COS 51-5
Anthropogenic invasion pressure improves prediction of invasive forest plant distribution in urban landscapes

Tuesday, August 11, 2015: 2:50 PM
338, Baltimore Convention Center
Amy J. Davis, US EPA, Office of Research and Development, Oak Ridge Institute for Science and Education, Research Triangle Park, NC
Kunwar Singh, Center for Geospatial Analytics, North Carolina State University, Raleigh, NC
Ross K. Meentemeyer, Forestry and Environmental Resources, North Carolina State University, Raleigh, NC
Jean-Claude Thill, Geography & Earth Sciences, UNC Charlotte, Charlotte, NC

Despite the pressing need to manage invasions in order to protect the conservation values of forests embedded within metropolitan landscapes, species distribution models (SDMs) of invasive forest plants in urban landscapes remain largely unexplored. Due to the spatial concentration of human activity in metropolitan areas, we expect that anthropogenic invasion pressure (AIP) is an important driver of invasions, distinct from dispersal from nascent foci. To test this hypothesis, we collected presence/absence data on a widespread forest invader, Ligustrum sinense, from 400 random plots located across 70 forest fragments stratified along an urban to rural gradient to capture the spectrum of heterogeneity that exists across the greater Charlotte, NC metropolitan area. We developed a base SDM containing only environmental predictors, and then evaluated the contribution of two spatially explicit metrics:  1) landscape scale AIP (based on the cumulative distance weighted probability of dispersal from residences), and 2) local scale neighborhood invasion pressure (representing dispersal from nascent foci) to improving model performance as indicated by rates of omission and commission, accuracy and area under the curve (AUC). We used generalized linear modeling to build our models. We also assessed the level of spatial dependence present in the residuals of each model.


Our results indicate that models that account for AIP have substantially higher accuracy than the environment only model, and that AIP is a better predictor of privet occurrence than neighborhood invasion pressure. The inclusion of AIP dramatically lowered the omission rate observed in the environment only model and was found to reduce residual spatial autocorrelation just as well as neighborhood invasion pressure, suggesting that AIP is an important driver of invasions by Chinese privet. Furthermore, our results show that high AIP increases the risk of invasion in environmentally suboptimal habitats, providing empirical support that AIP can overcome abiotic resistance to invasion, as has been demonstrated by others for neighborhood invasion pressure in experimental studies. Our approach demonstrates the importance of accounting for AIP in SDMs in order to accurately assess the risk of invasion of the forest understory in metropolitan landscapes by exotic invaders that are also heavily utilized by humans, such as ornamental plants thereby increasing the chances that the habitats at risk of invasion are not left unidentified.