Distribution of American pikas in a low-elevation lava landscape: conservation implications from the range periphery

Abstract

Predicting species distributions is increasingly important in conservation biology and, in the context of contemporary climate change, can be particularly informative for populations at the periphery of the range. Peripheral populations can exhibit unique patterns of habitat use in response to marginal conditions and can provide colonists adapted to novel or extreme environments. We conducted surveys for American pikas (Ochotona princeps) during 2007–2009 in 144 sites in Craters of the Moon National Monument and Preserve, Idaho. This species appears vulnerable to global climate change, and Craters of the Moon is an extensive area of low-elevation lava habitat situated on an interior edge of the species' range. We found pikas to be readily detectable when both direct and indirect sign were used. An estimate of detection probability from a subset of 72 sites that were visited twice was 0.92. We detected pikas in 31% of survey sites overall but only.at sites above 1,605 m. We used logistic regression to model the distribution of pikas as a function of elevation, substrate, and vegetation cover. Pikas were most likely to occur on structurally complex pahoehoe lava flows above 1,600 m. The odds of pika occurrence on pahoehoe lava was >10 times that of aa lava flows and increased by 2 times for each SD increase in elevation. Pikas were also more likely to occur on lava flow sites with higher structural complexity and forb cover. An area of pahoehoe lava encompassing >250 km2 in the northern portion of Craters of the Moon contained 91% of pika detections and all predicted site-occurrence probabilities >0.38, an optimal cutoff value determined by examining model receiver operating characteristic curves. Craters of the Moon may provide long-term refugia for the species, given the extent of lava habitat there. However, the importance of elevation in our models suggests that accelerated climate change could erode suitable pika habitat in the park. Most research on pikas has relied on censuses or nonrandom convenience surveys, but we demonstrate an efficient probabilistic sampling approach that has broad application for pika monitoring and research.