More complex distribution models or more representative data

By Jorge M. Lobo. 2008

152 citations recorded by the source corpus.

Distribution models for species are increasingly used to summarize species’ geography in conservation analyses. These models use increasingly sophisticated modeling techniques, but often lack detailed examination of the quality of the biological occurrence data on which they are based. I analyze the results of the best comparative study of the performance of different modeling techniques, which used pseudo-absence data selected at random. I provide an example of variation in model accuracy depending on the type of absence information used, showing that good model predictions depend most critically on better biological data.

Subfields: Artificial Intelligence, Ecological Modeling, Epidemiology

View Paper: https://www.semanticscholar.org/paper/d834b373d841ab2d0cedd02beb9b1972c55a2504

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

In Epidemiology

Citations in the corpus

Use of niche models in invasive species risk assessments (2011, background): “Variables thus identified can then be prioritized for inclusion in ecological niche models to estimate the potential of colonization, but inclusion of variables that are closely related to contingent effects such as dispersal barriers and biotic interactions should be avoided (Lobo 2008b).”
Not as good as they seem: the importance of concepts in species distribution modelling (2008, methodology): “Here, absence data must come from environmental conditions that are known to be unsuitable for the species (Chefaoui & Lobo, 2008).”
High uncertainty in the effects of data characteristics on the performance of species distribution models (2021, methodology): “…studied region (Relative Occurrence Area, ROA) determines predictive success, as the rate of well-predicted absences increases when the extent of analyses is much larger than the species distribution, hence inflating performance metrics (Lobo, 2008; Lobo et al., 2008; Moudrý and Šímová, 2013).”
A strict maximum likelihood explanation of MaxEnt, and some implications for distribution modelling (2013, methodology): “Almost all PO data sets used for DM are, however, strongly biased (Kodric-Brown & Brown 1993, Vaughan & Ormerod 2003, Kadmon et al. 2004, Edwards et al. 2006, Hortal et al. 2008, Lobo 2008, Robertson et al. 2010, Wolmarans et al. 2010, McCarthy et al. 2011).”
Modelling and predicting fungal distribution patterns using herbarium data (2008, methodology): “By including absence data in the data set, we will probably reduce bias and improve model predictions (Lobo, 2008).”
Cross-validation of species distribution models: removing spatial sorting bias and calibration with a null model. (2012, background): “Key words: AUC; Bioclim; cross-validation; MaxEnt; model evaluation; niche model; pairwise distance sampling; spatial autocorrelation; spatial sorting bias; species distribution model.”
Discrimination capacity in species distribution models depends on the representativeness of the environmental domain (2013, background): “As pointed out by some authors, data quality is probably the most important factor influencing general model performance, an aspect to which much more effort and resources should be devoted (Lobo, 2008; Jiménez-Valverde et al., 2010; Feeley & Silman, 2011; Rocchini et al., 2011).”
Effects of the number of presences on reliability and stability of MARS species distribution models: the importance of regional niche variation and ecological heterogeneity (2010, background): “As already pointed out by Lobo (2008), we think that adding new data will be far more beneficial than developing more complex techniques (Lobo 2008).”
Unprotecting the rare species: a niche‐based gap analysis for odonates in a core Cerrado area (2011, background): “, 2006; Brown & Lomolino, 2008) and support the plea for augmenting the quantity and quality of biogeographical data for conservation purposes (Gaston & Rodrigues, 2003; Hortal & Lobo, 2005; Romo et al., 2006; Lobo, 2008).”
Using species distribution modelling to disentangle realised versus potential distributions for rare species conservation (2013, background): “Yet, when no absence data is available, what we get are distributions close to potential ranges for the species (Jiménez-Valverde et al., 2008; Lobo, 2008).”
The effect of the extent of the study region on GIS models of species geographic distributions and estimates of niche evolution: preliminary tests with montane rodents (genus Nephelomys) in Venezuela (2010, methodology): “Extending recently developed ideas (Bahn & McGill, 2007; Chefaoui & Lobo, 2008), this study addresses how the extent of the study region may influence results in correlative modelling.”
Geographical patterns in prediction errors of species distribution models (2011, background): “increase the probability of making false presence errors (Lobo, 2008) + Relative occurrence area”
Uncertainty associated with survey design in Species Distribution Models (2014, methodology): “When this assumption is violated by biases in the collection of data, the accuracy of model predictions can be affected (Hortal et al., 2008; Lobo, 2008; Loiselle et al., 2008; Rocchini et al., 2011).”
Modelling potential distribution of the threatened tree species Juniperus oxycedrus: how to evaluate the predictions of different modelling approaches? (2011, unknown): “Consequently, in evaluating presence-only models, the ability to give a realistic prediction of species occupancy potential in the study area is considered to be more important than model accuracy (Lobo 2008; Lobo et al. 2008).”
Accounting for uncertainty when mapping species distributions: The need for maps of ignorance (2011, background): “Field efforts by naturalists were largely biased to classical localities of particular natural beauty, localities of high species richness in the past, and to areas near the experts’ residences and/or research centres (Dennis and Thomas, 2000; Lobo, 2008a, 2008b; see Table 1).”
Profile or group discriminative techniques? Generating reliable species distribution models using pseudo‐absences and target‐group absences from natural history collections (2010, background): “Bias may influence the reliability of the generated models, but in this study, we focus on the importance of generating reliable absences as a key strategy for producing high-quality models (Lobo, 2008).”