cartoon of the northeast Pacific with large-scale environmental and biological variation [adapted from findings and reviews of Chelton and Davis (1982), Polovina et al. (1995), Francis et al. (1998), Hare et al. (1999), and Chavez et al. (2003)]. The two maps are designated as (a) cooler conditions and (b) warmer conditions. The identifiers, cooler and warmer conditions, refer to sea-surface temperature (SST) resulting from the conditions shown. The indices shown are the Multivariate El Nin˜oSouthern Oscillation Index (MEI), Pacific Decadal Oscillation (PDO), Northern Oscillation Index (NOI), and Aleutian Low Pressure Index (ALPI). The arrows beside each index name show its sign difference from mean values across the 1974–2002 time series, and the relationships to one another are based on the results of this work (Table 2). The Aleutian Low and North Pacific High are shown as a red L and blue H. The size of the letters (L and H) is a representation of strength of the system relative to its average condition (not to one another). The Alaskan gyre is shown in its approximate location, and the font size indicates relative strength of the current. The green arrows refer to the Alaska and California Currents. The thickness of the arrows is a representation of strength of the current relative to its average condition (not to one another). In the blue box are the resulting physical changes (green) and biological changes (brown) resulting from each system state (cooler or warmer) to the coastal California Current System (CCS), coastal Gulf of Alaska (GOA), and Alaskan gyre waters. Arrows beside each variable indicate the relative direction of change in conditions. Each population of salmon considered in this study is shown along the coast. Coho salmon are labeled as c, ocean-type Chinook salmon are labeled as oc, and stream-type Chinook are labeled as sc. Arrows indicate direction of change in size at return under the considered condition based on results of Table 2 (‘–’ indicates no significant correlations reported). 2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
Covariation between salmon size and environment
1OW, as the fish distribute across a larger geographic region, larger-scale indices are required to resolve variation in growth. For management and modeling purposes it is important to quantify all appropriate spatial and temporal patterns that drive the dynamics of returning populations. While we did not examine
(a)
Cooler conditions MEI PDO ALPI
the correlation of survival and abundance to the largescale indices used here, existing work (e.g. Hare et al., 1999; Botsford and Lawrence, 2002; Logerwell et al., 2003) shows that the relationship of survival to these environmental forces is similar to that which we noted for fish size.
140
o
NOI c-
AK sc
L
BC-Puget c oc
gyre Vancouver Island
CCS
GOA
gyre
Upwelling Sea surface temp (north) Stream flow (south) Stratification Mixed layer depth Nutrients Primary production Zooplankton Salmon production
(b)
Warmer conditions MEI PDO ALPI
50o
c
H
250 500
1,000
Kilometers
NOI AK sc
gyre
BC-Puget c oc Vancouver Island
CCS
Col oc sc WOC c oc
0
c-
L
GOA
Upwelling Sea surface temp (north) Stream flow (south) Stratification Mixed layer depth Nutrients Primary production Zooplankton Salmon production
2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
77
gyre
H
c
Col oc sc WOC c oc
Canada USA Mexico
N
78
B.K. Wells et al.
Chinook salmon have a broader age structure at return than coho salmon. This variation in age at return suggests that age-specific growth rates determined from returning fish (as done here) may vary little across years because the growth rate is partially responsible for determining the probability of returning at any given age. Indeed, cursory examination of the age structure of Chinook salmon, relative to the same environmental indices examined here, suggests that the diversity and average age of returning fish may result from balancing growth and natural mortality of aging fish (B.K. Wells and C.B. Grimes, unpublished data). We have noted broadening of age distribution most dramatically for stream-type Columbia river fish. Concomitant with the broadening age distribution, these Chinook salmon had the least variation in size at age 3 across years of all the fish examined (Fig. 1c). Such diversity in age at return can reduce variability in recruitment