- Top of page
Climate change poses a primary threat to species across the globe (Sala et al. 2000; Thomas et al. 2004). Although it is clear that climate affects species’ abundance and distribution, the particular components of climate that most affect animal performance are generally unknown (Lawton 1995), which makes forecasting specific effects of climate change difficult. Both climatic and biological systems are complex, and relationships may appear at multiple spatial and temporal scales. Identifying the scale at which climatic factors best predict population dynamics can help to elucidate specific mechanisms driving the correlations, and potential consequences of climate change.
Recent studies (Hallett et al. 2004; Stenseth & Mysterud 2005) have described cases in which global scale climate phenomena, such as the North Atlantic Oscillation, better explain population dynamics than local weather variables. This counter-intuitive result demonstrates that the appropriate scale at which to describe the impact of climate on populations needs explicit investigation. The question of scale extends further to the population level, because differential responses to particular climatic phenomena can occur within species (Mueter, Peterman & Pyper 2002). Differential responses to climate can result from variation in behaviour or habitat, and often have a genetic component (Hilborn et al. 2003; Brannon et al. 2004). Understanding these population differences benefits at-risk species because it can help managers understand population fluctuations and trends, and prioritize actions aimed at maintaining diversity within a metapopulation.
Changes in Pacific salmon Oncorhynchus spp. abundance are strongly correlated with changes in the weather (Mantua et al. 1997; Mueter et al. 2002). Previous studies have focused on ocean conditions and broad scale indices such as the Pacific Decadal Oscillation. Here, we instead focus on a much finer scale by examining how juvenile survival in freshwater varies in relation to climate among closely related populations. We chose this life stage for several reasons. First, salmon exhibit diverse behaviour and morphology among closely related populations, which is often related to physical variation in freshwater spawning and rearing areas (Taylor 1990). Second, recent studies have identified juvenile survival as the most important stage for recovery of some threatened populations (Kareiva, Marvier & McClure 2000; Zabel et al. 2006), so the freshwater stages are crucial from a conservation standpoint. Third, stream temperatures are already approaching critically high levels for some endangered salmonids (Donato 2002), so further increases may have catastrophic effects (Eaton & Scheller 1996). Finally, we have detailed juvenile survival data for 18 populations of threatened Snake River spring–summer Chinook salmon O. tshawytscha (Walbaum) over 11 years and corresponding climatological data. We used this data to ask whether natural groups of populations exist in terms of their response to climate and, if so, what factors led to the groupings.
The populations we examined rear in streams primarily fed by snow melt from the neighbouring mountains, which results in stream flow and water temperature that are highly variable from year to year. This led to our primary hypothesis that water temperature and stream flow significantly affect juvenile survival in these populations. Several observations further support this hypothesis. First, as ectotherms, temperature regulates fish development and growth rates (Fry 1967), and has played an important part in the evolution of life-history strategies in salmon (Brannon et al. 2004). Second, stream temperature and stream width, which is often correlated with flow, predict salmonid distribution very well in the western United States at both large (Keleher & Rahel 1996) and small spatial scales (Taylor 1988; Torgersen et al. 1999) and third, temperature and flow are often linked to salmon production (Bartholow et al. 1993; Jager et al. 1997), presumably because they affect a variety of habitat characteristics, such as substrate, risk of scouring, oxygen concentration, energetic costs and habitat area. Furthermore, stream temperature and flow are likely to change over this century because they are intimately related to larger-scale processes shown in flux by general circulation models, especially changes in air temperature and precipitation. In the Pacific North-west, global warming will likely raise stream temperatures (Eaton & Scheller 1996; Mohseni, Stefan & Eaton 2003) and shift the timing and magnitude of stream flows (Hamlet & Lettenmaier 1999; Mote et al. 2003).
Here we propose a three-step approach for assessing population variation in response to climate and identifying the best scale at which to describe effects of climate: (1) aggregate survival estimates across all populations, and relate basin-wide survival to prevailing climate patterns at multiple spatial and temporal scales; (2) identify groups of populations that have similar patterns of survival, and relate within-group survival to prevailing climate patterns; and (3) attempt to uncover features within groups that can explain their similar responses.
