© 2006 International Council for the Exploration of the Sea
Climate regime shifts and community reorganization in the Gulf of Alaska: how do recent shifts compare with 1976/1977?
Alaska Fisheries Science Center, National Marine Fisheries Service 301 Research Court, Kodiak, AK 99615, USA
*Correspondence to M. A. Litzow: tel: +1 907 481 1723; fax: +1 907 481 1703. e-mail: mike.litzow{at}noaa.gov.
Climate regime shifts have recently occurred in the North Pacific (1998–1999) and the Arctic (2000), but the nature of biological reaction to these events is poorly understood. An index of local climate (1960–2005), and data from commercial fishery catches (1960–2004) and from small-mesh trawl surveys (1972–2005) are used to assess the impacts of these climate events in the Subarctic Gulf of Alaska. Non-linear regression showed that survey catch composition strongly responded to local climate at lags of 2 and 4 years, providing evidence of rapid ecological response to climate change in the system. A sequential regime shift detection method identified rapid change in local climate, and in survey and commercial catches following the well-documented regime shift to a positive state of the Pacific Decadal Oscillation (PDO) in 1976/1977. However, the analysis failed to detect the 1998/1999 regime shift in local climate, or in survey or commercial catches. This result is consistent with the view that the 1998/1999 climate regime shift did not represent a reversion to a negative PDO state. Local temperature increased and local sea level pressure decreased in the Gulf of Alaska during the years 2001–2005, consistent with anthropogenic warming and recent spatial reorganization in Arctic climate. There was no evidence of community reorganization following this climate event. Further observation will be required to evaluate the persistence of this new climate pattern, and the nature of community reaction to it.
Keywords: climate change, community ecology, North Pacific, Pacific Decadal Oscillation, regime shift, Subarctic, Victoria Pattern
Received 19 December 2005; accepted 8 June 2006.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
Shifts between decade-scale climate regimes produce sudden taxonomic reorganization in continental shelf ecosystems (Anderson and Piatt, 1999; Beaugrand, 2004). However, both the mechanisms producing climate regime shifts and their links with ecological change are poorly understood (Mantua and Hare, 2002; Chavez et al., 2003). As a result, there is currently no ability to forecast decade-scale climate variability and resulting biological change, in contrast to some annual-scale phenomena such as ENSO (El Niño – Southern Oscillation) events (Mantua and Hare, 2002; Bond et al., 2003). Climate regime shifts have important implications for the management of exploited populations (Chavez et al., 2003), so early recognition of the biological effects of climate regime shifts is an important goal.
During the 20th century, the Pacific Decadal Oscillation (PDO) was the dominant source of decade-scale variability in North Pacific climate (Mantua and Hare, 2002; Chavez et al., 2003). The state of the PDO is measured by an index that provides a temporal measure (the leading principal component) of changes in the spatial organization of Pacific Ocean sea surface temperature (SST) anomalies polewards of 20°N (Mantua and Hare, 2002). During the 20th century the PDO index showed strong interannual autocorrelation, with periods of generally positive or negative PDO values persisting for decades. Positive PDO values correspond to warm SST along the west coast of the Americas, and cool SST in the central North Pacific, with low winter sea level pressure (SLP) over the North Pacific producing enhanced counter-clockwise winds, and negative PDO values correspond to opposite spatial patterns (Mantua and Hare, 2002). Abrupt switches in the sign of the PDO index around 1925, 1947, and 1976/1977 were indicators of regime shifts between periods of relatively stable climate, and were associated with profound changes in community ecology on a basin-wide spatial scale (Mantua and Hare, 2002; Chavez et al., 2003). Effects of the 1976/1977 shift to a positive PDO state were particularly well documented, and in the Bering Sea and Gulf of Alaska this regime shift apparently contributed to declines in abundance for one species group (shrimp, Pandalidae; Pacific herring, Clupea pallasii; capelin, Mallotus villosus; Pacific sandfish, Trichodon trichodon; Pacific tomcod, Microgadus proximus), and increases in abundance for another (walleye pollock, Theragra chalcogramma; Pacific cod, Gadus macrocephalus; arrowtooth flounder, Atheresthes stomias; flathead sole, Hippoglossoides elassodon; salmon, Oncorhynchus spp.) (Mantua et al., 1997; Anderson and Piatt, 1999; Mueter and Norcross, 2000; Hunt et al., 2002). Overfishing of crustaceans may also have contributed to this community transition (Orensanz et al., 1998). A subsequent, weaker climate regime shift in 1989 was not a reversal of the PDO signal, and did not return the Bering Sea and Gulf of Alaska to pre-1976/1977 conditions (Hare and Mantua, 2000).
