On the identification of high mortality rate hotspots.

Borda-de-Agua, L., Ascensão, F., Barrientos, R. y Pereira, H.M. On the identification of high mortality rate hotspots. 2016 IENE International Conference. Programme and abstracts. Lyon: Infrastructure & Ecology Network Europe. 2016, Vol. 3.11, Num. 389, pag. 283. ISSN 978-2-37180-153-0

One of the main tasks in road ecology is to identify hotspots of high mortality so that one can devise and implement mitigation measures. A common strategy to identify hotspots is to divide a road into several segments and determine when the number of collisions is above a given threshold, reflecting a desired significance level obtained assuming a probability distribution (often the Poisson). The problem of this approach when applied to each segment individually is that the probability of identifying false hotspots is very high, i.e., the probability of making a type I error is very high. For instance, if we establish the threshold based as the top of a 95% confidence interval, then one should expect to incorrectly identify just by chance five false hotspots in every 100 segments. Although one may argue that such overly cautionary approach may be beneficial from a biological conservation perspective, it may lead to the waste of resources and, probably worse, it may raise doubts on the methodology adopted and the credibility of those suggesting it. The problem of multiple comparison occurs in several scientific areas and several corrections have been suggested. Here, we apply three different approaches to the identification of hotspots: a method similar to that of the Bonferroni correction; the false discovery rate (FDR); and a Bayesian approach that consists of a hierarchical Poisson model. The Bonferroni approach reduces the probability of type I errors, yet the probability of type II errors (rejecting a true hotspot), is very high and thus this procedure has low power. FDR method increases the power of the test while keeping the probability of identifying false hotspots low. The Bayesian approach uses the information obtained from all segments to infer the probability of a segment being a hotspotand avoids some of the problems inherent to the two previous approaches. We discuss the application of these three methods and give recommendations to the identification of hotspots with a view to providing a wide range of practitioners’ procedures that are reliable and simple to use in real situations.

Datos y Recursos

Metadatos
Descargar
DCAT-AP/NTI-RISP

Información básica
Tipo de recurso Texto
Fecha de creación 17-09-2024
Fecha de última modificación 17-09-2024
Mostrar histórico de cambios
Identificador de los metadatos e6d4955a-6b45-5c33-ad80-cc727b34ea05
Idioma de los metadatos Español
Temáticas (NTI-RISP)
Categoría del conjunto de alto valor (HVD)
Categoría temática ISO 19115 biota
Identificador alternativo
URI de palabras clave
Codificación UTF-8
Información espacial
Identificador INSPIRE ESPMITECOIEPNBFRAGM623
Temas INSPIRE
Identificador geográfico España
Sistema de Referencia de Coordenadas
Tipo de representación espacial
Extensión espacial
"{\"type\": \"Polygon\", \"coordinates\": [[[-18.16, 27.64], [4.32, 27.64], [4.32, 43.79], [-18.16, 43.79], [-18.16, 27.64]]]}"
Resolución espacial del dataset (m)
Procedencia
Declaración de linaje
Perfil de Metadatos
Conformidad
Conjunto de datos de origen
Frecuencia de actualización
Fuentes
  1. 2016 IENE International Conference. Programme and abstracts. Vol. 3.11
  2. Num. 389
  3. pag. 283
Propósito
Pasos del proceso
Cobertura temporal (Inicio)
Cobertura temporal (Fin)
Notas sobre la versión
Versión
Vigencia del conjunto de datos
Parte responsable
Nombre del autor Borda-de-Agua, L., Ascensão, F., Barrientos, R. y Pereira, H.M.
Nombre del mantenedor
Identificador del autor
Email del autor
Web del autor
Identificador del mantenedor
Email del mantenedor
Web del mantenedor