- Top of page
Our study demonstrates that different aspects of climate appear to be important at different scales of analysis, and consequently, the importance of considering variation among populations when examining the response of species to climate. Although we could identify a significant basin-wide response of Chinook salmon to climate indicators, when we examined relationships at the population scale, we found improved fits and revealed striking differential responses to climate. From a conservation standpoint, variation is important because diversity is a critical component of metapopulation stability, and hence, viability (McElhany et al. 2000). Distinct ecotypes of Bristol Bay sockeye salmon, for example, have allowed that aggregation of populations to maintain high abundance over several decades, despite wide fluctuations in the abundance of individual populations (Hilborn et al. 2003). From a management perspective, it is useful to recognize that certain groups of populations are likely to show similar interannual fluctuations in response to climate. With this information, we can better parameterize population viability models, identify environmental characteristics that should be monitored more closely, assess impacts of potential interventions, and target particular habitat restoration efforts to the populations that will most benefit from them.
When we explored characteristics of individual sites within clusters, we found intrinsic habitat features that were consistent within the clusters, suggesting that physical, site-specific characteristics may result in populations that respond differentially to climate. In general, juvenile survival in wider and warmer streams was negatively related to temperature, while survival in narrower and cooler streams was positively related to flow. As temperature increases above optimal levels, developmental processes in salmonids are impaired and predation risk increases (Marine & Cech 2004), and prey availability may decline (Bisson & Davis 1976). Flow may act upon survival by several mechanisms. We have observed that as flow decreases in the autumn, streams narrow and potential habitat decreases, which can concentrate predators. Also, these fish typically undergo an autumn migration to overwintering habitat, and reduced flow during this period may alter the timing or reduce success of the migration. Thus it is plausible that juveniles from cooler and narrower streams escape the detrimental effects of summer temperatures but suffer increased mortality due to loss of habitat. Juveniles from warmer and wider streams are not as susceptible to habitat loss, but are potentially more vulnerable to elevated summer stream temperatures.
Climate forecasts clearly indicate that higher summer temperatures and changing hydrological regimes are coming to the Pacific Northwest (Mote et al. 2003). These conditions may lower juvenile survival and affect adult migratory behaviour (Quinn & Adams 1996) and fertility (King & Pankhurst 2004). Despite the high elevation, stream temperatures in the Salmon River basin already routinely exceed the 13 °C maximum daily temperature thresholds for salmonids set by the Idaho Department of Environmental Quality (Donato 2002). Alarmingly, annual air temperatures have been climbing steadily at about 1·2 °C per decade since 1992 (Fig. 5). Our results confirm that temperature is negatively correlated with survival, both because winter and spring air temperatures affect the rate at which the snow pack melts and consequently the timing and magnitude of stream flows, and because summer air temperatures affect stream temperatures. Thus, although ocean conditions have improved since 1998 (Zabel et al. 2006), the realized benefit may be reduced because of deteriorating freshwater conditions.
Figure 5. Annual mean air temperature in the Salmon River Basin (average of 7 weather stations), measured from July to June to match periods of estimated survival. Regression line shows a warming rate of 1.2 °C per decade.
Download figure to PowerPoint
As with most endangered species, many factors reduce the long-term viability of these fish (McClure et al. 2003). Ecologists have long recognized the importance of maintaining diverse habitats and multiple independent populations, but the added threats from climate change make this goal even more imperative. Managers can sometimes influence stream flow and temperature, for example, by reducing water diversions or encouraging riparian cover, so identifying the most important environmental forcing factors for a particular population can have conservation benefits. However, we still do not know the specific mechanisms by which these factors affect survival, so more study is warranted. Understanding the mechanisms by which habitat heterogeneity confers stability to existing populations or metapopulations is essential because their stabilizing effects may unravel as the climate changes. In salmon, if temperatures in streams exceed the critical maximum, such streams will become uninhabitable, leading to the distribution shifts predicted by larger-scale correlation analyses (Eaton & Scheller 1996). Salmon are often considered ‘keystone’ species because they transfer marine nutrients to otherwise nutrient-poor environments, and provide a crucial food resource for many vertebrates (Willson & Halupka 1995), so their decline or extinction affects both community and ecosystem properties (Scheuerell et al. 2005). Many pieces of the puzzle are still unresolved, such as the extent to which behavioural change or thermal adaptation might change the impact of climatic factors over time, and the relative importance of or interactions between density and environmental factors. But clearly, climate change poses a significant risk to this and many other species, and understanding the role of population variation in a diversity of situations is key to understanding these risks.