Recent interest has focused on the possibility of two regime shifts with the potential to affect Pacific Ocean climate and biology: a North Pacific regime shift that may have taken place in 1998 and 1999, and a reorganization of Arctic climate patterns in 2000 and 2001. Several authors have proposed that the 1998/1999 shift may have been a transition to conditions consistent with a negative PDO state (Chavez et al., 2003; Peterson and Schwing, 2003). However, variability in SST since 1998/1999 has been summarized by the second principal component (the "Victoria Pattern") instead of the first principal component (the PDO), indicating that the 1998/1999 shift was orthogonal to change in the PDO (Bond et al., 2003). Positive values of the Victoria Pattern are characterized by a north–south dipole in North Pacific SLP, a spatial pattern that is distinct from either positive or negative states in the PDO (Bond et al., 2003). Initial biological responses to the 1998/1999 regime shift have been observed in the California Current and off the coast of Peru (Chavez et al., 2003; Peterson and Schwing, 2003), but the spatial extent of biological response in the North Pacific remains unknown. The Subarctic Pacific may also be susceptible to climate change associated with the appearance during the years 2000–2005 of spatial patterns in Arctic temperature and sea level pressure reminiscent of conditions observed during the 1930s (Overland and Wang, 2005). This new pattern in Arctic climate has been associated with transition from Arctic to Subarctic conditions in the Bering Sea, resulting in rapid ecological reorganization (Grebmeier et al., 2006). Climate anomalies associated with the pattern apparently project onto the Subarctic Pacific (Overland and Wang, 2005), suggesting the possibility that this shift may have a greater impact in that region than the 1998/1999 Victoria Pattern shift, which apparently had little impact in the Subarctic Pacific (Bond et al., 2003). However, the hypothesis of a biological response to either event has not been tested for the Gulf of Alaska.
The goal of this paper is to test the hypothesis that either the 1998/1999 Victoria Pattern regime shift or the 2000/2001 Arctic regime shift has contributed to a community-wide reorganization in the Gulf of Alaska similar in magnitude to the reorganization observed post-1976/1977. I use a time-series of catch data from small-mesh trawl surveys conducted in Gulf of Alaska bays during the years 1972–2005 to compare biological response to the different climate shifts. This time-series has previously been instrumental in documenting biological effects of the 1976/1977 regime shift (Anderson and Piatt, 1999; Mueter and Norcross, 2000). An important step in understanding local ecological responses to variability in large-scale climate indices is to understand interactions between local ecology, local climate, and the climate index of interest (Stenseth et al., 2003). I therefore used an index of local climate conditions to test for local response to large-scale climate indices (the PDO and the Victoria Pattern). Further, I examined the time-lag of ecological response to climate forcing by comparing the effect of the local climate index on small-mesh trawl catch composition at lags of 0–6 years. Because sampling began in 1972, the small-mesh time-series contains relatively few observations from before the 1976/1977 regime shift. Therefore, it was necessary to construct an index from Gulf of Alaska commercial catches of five taxa (shrimp; red king crab, Paralithodes camtschaticus; Pacific cod; Pacific halibut, Hippoglossus stenolepis; and salmon) known to have been affected by the 1970s regime shift (Mantua et al., 1997; Anderson and Piatt, 1999), in order to make inferences about community stability before the 1976/1977 regime shift.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Small-mesh trawl surveys have been conducted in the Gulf of Alaska annually by the Alaska Department of Fish and Game and the US National Marine Fisheries Service with standardized methods since 1972 (n = 8215 hauls). Surveys utilize high-opening bottom trawls with 32 mm stretched mesh throughout the net, and sample in bays at mean bottom depth of 114 m (range 18–226 m). Sampling locations are randomly selected from within-bay strata that were designed to reflect known concentrations of commercially exploited shrimp populations. Although the survey was designed to assess shrimp populations, a wide variety of taxa are captured, and several hundred taxa have been identified in catches (Anderson and Piatt, 1999).
Sampling in different years was in different bays and at different times of year, so spatial and seasonal variability in sampling confound interannual trends in catch composition. To control this variability, the analysis was restricted to hauls taken from July to October in the seven bays that were most consistently sampled (Marmot, Kiliuda, Two-headed Gully, Alitak, Chignik-Castle, Kuiukta, and Pavlof; Figure 1). Data from some early surveys with poor identification of non-shrimp taxa were excluded from analysis. The restricted data set contains 1802 hauls, with different bays being sampled in each year (Table 1).
