Quantitative detection of ecological boundaries and ecotones

Abstract

Quantitative detection of ecological boundaries and ecotones. Quantitative boundary-detection techniques are the main tool to study the  form  and  dynamics  of  ecotones  or  transitions  between  different  communities.  Geographic  boundary  analysis  is  the  detection  and statistical evaluation of boundary significance through the use of randomization tests. A boundary is the location in space where the rate of
change of a set of variables is the highest. To detect boundaries along transects the moving split-window technique has been very used. This method consists of calculating the dissimilarity between the two halves of a moving window which moves along the transect. To detect boundaries  of  quantitative  lattice  data  the  lattice-wombling  algorithm  has  been  widely  applied.  The  rate  of  change  in  values  among  four adjacent sampling locations forming a square is estimated by computing the first partial derivative in two directions. In the case of irregularly-spaced quantitative data and qualitative variables, the triangulation-wombling and the categorical-wombling algorithms are used, respectively.
There  are  boundary  statistics  which  describe  the  number  and  size  of  detected  boundaries.  To  study  the  degree  of  spatial  association between two sets of boundaries, several overlap statistics are defined which quantify how much spatial overlap exists and which is the distance between two types of boundaries. The significance of boundary and overlap statistics is evaluated using restricted randomization tests  of  boundaries  which  take  into  account  certain  degree  of  their  spatial  autocorrelation.  Other  boundary-detection techniques include wavelets. Analysing and comparing the location of boundaries types is relevant to understand complex ecological changes.