Inferring fine scale wild species distribution from spatially aggregated data - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2022

Inferring fine scale wild species distribution from spatially aggregated data

Résumé

In spatial ecology, huge amount of aggregated spatial data such as hunting data or fishermen declarations data offer possibilities to map wild species distribution at fine scale by combining them with high resolution data. However, this requires to properly handle the difference in spatial resolution between the different data sources. Such issue is often referred as the change of support (COS) problem. In ecological applications, accounting for COS can be challenging as the observations data can be complex (e.g. zero-inflated positive continuous data) and this can complicate the way COS is handled. In this paper, we develop a hierarchical approach that allows (1) to handle COS for a mixture of zero-inflated positive continuous data and (2) to combine fine scale data and aggregated data. We develop and apply the approach based on a fishery application where fishermen declarations data are registered at rough scale, but are used in combination with scientific survey data that are exactly geolocalized to infer fine scale species distribution. We compare (1) a rough but standard way to refine the resolution of declarations by proportionally reallocating declarations on fishing locations and (2) a model that handle COS by explicitly modeling the probability distribution of the aggregated declaration conditionally upon the exact locations observation. The rough approach leads to a loss of the species-habitat relationship, to smoothed maps of species distribution and to an overweighted contribution of declarations data into inference in comparison with scientific data. By contrast, the COS approach allows to provide unbiased estimates of the habitat effect and more accurate spatial predictions. Furthermore, scientific data contributes in a more significant way to inference. This approach is a valuable contribution for a wider use of spatially aggregated data in spatial ecology in order to properly integrate datasets that do not have the same spatial resolution to make fine scale inferences of species distribution.
Fichier principal
Vignette du fichier
draft_cos.pdf (4.21 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03878990 , version 1 (30-11-2022)
hal-03878990 , version 2 (09-01-2023)

Identifiants

  • HAL Id : hal-03878990 , version 1

Citer

Baptiste Alglave, Kasper Kristensen, Etienne Rivot, Mathieu Woillez, Youen Vermard, et al.. Inferring fine scale wild species distribution from spatially aggregated data. 2022. ⟨hal-03878990v1⟩
95 Consultations
131 Téléchargements

Partager

Gmail Facebook X LinkedIn More