|
|
To illustrate broad trends in community composition through time, log-transformed catch per unit effort (cpue; kg km towed–1) was plotted for five common taxa (capelin; pink shrimp, Pandalus borealis; arrowtooth flounder; Pacific cod; jellyfish, Scyphozoa) known to be sensitive to climate change (Anderson and Piatt, 1999). These trends included 2005 data (n = 58 hauls) that became available while this paper was in review, and that were not included in other analyses. Because trends among different taxa are not independent events, I used non-metric multidimensional scaling (NMDS) to summarize variability in abundance of the 30 most common taxa (98.8% of the total catch) for hypothesis testing. NMDS summarizes variability in catch composition in a restricted number of variables in a manner conceptually similar to principal components analysis, but it is more robust to the presence of large numbers of zero catches that characterize trawl survey data. Bray–Curtis dissimilarities and NMDS ordination were used to compare each haul in a matrix of 1744 rows (hauls, 1972–2004) and 30 columns (taxa). I fourth-root transformed cpue data before analysis, and the final configuration from the ordination was rotated so that the first axis corresponded to the axis of maximum variation. NMDS axes are orthogonal to each other, so each can be treated as an independent variable during analysis. The first three NMDS axes explained 34%, 24%, and 20% of variability in catch composition, respectively. Spearman rank correlation was used to compare the cpue of individual taxa with axis scores from individual hauls in order to interpret axis variation, with |r|
0.35 being set as the cut-off for interpreting axis variation as corresponding to variation in cpue of an individual taxon. For detailed NMDS methods, see Mueter and Norcross (1999, 2000). Study bays were located approximately 40–650 km from each other (Figure 1), and trends in catches are correlated among bays (Mueter and Norcross, 2000). Individual bays could not therefore be treated as independent sampling units, so NMDS axis scores were averaged among bays to develop a measurement of mean community state across the study area. Mean bay–year values of NMDS axes 1–3 were estimated as the mean axis score of every haul set in a given bay during a given year. However, before these bay–year axis values could be used to test the hypothesis of a recent community transition, values for missing bay–year combinations had to be estimated; failing to estimate missing values would introduce bias based on different bays being sampled in different years. Multiple imputation is a well-established technique for estimating missing values in multivariate data (Schafer, 1997), but methods for applying multiple imputation to time-series data are currently under development and not widely available (Hopke et al., 2001). Instead, values for axis scores in missing bay–year combinations were estimated with predicted values from autoregressive models of temporal variability (i.e. single imputation). Autoregressive modelling corrects regression estimates for the autocorrelation that is typical of time-series data (Fuller, 1978).
Linear, logarithmic, quadratic, and cubic models of temporal variability were compared for each NMDS axis in each bay using Akaike's Information Criterion (AIC), which selects the best model from a set of candidates based on the ability to explain variability in response variables with a minimal number of explanatory variables (Burnham and Anderson, 1998). Backward stepwise autoregression was used to select the autoregressive variables appropriate to each model (p to stay = 0.1). Synthetic time-series were then constructed to allow the hypothesis of a recent community transition to be tested. These time-series consisted of the grand mean of annual bay mean axis scores, and when these were missing, predicted axis scores from the selected models. Two time-series were constructed for each NMDS axis: one for all seven study bays, beginning in 1976, the first year that all bays were sampled, and another for a restricted set of bays (Chignik-Castle, Kuiukta, Pavlof) that have been sampled since 1973.
Community composition was also summarized with the first principle component of log-transformed mass of Gulf of Alaska commercial catches of salmon (all species combined), cod, halibut, king crab, and shrimp for the years 1960–2004 (data available at www.cf.adfg.state.ak.us/geninfo/pubs/pubshome.php and www.afsc.noaa.gov/refm/stocks/assessments.htm).
Local climate conditions for the years 1960–2005 were summarized with the first principal component of four measures of local climate: mean summer (JJA) and winter (DJF) NCEP/NCAR reanalysis SLP in four 2.5° x 2.5° blocks centred over the study area (Kalnay et al., 1996; www.cdc.noaa.gov/index.html), and mean summer (JJA) and winter (DJF) HadCRUT2 combined land and sea surface temperature anomalies for a 5° x 5° block centred over Kodiak Island (Rayner et al., 2003; www.cru.uea.ac.uk/cru/data/temperature/#datdow). Surface temperature and SLP were selected as variables for building the local climate index because they are tightly coupled in basin-scale climate patterns (Bond et al., 2003), have been implicated as contributing to previous community reorganization in the Gulf of Alaska (Anderson and Piatt, 1999), and are available for the entire time period over which community state was examined here. Winter seasons were labelled for the year corresponding to January.
