Minimum required number of specimen records to develop accurate species distribution models

Abstract

Species distribution models (SDMs) are widely used to predict the occurrence of species. Because SDMs generally use presence-only data, validation of the predicted distribution and assessing model accuracy is challenging. Model performance depends on both sample size and species' prevalence, being the fraction of the study area occupied by the species. Here, we present a novel method using simulated species to identify the minimum number of records required to generate accurate SDMs for taxa of different pre-defined prevalence classes. We quantified model performance as a function of sample size and prevalence and found model performance to increase with increasing sample size under constant prevalence, and to decrease with increasing prevalence under constant sample size. The area under the curve (AUC) is commonly used as a measure of model performance. However, when applied to presence-only data it is prevalence-dependent and hence not an accurate performance index. Testing the AUC of an SDM for significant deviation from random performance provides a good alternative. We assessed the minimum number of records required to obtain good model performance for species of different prevalence classes in a virtual study area and in a real African study area. The lower limit depends on the species' prevalence with absolute minimum sample sizes as low as 3 for narrow-ranged and 13 for widespread species for our virtual study area which represents an ideal, balanced, orthogonal world. The lower limit of 3, however, is flawed by statistical artefacts related to modelling species with a prevalence below 0.1. In our African study area lower limits are higher, ranging from 14 for narrow-ranged to 25 for widespread species. We advocate identifying the minimum sample size for any species distribution modelling by applying the novel method presented here, which is applicable to any taxonomic clade or group, study area or climate scenario.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

An early suitability assessment of two exotic Ophraella species (Coleoptera: Chrysomelidae) for biological control of invasive ragweed in Europe

“…The predictive power of the models was tested using the area under the receiver-operator characteristic function (area under the curve criteria, AUC) for evaluation (Pearce & Ferrier, 2000). Given the high AUC values (> 0.8) (van Proosdij et al, 2016) of each modelling approach for all species, the predictions of the models are robust. In order to visually overlay the insects distributions on that of A. artemisiifolia, the probabilities of presence of the insects were then binary transformed into presences and absences for each raster pixel (see details in Sun et al, 2017).…”

Planning priority conservation areas under climate change for six plant species with extremely small populations in China

“…These projections kept the non-climatic variables constant into the future, with only the climate variables changing in accordance with these scenarios (Wang et al 2016). Maxent is appropriate for this type of modelling for a variety of reasons: (1) it can be used with small sample sizes, which drastically impact both the performance and the adjustment of ENM (Pearson et al 2007, Merow et al 2013, Fourcade et al 2014, Proosdij et al 2016; (2) It is insensitive to multicollinearity amongst predictors, which can impede the analysis of species-environment relationships in multiple regression settings; finally, (3) it provides the relative contribution of each variable as an output (Pearson et al 2007, Merow et al 2013, Phillips et al 2017. All grid cells were assumed to be possible distribution space with maximum entropy (Phillips et al 2006, Merow et al 2013.…”

Proposed biodiversity conservation areas: gap analysis and spatial prioritization on the inadequately studied Qinghai Plateau, China

“…This system collected the latest endangered species information including mammals, birds, reptiles, amphibians, fish species or subspecies in China. Both the theoretical and practical simulations show that when the number of species presence points is greater than 14, the species distribution model can produce a better simulation result of species habitat (Proosdij et al 2015). Therefore, we excluded these species with less than 15 presence points from the two databases, and obtained species presence data for 59 key rare and endangered species (Table 2).…”