A first formal classification of the Hardeveld vegetation in Namaqualand, South Africa

Abstract

Questions The Succulent Karoo is a winter‐rainfall desert in southern Africa with a highly diverse flora. Which environmental factors differentiate vegetation composition at the plot level: How can compositional pattern be best represented by a formal phytosociological classification. Location Soebatsfontein communal area and surroundings in the Namaqualand Hardeveld Bioregion of the Succulent Karoo Biome, Northern Cape Province, South Africa. Methods We recorded vascular plant species composition, together with a range of structural and soil parameters as well as grazing intensity, in 355 plots of 100 m² size. This sample covered all vegetation types of the study area except those of river beds and anthropogenic sites. Vegetation classification was carried out using modified TWINSPAN followed by manual re‐arrangement of a subset of plots with the aim of increasing floristic distinctiveness of the vegetation units. For the final vegetation units, we determined diagnostic species and environmental parameters, according to phi‐values and ANOVA, respectively, and translated these into a phytosociological classification. Results The major factor influencing species composition and richness was soil salinity. At a broad scale, the less saline sites could be differentiated into sandy plains and rocky hills, and the saline sites into quartz fields, the centres of heuweltjies (termitaria) and the matrix between. We translated these patterns into two phytosociological classes with four orders, six alliances and 16 associations. Most of the syntaxa are very well defined in floristic terms, and all except the Othonnion cylindricae are new to science. The Hermannio trifurcae‐Zygophylletea morgsanae represents the zonal vegetation of marginally saline soils, the Didelto carnosae‐Cephalophylletea inaequalis the vegetation of saline soils. The vegetation of the extremely saline quartz patches probably belongs to a third class not described here due to the scarcity of representative plots. Conclusions Our phytosociological approach allowed the various degrees of ecological and floristic similarity to be represented by syntaxa of different ranks. A comparison with previous classifications from the Succulent Karoo revealed a high spatial turnover, with only two of our alliances and none of our associations coinciding with those found in previously studied areas of the Succulent Karoo. Our formal classification has the potential to be a powerful tool for local land use, ecological comparison and conservation, and to contribute to a future national phytosociological classification of South Africa.

A higher‐level classification of the Pannonian and western Pontic steppe grasslands (Central and Eastern Europe)

“…To improve the consistency between the TWINSPAN‐based groups and our formal definitions of the orders, we adjusted the order assignment of the relevés using the total cover of diagnostic species (see Luther‐Mosebach et al. for a similar approach). Relevés showing a mismatch between the total cover of diagnostic species and the initial, TWINSPAN‐based order assignment were manually re‐assigned (note that this step is similar to the “refined ordination” of the TWINSPAN algorithm).…”

A comparative framework for broad‐scale plot‐based vegetation classification

“…Li et al 2013). For example, when diagnostic species are calculated from the results of clustering, re-assignment of the plots might be necessary in order to achieve a consistent classification (Willner 2011;Luther-Mosebach et al 2012).…”

GRIMP: A machine‐learning method for improving groups of discriminating species in expert systems for vegetation classification

“…Dengler et al (2006) proposed to identify vegetation types in vegetation-plot databases based on a comparison of summed ordinal-transformed covers of diagnostic species of different vegetation types that occurred in each analyzed plot. The plot was assigned to that vegetation type whose diagnostic species had the highest summed cover in this plot (see also Michl, Dengler, &Huck, 2010 andLuther-Mosebach et al, 2012 for further application examples). Willner (2011) used essentially the same classification method, but with non-transformed percentage cover values of species.…”