Mapping sea bird densities over the North Sea: spatially aggregated estimates and temporal changes

Abstract

SUMMARYIn the Dutch sector of the North Sea, sea bird densities are recorded bi-monthly by using airborne striptransect monitoring. From these data we try to estimate: (i) high-resolution spatial patterns of sea bird densities; (ii) low-resolution spatial-average bird densities for large areas; and (iii) temporal changes in (i) and (ii), using data on Fulmaris glacialis as an example. For spatial estimation, we combined Poisson regression for modelling the trend as a function of water depth and distance to coast with kriging interpolation of the residual variability, assuming spatial (co)variances to be proportional to the trend value. Spatial averages were estimated by block kriging. For estimating temporal differences we used residual cokriging for two consecutive years, and show how this can be extended to analyse trends over multiple years. Approximate standard errors are obtained for all estimates. A comparison with a residual simple kriging approach reveals that ignoring temporal cross-correlations leads to a severe loss of statistical accuracy when assessing the significance of temporal changes.This article shows results for Fulmaris glacialis monitored during

A spatio-temporal climate-based model of early dengue fever warning in southern Taiwan

“…In particular, this study assumes that the DF incidence over the study area can be modeled as a space-time Poisson process in which the mean of DF incidence cases is expressed as the product of a purely temporal term (driven primarily by climatic variation) and a space-time term (accounting for ratios between the recorded incidences and climate-driven incidences obtained by pure time-series modeling). Poisson regression models have been applied in spatial geostatistics studies (Diggle et al 1998;Pebesma et al 2005). In the present work, we use the Poisson model in a composite space-time domain.…”

Spatial prediction of species’ distributions from occurrence-only records: combining point pattern analysis, ENFA and regression-kriging

“…Hy-63 brid spatial/niche-analysis SDMs have been suggested 64 also by Allouche et al (2008). Pebesma et al (2005) 65 demonstrates that geostatistics is fit to be used with 66 spatio-temporal species occurrence records. Analysis of 67 spatial auto-correlation and its use in species distribu-68 tion models is now a major research issue in ecology and 69 biogeography ( Engler et al (2004) suggested a hybrid approach to 72 spatial modeling of occurrence-only records -a combi-73 nation of Generalized Linear Model (GLM) and ENFA.…”

“…7) can be indeed reproduced 54 also from a representative sample (n=721). 55 We proceed with preparing the environmental predic-56 tors and testing their correlation with the density val-57 ues. We can extend the original single auxiliary map 58 (DEM) by adding some hydrological parameters: slope, 59 topographic wetness index and altitude above channel 60 network (all derived in SAGA GIS).…”

How Many Butterflies Are There in a City of Circa Half a Million People?

“…This method results in a continuum of interpolated information conformed by predicted pixels. Due to its robustness, this method is widely used in several disciplines and is starting to become common in ecological studies [33,[52][53][54]. To generate interpolated information based on our butterfly surveys, we square root transformed the observed abundances per sampling site in order to comply with normality and skewness [55].…”