The distribution and abundance of bur chervil, Anthriscus caucalis, within canyon grassland systems of Idaho

As ecology is called upon to inform invasive plant management and its policies, it is imperative to develop comprehensive approaches for identifying ecological mechanisms of plant invasions across spatial scales. Invasive plant management routinely takes place at intermediate stages of the invasion process, between rapid population increases and negative ecosystem impacts. Thus, the established invader whose potential spatial distribution is unknown at either landscape- or regional-scales presents a significant challenge to land managers. This dissertation is a compilation of observational studies that examine the potential impact of Anthriscus caucalis, an established exotic annual forb within canyon grassland ecosystems, at multiple spatial scales. An intended outcome of this dissertation is to produce a study framework that may contribute to risk assessment, or assessment of potential impacts, of established invasive plants. A potential framework for ecological risk assessment within Idaho's current Noxious Weed Law is explored in Chapter I.;Population level processes are at root of all proximate and ultimate causes of species distributions. Therefore, a four-year comparative demography study was conducted across two bunchgrass and two shrub community types. Intra-annual censuses focused on aboveground measures of abundance and associated vital rates, fecundity and juvenile survival. Results indicate that high shrub communities provide the most suitable habitat for A. caucalis growth and survival. Spatial and temporal variance of juvenile survival and fecundity was lower within high shrub communities, and mean fecundity was significantly higher, compared to bunchgrass communities. Population matrix models were used to simulate stochastic transient population growth rates under various dispersal rate scenarios. High shrub communities maintained positive growth rates under dispersal rate assumptions up to 60%. In comparison, growth rates in two of four bunchgrass communities declined (log A4 < 0) under an assumption of 0% seed dispersal. Observed patterns suggest that high shrub communities are likely source populations for population spread at a landscape scale.;Species distribution models (SDMs) may be utilized to predict the realized distributions of invasive plants at intermediate stages of invasions. Inclusion of predictor variables that account for dynamic processes influencing invasive spread improve model performance. In a companion study, landscape-level A. caucalis distribution survey data was utilized to fit SDMs using environmental, remote sensing and land-use history data as predictor variables, and then compared to a spatially-explicit mechanistic model using classified plant community maps and source-to-sink metrics (distance and relative area) as predictor variables. Using the mechanistic model, A. caucalis distributions were best described as a function of plant community type and the relative area of source habitat, suggesting that source-to-sink dispersal is a primary ecological process underlying A. caucalis invasive spread. Using a SDM, A. caucalis distributions were best described as a function of the best-fit vegetation index (NDVI) and grazing history. Model validations resulted in 65 to 85% accuracy across SDM and mechanistic models. Notably, SDM predictions were more accurate in bunchgrass communities where expected probabilities were low. Study results demonstrate the utility of SDMs for guiding management of invasive plants at intermediate stages of invasion, which includes: 1) identification of potential areas of new source populations at the invasion front, and 2) identification of potential areas of new sink populations contributing to invasive spread in relation to the proximity of source habitats.