Using occurrence records to model historic distributions and estimate habitat losses for two psammophilic lizards Cameron W. Barrows a, * , Kristine L. Preston a , John T. Rotenberry a,b , Michael F. Allen a a Center for Conservation Biology, University of California Riverside, Riverside, CA 92521-0334, USA b Department of Biology, University of California Riverside, Riverside, CA 92521-0334, USA ARTICLE INFO Article history: Received 30 January 2008 Received in revised form 5 May 2008 Accepted 11 May 2008 Keywords: Niche models Mahalanobis D 2 Conservation planning Landscape scale Aeolian sand Coachella Valley fringe-toed lizard Flat-tailed horned lizard ABSTRACT Estimating historic distributions of species is a critical step in evaluating current levels of habitat loss, evaluating sites for potential restoration and reintroductions, and for conser- vation planning at a landscape scale. However historic distributions can be difficult to esti- mate objectively because substantial habitat changes may have occurred prior to comprehensive surveys. As a means to address this question, we evaluated a novel approach by creating spatial niche models for two species of psammophilic lizards. Using a partitioned Mahalanobis D 2 analysis and abiotic variables that were independent of anthropogenic change, we created niche models for the federally threatened Coachella Val- ley fringe-toed lizard (Uma inornata) and for the flat-tailed horned lizard (Phrynosoma mcal- lii). The niche models estimated that within the Coachella Valley there were originally 32,164 ha of potential habitat for the fringe-toed lizard and 33,502 ha of potential habitat for the horned lizard. After screening these estimates of historic habitat for current condi- tions that would render that potential habitat unsuitable, we calculated a 91–95% loss of potential habitat for the fringe-toed lizard and an 83–92% loss for the horned lizard. Unlike the fringe-toed lizard, the horned lizard also occurs outside the Coachella Valley. Conduct- ing a similar analysis throughout its range would provide an objective estimate of the total habitat loss experienced by this species. This information could be used to address whether granting it federal or state protection is warranted. For species whose distributions can be modeled with abiotic variables such as soils, elevation, topography, and climate, this approach may have broad applications for resolving questions regarding their current lev- els of habitat loss and regional conservation planning. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Modeling a species’ habitat can identify variables that con- strain its distribution across a heterogeneous landscape (Rotenberry et al., 2002, 2006; Harvey and Weatherhead, 2006; Syartinilia, 2008). Additionally, identifying habitat rela- tionships allows development of quantitative and qualitative estimates of a species’ current and historical distributions, knowledge that could serve to identify, and assess conserva- tion priorities (Martı ´nez et al., 2006; Ortega-Huerta, 2007; Pa- war et al., 2007; Lassalle et al., 2008), recovery sites for a threatened or endangered species and to quantify the extent of habitat loss a species has undergone. These functions can contribute important information contributing to decisions 0006-3207/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2008.05.006 * Corresponding author: Tel.: +1 760 834 0594. E-mail address: cbarrows@ucr.edu (C.W. Barrows). BIOLOGICAL CONSERVATION xxx (2008) xxx – xxx available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/biocon Please cite this article in press as: Barrows, C.W. et al, Using occurrence records to model historic distributions and ..., Biol. Conserv. (2008), doi:10.1016/j.biocon.2008.05.006 ARTICLE IN PRESS