Do pseudo-absence selection strategies influence species distribution models and their predictions? An information-theoretic approach based on simulated data

By M. Wisz, Antoine Guisan. BMC Ecology, 2009

307 citations recorded by the source corpus.

BackgroundMultiple logistic regression is precluded from many practical applications in ecology that aim to predict the geographic distributions of species because it requires absence data, which are rarely available or are unreliable. In order to use multiple logistic regression, many studies have simulated "pseudo-absences" through a number of strategies, but it is unknown how the choice of strategy influences models and their geographic predictions of species. In this paper we evaluate the effect of several prevailing pseudo-absence strategies on the predictions of the geographic distribution of a virtual species whose "true" distribution and relationship to three environmental predictors was predefined. We evaluated the effect of using a) real absences b) pseudo-absences selected randomly from the background and c) two-step approaches: pseudo-absences selected from low suitability areas predicted by either Ecological Niche Factor Analysis: (ENFA) or BIOCLIM. We compared how the choice of pseudo-absence strategy affected model fit, predictive power, and information-theoretic model selection results.ResultsModels built with true absences had the best predictive power, best discriminatory power, and the "true" model (the one that contained the correct predictors) was supported by the data according to AIC, as expected. Models based on random pseudo-absences had among the lowest fit, but yielded the second highest AUC value (0.97), and the "true" model was also supported by the data. Models based on two-step approaches had intermediate fit, the lowest predictive power, and the "true" model was not supported by the data.ConclusionIf ecologists wish to build parsimonious GLM models that will allow them to make robust predictions, a reasonable approach is to use a large number of randomly selected pseudo-absences, and perform model selection based on an information theoretic approach. However, the resulting models can be expected to have limited fit.

Subfields: Ecological Modeling, Genetics, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/6ab938b032ea6777ed0a6455252d4c306cc3c1fb

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 Genetics

In Nature and Landscape Conservation

Citations in the corpus

The Effects of Sampling Bias and Model Complexity on the Predictive Performance of MaxEnt Species Distribution Models (2013, background): “A number of other studies have shown the importance of the background data selection method through a variety of SDM techniques [19,20,21,22,23,52] and studies have begun to also evaluate MaxEnt [12,14,53,54,55].”
Selecting pseudo‐absences for species distribution models: how, where and how many? (2012, methodology): “When using regression techniques (GLM,GAMandMARS), the best strategy was to randomly generate the pseudo-absences data, which supported results from Wisz & Guisan (2009).”
Plant extinction risk under climate change: are forecast range shifts alone a good indicator of species vulnerability to global warming? (2012, methodology): “For models requiring records of absence as well as presence, pseudo-absences were chosen randomly from locations with low climatic suitability (Wisz & Guisan, 2009), i.e., outside the observed 90th percentile for each climate variable.”
Protected areas network is not adequate to protect a critically endangered East Africa Chelonian: Modelling distribution of pancake tortoise, Malacochersus tornieri under current and future climates (2020, background): “…is biased, one cannot differentiate whether species are observed in particular environments because those locations are preferable or because they receive the largest search effort (Phillips et al. 2009, Sastre and Lobo 2009, Wisz and Guisan 2009, Newbold et al. 2010, Chakraborty et al. 2011).”
Diversity and distribution of benthic macrofauna in the Baltic Sea Data inventory and its use for species distribution modelling and prediction (2010, unknown): “We assumed that our data contains reliable “true absences” sensu Wisz and Guisan (2009), as species we are considering can be regarded as discriminatory, common and relatively well known for the Baltic Sea.”
Where will conflicts between alien and rare species occur after climate and land-use change? A test with a novel combined modelling approach (2011, methodology): “…significant predictors for each species were tested for pair-wise correlations using Spearman’s rho correlation coefficient using the results of the best models obtained by data dredge, and only predictors with correlation lower than 0.7 were considered (Elith et al. 2006; Wisz and Guisan 2009).”
A 40‐year, continent‐wide, multispecies assessment of relevant climate predictors for species distribution modelling (2014, background): “…given that confirmed absences are usually difficult to obtain, especially for mobile species (Mackenzie & Royle, 2005), pseudo-absences are widely used, adding yet another source of uncertainty (Chefaoui & Lobo, 2008; VanDerWal et al., 2009; Wisz & Guisan, 2009; Barbet-Massin et al., 2012a).”
A framework for species distribution modelling with improved pseudo-absence generation (2015, background): “In fact, several previous studies addressing this issue (e.g. Hengl26 et al., 2009; Wisz and Guisan, 2009; Stokland et al., 2011; Senay et al., 2013)27 propose contradictory solutions.”
Measuring the relative effect of factors affecting species distribution model predictions (2014, methodology): “.,x5 by fitting a GLM to real data, projecting it throughout the study area and then treating the resulting spatial predictions as the true distributions (see Wisz & Guisan 2009; Albert et al. 2010).”
The effects of small sample size and sample bias on threshold selection and accuracy assessment of species distribution models (2012, background): “A more complete discussion on how pseudo-absences are selected, and their utility in species distribution modeling can be found in Wisz and Guisan (2009).”
Exploring consensus in 21st century projections of climatically suitable areas for African vertebrates (2011, background): “Yet, random selection has been shown to result in higher predictive power than strategies that select pseudo-absences from low-suitability regions (Wisz & Guisan, 2009).”
The effect of sample size on the accuracy of species distribution models: considering both presences and pseudo‐absences or background sites (2019, methodology): “ENFA and Maxent were included as both of these profile techniques (Wisz and Guisan 2009) also use background data.”
Threats to cassava production: known and potential geographic distribution of four key biotic constraints (2011, methodology): “This limitation was addressed using the concept of pseudo-absence data (Benito de Pando and Peñas de Giles, 2007; Wisz and Guisan 2009; Phillips et al. 2006; Phillips and Dudik 2008).”
Benthos distribution modelling and its relevance for marine ecosystem management (2015, background): “The spatial bias is particularly problematic if presence-only or pseudo-absence/background data are used for modelling (Wisz and Guisan, 2009; Lobo et al., 2010).”