Scale and spatial analysis

Abstract

Scale and spatial analysis. The issue of scale underpins any challenge of spatial analysis in ecology because ecological heterogeneity strongly  depends  on  the  scale.  An  accurate  view  of  scale  needs  the  discrimination  between  the  ecological,  sampling  and  analysis dimensions of the scale concept. Our ability to infer ecological mechanisms depends on how sampling and analysis scales fit to the real spatial dimensions of ecological phenomena. The major analytical concerns of ecologists respecting to scale are: i) the variance allocation of ecological parameters across scales, and ii) the change in covariation among variables across scales. For both concerns, we may consider
space explicitly by means of analysis units exclusively defined by spatial features such as position, extent and distance, but also implicitly, by using levels of biological and structural heterogeneity that necessarily associate to differences in spatial features (e.g. microhabitats, habitats). A frequent constraint emerging in scaling approaches is the reduction of sample size towards wider scales, with the consequent decrease in our ability to detect significant variation and co-variation. The recently developed Analysis of Principal Coordinates of Neighbour Matrices overcomes this constraint, enabling to dissect the spatial variance of ecological parameters as well as to generate fitted values amenable as response variables in co-variation analyses, with homogeneous analytical power across different spatial scales.