Factors predisposing short-tussock grasslands to Hieracium invasion in Marlborough, New Zealand

The effects of environment and management on the composition of short-tusso ck grasslands and the abundance of the invasive weed Hieracium pilosella we re investigated in two small catchments. Species composition and site facto rs were recorded on a total of 182 plots and the management history of each catchment was reviewed. H. pilosella was present on >80% of all plots, but was at an early stage of invasion in one catchment (<5% cover) and dominan t in the other (25% cover). Classification and ordination revealed strong b etween-catchment differences in community composition that reflected differ ences in environment (soil fertility and rainfall), disturbance history (an imal populations and burning), and the stage of invasion by H. pilosella. I n both catchments H. pilosella tended to be least abundant on the wettest, driest, and most fertile soils. However, such relationships were weak. Gene ralised additive models and regression showed that in the earlier stage of invasion individual site factors explained less than 20% of the variation i n H. pilosella cover. Topographic position and slope (both indicative of so il moisture) were the most significant combined predictors, but together ex plained only 32% of the variation. In the later stage of invasion individua l factors explained up to 33% of the variation. Topsoil sulphur, slope, and topsoil calcium were the most significant combined predictors, but togethe r explained only 53% of the variation. Between-catchment comparisons highli ghted the inter-related roles of environment, disturbance history, geograph ic location, availability of H. pilosella propagules, and stage of invasion in more fully explaining the abundance of H. pilosella. Of five models tha t have been proposed for Hieracium invasion, the: "grassland decline" model best incorporated the inter-related factors that influence spatial and tem poral variation in H. pilosella abundance in the study area. This model con centrates on identifying predisposing and trigger factors that increase the likelihood of invasion and accounts for multiple causes and interactions b y specifying five key factors that influence the ability of a plant species to invade existing vegetation: environment, disturbance, vegetation struct ure and composition, life history attributes of the invader, and the availa bility of invading propagules. The model potentially provides a comprehensi ve framework for evaluating the causes of Hieracium invasion, targeting res earch effort, and developing sustainable management strategies.