Martinson-Stokes-Scarnecchia April152008 Edited in red by stokes
Effects of North Pacific climatic-oceanic regimes, body size,and salmon abundanceon the growth ofsockeyesalmon, 1925-1998
Ellen C Martinson, John H Helle, Dennis L Scarnecchia, and HoustonH Stokes
Ellen C Martinson,Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA, 17109 Point Lena Loop Road, Juneau, Alaska 99801, . tel. (907) 789-6604. fax. (907) 789-6094. Corresponding Author.
John H Helle,Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA, 17109 Point Lena Loop Road, Juneau, Alaska99801, . tel. (907) 789-6038. fax. (907) 789-6094.
Dennis L Scarnecchia, Department of Fish and Wildlife Resources, University of Idaho, Moscow, Idaho83844, . tel. (208) 885-5984. fax. (208) 885-5534.
Houston HStokes, Department of Economics,University of Chicago, 601 S. Morgan Street Chicago, Illinois60607, USA. . tel.(312) 996-0971. fax.(312) 996-3344.
Abstract:To investigatehow marine growth and the relationship between marine growth and sockeye salmon abundances was influenced byclimate regimes and shifts, body size at the start of the growing season, and abundances of sockeye salmon weused multivariate adaptive regression spline threshold modelingfor a 75 year time series from 1924 to 1998. Marine growth during thejuvenile, immature, and maturinglife stagewas estimated from increments on scales of adult sockeye salmon that returned to spawn at KarlukRiver and Lake on Kodiak Island, Alaska.Intra-specific density-dependent growth was inferred frominverse relationships between growth and sockeye salmon abundance andoccurred in all marine life stages, during the cool regime, at lower abundance levels, and at smaller body sizes at the start of the juvenile life stage. A positive relationship betweenimmature growth andsockeye salmon abundances and reduced density-dependent relationship in juvenile and maturing growth during the warm regimes that favored the survival of Alaska salmon indicate that processes influencing the survival of Pacific salmon in the Central North Pacific Ocean are reflected in the scale growth of sockeye salmon. We question whether a scenario of a shift to a cold regime or extreme warm regime at higher population abundances could drastically reduce the marine growth of salmon and increase competition for resources.
Keywords: sockeye salmon, growth,density-dependent, regime
INTRODUCTION
In the last quarter of the twentieth century, considerable research was conducted to assess the effects of variations in climatic and oceanic factorson production and yield of salmonid fishes. Although studies earlier in the century had focused on dominant freshwater factors(Neave 1949, Shapovalov & Taft 1954), evidence from numerous studies suggested that broad climatic and oceanic factors had been inadequately considered (Ricker 1976).In the later part of the century, yield and survival rates of Pacific salmon (Oncorhynchus spp.) were also linked to fluctuations in the regional and basin-scale variations in climate and oceanic conditions (Royal & Tully 1961, Cushing 1971, Scarnecchia 1981, Beamish 1993, Mueter et al. 2005). Climatic and oceanic variations have also been associated with fluctuations in Atlantic salmon abundance and catches in Iceland (Scarnecchia 1984, Scarnecchia et al. 1988),Ireland (Boylan & Adams 2006), Norway and Scotland (Friedland et al. 2000).
During the twentieth century, climatic and oceanic conditions in the North Pacific underwent large fluctuations, with two distinct warm regimes (1925-46 and 1977-98) and a cool regime (1947-76) (Mantua & Hare 2002). The warm regimes were characterized by increased winter storm activity and atmospheric circulation in the North Pacific Ocean, higher precipitation in coastal regions, increased offshore upwelling of nutrient rich waters, and above-normal coastal sea-surface temperatures. The cool regime was characterized by the opposite conditions (Trenberth Hurrell 1994).
