For a plague epizootic to occur, susceptible hosts, flea vectors and the plague bacterium must coincide in space, along with the proper biotic and abiotic conditions to support disease transmission.  Using a long-term data set (1981-2006) monitoring black-tailed prairie dog (Cynomys ludovicianus) colony size and extinction on the shortgrass steppe in Northeastern Colorado, we investigated the link between climatic triggers such as spring and summer temperature and precipitation, the microclimate of soils, and the spatial configuration of both extant and plaguing black tailed prairie dog colonies and the probability that individual colonies will go extinct from plague.  We tested for spatial and temporal autocorrelation using spatial logit association models and for temporal autocorrelation using autoregressive models.  We also tested for localized epizootic hotspots using spatial logit association models.  We found that globally, the probability of an outbreak on a colony is spatially independent of the status of its neighbors given the colony connectivity observed on the study area.  Locally, however, hotspots of high plague activity cluster spatially.  We also found no temporal trend in the patterns of extinction over time which supports the idea that extinctions are based on current conditions and not past events.  Logistic regression analyses show that plague occurrence in prairie dogs was strongly associated with two spatial configuration variables, connectivity to extant colonies and connectivity to plagued colonies, and with the 15 bar soil moisture content of prairie dog colonies.  We also found a weak positive association between early spring temperature and plague outbreaks, and a strong negative association between summer temperature and epizootics, but found no link between present or past precipitation and plague occurrence.  As such, our models do not support the trophic-cascade model of plague, but rather, a direct climate effects hypothesis for plague occurrence.