by expanding reproductive investment of large cohorts (Secor, 2000). Further, results presented in Fig. 5 of Magnusson and Hilborn (2003) show that over the last 30 yr, hatchery Chinook salmon population dynamics have been driven by relatively few successful, abundant cohorts. During the same time period coho salmon showed less punctuated survival (Magnusson and Hilborn, 2003). We have diverged from the primary target of discussing growth to impart how significant the results presented herein are, given that Chinook salmon have adapted to environmental variation with an unstable age structure. In summary, we have demonstrated that size at return of salmon along the northeast Pacific coast is related to large-scale ocean environmental forces. To date, attention has been directed largely at survival models, but variation in growth-related vital rates (e.g. fecundity, egg size, and maturity) and how they covary with the environment must be understood to properly model escapement, recruitment, and stock biomass. Specifically, we have made a first critical pass at defining variables that may be appropriate candidates for use in adaptive management models such that growthrelated variables can be modeled in advance of fish returning. ACKNOWLEDGEMENTS We thank R. Fisher, J. Irvine, B. MacFarlane, N. Mantua, G. Watters, C. Wells and anonymous reviewers for their useful suggestions. Also, thanks to the Pacific Salmon Commission for management of the CWT database and to all the facilities that
participate in the collection of the data. Funding was provided by the NOAA Fisheries and the Environment Program. REFERENCES Beamish, R.J. and Bouillon, D.R. (1993) Pacific salmon production trends in relation to climate. Can. J. Fish. Aquat. Sci. 50:1002–1016. Beamish, R.J. and Mahnken, C. (2001) A critical size and period hypothesis to explain natural regulation of salmon abundance and the linkage to climate and climate change. Prog. Oceanogr. 49:423–437. Beamish, R.J., Riddell, B.E., Neville, C.-E.M., Thomson, B.L. and Zhang, Z. (1995) Declines in chinook salmon catches in the Strait of Georgia in relation to shifts in the marine environment. Fish. Oceanogr. 4:243–256. Beamish, R.J., Neville, C.M. and Cass, A.J. (1997) Production of Fraser River sockeye salmon (Oncorhynchus nerka) in relation to decadal-scale changes in the climate and the ocean. Can. J. Fish. Aquat. Sci. 54:543–554. Beamish, R.J., Noakes, D.J., McFarlane, G.A., Klyashtorin, L., Ivanov, V.V. and Kurashov, V. (1999) The regime concept and natural trends in the production of Pacific salmon. Can. J. Fish. Aquat. Sci. 56:516–526. Beamish, R., Noakes, D., Mcfarlane, G., Pinnix, W., Sweeting, R. and King, J. (2000) Trends in coho marine survival in relation to the regime concept. Fish. Oceanogr. 1: 114–119. Beamish, R.J., Mahnken, C. and Neville, C.M. (2004) Evidence that reduced early marine growth is associated with lower survival of coho salmon. Tran. Am. Fish. Soc. 133: 26–33. Bigler, B.S., Welch, D.W. and Helle, J.H. (1996) A review of size trends among North Pacific salmon (Oncorhynchus spp.). Can. J. Fish. Aquat. Sci. 121:593–602. Botsford, L.W. and Lawrence, C.A. (2002) Patterns of covariability among California Current, Chinook salmon, coho salmon, Dungeness crab, and physical oceanographic conditions. Prog. Oceanogr. 53:283–305. Brodeur, R.D., Fisher, J.P., Teel, D.J., Emmett, R.L., Casillas, E. and Miller, T.W. (2004) Juvenile salmonid distribution, growth, condition, origin, and environmental and species associations in Northern California Current. Fish. Bull. 102:25–46. Chavez, F.P., Ryan, J., Lluch-Cota, S.E. and Niquen, M. (2003) From anchovies to sardines and back: multidecadal change in the Pacific Ocean. Science 299:217–221. Chelton, D.B. and Davis, R.E. (1982) Monthly mean sea level variability along the west coast of North America. J. Phys. Oceanogr. 12:757–784. Francis, R.C., Hare, S.R., Hollowed, A.B. and Wooster, W.S. (1998) Effects of interdecadal climate variability on the oceanic ecosystems of the NE Pacific. Fish. Oceanogr. 7: 1–21. Godfrey, H., Henry, K.A. and Machidori, S. (1975) Distribution and abundance of coho salmon in offshore waters of the North Pacific Ocean. Int. N. Pac. Fish. Comm. Bull. 31: 80 pp. Hare, S.R., Mantua, N.J. and Francis, R.C. (1999) Inverse production regimes: Alaskan and West Coast Salmon. Fisheries 24:6–14.
