Estimating spatially varying occupancy trends for multi-scale conservation: An amphibian case study in the midwestern United States
Abstract
Species population and distribution trends are essential inputs in the development of conservation plans. While trend estimates at broad spatial scales provide fundamental information, such estimates may mask spatial heterogeneity occurring at local levels. Local variations in occupancy trends can be particularly meaningful for taxonomic groups such as amphibians that are experiencing precipitous large-scale declines as a result of localized and variable patterns of habitat loss/degradation, climate change and disease. Here we develop a spatial occupancy model that provides nuanced inferences on multi-scale occupancy trends. Using volunteer-collected, participatory science data, we apply our framework to estimate multi-scale trends for 11 anuran species across Minnesota, USA from 1996 to 2015. We found substantial fine-scale spatial variation in occupancy trends across all 11 species, suggesting localized drivers of anuran occupancy dynamics across the state. Spatial variation in trends was highly species-specific, with some species having positive trends that varied in magnitude across the state (e.g. boreal chorus frog [Pseudacris maculata]), and others that showed both positive and negative trends across the state (e.g. northern leopard frog [Lithobates pipiens]). Statewide projections of occupancy trends provided fine-scale insights on anuran distribution changes that were not possible when fitting simpler models. Our Bayesian approach allows trends to be summarized across areal units (e.g. ecological strata, management units) that are directly relevant for management and conservation planning. For example, the statewide trend for American toad (Anaxyrus americanus) was effectively zero, but finer scale analyses across ecological strata (used for statewide wildlife conservation planning) revealed multiple areas of both localized increases and declines. Synthesis and applications. Our spatial modelling approach allows practitioners to estimate trends across multiple spatial scales to inform local and regional management priorities, particularly when summarizing trends within key areal units used for conservation planning. Our framework can be used to identify hotspots of decline across multiple sympatric species for targeted monitoring or management actions. Further, estimation of spatially varying trends provides an improved ability to assess effects of protected areas or management actions, inform spatial prioritization and hypothesize drivers of occupancy trends.
