Spatial heterogeneity in the determinants of woody plant invasion of lowland heath.
Questions: 1. What is the scale and extent of spatial variability in factors affecting Betula invasion of heaths? 2. How much effect does each factor have on within-patch patterns of invasion? 3. How can this understanding aid in managing Betula invasions? Location: Lowland heath of southern England. Methods: Determinants of Betula (both B. pubescens and B. pendula) invasion: biomass density, necromass density, mean vegetation height, P-availability, soil water content and total Betula seed bank density, were measured at two sites on a 5-ha sampling grid. Spatial pattern was assessed using geostatistics. Contributions of each determinant to within-site heterogeneity in predicted Betula seedling densities were estimated by varying variables over their full and interquartile ranges in a statistical model derived from experimental data. Results: Salient spatial trends were revealed: strong autocorrelation over distances of <50 m for soil factors and more extensive autocorrelation (0 to >150 m) in vegetation variables and Betula seed bank densities. The latter resulted in single across-site gradients, the former small, distinct patches. All patterns were overlain with variance that was present at distances of <17.6 m. Variables displaying spatial pattern also accounted for within-site heterogeneity in predicted Betula seedling densities but their relative contribution to this varied between sites. Conclusions: Identifiable spatial autocorrelation in factors controlling patch-scale invasion patterns allows managers to target invasion prone patches, potentially reducing management intensities. Furthermore, management effort may be optimised by spatially de-coupling Betula seed from safe-sites. This plan may adaptable to the management of other weeds and openland ecosystems.