The longest synthetic time-series (1973–2004) for each NMDS axis was compared with the local climate series to test for climate effects on community composition. Separate analyses were performed for climate effects at lags of 0–6 years in order to assess the temporal scale at which community ecology responds to climatic variability in this system. North Pacific biological time-series typically show non-linear dynamics (Hsieh et al., 2005), so I used four-parameter sigmoidal regression in this analysis.
Correlation between the local climate index and the PDO and Victoria Pattern indices was used to assess response of local climate to large-scale variability in Pacific climate. Autocorrelation in time-series inflates estimates of degrees of freedom in correlation tests and increases the chance of a type-I error, so I used the Modified Chelton method (Pyper and Peterman, 1998) to control for autocorrelation. These results are reported with adjusted sample sizes (n') and probability values (p'). PDO values were calculated as annual means of November–March monthly values (jisao.washington.edu/pdo/PDO.latest), and November–March Victoria Pattern data were obtained from N. Bond and M. Spillane (University of Washington, pers. comm.). PDO and Victoria Pattern indices are calculated as the first and second principal components, respectively, of HadCRUT2 5° x 5° sea surface temperature anomalies (Rayner et al., 2003) for the North Pacific polewards of 20°N.
Finally, sequential t-test analysis of regime shifts (STARS; Rodionov, 2004) was used to test for transitions in time-series. STARS uses a t-test to determine whether sequential observations in a time-series represent statistically significant departures from mean values observed during the preceding period of a pre-determined duration. STARS is appropriate for analysis towards the end of time-series, so allows for timely detection of shifts. STARS results are determined by the cut-off length for proposed regimes (L), and the Huber weight parameter (H), which defines the range of departure from the observed mean (in standard deviations) beyond which observations are considered as outliers. L was set here to 10, and H to 1, but exploratory analyses with L = 5 or 10 and H = 1 or 2 produced identical conclusions.
Because the harm of a type-II error is potentially high in tests for incipient community reorganization, 
was set to 0.1 for all STARS tests, but to 0.05 for all non-STARS analysis. Because single imputation of missing values introduces uncertainty into analysis of synthetic time-series (Schafer, 1997), separate analyses were conducted with data from Pavlof Bay, the only bay without missing observations, as a way of checking the general validity of results from the imputed, area-wide time-series.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Climate change
The first principal component (PCclimate) explained 51.3% of the variability in the four climate variables. PCclimate positively weighted surface temperature (winter eigenvector = 0.633, summer eigenvector = 0.443), and negatively weighted SLP (winter eigenvector = –0.596, summer eigenvector = –0.220). Positive values in PCclimate can therefore be thought of as indicating warm years with increased incidence of storms, especially in winter, while negative PCclimate values indicate cold years with decreased storminess. PCclimate was positively correlated with the PDO index (n' = 33.4, r = 0.682, p' < 0.0001), but it was not correlated with the Victoria Pattern (n' = 42.8, r = 0.199, p' = 0.095). STARS analysis indicated a shift to warmer temperatures and lower SLP in local climate following the 1976/1977 regime shift (p = 0.006), and a further increase in temperature and decrease in SLP beginning in 2001 (p = 0.012; Figure 2). Neither the 1988/1989 nor the 1998/1999 regime shifts appeared in the local climate signal (p > 0.1).
|
0.012).Community response
Following the 1976/1977 PDO regime shift, catch per unit effort (cpue) data for representative taxa showed either rapid decline (capelin, pink shrimp), or rapid increase (arrowtooth flounder, Pacific cod, jellyfish; Figure 3). Log (cpue) for all five taxa remained steady from the mid-1980s to 2005, with the exception of Pacific cod, which apparently declined in abundance across the study area from 2000 to 2005 (Figure 3).
|
Of the three leading NMDS axes, only axis 1 showed coherent temporal variability across all study bays (Figure 4). Axes 2 and 3 scores mostly varied with depth and among bays, respectively. In modelling temporal variability of NMDS scores for each bay, AIC resulted in the selection of quadratic models of axis 1 for four bays, and linear and cubic models for two other bays. Selected models showed strong temporal effects on axis 1 scores and close fits with available data (Table 2). Selected models for axes 2 and 3 generally showed weaker temporal effects than axis 1 scores. I could find no effect of local climate (PCclimate) on axes 2 and 3 at lags of 0–6 years, either with four-parameter sigmoidal regression (p > 0.4), or with linear, logarithmic, quadratic, or cubic regression (p > 0.2); these axes were therefore dropped from further analysis of climate–ecosystem linkage.
