Assessing Predictive Invasive Species Distribution Methods

Advances in ecological modeling and decision-support tools are providing new opportunities to predict and manage the spread of invasive species, especially in vulnerable freshwater ecosystems. In their study "Performance of alternative methods for generating species distribution models for invasive species in the Laurentian Great Lakes," Exponent ecological and biological scientist Victoria Prescott, Ph.D., PMP, and coauthors assess three tools — the Risk Assessment Mapping Program (RAMP), boosted regression trees (BRTs), and maximum entropy (MaxEnt) — for their ability to forecast the potential distributions of 30 aquatic nonnative species across the basin. These approaches use climate and known occurrence data to estimate suitable habitats, with implications for conservation and resource management.

The authors found that machine learning methods (BRTs and MaxEnt) yielded high true skill statistic (TSS) values, indicating strong predictive accuracy for where species are likely to occur. However, these models tended to underestimate the full range of suitable habitats, sometimes missing areas where invasive species are already established. In contrast, RAMP forecasted much broader areas as climatically appropriate but produced lower TSS scores, reflecting a tendency to overestimate potential distributions. This discrepancy arises partly from differences in climate suitability thresholds and the algorithms used, with RAMP applying a fixed threshold and machine learning methods tailoring thresholds to model outputs.

The study highlights that while machine learning models require more expertise and time, their conservative predictions may lead to under-preparedness for actual invasions. RAMP, despite its lower precision, may be more suitable for rapid, risk-averse management decisions, providing broader guidance for agencies tasked with preventing or mitigating new invasions. The authors recommend further assessment of RAMP and call for consistent use of modeling tools across jurisdictions to support standardized decision-making and efficient resource allocation in invasive species management.

From the publication: "Accurate species distribution models can potentially provide useful information to agencies because they indicate how much effort to expend for controlling a given nonnative species."