A joint distribution framework to improve presence‐only species distribution models by exploiting opportunistic surveys. (29th April 2022)
- Record Type:
- Journal Article
- Title:
- A joint distribution framework to improve presence‐only species distribution models by exploiting opportunistic surveys. (29th April 2022)
- Main Title:
- A joint distribution framework to improve presence‐only species distribution models by exploiting opportunistic surveys
- Authors:
- Escamilla Molgora, Juan M.
Sedda, Luigi
Diggle, Peter
Atkinson, Peter M. - Abstract:
- Abstract: Aim: The availability of data related to species occurrences has favoured the development of species distribution models using only observations of presence. These data are intrinsically biased by the sampling effort. Presence‐only (PO) species distribution models (SDM) typically account for this effect by introducing additional data considered to be related with the sampling. This approach, however, does not allow the characterisation of the sampling effort and hinders the interpretation of the model. Here, we propose a Bayesian framework for PO SDMs that can explicitly model the sampling effect. Location: Mexico. Taxon: Pines, flycatchers (family Tyranidae), birds and plants. Methods: The framework defines a bivariate process separable into ecological and sampling effort processes. PO data are conceived of 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 to account for a 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 and another for the prediction of flycatchers. Results: In both caseAbstract: Aim: The availability of data related to species occurrences has favoured the development of species distribution models using only observations of presence. These data are intrinsically biased by the sampling effort. Presence‐only (PO) species distribution models (SDM) typically account for this effect by introducing additional data considered to be related with the sampling. This approach, however, does not allow the characterisation of the sampling effort and hinders the interpretation of the model. Here, we propose a Bayesian framework for PO SDMs that can explicitly model the sampling effect. Location: Mexico. Taxon: Pines, flycatchers (family Tyranidae), birds and plants. Methods: The framework defines a bivariate process separable into ecological and sampling effort processes. PO data are conceived of 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 to account for a 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 and another for the prediction of flycatchers. Results: In both case studies, at least one of the proposed models achieved higher predictive accuracy than MaxEnt. The model III fit best when the sampling effort was informative, while model II was more suitable in cases with a high proportion of non‐sampled sites. Main Conclusions: Our approach provides a flexible framework for PO 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 of MaxEnt. Resumen: Objetivo: La disponibilidad de datos relacionados a la ocurrencia de especies ha favorecido el desarrollo de modelos de distribución de especies que solo usan observaciones de presencia. Estos datos están intrínsicamente sesgados por el esfuerzo de muestreo. Como consecuencia, los modelos de distribución de especies de solo presencia típicamente modelan este efecto introduciendo datos adicionales relacionados con un esfuerzo de muestreo genérico. Sin embargo, esta metodología no permite modelar el esfuerzo de muestro explícitamente, dificultando la interpretación del modelo. En este trabajo proponemos un marco de trabajo bayesiano para modelos de distribución de especie de solo presencia que pueden modelar explícitamente el esfuerzo de muestreo. Ubicación: México. Taxa: pinos, tiránidos, aves y plantas. Métodos: El marco de trabajo define un proceso bivariado separable en dos procesos, uno ecológico y otro para el esfuerzo de muestreo. Los datos de solo presencia se consideran como observaciones incompletas donde algunos datos de presencia han sido filtrados. Un principio de selección es usado para clasificar observaciones entre presencias, datos faltantes y ausencias relativas a la especie de interés y muestreo. El marco brinda tres alternativas para modelar la estructura de correlación espacial: efectos aleatorios espaciales independientes (modelo I), efecto aleatorio espacial común (modelo II) y efectos aleatorios espaciales correlacionados (modelo III). El marco fue comparado contra el algoritmo de máxima entropía usando dos casos de estudio: uno para predecir pinos y otro para predecir presencia de tiránidos. Resultados: En los dos casos de estudio al menos uno de los modelos propuestos obtuvo una exactitud predictiva mayor que MaxEnt. El modelo III obtuvo un ajuste óptimo cuando el esfuerzo de muestreo fue informativo, mientras que el modelo II resultó ser más adecuado en casos con una alta proporción de sitios no muestreados (datos faltantes). Conclusiones principales: Proponemos un marco de trabajo flexible para modelos de distribución de especies basados en datos de solo presencia ayudado por un proceso de esfuerzo de muestreo informado por observaciones independientes de estudios de campo. Para los dos casos de estudio, el marco obtuvo mayores exactitudes predictivas que el modelo más optimizado de MaxEnt. … (more)
- Is Part Of:
- Journal of biogeography. Volume 49:Number 6(2022)
- Journal:
- Journal of biogeography
- Issue:
- Volume 49:Number 6(2022)
- Issue Display:
- Volume 49, Issue 6 (2022)
- Year:
- 2022
- Volume:
- 49
- Issue:
- 6
- Issue Sort Value:
- 2022-0049-0006-0000
- Page Start:
- 1176
- Page End:
- 1192
- Publication Date:
- 2022-04-29
- Subjects:
- aggregated areal data -- conditional autoregressive models -- maximum entropy benchmark -- presence‐only data -- sampling bias -- sampling effort -- species distribution models
Biogeography -- Periodicals
578.09 - Journal URLs:
- http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1365-2699 ↗
http://onlinelibrary.wiley.com/ ↗ - DOI:
- 10.1111/jbi.14365 ↗
- Languages:
- English
- ISSNs:
- 0305-0270
- Deposit Type:
- Legaldeposit
- View Content:
- Available online (eLD content is only available in our Reading Rooms) ↗
- Physical Locations:
- British Library DSC - 4952.900000
British Library DSC - BLDSS-3PM
British Library HMNTS - ELD Digital store - Ingest File:
- 21517.xml