The Effects of Sampling Bias and Model Complexity on the Predictive Performance of MaxEnt Species Distribution Models

By M. Syfert, Matthew J. Smith, David A. Coomes. PLoS ONE, 2013

616 citations recorded by the source corpus.

Species distribution models (SDMs) trained on presence-only data are frequently used in ecological research and conservation planning. However, users of SDM software are faced with a variety of options, and it is not always obvious how selecting one option over another will affect model performance. Working with MaxEnt software and with tree fern presence data from New Zealand, we assessed whether (a) choosing to correct for geographical sampling bias and (b) using complex environmental response curves have strong effects on goodness of fit. SDMs were trained on tree fern data, obtained from an online biodiversity data portal, with two sources that differed in size and geographical sampling bias: a small, widely-distributed set of herbarium specimens and a large, spatially clustered set of ecological survey records. We attempted to correct for geographical sampling bias by incorporating sampling bias grids in the SDMs, created from all georeferenced vascular plants in the datasets, and explored model complexity issues by fitting a wide variety of environmental response curves (known as “feature types” in MaxEnt). In each case, goodness of fit was assessed by comparing predicted range maps with tree fern presences and absences using an independent national dataset to validate the SDMs. We found that correcting for geographical sampling bias led to major improvements in goodness of fit, but did not entirely resolve the problem: predictions made with clustered ecological data were inferior to those made with the herbarium dataset, even after sampling bias correction. We also found that the choice of feature type had negligible effects on predictive performance, indicating that simple feature types may be sufficient once sampling bias is accounted for. Our study emphasizes the importance of reducing geographical sampling bias, where possible, in datasets used to train SDMs, and the effectiveness and essentialness of sampling bias correction within MaxEnt.

Subfields: Ecological Modeling, Ecology, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/c6058192895800da0de71af26c204c03c5655d7f

How researchers see this paper

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

In Ecological Modeling

In Ecology

In Nature and Landscape Conservation

Citations in the corpus

Comparing sample bias correction methods for species distribution modeling using virtual species (2021, background): “Yet, other work suggests that the FactorBiasOut is superior, especially when targetgroup background data are incorporated (e.g., Syfert et al. 2013, Fourcade et al. 2014, Vollering et al. 2019).”
High uncertainty in the effects of data characteristics on the performance of species distribution models (2021, background): “…studies found no relationship between marginality and SDM performance (Newbold et al., 2009; Pöyry et al., 2008), and others report opposite results about the influence of data prevalence and range size (Marmion et al., 2009; Sor et al., 2017; Syfert et al., 2013; van Proosdij et al., 2016).”
MaxEnt versus MaxLike: empirical comparisons with ant species distributions (2013, methodology): “This method, known as ‘‘target-group background’’ (Phillips et al. 2009), has been shown to generally improve performance of MaxEnt models when averaged across all species (e.g., Mateo et al. 2010, Syfert et al. 2013), but not necessarily for all species or regions (Phillips et al. 2009).”
The importance of correcting for sampling bias in MaxEnt species distribution models (2013, methodology): “Alternatively, background data can be manipulated (Phillips et al., 2009; Syfert et al., 2013): Recent releases of MaxEnt allow the inclusion of so-called bias files that allow the user to choose background data with the same bias as occurrence data.”
Using species distribution models to inform IUCN Red List assessments (2014, methodology): “Their selection was based on a combination of correlation, principal components and cluster analyses (Syfert et al. 2013) to minimise the correlation between the layers, and expert judgement based on the ecology of the taxa.”
Estimating optimal complexity for ecological niche models: A jackknife approach for species with small sample sizes (2013, methodology): “Syfert et al. (2013) also selected simpler features (LQ) for modeling species with a large number of occurrence records (over 400).”
Mapping Species Distributions with MAXENT Using a Geographically Biased Sample of Presence Data: A Performance Assessment of Methods for Correcting Sampling Bias (2014, background): “Second, the empirical bias caused by sampling intensity [41] was never tested and no more than two correction techniques have been evaluated at the same time whereas many more have been proposed or used in the literature [47,49,55].”
Accounting for spatially biased sampling effort in presence‐only species distribution modelling (2015, background): “Syfert et al. (2013) demonstrated the importance and efficacy of the bias-file option within a Maxent framework and found that correcting for bias had more of an impact on model fit than the type of response curves permitted to train the model.”