Making better M axent models of species distributions: complexity, overfitting and evaluation
2013/12/06 by Aleksandar Radosavljevic, Aleksandar Radosavljević, Robert P. Anderson · 45 citations
Environmental Science · #Ecology and Vegetation Dynamics Studies #Species Distribution and Climate Change #Wildlife Ecology and Conservation
paper · doi:10.1111/jbi.12227
openalex publication_date 2013/12/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/31
Abstract
Abstract Aim Models of species niches and distributions have become invaluable to biogeographers over the past decade, yet several outstanding methodological issues remain. Here we address three critical ones: selecting appropriate evaluation data, detecting overfitting, and tuning program settings to approximate optimal model complexity. We integrate solutions to these issues for Maxent models, using the Caribbean spiny pocket mouse, H eteromys anomalus , as an example. Location N orth‐western S outh A merica. Methods We partitioned data into calibration and evaluation datasets via three variations of k ‐fold cross‐validation: randomly partitioned, geographically structured and masked geographically structured (which restricts background data to regions corresponding to calibration localities). Then, we carried out tuning experiments by varying the level of regularization, which controls model complexity. Finally, we gauged performance by quantifying discriminatory ability and overfitting, as well as via visual inspections of maps of the predictions in geography. Results Performance varied among data‐partitioning approaches and among regularization multipliers. The randomly partitioned approach inflated estimates of model performance and the geographically structured approach showed high overfitting. In contrast, the masked geographically structured approach allowed selection of high‐performing models based on all criteria. Discriminatory ability showed a slight peak in performance around the default regularization multiplier. However, regularization levels two to four times higher than the default yielded substantially lower overfitting. Visual inspection of maps of model predictions coincided with the quantitative evaluations. Main conclusions Species‐specific tuning of model parameters can improve the performance of Maxent models. Further, accurate estimates of model performance and overfitting depend on using independent evaluation data. These strategies for model evaluation may be useful for other modelling methods as well.
Citations
Cited by
- Habitat Suitability for a Unique, Fully Marine American Manatee ( Trichechus manatus ) Population Primarily Depends on Presence of Submarine Freshwater Springs
- Linking community structure and climate vulnerability in desert plant assemblages of southern California
- Spatial patterns of bark beetle–induced pine mortality on a Mediterranean island: Implications for forest impact assessment
- Climate Change Favors African Malaria Vector Mosquitoes
- Mapping Risk Zones: niche overlap between Batrachochytrium dendrobatidis and endemic amphibians in Mexico’s Central-Southern Highlands
- When Good Fit Goes Bad: Identifying and Minimising Overfitting in Ecological Niche Models
- Co-occurrence patterns of introduced terrestrial flatworms in metropolitan France revealed by citizen science
- Impact of Climate Change on the Multiple Facets of Forest Bird Diversity in a Biodiversity Hotspot Within the Atlantic Forest
- Slipped through the narrow door: Effects of climate change on the distribution and conservation of a vulnerable tiger gecko, Goniurosaurus luii, from China and Vietnam
- Ecological niches and climate-driven range shifts in Hemorrhois snakes: implications for biogeography
- Climate change favours African malaria vector mosquitoes
- Modelling the present and future distribution of Ambystoma altamirani in the Transmexican Volcanic Belt, Mexico
- Decadal changes in environmental suitability for the margay ( Leopardus wiedii ) under anthropogenic pressure in the Yucatán Peninsula
- An Ecological Modelling Approach to Support Peru Wildlife Conservation Planning Based on Geospatial Datasets and Remote Sensing Information
- Ground‐Truthing of MaxEnt Models Reveals Poor Predictive Accuracy for Lizards in the Mackenzie Basin, New Zealand
- The risk of underestimating generation length for extinction risk assessments
- Contrasting invasion potential of Ciona robusta and C. intestinalis in Subantarctic and Antarctic ecosystems under warming scenarios: insights of ecological niche models and physiological data
- Nutrient‐Driven Expansion of Reef Fish Habitat Suitability in Indonesian Waters: A 24‐Year Analysis Reveals Iron and Phosphate as Dominant Drivers
- From research to conservation: Site selection for habitat restoration of a narrowly distributed and critically endangered butterfly
- Habitat suitability and niche overlap of endemic and data-deficient catfishes: A prediction modelling approach for river basin scale conservation in peninsular India
- Where the Black-Horned Capuchins Roam: Present and Future Ranges of Sapajus nigritus and Their Overlap with Pine Plantations
- Comprehensive spatial risk assessment using community-based species distribution models stratified by ant damage types
- Weak association between ecological niche divergence and speciation in galagos (Primates: Galagidae)
- Climatic Suitability from MaxEnt Models Reflects Growth Performance in European Forest Trees
- Sampling and modelling rare species: Conceptual guidelines for the neglected majority
- Protected areas mitigate the reduction in habitat suitability for swans under climate change: A case study in the Yellow River Basin
- Integrating biotic factors into niche modeling: a case study with Rhipicephalus (Boophilus) microplus in the Americas
- Spatial filtering to reduce sampling bias can improve the performance of ecological niche models
- Adapting distribution patterns of desert locusts, Schistocerca gregaria in response to global climate change
- Climatic and geometric controls on the global distribution of surge-type glaciers: implications for a unifying model of surging
- Nationwide mapping of pine mistletoe hotspots for adaptive management: MaxEnt modelling in Polish forests
- Predicting the Current and Future Distribution of the Invasive Weed Ageratina adenophora in the Chitwan–Annapurna Landscape, Nepal
- Range Shift and Climatic Refugia for Alpine Lichens Under Climate Change
- Diversity and spillover risk of swine acute diarrhea syndrome and related coronaviruses in China and Southeast Asia
- Modeling the potential global distribution of the invasive Jack Beardsley mealybug (Hemiptera: Pseudococcidae) under climate change
- Prioritizing stream network connectivity for species conservation and sustainable planning
- Global climate model comparisons of niche evolution in turritelline gastropods across the Cretaceous–Paleogene mass extinction
- Implications for the distributional range of the European bark beetles under future climate change
- Random forest and spatial cross-validation performance in predicting species abundance distributions
- ENVIREM: an expanded set of bioclimatic and topographic variables increases flexibility and improves performance of ecological niche modeling
- Cross‐validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure
- Combining multiscale replication in network and landscape-genetic analyses to assess functional connectivity and population resilience in Parnassius clodius butterflies
- The challenge of modeling niches and distributions for data‐poor species: a comprehensive approach to model complexity
- Projecting climate change impacts on the distribution of the most captured tuna species on Atlantic waters
- Range-wide mitochondrial phylogeography of Sorex monticola: evolutionarily distinct clades occupy divergent abiotic niches
Related