Interacting gradients of selection and survival probabilities to estimate habitat quality: An example using the Gray Vireo (Vireo vicinior)

Abstract Common approaches to estimating habitat quality are habitat suitability indices, which are often subjective. We propose a quantitative definition of habitat quality as the percentage of selected habitat that has a high probability of contributing to population growth. Using this definition, we created a habitat quality index based on spatial projections of habitat selection and survival probabilities, and the interaction of those probability gradients. We used nest-site selection and nest survival of Gray Vireos (Vireo vicinior) to represent habitat selection and survival probabilities of a sensitive life stage that is likely to increase population size. We used data from 173 Gray Vireo nests in central New Mexico to estimate selection and survival probabilities. Generalized linear mixed-effect models (GLMM) and logistic exposure models (LEM) were used to estimate nest-site selection and daily nest survival, respectively. We projected top-ranked GLMM and LEM models in a geographic information system (GIS) to spatially display relative probabilities of selection and survival. We converted these continuous probabilities into binary rasters representing high and low selection and survival. Multiplying these two rasters in ArcGIS provided us with a raster of four categories: areas with i) low selection and low survival, ii) low selection and high survival, iii) high selection and low survival, and iv) high selection and high survival. We determined the spatial area of each category and calculated percentage of selected habitat where survival was likely to occur, which we termed a habitat quality index (HQI). At our study site, Gray Vireos had a HQI of 0.85, suggesting that approximately 85% of the highly selected habitat had a high probability of contributing to Gray Vireo population growth through nest survival, and that few ecological traps exist in this landscape. These methods provide a quantifiable, continuous index of habitat quality, with spatial projections of potential ecological trap locations.