|
|
NMDS axis 1 positively weighted taxa that increased after the 1976/1977 regime shift (jellyfish, arrowtooth flounder, walleye pollock, flathead sole), and negatively weighted taxa that declined following the regime shift (pink shrimp; capelin; coonstripe shrimp, Pandalus hypsinotus; humpy shrimp, P. goniurus; sidestripe shrimp, Pandalopsis dispar; Pacific sandfish; red king crab; sculpins, Cottidae, Psychrolutidae, Hemitripteridae; Figure 5). Similar NMDS results were obtained in a previous analysis of small-mesh trawl survey data from bays of Kodiak Island (Mueter and Norcross, 2000).
|
The imputed axis 1 community time-series beginning in 1973 (Pavlof, Kuiukta, Chignik-Castle bays) showed strongest effects (i.e. highest r2 values) of local climate at lags of 2 and 4 years (Figure 6). Analysis of both the non-imputed time-series for Pavlof Bay and the imputed time-series for all study bays produced similar results, with strongest effects at lags of 2 and 4 years, although these results are not illustrated.
|
The first principle component (PCcatch) explained 63.9% of the variance in commercial fishing catches. PCcatch negatively weighted catches of shrimp (eigenvector = –0.513) and king crab (eigenvector = –0.538) and positively weighted catches of salmon (eigenvector = 0.525), halibut (eigenvector = 0.205), and cod (eigenvector = 0.360). PCcatch was highly correlated with NMDS axis 1 values averaged among the three bays sampled between 1973 and 2004 (n' = 5.0, r = 0.945, p' = 0.008).
The 1976/1977 regime shift produced rapid ecological change that produced reorganization in the composition of both commercial and survey catches (Figure 7). PCcatch showed a transition to a more positive state (lower shrimp and king crab catches, higher cod, halibut, and salmon catches) during the period 1978–1989; this change was detected by STARS as shifts in 1980 (p = 0.0002) and 1988 (p = 0.001, Figure 7). NMDS axis 1 scores in survey catches transitioned from a negative to a positive state in all study bays (Figure 7). The number of years required to detect community reorganization in a particular bay after the regime shift was a function of the number of years of sampling that occurred before the shift (Pearson correlation of the number of years sampled prior to 1977 and the number of years after 1977 until shift was detected, n = 7 bays, r = –0.953, p = 0.0009). When axis 1 scores were averaged among all bays (including imputed and observed data), STARS detected a transition in 1983 (p < 0.0001). This result was confirmed by analysis of the non-imputed Pavlof Bay time-series, where a continuous community transition following the 1976/1977 regime shift was detected by STARS as transitions in 1979 (p < 0.0001) and 1988 (p < 0.0001, Figure 7).
|
Following both the 1998/1999 regime shift and the 2000/2001 warming event in local climate, no transition was detected in PCcatch, NMDS axis 1 scores averaged across all study bays, or the NMDS axis 1 time-series from Pavlof Bay (p > 0.1, Figure 7).
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The community reorganization that followed the 1976/1977 PDO shift was dramatic and sudden (Anderson and Piatt, 1999; Mueter and Norcross, 2000). STARS analysis identified this ecological transition in commercial catches throughout the Gulf of Alaska, and in small-mesh trawl survey catches in all seven study bays (Figure 7). The same analysis failed to identify any community reorganization following the 1988/1989 regime shift, consistent with observations that the regime shift was much less pervasive than the 1976/1977 event (Hare and Mantua, 2000). STARS analysis also failed to identify any pervasive community transitions following the 1998/1999 regime shift (Figure 7). The local climate index also did not show the effect of the 1998/1999 event (Figure 2), and was not correlated with the Victoria Pattern index. Survey catch composition showed the strongest response to local climate at lags of 2 and 4 years (Figure 6), and community reorganization was detected more quickly in bays with more years of sampling prior to 1977, so the NMDS analysis through 2004 presented here is presumably extensive enough (i.e. >20 years of sampling before and 5 years of sampling after the 1998/1999 regime shift) to detect the onset of a reorganization similar to that following the 1976/1977 regime shift. Missing values in time-series in this study were estimated with single imputation, which introduces uncertainty associated with estimated values (Schafer, 1997). However, because the alternate hypothesis proposes dramatic community changes such as those observed post-1976/1977, and sampling frequency in recent years has been high (Table 1), I conclude that the estimation of missing values was unlikely to reduce the ability to detect recent community reorganization. Moreover, results from the non-imputed Pavlof Bay time-series (Figure 7) were similar to those from the imputed time-series, so support the area-wide results from the imputed data set. The possibility also exists that community reorganization in the Gulf of Alaska following the 1998/1999 regime shift was orthogonal to reorganization following the 1976/1977 event, i.e. that a recent reorganization was reflected in an NMDS axis other than axis 1. However, the second and third NMDS axes did not show coherent temporal trends (Figure 4) or response to climate change at lags of 0–6 years. Axes subsequent to the first three explain progressively less of the overall variation in catch composition, so are unlikely to reflect major community reorganization.
