Estimating soil properties in heterogeneous land‐use patches: a Bayesian approach

Cities provide unique opportunities for integrating humans into ecology. Using data from a socio‐ecological inventory of metropolitan Phoenix, Arizona, we explore the contribution of human‐related variables to explaining observed variation in soil nitrate‐N (NO 3 N) and total carbon (C) concentrations across the city, agricultural fields, surrounding desert, and mixed regions. Conventional modeling approaches in such a setting would lead to examination of spatial relationships over the entire study area or on subsets of the data independently. However, the spatial relationships for NO 3 N and C may be different in each of these regions. Here we estimate the correlation coefficients for influential variables toward soil NO 3 N and C across the entire region, while at the same time accounting for potentially differing spatial patterns in each of these regions. Soil NO 3 N shows markedly greater spatial autocorrelation in the desert regions, while the soil C shows varying amounts of spatial relationships in the different regions. Copyright © 2006 John Wiley & Sons, Ltd.