The climate and oceanic variations have been linked to concurrent variations in Pacific salmon production, which was higherin Alaskaduring the warm regimes and lower during the cool regime (Eggers et al. 2003) in that climate during the first year at sea is important in determining survival.Alaska salmon stocks fluctuate in phase with decadal scale fluctuations in North Pacific sea surface temperatures at a 1 year lag for Alaska pink salmon and 2 and 3 year lag for Alaska sockeye salmon indicating that climate during the first year at sea is important in determining survival (Mantua et al. 1997). Conversely, salmon production in Washington and Oregonresponded favorably to cool regimes in part due to increased coastal upwelling (Scarnecchia 1981). Warm regime conditions that suppressed coastal upwelling along the continental US did not favor the survival more southerly salmonstock inhabiting the region (Hare & Francis 1994, Hare & Mantua 2000).
Several studies also supportthe idea that climatic and oceanic conditions can affect salmon carrying capacity (Myers et al. 2001, Kaeriyama 2007), manifested as density-dependent survival and growth responses to food resource limitations (Salo 1988,Fukuwaka & Suzuki 2000). The density-dependent responses to increased population abundance were associated with the 1976-77 shift to a warm regime (Ishida et al. 1993, Helle & Hoffman 1995, Bigler et al. 1996), was followed by shifts to larger body size salmon from Oregon north to Western Alaska in the mid-1990s (Helle et al 2007).
From the mid-1970s to the mid-1990s, theincreases in overall salmon production coincided with the decreased growth, decreased size at maturity, and increased age at maturity of many North American salmon populations(Ishida et al. 1993, Helle Hoffman 1995, Bigler et al. 1996). In situ density-dependent growth was observed by inverse relationships between local densities of salmon and individual’s body weight, feeding rates, and the volume of prey in stomachs (Fukuwaka & Suzuki 2000, Kaeriyama et al. 2000, Ishida et al. 2002). In natural conditions, density-dependent growth may be manifested through competition for food within and among salmon species.Behavioral responses to competition that can reduce growth include reduced feeding rate, switching from high- to low-quality prey, and changing predator and prey distributions (Tadokoro et al. 1996,Azumaya Ishida 2000,Davis et al. 2000, Fukuwaka Suzuki 2000).Climatic and oceanic variations can also potentially influence density-dependent competition by altering salmon distribution (Rogers 1980), changing the latitudinal boundary of the summer feeding zones (Aydin et al. 2000), and increasing overlap in the diets of O. nerkaand pink O. gorbuschasalmon with chum O. keta and coho O. kisutchsalmon (Kaeriyama et al. 2004).
The potential for intra- and inter-specific competition among Pacific salmon stems from their high degree of overlap in distribution and feeding in the marine environment. Juvenile salmon distribute in coastal continental shelf waters during the summer growing season (Myers et al. 1996). Diet overlap among the five anadromous salmon species is highest among sockeye and pink salmon (Auburn & Ignell 2000). As juveniles, pink and sockeye salmon fed primarily on euphausiids in nearshore habitats, fish on the shelf, and euphausiids on the slope. Immature sockeye from central and southern Alaska distribute and feed with other salmon from North America and Asia in the Central North Pacific Ocean (Kaeriyama et al. 2004). In offshore waters, the major prey items of sockeye salmonincluded euphausiids, copepods, hyperiid amphipods, and large squid; and large squid for pink salmon (Davis 2003). Maturing sockeye salmon from southern Alaska distributed more eastward and fed primarily with immature and maturing salmon in offshore waters, and with juvenile salmon in coastal waters as they return to their natal stream to spawn (Kaeriyama et al. 2004).
To investigate if climatic and oceanic variations and regimes, salmon populationsizes, and body size at the start of the growing seasoninfluence the marine growth of salmon at varied life history stages, we examined scale growth of adult sockeye salmon O. nerka from the Karluk River, Kodiak Island, Alaska over a 74-year period in relation to marine abundance of sockeye salmon in central and southeast Alaska based on harvest statistics (1925-1998). Understanding the density-dependent interactions among sockeye salmonduringthe marine juvenile, immature, and maturing life history stagesand among ocean regimes will provide insight into the influence of climate change on the carrying capacity of salmon in the North Pacific Ocean.
