Trites and Larkin, 1996). The dashed line shows the division between the declining

Transcription

1 Fig. 1. Locations of major geographic features cited in the text. The inserted graph shows estimated numbers of Steller sea lions (all ages) in Alaska from 1956 to 2000 (based on Trites and Larkin, 1996). The dashed line shows the division between the declining (western) and increasing (eastern) population.

2 Fig. 2. Conceptual model showing how sea lion numbers might be affected by ocean climate through bottom-up processes. Water temperatures, ocean currents and other climatic factors determine the relative abundances of fish available to eat, which in turn affects sea lion health (proportion of body fat, rates of growth and at a cellular level oxidative stress). These three primary measures of individual health ultimately determine pregnancy rates, birth rates, and death rates (through disease and predation). Also shown are the effects of human activities that can directly or indirectly affect sea lion numbers. (From Trites and Rosen, in review). 2

3 Figure 3. Diets, population trends and sizes for Steller sea lions at 33 rookeries in Alaska during the 1990s. Estimated numbers of sea lions in 2002 (N) and population growth rates (λ) were determined from linear regressions of log-transformed counts of pups and non-pups conducted from Trend and numbers were calculated independently for pups and non-pups, and then averaged. Average population trends ranged between 0.85 and 1.05, with values <1 indicating declines. Bar heights are proportional to the maximum average N and λ, with solid vertical lines denoting distinct geographic shifts in population sizes and trends. Diet data for summer (S) and winter (W) are from Sinclair and Zeppelin (1998), with circles representing the split-sample frequencies of occurrence of prey (proportional to area of circle). The frequencies sum to 100% and were calculated for the 9 principal species shown as well as for less important species not shown (grouped into 6 prey types: flatfish, forage fish, gadids, hexagrammids, other, rockfish). The zig-zag lines indicate seasonal geographic changes in diet (determined through cluster analysis of the summer data by Sinclair and Zeppelin, 2002). 3

4 (a) (b) Figure 4. Predicted realized habitat of Steller sea lions during summer (a. breeding season) and winter (b. non-breeding season). Habitat suitability is shaded from black through blue to green, yellow and red (areas with probabilities 0.75 are red). Regions with the highest probabilities of encountering sea lions had the highest long-term average census counts from 1956 to (From Gregr and Trites in review). 4

5 Figure 5. Time series and factor loadings of the first four common trends (similar to principal components, but without the constraint of orthogonality) from a state-space decomposition of COADS sea surface temperature. These data are from COADS 1 summaries at the colored boxes over the period From Bograd et al. (2004). 5

6 Figure 6. Nonlinear neural-net based PCA results from a multivariate dataset of 22 physical indices representing both large and local environmental processes. The explained variance in NL PC1 from this analysis is 32%, while PC1 from a linear analysis accounts for 25% of the variance in these data. NLPC2 has more interannual variability than NLPC1 and no hint of interdecadal climate regime shifts. The error bars for the NLPC scores are obtained from a jittering technique that involves the introduction of small errors into the data matrix prior to computation of the NL PCs. From Marzban et al. (2004). 6

7 Figure 7. Sea surface temperature (top panels) and sea level pressure and surface wind anomalies (bottom panels) for winter periods before and after the regime shift, and for the most recent period. From Peterson and Schwing (2003). 7

8 Figure 8. (top) Mean winter (Dec-May) Ekman pumping in the Gulf of Alaska for the period from the NCAR/NCEP reanalysis. (bottom) Change in mean winter Ekman pumping for the period relative to From Capotondi et al. (2004). 8

9 Figure 9. Modeled Gulf of Alaska coastal discharge based on National Weather Service temperature and precipitation data (blue) with glacial ablation (red) since 1980 and the same model using NCEP Reanalysis data (magenta). A 5-year butterworth filter is applied to the data. From Royer (2004). 9

10 Figure 10. Trends in Gulf of Alaska coastal freshwater discharge, showing seasonal changes in the amplitude and phase of the discharge. From Royer (2004). 10

11 Figure 11. Wavelet analysis of the Gulf of Alaska coastal discharge, revealing changes in the low frequency variations of the signal. From Royer (2004). 11