Whole Effluent Toxicity (WET) testing has been identified as one of the tools in the management of complex effluents in aquatic ecosystems. In South Africa, toxicity testing has not been required for regulatory purposes. Recently, the Department of Water Affairs and Forestry has adopted WET testing as a tool to evaluate the suitability of hazardous effluent for discharge into receiving environments. This has necessitated suitable procedures to be established for use in the South African situation. With the implementation of the new National Water Act (No 36 of 1998), industries have to comply with set standards to protect the aquatic environment. However, the South African Water Quality Guidelines for Aquatic Ecosystems have been set using international toxicity data, and it is not known if they are comparable with South African conditions. The aim of this study was to investigate the tolerances of selected indigenous riverine invertebrates to complex saline effluents. The study investigated the effects of kraft mill effluent to Tricorythus tinctus, a tricorythid mayfly from the Sabie River, Mpumalanga, and the effects of a textile effluent to baetid mayflies of the Buffalo River, Eastern Cape. Indigenous riverine invertebrates were chosen as test organisms, as there is no toxicity data in South Africa which could be used to evaluate the level of protection afforded by the South African Water Quality Guidelines for Aquatic Ecosystems. The use of indigenous riverine invertebrates added the challenge of variability of a wild population, and the use of a complex effluent as toxicant added the variability of effluent composition. In this study, WET testing was used to determine the dilution of whole effluents required for discharge. Hazard-based guidelines were developed for the disposal of kraft and textile effluents. The level of environmental hazard posed by different effluent concentrations was ranked, and was related to the River Health Class. This indicated effluent concentrations that may be allowed to enter the aquatic environment, e.g. 3% effluent concentration guideline for both general kraft effluent and general textile effluent for the protection of a Class A river. This approach could contribute to the use of an Environmental Risk Assessment, approach for the management of complex effluents. A number of acute 96 hour toxicity tests were conducted following an unreplicated regression design, using kraft and textile effluents as toxicants, mayfly nymphs as test organisms, and river water as diluent and control. Test organisms were sampled from unimpacted, flowing-water riffle areas, and were exposed in recirculating artificial streams (or channels) to a range of effluent concentrations. Mortality was selected as end-point and observed twice daily. The experimental results showed the variability and acute toxicity of both kraft and textile mill effluents. Baetids were more sensitive (mean LC50=16% effluent concentration) to General Textile Effluent (GTE), but less sensitive to Post Irrigation Textile Effluent (PITE). Textile effluent (PITE) held in a holding dam were therefore less variable and less toxic; suggesting that stabilization of the effluent could have contributed to reduced toxicity. Effluent composition, e.g. higher calcium levels, may also have contributed to lowering toxicity. T. tinctus was sensitive to kraft effluents, but showed less variable responses to Irrigation Kraft Effluent than General Kraft Effluent. Toxicity test data indicated that GKE, IKE and GTE should not enter the aquatic environment without treatment, as they can cause adverse effects to aquatic biota. Both kraft and textile effluents must therefore be treated before discharge. Different responses to different effluent batches were probably due to effluent variability. The use of indigenous organisms, and not a standard laboratory organism, could also have contributed to variability. A hazard-based approach could be useful, as it will provide a consistent basis for deciding on the acceptability of impacts, while allowing natural site-specific differences to be taken into account.

Water resources are under ever-increasing pressure to meet the demands of various water users both nationally and internationally. The process of anthropogenically-induced salinisation serves to exacerbate this pressure by limiting the quantity and quality of water available for future use. Water quality guidelines provide the numerical goals which water resource managers can use to adequately manage and protect aquatic ecosystems. Various methods which have been developed and used internationally to derive such guidelines are discussed. Acute toxicity tests were conducted using two inorganic salts, NaCl and Na₂SO₄. Field collected, indigenous, freshwater macroinvertebrates were used as tests organisms. Data generated from these tests contributed to the expansion of the currently limited toxicological database of response data for indigenous organisms and the suitability of using such organisms for future testing was discussed. Salt sensitivities of indigenous freshwater invertebrates were compared those of species sourced from an international toxicological database and were found to have similar ranges of tolerances to NaCl and Na₂SO₄. Species sensitivity distributions (SSDs), a method of data extrapolation, were derived using different types of toxicological data, and hence different guideline values or protective concentrations were derived. These concentrations were equated to boundary values for South Africa’s ecological Reserve categories, which are used to describe degrees of health for aquatic ecosystems. Provisional results suggest that using only acute toxicity data in guideline derivation provides ecosystem protection that is under-protective. Chronic toxicity data, which include endpoints other than mortality, provide the most realistic environmental protection but lack data confidence due to small sample sizes (acute tests are more readily conducted than chronic tests). The potential contribution of sub-chronic data to guideline derivation is highlighted as these data are more readily extrapolated to chronic endpoints than acute data and sub-chronic tests are not as complex and demanding to conduct as chronic tests.