Modeling habitat suitability across different levels of invasive plant abundance

Predicting where invasive plants are likely to spread and become abundant is critical for informing invasive plant management. Species distribution models are a key tool for informing the geography of invasion risk, but most distribution models are limited by their use of presence data, including no information on invader population abundance. In this study, we ask how habitat suitability varies for different levels of abundance for three invasive plants: stiltgrass (Microstegium vimineum), sericea lespedeza (Lespedeza cuneata), and privet (Ligustrum sinense). For each species, we used an ensemble distribution modeling approach to compare suitability for invasion estimated from subsets of point location data: all presences vs. locations with percent cover ≥ 1%, ≥ 5%, ≥ 10%, ≥ 25%, and ≥ 50%. For all species, the total area predicted as suitable for abundant populations was 32%–68% less than the area predicted as suitable for presence. For stiltgrass and sericea lespedeza, the area suitable for invasion decreased when predicted from higher levels of abundance, whereas for privet, suitable area was similar across abundance levels. Stiltgrass and sericea lespedeza are therefore likely to become highly abundant in a smaller portion of their ranges, while privet could become highly abundant anywhere it can establish at low abundance. Different environmental predictors explained suitability for presence versus abundance, suggesting the environmental niche associated with presence differs from that associated with high population abundance. Analyses of more species and growth forms are still needed, but our results combined with previous studies consistently show that fitting distribution models to point locations with ≥ 5–10% cover refines range maps and can produce a more targeted assessment of invasion risk.