2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
Covariation between salmon size and environment
Healey, M.C. (1982) Timing and relative intensity of sizeselective mortality of juvenile chum salmon (Oncorhynchus keta) during early sea life. Can. J. Fish. Aquat. Sci. 39:952– 957. Healey, M.C. (1991) Life history of Chinook salmon. In: Pacific Salmon Life Histories. C. Groot & L. Margoli (eds) Vancouver: University of British Columbia Press, pp. 311–394. Hobday, A.J. and Boehlert, G.W. (2001) The role of coastal ocean variation in spatial and temporal patterns in survival and size of coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 58:2021–2036. Hollowed, A.B., Hare, S.R. and Wooster, W.S. (2001) Pacific basin climate variability and patterns of Northeast Pacific marine production. Prog. Oceanogr. 49:257–282. Johnson, S.L. (1988) The effects of the 1983 El Nin˜o on Oregon’s coho (Oncorhynchus kisutch) and Chinook (O. tshawytscha) salmon. Fish. Res. 6:105–123. Kaeriyama, M., Nakamura, M., Edpalina, R. et al. (2004) Change in feeding ecology and trophic dynamics of Pacific salmon (Oncorhynchus spp.) in the central Gulf of Alaska in relation to climate events. Fish. Oceanogr. 13:197–207. Levin, S.A. (1992) The problem of pattern and scale in ecology. Ecology 73:1943–1967. Levitus, S., Antonov, J.I., Wang, J., Delworth, T.L., Dixon, K.W. and Broccoli, A.J. (2001) Anthropogenic warming of Earth’s climate system. Science 292:267–270. Logerwell, E.A., Mantua, N.J., Lawson, P., Francis, R.C. and Agostini, V. (2003) Tracking environmental processes in the coastal zone for understanding and predicting Oregon coho (Oncorhynchus kisutch) marine survival. Fish. Oceanogr. 12:1–15. Magnusson, A. (2002) Survival Rates of Coho (Oncorhynchus kisutch) and Chinook Salmon (O. tshawytscha) Released from Hatcheries on the U.S. and Canadian Pacific coast 1972–1998, with Respect to Climate and Habitat Effects, . Washington: University of Washington,115 pp. Magnusson, A. and Hilborn, R. (2003) Estuarine influence on survival rates of Coho (Oncorhynchus kisutch) and Chinook salmon (Oncorhynchus tshawytscha) released from hatcheries on the US Pacific Coast. Estuaries 26:1094–1103. Mantua, N.J., Hare, S.R., Zhang, J.M., Wallace, J.M. and Francis, R.J. (1997) A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Am. Meteror. Soc. 78:1069–1079. Marine, K.R. and Cech, J.J. Jr (2004) Effects of high water temperature on growth, smoltification, and predator avoidance in juvenile Sacramento River Chinook salmon. N. Am. J. Fish. Manage. 24:198–210.
2006 Blackwell Publishing Ltd, Fish. Oceanogr., 15:1, 67–79.
79
Mueter, F.J., Peterman, R.M. and Pyper, B.J. (2002a) Opposite effects of ocean temperature on survival rates of 120 stocks of Pacific salmon (Oncorhynchus spp.) in northern and southern areas. Can. J. Fish. Aquat. Sci. 59:456–463. Mueter, F.J., Ware, D.M. and Peterman, R.M. (2002b) Spatial correlation patterns in coastal environmental variables and survival rates of salmon in the North-east Pacific Ocean. Fish. Oceanogr. 11:205–218. Newman, M., Compo, G.P. and Alexander, M.A. (2003) ENSO-forced variability of the Pacific decadal oscillation. J. Climate 16:3853–3857. Pearcy, W.G. and Schoener, A. (1987) Changes in the marine biota coincident with the 1982–1983 El Nin˜o in the Northeastern Subarctic Pacific Ocean. J. Geophys. Res. 92:14417–14428. Polovina, J.J., Mitchum, G.T. and Evans, G.T. (1995) Decadal and basin-scale variation in mixed layer depth and the impact on biological production in the Central and North Pacific, 1960–88. Deep Sea Res. I. 42:1701–1716. Pyper, R.J. and Peterman, R.M. (1998) Comparison of methods to account for autocorrelation in correlation analyses of fish data. Can. J. Fish. Aquat. Sci. 55:2127–2140. Quinn, T.P., Vøllestad, L.A., Peterson, J. and Galluci, V. (2004) Influences of freshwater and marine growth on the egg sizeegg number tradeoff in coho and Chinook salmon. Tran. Am. Fish. Soc. 133:55–65. Reimers, P.E. (1973) The length of residence of juvenile fall chinook salmon in Sixes River, Oregon. Oregon Fish. Comm. Rep. 4:43 pp. Ricker, W.E. (1995) Climate change and northern fish populations. Can. Spec. Pub. Can. J. Fish. Aquat. Sci. 121:593–602. Ruggerone, G.T., Zimmerman, M., Myers, K.W., Nielsen, J.L. and Roger, D.E. (2003) Competition between Asian pink salmon (Oncorhynchus gorbuscha) and Alaskan sockeye salmon (O. nerka) in the North Pacific Ocean. Fish. Oceanogr. 12:209–219. Sandercock, F.K. (1991) Life history of coho salmon. In: Pacific Salmon Life Histories. C. Groot & L. Margoli (eds) Vancouver: University of British Columbia Press, pp. 395–446. Schwing, F.B., Murphree, T. and Green, P.M. (2002) The Northern Oscillation Index (NOI): a new climate index for the Northeast Pacific. Prog. Oceanogr. 53:115–139. Secor, D.H. (2000) Spawning in the nick of time? Effect of adult demographics on spawning behaviour and recruitment in Chesapeake Bay striped bass. ICES J. Mar. Sci. 57:403–411. Ware, D.M. and McFarlane, G.A. (1989) Fisheries production domains in the Northeast Pacific Ocean. Can. Spec. Pub. Can. J. Fish. Aquat. Sci. 108:359–379. Wolter, K. and Timlin, M.S. (1998) Measuring the strength of ENSO – how does 1997/98 rank? Weather 53:315–324.
