Effect of annual weather variation vs short-term weather extremes on population change of the alpine butterfly (Parnassius smintheus) in the Rocky Mountains

We use estimates of annual population change (Rt) from 21 sub-populations of the alpine butterfly, Parnassius smintheus, over a 21-year interval to determine the relative importance of annual variation in weather for each month and short-term extremes in weather within months in determining population change. Variables include monthly estimates of mean, maximum and minimum temperature as well as extreme maxima and minima in each month. Snow variables include monthly snow amounts and date of spring snow melt. A combination of mixed effects models and regression trees were used to identify patterns of combination of significant interacting variables. Although population size dominated the effects on Rt in all models, there were key weather variables related to population change. In particular, particularly cold Novembers, and extreme cold events within November, were associated with population decline, as was early snow melt in spring. The strong effect of population size on Rt (density-dependent population growth) was only apparent in years with relatively benign November weather. Results support the general pattern reported earlier of population declines in both especially cold winters and especially warm winters, likely through differing respective mechanisms.