Spatial autocorrelation and the interpretation of patterns of tree seed and seedling predation by rodents in old‐fields

Abstract

Spatial autocorrelation is ubiquitous in ecological systems. However, consideration of this phenomenon in seed and seedling predation studies has been limited, perhaps explaining some of the variation observed between studies and the difficulty in generalizing about the impacts of herbivores on successional processes. I examined the prevalence and patterns of spatial autocorrelation in seed and seedling survival between neighboring tree propagules in old‐fields. All experiments were conducted within enclosures in which I manipulated the density of meadow voles (Microtus pennsylvanicus). Voles, and competitively inferior white‐footed mice (Peromyscus leucopus), are the dominant consumers of tree seedlings and seeds in old‐fields in the northeastern United States, respectively. An intermediate level of spatial autocorrelation was present in seed and seedling predation trials. Furthermore, the frequency of spatial autocorrelation increased under low‐vole density conditions and in study sites with greater heterogeneity in vegetation cover. Survival of neighboring tree propagules was generally positively autocorrelated, suggesting that foraging by mice and voles contributes to a clumped pattern of tree propagule establishment in old‐fields. A review of published predation studies found that only 8.3% attempted to test for spatial autocorrelation and that the minimum average spacing used between replicates is well below the average autocorrelation distance of 8.1±6.3 m observed in the current study. Patterns of autocorrelation appear somewhat predictable based on knowledge of the ecology of mice and voles and to provide insights into the factors influencing their foraging behavior and potential impacts on old‐field succession.