A statistical explanation of MaxEnt for ecologists

By Jane Elith, Steven J. Phillips, Trevor J. Hastie, Miroslav Dud'ik, Yung En Chee, C. Yates. 2011

6,009 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Ecological Modeling

View Paper: https://www.semanticscholar.org/paper/42415bb09fea53b08a32ca4ad11ecc98745a7fa7

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Ecological Modeling

Citations in the corpus

Variation in niche and distribution model performance: The need for a priori assessment of key causal factors (2012, methodology): “More specifically, Maxent minimizes the Kullback–Leibler divergence div(f1, f ) = ∫ z f1(z) ln[f1(z)/f (z)] dz, between these two distributions, subject to the means restriction (Elith et al., 2011).”
The Effects of Sampling Bias and Model Complexity on the Predictive Performance of MaxEnt Species Distribution Models (2013, background): “(known as ‘‘feature types’’) are: Linear (L), Quadratic (Q), Hinge (H), Product (P), Threshold (T) and categorical (see [26,27] for detailed descriptions).”
Ocean‐scale prediction of whale shark distribution (2012, methodology): “…MaxEnt models more comparable to the GLM/ GLMM, we used the same data sets for presences and pseudoabsences to generate similar models per season with restricted settings (i.e. we fitted only linear and quadratic features) in MaxEnt (for details on MaxEnt features options see Elith et al., 2011).”
Comparing sample bias correction methods for species distribution modeling using virtual species (2021, background): “…2021 v Volume 12(3) v Article e03422 observations of multiple species within a target group of similar taxa whose locality data were collected in the same way as the modeled species, and has shown promise for reducing sampling bias (Phillips and Dudik 2008, Elith et al. 2011, Ranc et al. 2016).”
Predicting the distribution of vulnerable marine ecosystems in the deep sea using presence-background models (2014, background): “These results are then provided as a map of habitat suitability values, where the value within each grid cell is a function of the environmental variables at that cell (Elith et al., 2011).”
Predictor complexity and feature selection affect Maxent model transferability: Evidence from global freshwater invasive species (2020, methodology): “Models built using only hinge features tend to produce relatively smooth fitted response curves, similar to a generalized additive model (GAM) (Elith et al., 2010, 2011).”
Assessing species climatic requirements beyond the realized niche: some lessons mainly from tree species distribution modelling (2017, result): “In comparison to other options, hinge features provide smoother model fits (Elith et al. 2011).”
Mapping Species Distributions with MAXENT Using a Geographically Biased Sample of Presence Data: A Performance Assessment of Methods for Correcting Sampling Bias (2014, methodology): “species response to abiotic environmental factors [14]) and project it onto the geographical space to derive the probability of presence for any given area or, depending on the method, the likelihood that specific environmental conditions are suitable for the target species [15].”
Predicting distributions of invasive species (2013, background): “15 presence records will be affected similarly, so these impacts should not be used to argue against using absence data (Elith et al., 2011).”
Sky islands of southwest China. I: an overview of phylogeographic patterns (2014, neutral): “By coupling genetic data, coalescent and SDM [102], one can generate alternative phylogeographic hypotheses (using SDM), test the hypotheses (using genetic data and coalescent) and explain why species respond to climate change differently [103].”
Creating individual accessible area hypotheses improves stacked species distribution model performance (2018, positive): “Many algorithms can be used to provide such indices of environmental similarity (Chefaoui & Lobo, 2008; Elith et al., 2006, 2011; Peterson et al., 2011; Phillips et al., 2009), as reviewed by Franklin (2010), which in turn can then be thresholded to create binary correlative models that predict a…”
Phylogeography of a widespread Asian subtropical tree: genetic east–west differentiation and climate envelope modelling suggest multiple glacial refugia (2014, methodology): “3k (Phillips et al., 2006; Elith et al., 2011) (see Appendix S1 for details of the climate envelope modelling).”
How much does climate change threaten European forest tree species distributions? (2018, methodology): “For that reason, the prevalence of background points makes the model more conservative because the model requires a stronger signal of presences than would be the case for equal proportions of presences and pseudoabsences (Elith et al., 2011).”
No silver bullets in correlative ecological niche modelling: insights from testing among many potential algorithms for niche estimation (2015, background): “For these correlative approaches, numerous mathematical and statistical algorithms have been developed to estimate ecological niches in highly dimensional environmental spaces and over complex geographic landscapes (Nix 1986; Hastie & Tibshirani 1990; Carpenter, Gillison &Winter 1993; Stockwell & Pe”