2.5 The influence of climate change on size of S. araneus
We examined the weather data from Białowieża from 1952-2004 to explain changes in skull size in juvenile and adult S. araneus . We obtained the mean daily temperature (°C), rainfall (mm) and snow cover (mm) from the Białowieża Meteorological Station. We summed or averaged these daily measures for longer periods and computed the soil moisture deficit from monthly temperatures and precipitation using the Watbug program (Willmott 1977). Monthly deficits were summed for years or other periods as indicated below. As in the previous section, we used linear regression, tested for nonlinearity by adding quadratic terms, and smoothed the relationships with LOESS.
We averaged each of three skull measurements over each year and regressed averages against weather variables after weighting by sample size. We calculated the mean daily temperature, sum of the precipitation, and sum of the moisture deficit in the soil in the period preceding the collection of specimens (from their approximate time of birth; see Results for details). We checked for a temporal autocorrelation (correlation of residuals in neighbouring years) in juvenile shrews using the Durbin-Watson statistic. To account for the effect of significant autocorrelation, we corrected the regressions with the Cochrane-Orcutt procedure. We corrected all regressions using adult shrews because the smaller sample size (32 years) did not allow us to test for autocorrelation. The results are shown as the coefficientsr 2, the proportion of the skull size variation explained by weather variables.