TY - JOUR

T1 - A joint distribution framework to improve presence-only species distribution models by exploiting opportunistic surveys

AU - Molgora, Juan M. Escamilla

AU - Sedda, Luigi

AU - Diggle, Peter

AU - Atkinson, Peter M.

PY - 2021/6/30

Y1 - 2021/6/30

N2 - Aim We propose a Bayesian framework for modelling species distributions using presence-only biodiversity occurrences obtained from historical opportunistic surveys.Location Global applicability with two case studies in south-east Mexico.Methods The framework defines a bivariate spatial process separable into ecological and sampling effort processes that jointly generate occurrence observations of biodiversity records. Presence-only data are conceived as incomplete observations where some presences have been filtered out. A choosing principle is used to separate out presences, missing data and absences relative to the species of interest and the sampling observations. The framework provides three modelling alternatives for accounting the spatial autocorrelation structure: independent latent variables (model I); common latent spatial random effect (model II); and correlated latent spatial random effects (model III).The framework was compared against the Maximum Entropy (MaxEnt) algorithm in two case studies: one for the prediction of pines (Class: Pinopsida), using botanical records as sampling observations and another for the prediction of Flycatchers (Family: Tyranidae), using bird sightings as sampling records.ăResults In both case studies, at least one of the proposed models achieved higher predictive accuracy than MaxEnt. The model with correlated spatial effects fit best when the sampling effort was informative, while the one with a shared spatial effect was more suitable in cases with high proportion of non sampled sites.Main Conclusions Our approach provides a flexible framework for presence-only SDMs aided by a sampling effort process informed by the accumulated observations of independent and heterogeneous surveys. For the two case studies, the framework provided a model with a higher predictive accuracy than an optimised version MaxEnt.

AB - Aim We propose a Bayesian framework for modelling species distributions using presence-only biodiversity occurrences obtained from historical opportunistic surveys.Location Global applicability with two case studies in south-east Mexico.Methods The framework defines a bivariate spatial process separable into ecological and sampling effort processes that jointly generate occurrence observations of biodiversity records. Presence-only data are conceived as incomplete observations where some presences have been filtered out. A choosing principle is used to separate out presences, missing data and absences relative to the species of interest and the sampling observations. The framework provides three modelling alternatives for accounting the spatial autocorrelation structure: independent latent variables (model I); common latent spatial random effect (model II); and correlated latent spatial random effects (model III).The framework was compared against the Maximum Entropy (MaxEnt) algorithm in two case studies: one for the prediction of pines (Class: Pinopsida), using botanical records as sampling observations and another for the prediction of Flycatchers (Family: Tyranidae), using bird sightings as sampling records.ăResults In both case studies, at least one of the proposed models achieved higher predictive accuracy than MaxEnt. The model with correlated spatial effects fit best when the sampling effort was informative, while the one with a shared spatial effect was more suitable in cases with high proportion of non sampled sites.Main Conclusions Our approach provides a flexible framework for presence-only SDMs aided by a sampling effort process informed by the accumulated observations of independent and heterogeneous surveys. For the two case studies, the framework provided a model with a higher predictive accuracy than an optimised version MaxEnt.

U2 - 10.1101/2021.06.28.450233

DO - 10.1101/2021.06.28.450233

M3 - Journal article

JO - Biorxiv

JF - Biorxiv

ER -