These results lead to rejection of the hypothesis that the 1998/1999 regime shift contributed to rapid change in local climate and community state in the Gulf of Alaska. This finding is consistent both with the observation that positive Victoria Pattern index scores, which appeared after 1998 and 1999, indicate a north–south dipole in SLP, with the Gulf of Alaska and other regions of the Subarctic Pacific experiencing negative SLP anomalies similar to those prevalent since the 1976/1977 regime shift (Bond et al., 2003), and with more recent observations that the Victoria Pattern signal may have weakened in 2004 and 2005 (Goericke et al., 2005).
There is, however, evidence of a persistent increase in temperature and a decrease in SLP in the northwestern Gulf of Alaska during the years 2001–2005 (Figure 2). This observation is consistent both with anthropogenic warming, which has been more rapid in the Arctic and Subarctic than elsewhere in the world (IPCC, 2001), and with the emergence of a spatial pattern in Arctic climate characterized by decreased winter SLP over Subarctic North America, including the northwestern Gulf of Alaska (Overland and Wang, 2005). Change in Gulf of Alaska climate beginning in 2001 was presumably not a lagged effect of the 1998/1999 Victoria Pattern regime shift, because the Victoria Pattern index is derived from principal components analysis of North Pacific SST data, so reflecting change in the spatial organization of North Pacific climate with no time-lag (Bond et al., 2003). Although Pacific cod cpue apparently declined in the small-mesh survey during the years 2001–2005 (Figure 3), and STARS detected a shift in NMDS axis 1 in one study bay (Kiliuda Bay, Figure 7), there was no evidence of a geographically extensive community-wide reorganization in response to the 2001 warming event, although it is likely premature at this time to conclude that no such reorganization will take place. Attempting to predict future change in Gulf of Alaska climate is beyond the scope of this paper. However, to the extent that the warming observed during the period 2001–2005 is the product of anthropogenic warming, it can be expected to continue and intensify (IPCC, 2001).
In the simplest terms, NMDS axis 1 expresses the degree to which small-mesh catches are dominated by crustaceans and small pelagic fish (negative values) or groundfish (positive values; Figure 5). As such, this axis identifies two alternate stable states that have been observed repeatedly in Pacific and Atlantic Subarctic/boreal continental shelf ecosystems; switches between "crustacean/small pelagic fish" communities and "groundfish" communities have been observed in the Bering Sea (Hunt et al., 2002), Gulf of Alaska (Anderson and Piatt, 1999), Scotian Shelf (Choi et al., 2004), and North Sea (Cushing, 1980, 1984). Switches between the two states are apparently modulated by the relative strength of demersal and pelagic secondary production (Hunt et al., 2002; Choi et al., 2004), and the strength of top-down ecosystem control (Worm and Myers, 2003; Frank et al., 2005). Transition back to a crustacean/small pelagic fish state is an unlikely result of the current warming in the Gulf of Alaska, because dominant taxa in this state, such as capelin and pandalid shrimp, are associated with cold temperatures (Rose, 2005; Wieland, 2005). A more likely possibility is a transition to a community containing more temperate/warm-water species. Persistent warming in the Atlantic has resulted in rapid northern distribution shifts in temperate fish species, both in the Subarctic Northeast Atlantic during the years 1920–1940 (Rose, 2005), and in the North Sea during recent decades (Beare et al., 2004; Perry et al., 2005). Such northern distribution shifts can be sudden, with orders of magnitude increases in local abundance of warm-water species occurring at time scales of 1–2 years (Beare et al., 2004), so have the potential to create a community transition as profound and sudden as that following the 1976/1977 regime shift in the Gulf of Alaska.
Sudden transitions in exploited ecosystems can be economically and socially devastating to fishing communities (Hamilton et al., 2004). If high temperatures and low SLP do persist in the Gulf of Alaska, understanding the ecological effects of the climate change will be an important management goal. Given that a comprehensive ecological response to recent climate change in the Gulf of Alaska is not yet apparent, fisheries managers have the opportunity to prepare for the possibility of climate-forced ecological change before it occurs. Possible management priorities are suggested by research into transitions between alternative stable states in a variety of marine, freshwater, and terrestrial ecosystems. Transitions between ecosystem states are often triggered by sudden perturbations, but are also typically preceded by relatively long periods of declining ecosystem resilience, which is defined as the ability of an ecosystem to absorb perturbation while remaining in its current state (Scheffer et al., 2001; Carpenter and Brock, 2006). Developing the ability to measure and preserve ecosystem resilience should therefore become a focus of fisheries management (Hughes et al., 2005), especially in systems, such as the Gulf of Alaska, that are subject to rapid climate change.
