Northern Prairie Wildlife Research Center
We used daily rate of nest predation as the response variable in our analysis. Explanatory variables were area-year, search period when nest was found, and the interaction between these 2 variables. To examine interaction with search period, it was necessary to exclude (Appendix D) area-years for individual species unless at least 1 nest was found during each search period. We were left with 21-29 area-years for each of the 5 common species. Observations were weighted by the number of exposure days. Chi-squared statistics for each effect were calculated from Type III sums of squares (Johnson 1990).
Detection of a significant interaction between area-year and search period for 4 of the 5 common species prompted us to examine more closely the relation between drought conditions and daily rate of nest predation. To do this we grouped area-years into 3 wetness intervals by means of a centroid clustering procedure (SAS Institute, Inc. 1989). Intervals were based on percentage of seasonal wetlands in each area-year that were wet in May and on departure in total precipitation from the long-term average for that area during April through June. The 3 intervals were dry (<43% of wetlands wet and precipitation <-4% of average), moderate (>70% of wetlands wet and precipitation <-11% of average), and wet (>46% of wetlands wet and precipitation >7% of average). Within each wetness category (dry, n = 11; moderate, n = 7; wet, n = 13), we used linear contrasts to examine interactions between area-year and search period by species, again with daily rate of nest predation as the response variable.