An evaluation of alternative algorithms for fitting species distribution models using logistic regression

By J. Pearce, Simon Ferrier. 2000

400 citations recorded by the source corpus.

Subfields: Ecological Modeling, Ecology, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/b75b7d1f59fbccc3496f8db21299a2822d52ef8a

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

Species-specific traits associated to prediction errors in bird habitat suitability modelling (2005, background): “For example, among groups of animal species, the success of several modelling techniques relates inversely with spatial variability (mobility and nomadism) and niche width, but there are some effects which are not consistent across all biological groups ( Pearce and Ferrier, 2000; Pearce et al., 20”
A strict maximum likelihood explanation of MaxEnt, and some implications for distribution modelling (2013, methodology): “Pearce & Ferrier (2000b) show that bi can be appropriately modelled as the response to ql (or ql ) by a logit-logit relationship, which ensures that both the RPPP values, the response bi, and the fitted values of the model, qi, are expressed on probability ([0,1]) scales: ln = β . ln + β0 Values…”
Modelling breeding habitat preferences of Bonelli’s eagle (Hieraaetus fasciatus) in relation to topography, disturbance, climate and land use at different spatial scales (2005, methodology): “For each logistic regression model, standard backward stepwise procedure was used ( Pearce and Ferrier 2000a ), including all variables at once and removing non-significant ones stepwise by Wald’s test (Johnson 1998).”
Effects of species and habitat positional errors on the performance and interpretation of species distribution models (2009, methodology): “In the majority of cases with mild errors, the changes in neither AUC nor deviance explained were sufficient to cause the researcher to reject the model (according to the thresholds for acceptable models based on AUC – see Swets, 1988; Hosmer & Lemeshow, 2000; Pearce & Ferrier, 2000).”
Presence-absence versus presence-only modelling methods for predicting bird habitat suitability (2004, background): “A major but variable role of prevalence on the predictive accuracy of habitat models has been stressed by several studies (Araújo and Williams 2000, Pearce and Ferrier 2000, Manel et al. 2001, Stockwell and Peterson 2002b).”
Novel methods improve prediction of species' distributions from occurrence data (2006, neutral): “While other studies indicate that small sample size negatively influences modelling success (Pearce and Ferrier 2000a, Harrell 2001, Stockwell and Peterson 2002, Kadmon et al. 2003), this was not evident here, perhaps because in general we had adequate localities for modelling (mean number of…”
Mapping Paleo-Fire Boundaries from Binary Point Data: Comparing Interpolation Methods (2007, background): “In general, values of 0.5 to 0.7 indicate poor discriminating ability, values of 40.7 to 0.9 indicate reasonable discriminating ability, and values 40.9 indicate excellent discriminating ability (Manel et al. 1999; Pearce and Ferrier 2000; Gibson, Wilson, and Aberton 2004).”
The effect of sample size on the accuracy of species distribution models: considering both presences and pseudo‐absences or background sites (2019, background): “Several factors have been found to affect the performance of SDMs, including sample size (Pearce and Ferrier 2000, Stockwell and Peterson 2002, Kadmon et al.”
Modelling nesting habitat as a conservation tool for the Eurasian black vulture (Aegypius monachus) in Dadia Nature Reserve, northeastern Greece (2004, methodology): “In each case, we used backward stepwise elimination to select the variables in the independent models (Pearce and Ferrier, 2000; Manel et al., 2001).”
A fuzzy classification technique for predicting species’ distributions: applications using invasive alien plants and indigenous insects (2004, background): “Currently, the most popular of these appear to be Generalized Linear Models (GLM; Austin et al, 1984; McCullagh & Nelder, 1989; Austin et al ., 1990; Austin et al ., 1994; Ferrier & Watson, 1997; Guisan et al ., 1998; Pearce & Ferrier, 2000 ) and Generalized Additive Models (GAM; Austin & Meyers, 1”
A generalized additive model for the spatial distribution of snowpack in the Spanish Pyrenees (2005, background): “This technique provides higher levels of predictive capacity than linear regressions or generalized linear models ( Pearce and Ferrier, 2000 ) and also a better understanding of the relationships between elements of environmental systems because it deals with non-linear data structures.”
Delimiting the geographical background in species distribution modelling (2012, methodology): “Calibration of the P values was assessed using Miller’s statistic, which is based on the hypothesis that the calibration line – perfect calibration – has an intercept of zero and a slope of one (for details see Miller et al., 1991; Pearce & Ferrier, 2000).”
Habitat suitability modelling of an invasive plant with advanced remote sensing data (2009, background): “A complicating factor, however, is that species distribution models and their evaluation are sensitive to the sample size (Pearce & Ferrier, 2000; Stockwell & Peterson, 2002; McPherson et al., 2004; Guisan et al., 2007), balance between presence and absence records (McPherson et al., 2004; Reese et…”
Effects of sample size on accuracy of species distribution models (2002, methodology): “In a multi-taxon evaluation of GAM and GLM approaches, sample sizes greater than 250 were needed to maximize accuracy (Pearce and Ferrier, 2000).”
Assessing the Roles of Patch Quality, Area, and Isolation in Predicting Metapopulation Dynamics (2002, methodology): “Because our goals were to mimic this data limitation and to avoid Type II errors while illustrating how the period of data collection may affect models, we did not apply Bonferroni corrections (see Pearce & Ferrier 2000).”
BIOMOD – optimizing predictions of species distributions and projecting potential future shifts under global change (2003, positive): “This does not require the calculation of a threshold to transform probability values from models to binary presence–absence form (Pearce & Ferrier, 2000; Thuiller et al. , in review-b).”