MATERIALS AND METHODS
Although actual fish length information was not available from salmon collected at sea, scales had been collected over the period 1925 to 1998 (with 7 years of missing data: 1945, 1947, 1958, 1965, 1966, 1969, and 1979) from the age 2.2 sockeye that returned to Karluk Lake on Kodiak Island, Alaska. Age was designated using the decimal method by Koo (1962) where the number to the left of the decimal is the number of winters spent in fresh water after emergence from the gravel and the number to the right of the decimal is the number of winters spent in saltwater. For example, age 0.3 represented a four year old fish. Marine grow was estimated from measurements on the scale.
Scale samples and preparation--For each year, from 30 to 50 scales per year were selected at equal time intervals though out the collection from the early run (May 1-July 21) spawning migration. From historical records, scales had been taken from the sockeye at a few rows above the lateral line and below the posterior insertion of the dorsal fin using a scrape method (1925-51)and forceps (1952-98) and assumed low variability in the body location sampled for scales among years (Scarnecchia 1979; Clutter and Whitesel 1956). One scale per fishhad beenplaced onto gummed cards with the reticulated side facing away from the card andimpressed ontoanacetate card using a hydraulic press at 100°C and224 psifor 3 minutes (Arnold 1951).
Scale impressions were viewed and scanned using an Indus microfiche reader Model 4601-11 with a 24 objective lens. Images of scales were copied from the reader screen with the Screenscan Microfiche PC Model high-resolution scanner hardware and saved as TIFF files using the ScreenScan Application software, version 1.00.0.8. Images were then imported into the Optimate image analysis software for measuring.
Scale measurements -- In using scale measurements to estimate marine growth of salmon, we assumed that a) growth along a specified radius of the scale was proportional to the growth in fish length(Dahl 1909), and b) the distance between adjacent annuli on a scale depicted one year of somatic growth (Fukuwaka & Kaeriyama 1997).
Scales were read for age and measured by the lead author. Scale measurements were taken along a reference line drawn from the focus to the edge of the scale along the longest anterior radial axis in millimeters (Narver 1968). Measurements were adjusted to the original scale size by dividing by 24. One scale was measured per fish and 30 to 50 scales were measured per year (N=69 years) for a total of 3,116 scales.
Growth during each year of marine residencewas estimated from the measured distances between adjacent annuli onthe scaleimage (Fig. 1). Total freshwater growth (FW), an indicator for body length at the start of the first marine year as juveniles, was estimated as the distance from the center of the focus to the center of the space between the last freshwater circulus and the first marine circulus. Growth in the first marine year (M1), an indicator of total growth during the juvenile stage,was estimated as the distance fromthe space between the last freshwater circulus and first marine circulus to the leadingedge of the first marine annulus. Second-year marine growth (M2), an indicator for immature growth, was estimated as the distance from the leading edge of the first marine annulus and the leading edge of the second marine annulus. Third-year marine growth (M3), an indicator for maturing growth, was estimated as the distance from the leading edge of the second marine annulus to the outer edge of the scale.Scales with reabsorbed edges and evidence for being regenerated were not measured.Mean values for M1, M2, and M3 growth were calculated for each brood. Mean growth for the seven years of missing scale data were estimated as points along a local ordinary least squared smoothing linefit to the datato satisfy the statistical analysis requirement of a complete time series.Because, body size at the start of the growing season may influence growth, we also created means for the scale radius at the start of the first marine year (FWt), at the start of the second marine year (L1t=FWt-1+M1t-1), and at the start of the third marine year (L2t=FWt-2+M1t-2+M2 t-1). Mean values of specified scale growth measurements were calculated by brood year and compared among broods to assess inter-annual variation in growth by age group and stock.