|
Acknowledgements |
|---|
I am indebted to all those who helped conduct small-mesh trawl surveys over the years, especially Paul Anderson and Dave Jackson. I also thank Alisa Abookire for help with NMDS analysis; Claire Armistead for making the study site figure; Jennifer Boldt and Dan Urban for sharing fisheries data; Nick Bond and Mick Spillane for sharing Victoria Pattern data; Franz Mueter and Dave Somerton for statistical advice; and Alisa Abookire, Kevin Bailey, Pierre Pepin, Sergei Rodionov, Dave Somerton, Dan Urban, and an anonymous reviewer for helpful comments on earlier versions of this manuscript. Climate data were made publicly available by the Climatic Research Unit, University of East Anglia, UK, and the NOAA Earth System Research Laboratory, Boulder, CO, USA.
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Anderson P.J. and Piatt J.F. (1999) Community reorganization in the Gulf of Alaska following ocean climate regime shift. Marine Ecology Progress Series 189:117–123.[CrossRef][Web of Science]
Beare D.J., Burns F., Greig A., Jones E.G., Peach K., Kienzle M., McKenzie E., Reid D.G. (2004) Long-term increases in prevalence of North Sea fishes having southern biogeographic affinities. Marine Ecology Progress Series 284:269–278.[CrossRef][Web of Science]
Beaugrand G. (2004) The North Sea regime shift: evidence, causes, mechanisms and consequences. Progress in Oceanography 60:245–262.[CrossRef][Web of Science]
Bond N.A., Overland J.E., Spillane M., Stabeno P. (2003) Recent shifts in the state of the North Pacific. Geophysical Research Letters 30:2183.[CrossRef]
Burnham K.P. and Anderson D.R. (1998) Model Selection and Inference: A Practical Information-Theoretic Approach(Springer, New York) 353 pp.
Carpenter S. and Brock W. (2006) Rising variance: a leading indicator of ecological transition. Ecology Letters 9:311–318.[CrossRef][Medline]
Chavez F.P., Ryan J., Lluch-Cota S.E., Niquen M. (2003) From anchovies to sardines and back: multidecadal change in the Pacific Ocean. Science 299:217–221.
Choi J.S., Frank K.T., Leggett W.C., Drinkwater K. (2004) Transition to an alternate state in a continental shelf ecosystem. Canadian Journal of Fisheries and Aquatic Sciences 61:505–510.
Cushing D.H. (1980) The decline of the herring stocks and the gadoid outburst. Journal du Conseil International de l'Exploration de la Mer 39:70–81.
Cushing D.H. (1984) The gadoid outburst in the North Sea. Journal du Conseil International de l'Exploration de la Mer 41:159–166.
Frank K.T., Petrie B., Choi J.S., Leggett W.C. (2005) Trophic cascades in a formerly cod-dominated ecosystem. Science 308:1621–1623.
Fuller W. (1978) Introduction to Time Series(John Wiley, New York) 728 pp.
Goericke R., Venrick E., Mantyla A., Bograd S.J., Schwing F.B., Huyer A., Smith R.L., Wheeler P.A., Hooff R., Peterson W.T., Chavez F., Collins C., Marinovic B., Lo N., Gaxiola-Castro G., Durazo R., Hyrenbach K.D., Sydeman W.J. (2005) The state of the California current, 2004–2005: still cool? CalCOFI Reports 46:32–71.
Grebmeier J.M., Overland J.E., Moore S.E., Farley E.V., Carmack E.C., Cooper L.W., Frey K.E., Helle J.H., McLaughlin F.A., McNutt S.L. (2006) A major ecosystem shift in the northern Bering Sea. Science 311:1461–1464.
Hamilton L., Haedrich R., Duncan C. (2004) Above and below the water: social/ecological transformation in northwest Newfoundland. Population and Environment 25:195–215.[CrossRef][Web of Science]
Hare S.R. and Mantua N.J. (2000) Empirical evidence for North Pacific regime shifts in 1977 and 1989. Progress in Oceanography 47:103–145.[CrossRef][Web of Science]
Hopke P.K., Liu C., Rubin D.B. (2001) Multiple imputation for multivariate data with missing and below-threshold measurements: time-series concentrations of pollutants in the Arctic. Biometrics 57:22–33.[CrossRef][Web of Science][Medline]
Hsieh C.H., Glaser S.M., Lucas A.J., Sugihara G. (2005) Distinguishing random environmental fluctuations from ecological catastrophes for the North Pacific Ocean. Nature 435:336–340.[CrossRef][Medline]
Hughes T.P., Bellwood D.R., Folke C., Steneck R.S., Wilson J. (2005) New paradigms for supporting the resilience of marine ecosystems. Trends in Ecology and Evolution 20:380–386.[CrossRef]
Hunt G.L., Stabeno P., Walters G., Sinclair E., Brodeur R.D., Napp J.M., Bond N.A. (2002) Climate change and control of the southeastern Bering Sea pelagic ecosystem. Deep-Sea Research Part II: Topical Studies in Oceanography 49:5821–5853.[CrossRef]
IPCC. (2001) Climate Change 2001: The Scientific Basis(Cambridge University Press, Cambridge, UK) 881 pp.
