Using species distribution models at local scale to guide the search of poorly known species: Review, methodological issues and future directions

Abstract

Among specific applications of species distribution models (SDMs), the use of SDMs probabilistic maps for guiding field surveys is increasingly applied. This approach is particularly used for poorly known and/or cryptic species in order to better assess their distribution. One of the most interesting aspects of these applications is that predictions could be clearly validated by real data, subsequently obtained in the field. Despite this important difference from other applications, to our knowledge, the efficiency of different algorithms, metrics for model evaluation and algorithm-specific settings have not yet been sufficiently investigated. This research performs a literature survey to investigate which species, study area characteristics, variables, techniques and settings were used or suggested by previous authors. We then applied the most common approaches to guide field surveys for a set of 70 vascular plants in an endemic-rich area of Sardinia (Italy) of approx. 9000 ha, the flora of which was deeply investigated during the last two years. Our main aims were: (1) to use pre-model records for predicting the potential occurrence of plant species with different sample size, detectability and habitat preference, (2) to apply results for guiding searches for new populations of poorly known species, (3) to calculate the model performance according to independent real presence/absence data (testAUC) and (4) to compare different modelling data input and settings on the testAUC basis. By emphasizing the importance of field verification, both the review and the worked example supported the reliability of SDMs for a wide range of species to understand where species could potentially be present and therefore to optimise resources for the search of new species localities. This study may help understand and summarise the most applied methodological approaches and to highlight future directions for this practical application. Without underrating the importance of most common recommendations, practitioners are encouraged to test the ability of this SDMs' application with their own data. Indeed, large gaps in species' types (e.g. insects) and in regions covered by these kind of studies (e.g. many African and Asian territories) were found. Furthermore, eventual biases due to lack of data, experience or staff, have in this experimental case less irreparable consequences than others, such as conservation assessments based on future projections, which cannot be otherwise adjusted by explicit data from ground validation.

Modelling the current and future biodiversity distribution in the Chilean Mediterranean hotspot. The role of protected areas network in a warmer future

“…However, the data problem is a very complicated setback for most areas in the world (Cardoso, Erwin, Borges, & New, ; Scherson, Fuentes‐Castillo, Urbina‐Casanova, & Pliscoff, ). SDMs have accounted for paucity of georeferenced data (Fois, Cuena‐Lombraña, Fenu, & Bacchetta, ), and as this study shows the possibility of using them to extrapolate occurrences in areas where fewer locality information is found, taking into account limitations regarding number of independent occurrences and uneven distribution of species information in the datasets. However, not enough DNA data are yet available—at least for plants—for doing phylogenies of floras of very large areas at the species level.…”

“…SDMs have accounted for paucity of georeferenced data (Fois, Cuena-Lombraña, Fenu, & Bacchetta, 2018), and as this study shows the possibility of using them to extrapolate occurrences in areas where fewer locality information is found, taking into account limitations regarding number of independent occurrences and uneven distribution of species information in the datasets.…”

A field‐validated species distribution model to support management of the critically endangered Poweshiek skipperling (Oarisma poweshiek) butterfly in Canada

“…Anderson et al (2016) found a correlation of only 10% between presence/absence of their predicted species and Maxent-predicted probability of occurrence and concluded that their model failed. Fois, Cuena-Lombraña, Fenu, and Bacchetta (2018) tested Maxent models against independent datasets and concluded reliability was low. Although West, Kumar, Brown, Stohlgren, and Bromberg (2016) found that detection of their target species was more frequent in higher predicted habitat suitability classes, they did not statistically quantify prediction accuracy.…”

Essential environmental variables to include in a stratified sampling design for a national-level invasive alien tree survey

“…A set of physiologically relevant environmental variables that have been shown to correlate with species abundance were included, namely climatic, topographic (terrain) and soil related variables (Williams et al 2012, Hageer et al 2017, Fois et al 2018. The climatic variables were obtained from the WorldClim2 dataset (Fick & Hijmans 2017).…”

“…IAP distribution, but rather to use modelling to support the development of a stratification that could be used in a sampling design (Särndal 2010) in order to quantify IAP abundance based on a representative grid of empirical sampling points. Similar approaches have been used to guide surveys for example where field surveys are limited due to a lack of resources (Fois et al 2018). In these cases post-model field surveys were targeted on where a high probability of occurrence was predicted but without pre-model field data (Peterman et al 2013).…”