Topographical units and soil types prove more efficient for vegetation sample site placement than Land Type units in semi-arid savanna, North West province, South Africa

Abstract

Land Type mapping units were used to stratify the Heritage Park corridor, in the North West province, South Africa, into eight Land Type units prior to placing 109 floristic sample sites. At each site, species composition and relative frequency data were recorded for the herbaceous and woody layer. The floristic data (species presence at each site) were grouped into Land Types, topographical units and broad soil types. Each group was analysed independently using multivariate detrended correspondence analysis (DCA) and the mean similarity test. The floristic data in each Land Type showed a 42% range of similarity and considerable overlap of polygons on the DCA diagram. However, floristic data classed in topographical units reduced the range of percentage similarity to 29%. Before the Land Types map could be completely dismissed, the broader soil descriptions were compared and the percentage similarity of the four different soil forms was 28%. The detailed combination of terrain types, soil forms and climate zones constituting Land Type units was less suitable for the stratification of vegetation units than either of the broader soil types or topographic units when stratifying large savanna areas for vegetation surveys.