Appendix 1. Interaction of NDVI effects with latitude
Separate regressions for each species in each strata (~100 regressions) for the effect of Start of Season
- The model for each stratum is y[t]~ beta0 + beta1 x y[t-1] + beta2 x SOS[t-1], where y is log(density)
- Note how there are very few significant results (each data point is the estimate for the SOS effect and the error bars are 95% CI), but for some species, they tend to get more positive with latitude. Those are the species which had significant ‘SOS x Latitude’ interactions in the manuscript.
Appendix 2. Maps of mean population size (1982-2013) for each species in the analysis
Red fill means the strata was excluded from the analysis for having >=5 years of estimated N=0
Appendix 3. Weighting the NDVI time series by scaup population estimates to look for a WBF-scale change that could explain the 1983-2006 scaup decline
Appendix 4. Comparison of variability in annual growth rates between PPR and WBF duck populations
Selected species which have substantial populations in both PPR and WBF
CV.obs and CV.SSM are the coefficient of variation in the raw data and the state-space model estimates. SD.pro is the estimate of variation in the annual growth rates from the state-space Gompertz mdoel. SD.obs is the estimate of observation error (not pertinent)
A caption
Appendix 5. Correlation between NDVI start of season and mean spring temp
Using NCEP reanalysis I data from 1982-2013, summarized to WBF polygon for April & May, and the stratum-area-weighted mean SOS for the study area
