Figure 2

Characterisation of the impala population sub-units. The spatial structure has been identified by a hierarchical clustering analysis applied on all the locations of marked impalas monitored over the study area during the study period (2002 and 2003); a) distribution of the cluster criterion value (i.e. the dendrogram node height) in relation to the numbers of clusters; b) box plots of latitude values (interquartile range and outliers around the median) of locations of impalas for the two spatial clusters in each season (one unit along the y axis is equivalent to 1000 m); c) distribution of locations assigned to the two spatial clusters (north and south units). Group size illustrates the number of marked impalas when observed in a common herd.