Integrating presence-only and presence-absence data to model changes in species geographic ranges: An example of yaguarundí in Latin America

Abstract

Anthropogenic changes such as land use and climate change affect species’ geographic ranges, causing range shifts, contractions, or expansions. However, data on range dynamics are insufficient, heterogeneous, and spatially and temporally biased in most regions. Integrated species distribution models (IDMs) offer a solution as they can complement good quality presence-absence data with opportunistically collected presence-only data, simultaneously accounting for heterogeneous sampling effort. However, these methods have seen limited use in the estimation of temporal change of geographic ranges and are not yet widespread as they have a steep learning curve. Here we present a generalisable model and case example to ease their adoption. Using data on presence-absence and presence-only on the yaguarundí (Herpailurus yagouaroundi), we modelled the species distribution at two time periods (2000-2013 and 2014-2021) using a Bayesian model based on Poisson point process in JAGS. Our model integrates the different types of data while accounting for varying sampling effort and spatial effect. We predicted the species range at the two time periods and quantified their changes. We found that between the two time periods, the yaguarundí has contracted its southern and northern range limits towards the equator, but expanded its area of distribution over the entire species’ range. Also, our results show that modelled geographic range (either pre or post) is not entirely consistent with the current expert range map from IUCN. Our modelling approach provides a working example with the potential to address data gaps and biases in other taxa and regions. Given the increasing number of incidental data being generated by community-derived initiatives in Latin America, IDMs can become a valuable source for species distribution modelling in the region. To our knowledge, this is the first application of the IDM approach with temporal dimension and over the entire species’ geographic range.

Spatio‐temporal integrated Bayesian species distribution models reveal lack of broad relationships between traits and range shifts

“…In recent years, SDMs have gained popularity, due to their ability to overcome many of the issues in estimating species ranges (Franklin, 2023;Kéry et al, 2013). The extension to spatiotemporal SDMs has shown promising results for estimating range shifts, representing the probability of transitions between time periods, taking into account spatial and temporal changes in sampling effort (Beale et al, 2013;Bled et al, 2013;Grattarola et al, 2023).…”

“…In recent years, SDMs have gained popularity, due to their ability to overcome many of the issues in estimating species ranges (Franklin, 2023; Kéry et al., 2013). The extension to spatio‐temporal SDMs has shown promising results for estimating range shifts, representing the probability of transitions between time periods, taking into account spatial and temporal changes in sampling effort (Beale et al., 2013; Bled et al., 2013; Grattarola et al., 2023). The recent development of the Integrated Nested Laplace Approximation (INLA) method and its associated R‐INLA package (Bakka et al., 2018; Lindgren & Rue, 2015) has made it possible to develop and run complex Bayesian SDMs with drastically reduced computation times, but similar accuracy, compared to other methods (Blangiardo et al., 2013), while accounting for common issues like spatial clumping and sparse data (Redding et al., 2017).…”

“…R‐INLA is continuously being developed, with the addition of the Stochastic Partial Differential Equation (SPDE) allowing efficient modelling of spatial autocorrelation (Lindgren et al., 2011), further enabling robust assessment of range shifts. The recent development of an INLA framework for data integration allows combining different data sources of species occurrence, potentially improving predictions for species with poor data availability (Grattarola et al., 2023; Isaac et al., 2020; Morera‐Pujol et al., 2023; Sadykova et al., 2017).…”

Integrated community models: A framework combining multispecies data sources to estimate the status, trends and dynamics of biodiversity

“…Integrated population models allow for a mechanistic understanding of species‐level processes primarily by combining demographic data with time‐series population‐ or site‐level count data (Brown & Collopy, 2013; Saunders et al, 2018). Integrated distribution models synthesize presence‐only, detection–nondetection and/or count data to estimate species distribution patterns and the effects of covariates on occurrence or abundance (Fletcher Jr. et al, 2019; Grattarola et al, 2023).…”

Reshaping biogeography: Perspectives on the past, present and future

“…In Reshaping Biogeography , we will bring more of these scientific ‘standpoints’ to biogeography in the remainder of this anniversary year. For example, in coming issues, Ariza et al (2023) will propose marine biogeographical divisions based on acoustic mapping; Grattarola et al (2023) will explore how distinct sampling efforts across species' datasets can be reconciled in integrated species distribution models; Franklin (2023) will look to the past, present and future of species distribution modelling; and Ung & Buttigieg (2023) will develop an ontology for comparative biogeography. We look forward to these and additional topics—and your responses to them or additional ideas—in future issues of this special section and beyond.…”