Main content area

Differences in landscape drivers of garlic mustard invasion within and across ecoregions

Urbanowicz, Christine, Pasquarella, Valerie J., Stinson, Kristina A.
Biological invasions 2019 v.21 no.4 pp. 1249-1258
Alliaria petiolata, agricultural land, ecological invasion, ecoregions, edge effects, habitats, indigenous species, invasive species, land use, landscapes, models, plants (botany), population distribution, probability, regression analysis, understory, Massachusetts
While landscape models of invasive plant distributions have potential for targeting management areas, we focus on two reasons such models may be limited in their application. First, models to date are biased towards explaining establishment in ruderal habitat but not spread of invasives into adjacent habitat, where they may most impact native plants. Second, models are usually developed across ecoregions or for a single ecoregion, and it is unclear how well models for different regions and spatial extents agree. Our aims were to (1) test how landscape variables explain garlic mustard (Alliaria petiolata) occurrence on forest edges and its incursion into forest understory habitat; (2) compare models constructed for two ecoregions, separately, and both ecoregions together. In 183 sites in two ecoregions in Massachusetts, we recorded whether garlic mustard occurred on the forest edge (edge occurrence) and in the understory (understory incursion). We used logistic regression to relate either edge occurrence or understory incursion to elevation and four variables describing land use surrounding each site: percent open land in 1830, percent developed and agricultural land in 2005, and forest-edge length in 2005. Elevation was negatively associated with edge occurrence within regions but positively associated across regions. Land use from 2005 explained edge occurrence and understory incursion in only one region. These results suggest that mechanisms driving garlic mustard distributions are scale- and region-dependent. Our findings also suggest that region-specific invasive distribution models are necessary, and we caution against implying probability of understory incursion from the probability of edge establishment.