Weather, landscape structure, and the population ecology of a threatened butterfly, Euphydryas editha bayensis

This dissertation documents population responses of the threatened Bay checkerspot butterfly, Euphydryas editha bayensis, in topographically complex habitat, and relates the responses to mechanisms driven by variable phenology of the butterfly and its hostplants. Using a Geographic Information System (GIS), we built terrain models and calculated insolation across slopes to quantify the dominant ecological gradient in the habitat. From 1985-96 at a 100 ha site, larval numbers ranged from 35,000 to 900,000. Four consecutive drought years (1987 to 1990) led to a 95% decrease in numbers. The distribution of larvae along the insolation gradient also changed--when numbers increased, the larval distribution shifted towards warmer slopes, and when numbers decreased, the distribution shifted toward cooler slopes. Cooler slopes provided refugia from the drought. Larval surveys across thousand of hectares from 1992-96 showed that larval densities in areas 500 meters apart fluctuated asynchronously.; Changes in larval numbers at the 100 ha site were predicted by the amount of time between peak adult flight and hostplant senescence, indicating that weather acts via phenology to produce population responses. We implemented a model of postdiapause development on the GIS, which can assess the combined impacts of weather and larval distributions on adult emergence dates. Under the same weather year, the different observed larval distributions at Kirby Canyon could alter the peak flight date by 12 days. Infrared thermometer measurements were used to quantify grassland topoclimates and calibrate a model of plant phenology. Hostplant phenology is largely controlled by temperature and insolation, and the time to senescence after flowering is strongly influenced by mean April air temperatures. The phenological models allow for explicit consideration of the impacts of climate change on phenology in complex terrain. The interactions between yearly weather, larval distributions, and hostplant phenology can lead to asynchronous dynamics across continuous habitat. Large, continuous, topographically diverse habitats provide the best chance for persistence of the butterfly. Because many species of insects depend on precise timing of their life cycles with host resources, the monitoring techniques, modeling exercises, and synthesis presented here may have broad applications for conservation and pest management.