Kalnay E., Kanamitsu M., Kistler R., Collins W., Deaven D., Gandin L., Iredell M., Saha S., White G., Woollen J., Zhu Y., Chelliah M., Ebisuzaki W., Higgins W., Janowiak J., Mo K.C., Ropelewski C., Wang J., Leetmaa A., Reynolds R., Jenne R., Joseph D. (1996) The NCEP/NCAR 40-year reanalysis project. Bulletin of the American Meteorological Society 77:437–471.[CrossRef][Web of Science]
Mantua N.J. and Hare S.R. (2002) The Pacific decadal oscillation. Journal of Oceanography 58:35–44.[CrossRef][Web of Science]
Mantua N.J., Hare S.R., Zhang Y., Wallace J.M., Francis R.C. (1997) A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society 78:1069–1079.[CrossRef][Web of Science]
Mueter F.J. and Norcross B.L. (1999) Linking community structure of small demersal fishes around Kodiak Island, Alaska, to environmental variables. Marine Ecology Progress Series 190:37–51.[Web of Science]
Mueter F.J. and Norcross B.L. (2000) Changes in species composition of the demersal fish community in nearshore waters of Kodiak Island, Alaska. Canadian Journal of Fisheries and Aquatic Sciences 57:1169–1180.
Orensanz J.M.L., Armstrong J., Armstrong D., Hilborn R. (1998) Crustacean resources are vulnerable to serial depletion – the multifaceted decline of crab and shrimp fisheries in the Greater Gulf of Alaska. Reviews in Fish Biology and Fisheries 8:117–176.[CrossRef][Web of Science]
Overland J.E. and Wang M. (2005) The third Arctic climate pattern: 1930s and early 2000s. Geophysical Research Letters 32:L23808.[CrossRef]
Perry A.L., Low P.J., Ellis J.R., Reynolds J.D. (2005) Climate change and distribution shifts in marine fishes. Science 308:1912–1915.
Peterson W.T. and Schwing F.B. (2003) A new climate regime in northeast Pacific ecosystems. Geophysical Research Letters 30:1896.[CrossRef]
Pyper B.J. and Peterman R.M. (1998) Comparison of methods to account for autocorrelation in correlation analyses of fish data. Canadian Journal of Fisheries and Aquatic Sciences 55:2127–2140.
Rayner N.A., Parker D.E., Horton E.B., Folland C.K., Alexander L.V., Rowell D.P., Kent E.C., Kaplan A. (2003) Global analyses of sea surface temperature, sea ice and night marine air temperature, since the late nineteenth century. Journal of Geophysical Research 108:4407.[CrossRef]
Rodionov S.N. (2004) A sequential algorithm for testing climate regime shifts. Geophysical Research Letters 31:L09204.[CrossRef]
Rose G.A. (2005) On distributional responses of North Atlantic fish to climate change. ICES Journal of Marine Science 62:1360–1374.
Schafer J. (1997) Analysis of Incomplete Multivariate Data(Chapman and Hall, New York) 448 pp.
Scheffer M., Carpenter S., Foley J., Folke C., Walker B. (2001) Catastrophic shifts in ecosystems. Nature 413:591–596.[CrossRef][Medline]
Stenseth N.C., Ottersen G., Hurrell J.W., Mysterud A., Lima M., Chan K-S., Yoccoz N.G., Ådlandsvik B. (2003) Studying climate effects on ecology through the use of climate indices: the North Atlantic Oscillation, El Niño Southern Oscillation and beyond. Proceedings of the Royal Society of London, Series B, Biological Sciences 270:2087–2096.[Medline]
Wieland K. (2005) Changes in recruitment, growth, and stock size of northern shrimp (Pandalus borealis) at West Greenland: temperature and density-dependent effects at released predation pressure. ICES Journal of Marine Science 62:1454–1462.
Worm B. and Myers R.A. (2003) Meta-analysis of cod–shrimp interactions reveals top-down control in oceanic food webs. Ecology 84:162–173.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||