Two ecological niche models for the conservation of the imperiled Spotted Sucker
North American Journal of Fisheries Management · 02.11.2026
DOI: 10.1093/najfmt/vqag094 ↗Kurz gefasst
Fachartikel in der Zeitschrift North American Journal of Fisheries Management, online erschienen am 02.11.2026. Verfasst von Anastasia M Umstott und 5 weiteren Autor:innen. Die Arbeit nennt 101 Literaturangaben. Gefördert von: Texas Comptroller of Public Accounts Office, Texas Parks and Wildlife Department.
Abstract (Originalsprache)
Objective
Advances in ecological niche modeling have improved our ability to identify critical habitats and model species–environment relationships in river networks. These advancements are valuable for species that occupy extensive river networks but have poorly defined distributions. We created ecological niche models using maximum entropy and random forest algorithms to compare model performance, map habitat suitability, and identify the key environmental predictors influencing the distribution of Spotted Sucker Minytrema melanops, a Species of Greatest Conservation Need in Texas freshwater systems.
Methods
We combined online records of Spotted Sucker with occurrence records from field surveys that were conducted between 2023 and 2024 throughout streams and main-stem channels within the Neches and Sabine River basins to capture the broad distribution of the species. Occurrences were assigned to discrete stream segments, which served as the spatial unit of analysis. Environmental predictors (e.g., hydrogeomorphic, land cover, climatic) were extracted from online databases and linked to individual segments. Models were created based on predictor selection and regularization for maximum entropy and pseudo-absence generation for random forest. Model outputs underwent spatial fivefold cross validation to evaluate performance and interpret environmental relationships.
Results
Both maximum entropy and random forest models exhibited high performance in predicting habitat suitability (area under the curve > 0.80) for Spotted Sucker. Although maximum entropy outperformed random forest based on the values for the true skill statistic, both models generated highly similar spatial predictions and consistently identified hydrologic predictors (i.e., catchment area, stream order, flow accumulation) as important drivers of habitat suitability. The variable response curves of these important predictors showed consistent relationships with probability of habitat suitability for Spotted Sucker in both the maximum entropy and random forest models.
Conclusions
Stream-segment-based ecological niche models using maximum entropy and random forest offer valuable insights into the strengths and limitations of predicting the distribution of data-limited freshwater fish species like Spotted Sucker. Although both models generated similar predictions, maximum entropy may be better suited for studies with limited presence data, whereas random forest excels at capturing complex multivariate relationships. Integrating the spatial predictions of both models may enhance ecological interpretation and guide conservation planning.
Förderung
- Texas Comptroller of Public Accounts Office
- Texas Parks and Wildlife Department
Zitieren
APA
Umstott, A. M., Ellard, J. K., Mangold, R. D., Perkin, J. S., Whisenant, A., & Montaña, C. G. (2026). Two ecological niche models for the conservation of the imperiled Spotted Sucker. North American Journal of Fisheries Management. https://doi.org/10.1093/najfmt/vqag094
Vancouver
Umstott AM, Ellard JK, Mangold RD, Perkin JS, Whisenant A, Montaña CG. Two ecological niche models for the conservation of the imperiled Spotted Sucker. North American Journal of Fisheries Management. 2026. doi:10.1093/najfmt/vqag094
BibTeX
@article{Umstott2026,
author = {Umstott, Anastasia M and Ellard, Johnathan K and Mangold, Rebecca D and Perkin, Joshuah S and Whisenant, Adam and Montaña, Carmen G},
title = {Two ecological niche models for the conservation of the imperiled Spotted Sucker},
journal = {North American Journal of Fisheries Management},
year = {2026},
doi = {10.1093/najfmt/vqag094}
}
EUVIMED