Predicting the Spatial Distribution of King Rails in an Agricultural Landscape

Abstract

Species distribution modeling is valuable for assessing rare or elusive species and in landscapes where accessibility is limited, such as agricultural lands. While birds' local habitat associations are often well known, knowledge of broad-scale relationships and the spatial patterns of environmental factors is generally lacking. In this study, we develop and validate a predictive habitat model for the King Rail (Rallus elegans) in the rice-growing region of southwestern Louisiana and identify important areas for this elusive bird. In a previous localized study we found it to be associated with canopy cover and ditches/streams around the perimeter of rice fields. Here, we used a geographic information system (GIS) to extrapolate these habitat associations to the region (>940 000 ha). We determined the proportion of rice fields within a 5-km radius and quantified canopy cover and ditches within a 1-km radius. In repeated call-back surveys from 2007 to 2009 we recorded King Rails at 44 of 155 locations. On this basis, accounting for imperfect detection, we modeled King Rail occurrence. The validation results showed the model had good predictive ability, and in 17% of the area the predicted probability of occurrence was high. Accounting for imperfect detection resulted in an increased probability of occurrence. Most importantly, the spatial pattern revealed the threat of encroaching woody vegetation and the vulnerability of rail habitat to hurricanes and further land-use changes. Our study illustrates that field research can be extrapolated to GIS-based spatial models, which can identify key spatial patterns in a species' distribution.