You are here:
PREDICTIONS IN AN INVADED WORLD - PART I: USING NICHE MODELS TO PREDICT DISTRIBUTIONS OF MARINE/ESTUARINE SPECIES AT THE HABITAT SCALE
Citation:
LEE, H. AND D. REUSSER. PREDICTIONS IN AN INVADED WORLD - PART I: USING NICHE MODELS TO PREDICT DISTRIBUTIONS OF MARINE/ESTUARINE SPECIES AT THE HABITAT SCALE. Presented at 5th International Bioinvasions Conference, Cambridge, MA, May 21 - 24, 2007.
Description:
Niche models can be used to predict the distributions of marine/estuarine nonindigenous species (NIS) over three spatial scales. The goal at the biogeographic scale is to predict whether a species is likely to invade a geographic region. At the regional scale, the goal is to predict which specific estuaries or oceanic segments within a biogeographic province are most likely to be invaded. At the habitat scale, the goals are to predict which community types are most vulnerable and to elucidate niche requirements of invaders. As an initial step in evaluating several niche models over these spatial scales, we are comparing model performance for species with known distributions on the Pacific Coast of the United States. Here we discuss nonparametric multiplicative regression (NPMR) to predict habitat-scale distributions. NPMR incorporates interactions among variables, allows qualitative and categorical habitat variables, and in contrast to some other niche models, utilizes absence as well as presence data. This effort utilized more than 660 benthic samples in estuaries and on the continental shelf from Alaska to California from the U.S. EPA's EMAP program. Distributions of 13 native, 10 nonindigenous and 5 cryptogenic benthic species were modeled using four quantitative and seven categorical habitat variables. Most species were adequately modeled with 3 or 4 habitat variables. Latitude was the most frequently included variable, being incorporated into 27 of the 28 species' models while salinity class was incorporated into 23 models. Qualitative estimates of the presence/absence of major habitat types (e.g., seagrass, marsh) were incorporated into more models than quantitative measures of depth or percent silt/clay. Results to date indicate that NPMR performs well at the habitat scale with a moderate to large sample size and that habitat-scale distributions of nonindigenous and cryptogenic species are predicted as well as those of native species.