Salmon abundance estimates
Information on salmon biomass was unavailable, as was information on the abundance, biomass, or catch per unit effort of juvenile, immature, and maturing salmon in the ocean. Therefore, the indexof sockeye salmon abundance (SSA Index) by cohort wasbased on estimates of commercial harvest (number of fish per year) in central and southeast Alaska management regions (Eggers et al. 2003). The central Alaska region included areas from CapeSuckling to UnimakPass; the southeast Alaskaregion included areas from British Columbia to CapeSuckling. In using commercial harvest to estimate salmon abundance, we assumed a constant marine mortality rate, a low contribution of harvest from sport and subsistence fisheries compared to the commercial fisheries, and a constant exploitation rate among years.
Marine growth versus salmon abundance -- It was hypothesized that intra-specific density-dependent growth would be manifested as negative relationships between the estimated marine growth based on scale measurements (M1-M3) andtheSSA Index lagged to the growth year of cohort. For the juvenile stage, the first-year marine scale growth (M1) in year t was related to the number of maturing sockeye salmon caught in the fishery in year t+2 (SSAM1),the abundance index for the juvenile sockeye in year t.For the immature stage, the second-year marine scale growth (M2) in year t was related to the number of maturing sockeye captured in the fishery in year t+1 (SSAM2), the abundance index for immature sockeye salmon in year t. For the mature stage, the third-year marine scale growth (M3) in year t was related to the number of maturing sockeye captured in the fishery in year t (SSAM3),the abundance index for maturing sockeye salmon in year t.Text and summary measures are given in Table 1.
Two-way scatter plots between scale growth (dependent variable) and sockeye salmon abundance indices (independent variable) were created for M1 against SSAM1, M2 against SSAM2, and M3 against SSAM3 (Fig.3).Plots were examined for negativegrowth-abundance relationships and changes in relationships associated with the three North Pacific Ocean climatic and oceanic regimes(ie. early warm (1925-46), cool (1947-76), and the late warm (1977-98) periods).Regime(called SHIFT) was included as a categorical variable in the models to test whether a change occurred in the growth-abundance relationship associated with the regime shift. To verify the presence of the regime shift we substitute YEAR in for SHIFT and allowed the model to automatically detect changes in the relationships between growth and predictor variables.
Statistical analyses -- To describe density-dependent growth of Karluk sockeye salmon during each marine life history stage we usedordinary least squares (OLS) and multivariate adaptive regression spline (MARS) methods. Individual models for the juveniles (M1), immatures (M2), and maturing (M3) growth were describedas a function of the 1) 1976-77 ocean regime shift (SHIFT), 2) autocorrelation lags at one and two years in the scale growth variable, 3) size at the start of the growing season of the cohort (FW, L1, L2), and 4) the index of sockeye salmon abundance (Table 1). Density-dependence was inferred from negative relationship between scale growth and population abundance.
The OLS and MARS model resultswere compared to measure any possible gains obtained by relaxing some of the restrictive assumptions of OLS.The ordinary least squares model of the form , where y was the dependent variable and were the independent variables, assumed that all effects are linear and that all variables are in the model for every period. The estimated coefficient for the ith input variable in a model measured the unit change of that input variable in explaining a change in the dependent variable y and whether the input variable was positively related or negatively related to the dependent variable. The statistical significance of the estimated coefficient was measured using the t-value . Some important restrictions of OLS include (a) that the effect, if found, was always present, (b) the effect was always the same size for a one unit change in the independent variable, and (c) unless the independent variable was transformed it was not related to other independent variables. The MARS technique allowed testing and relaxing of these restrictions.The MARS model allows the possibility that the effect of x on y can be impacted by an unknown threshold which alters the relationship.Friedman (1991) is a thorough basic MARS reference. Lewis and Stevens (1991) were early users of this approach. Stokes (1997) and Stokes and Lattyak (2006) provide additional information and examples. For example