On using stratified two-stage sampling for large-scale multispecies surveys - Office français de la biodiversité Accéder directement au contenu
Article Dans Une Revue Ecological Informatics Année : 2023

On using stratified two-stage sampling for large-scale multispecies surveys

Résumé

Designing a large-scale multispecies survey presents significant challenges due to its multivariate, multiscale, and often multiobjective nature. Because it is impractical to measure or observe all variables of interest at every location in the survey area, spatial sampling is crucial. To ensure a statistically sound survey, free of assumptions, randomization of the selection of spatial sampling units is required. In a large-scale survey, minimizing travel costs requires multistage sampling in which nearby sampling units are grouped within higher-level sampling units. The design may be supplemented by one or more strata systems — possibly nested — that intervene at specific sampling stages. This article focuses on stratified two-stage sampling that balances statistical efficiency and travel cost minimization. As presurvey knowledge, we assume that the spatial distribution of the species of interest is available for the sampled spatial domain, either from a spatial database or a published species distribution atlas. The multispecies nature of the survey sampling plan relies on these data. The primary sampling units are defined by the spatial grid resolution of the distribution data, while the presence data are used to stratify the PSUs to ensure that the sampling effort is concentrated where most species occur. In this article, we formally define the sampling coverage probability of a species distribution when the selection of samples is randomized according to a given probability mass function (i.e., a probability sampling plan or sampling design, in the statistical sense of the term). By maximizing the sampling coverage probability averaged over the species, the stratification and allocation of spatial sampling units concentrate observer effort where most species occur and minimize effort elsewhere. To guide and illustrate our ideas, we present as an example a multispecies waterbird program at the national scale in European France within the framework of probability sampling from finite populations.
Fichier principal
Vignette du fichier
EcolInform-2023-77-102229-Aubry.pdf (14.35 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04258681 , version 1 (25-10-2023)

Identifiants

Citer

Philippe Aubry, Gwenaël Quaintenne, Jeremy Dupuy, Charlotte Francesiaz, Matthieu Guillemain, et al.. On using stratified two-stage sampling for large-scale multispecies surveys. Ecological Informatics, 2023, 77, pp.102229. ⟨10.1016/j.ecoinf.2023.102229⟩. ⟨hal-04258681⟩

Collections

OFB
24 Consultations
19 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More