This study tested two key assumptions in developing regional biocriteria: (1) the accuracy of the ecoregion classification framework and (2) the accuracy of standardized qualitative sampling. Except for the Central Appalachians ecoregion, there was little or no correspondence of benthic macroinvertebrate distribution with the ecoregions or subregions of western Virginia. I found that it was more accurate to rearrange the subregions into three larger regions called bioregions: the forested hills and mountains (subregion 69a), valleys and plateaus (subregions 66c, 67a, and 67b), and the mountains (subregions 66a, 66b, 67c, and 67d). As an alternative to the ecoregion classification scheme, I classified my reference sites in biotic groups and then assessed the effect of several environmental variables on discriminating between the groups. There was a 69.8% correct classification rate using 14 environmental variables. Stepwise multiple discriminant analysis and graphical analysis showed that sampling date, slope, pH, habitat assessment score and distance to source were the best predictors of community structure. These environmental variables correctly classified 52.8% of the reference sites. These classification rates are comparable to rates published in similar studies. My study demonstrated that aggregations of subregions into bioregions and a biotic approach are more accurate classification schemes than ecoregions or subregions for biocriteria based on benthic macroinvertebrates. A second study evaluated the accuracy of a standardized qualitative sampling approach, commonly used in rapid bioassessments, in assessing the biological condition of lotic systems. I compared a typical standardized qualitative sampling method with subsampling with a typical quantitative sampling method on a stream with varying degrees of impairment. Although some metrics did respond to differences in sample abundance, overall the two methods made similar estimates of community composition. The two sampling methods made the same assessment an average 89% of the time using multimetric index developed for the Mid-Atlantic Region. I found no pattern showing one method was more accurate in making assessments of biological condition than the other. Given the greater time and costs associated with quantitative sampling methods, I conclude that standardized qualitative methods, are preferable for rapid bioassessment approaches to environmental assessment. / Ph. D.

This study records all known legume (Leguminosae/Fabaceae) species indigenous to South Africa, Lesotho and Swaziland to establish distribution patterns and optimum climatic and soil conditions for growth. The main purpose was to propose a list of legume species for further evaluation of their pasture potential. Collection data supplied by the National Herbarium (PRE) Computerised Information System were recorded to establish the distribution patterns of species based on the bioregions vegetation map. A total of 1 654 species are known to be indigenous, representing 24 tribes and 122 genera. The grouping of legume species into five main clusters and 16 Leguminochoria is ecologically described, with the highest legume species richness found in the Northern Mistbelt Forest. Key and diagnostic species are provided for each Leguminochorion. Soil pH and mean annual minimum temperature were found to be the main drivers for distinguishing between legume assemblages. The optimum climatic and soil conditions for growth are described as well as the available descriptive attributes for species recorded. Information on the range of tolerance of most species to abiotic factors is presented. Mean annual rainfall and soil pH are highly correlated with the distribution pattern of most species, followed by mean annual minimum temperature. Legume species adapted to a wide range of soil pH levels and low soil phosphorus levels are recorded. Existing data on the cultivation and grazing or browsing status of indigenous legumes were used to select 584 species found mainly in the Central Bushveld, Mopane and Lowveld Bioregions to be further evaluated for their pasture potential. Known characteristics were used to categorise species. Species contained in the tribe Phaseoleae are of special interest since it contains most of the genera with present-day agricultural value, i.e. Eriosema, Rhynchosia and Vigna species are listed as having high potential as pasture species. This study has shown that the descriptive and distribution data accumulated by botanists (notably taxonomists) could be of beneficial use in meeting agricultural objectives. Indigenous legumes are adapted to a wide range of soil and climatic conditions and represent a valuable but largely unexploited natural resource for pasture development and soil conservation practices. / Thesis (PhD)--University of Pretoria, 2013. / gm2014 / Plant Science / unrestricted