Water temperatures are a fundamental water quality component, and a key abiotic determinant of fish distribution patterns in rivers. A river 's thermal regime is the product of a multitude of thermal drivers and buffers interacting at different temporal and spatial scales, including, inter alia, air temperatures, flow volumes (including groundwater flows and lateral inputs from tributaries), channel geomorphology and riparian vegetation. "Healthy" river systems are self-sustaining, with adequate thermal variability to maintain biotic diversity. Temporal variability of flow volumes and water temperatures, and how these change along the longitudinal axis of a river, contribute towards a rivers "signature". Rivers that have had their signatures altered through anthropogenic impacts may no longer be sustainable, and require varying levels of management. Successful river management should include a quantification of these signatures , a definition of the "desired" state which management aims to achieve, associated "thresholds" of change or concern, and monitoring programmes. Such an approach requires flexibility and adaptability, as well as appropriate tools being available to natural resource managers. Indices, the utility of which are enhanced when included in predicative modelling systems, are a common means of assessing system variability and change. The degree of confidence placed in such tools depends on the level of fundamental science, and the degree of system understanding, underpinning them. This research contributes to the understanding of the ecological significance . of water temperatures in variable semi-arid river systems, using the Sabie River (Mpumalanga, South Africa) as a case study, and indices derived from biological indicators (Chiloglanis , Pisces: Mochokidae) to quantify the effects of cumulative changes in heat units against a hypothesised critical water temperature threshold. Hourly water temperatures for 20002002 collected at nine sites in the main rivers of the Sabie catchment, together with biannual surveys of relative abundances and community patterns of fish collected using standard electrofishing techniques, were used to provide the basis for a modelling system which aims to provide river managers with a tool for quantifying changes to the thermal regime of the Sabie River. This modelling system consisted of a suite of pragmatic models, including multiple linear regression models for simulating daily maximum water temperatures, and simple cause-and-effect relationships between biological indices (change In condition factor and change in the ratio of relative abundances of two species of Chiloglanis) and annual metrics of time-of-exposure to heat stress. It was concluded that changes in the thermal regimes of the rivers in the Sabie catchment are likely to lead to changes in fish distribution patterns, and a decline in river health. Inherent system variability suggests that management decisions will be made in the face considerable uncertainty. Indirect management of water temperatures may be possible through maintenance of flow volumes and flow variability. However, the most appropriate management approach for maintaining fish diversity within these rivers is to ensure that obstacles to fish migration are minimized, to maximise the ability of river biota to respond to thermal changes, by accessing suitable alternative habitats or refugia. Future research should focus on extending the time series of water temperatures from such river systems, and further understanding the drivers and buffers contributing to the thermal regimes of variable semi-arid river systems in South Africa. Additional testing of the validity of the hypothesized relationships between abiotic processes underpinning biotic patterns should be undertaken. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 2003.