MODELING PATCH OCCUPANCY BY FOREST RODENTS: INCORPORATING DETECTABILITY AND SPATIAL AUTOCORRELATION WITH HIERARCHICALLY STRUCTURED DATA

摘要: Abstract Widespread conversion of deciduous forests to agriculture in the midwestern United States has resulted landscapes whose remaining native habitats are highly fragmented, with well-documented consequences for wildlife community structure. We analyzed trap data 5 forest rodent species from 525 sites 35 study throughout upper Wabash River basin, which is dominated by agricultural use and drains >20% Indiana. used a recently developed likelihood approach multi-model inference obtain unbiased estimates occurrence probabilities when detection rates <1, we hierarchical generalized linear modeling evaluate random effects associated nested Inclusion spatial autocovariate term had important on model selection results 4 improved fit models. Eastern chipmunks (Tamias striatus), although previously described as sensitive fragmentation, were nearly ubiquitous our region an...