The Groot Marico catchment in the North West Province is a National Freshwater Ecosystem Priority Area (NFEPA) because it represents unique landscape features with unique biodiversity that are considered to be of special ecological significance. Three native freshwater species Amphilius uranoscopus, Chiloglanis pretoriae and Barbus motebensis, have high local conservation importance and B. motebensis is endemic to the catchment and is IUCN-listed as vulnerable. The main objective of this study is to contribute towards the effective conservation of these three species in the Groot Marico River system by assessing their genetic structure to determine whether tributary populations of the three species comprise of one genetic population or whether they are divided into genetically distinct subpopulations, in order to prioritise areas for conservation. The central null hypothesis was that there is no genetic differentiation between tributary populations (i.e., panmixia) of B. motebensis, A. uranoscopus and C. pretoriae in the Groot Marico catchment, North West Province. In total, 80 individuals per species were collected, targeting at least 10 individuals per population from a total of eight populations (seven tributaries and the Groot Marico main stem) and across the study area. Samples were collected by electrofishing and specimens were euthanized using an overdose of clove oil. A sample of muscle tissue was removed for genetic evaluation and the remainder of the specimens served as voucher specimens. For the genetic evaluation, mitochondrial (ND2, cyt b) and nuclear (S7) genes were used. Genetic techniques used were DNA extraction, polymerase chain reaction (PCR), purification and sequencing. From the 240 individuals collected, 123 sequences for B. motebensis, 111 sequences for A. uranoscopus and 103 sequences for C. pretoriae were analysed across all three genes. Statistical analysis included looking at cleaned sequences in order to obtain models using MODELTEST (version 3.06). Population structuring and phylogeographic analysis was performed in Arlequin (version 2000), TCS (version 1.2.1) and PAUP*. Results indicated that for B. motebensis the null hypothesis could be rejected as there were two distinct lineages (the Draai and Eastern lineages) that demonstrated significant divergence in both the ND2 and S7 genes, suggesting historical isolation. The low divergence in the mitochondrial cytochrome b gene (0% < D < 0.8%) suggests that this isolation is not very old and is probably not comparable to species level differentiation. The null hypothesis was also rejected for A. uranoscopus as there were also significant levels of differentiation between tributary populations resulting in the identification of two lineages (the Ribbok and Western lineages). However, for C. pretoriae, the null hypothesis could not be rejected as there was no genetic differentiation between tributary populations i.e., one panmictic population. Therefore, due to each species showing different genetic structuring within the tributary populations, more than one priority area for conservation needs to be implemented. These priority areas of conservation where therefore evaluated based on the current conservation status of the species (B. motebensis being vulnerable on the IUCN Red List), the number of Evolutionary Significant Units for each species and the overall genetic diversity of all three species in the Groot Marico catchment. In total, four tributary populations were conservation priorities areas, these were the Draai, Vanstraatens, Ribbok and Kaaloog tributaries. The Draai, Vanstraatens and Kaaloog tributaries were selected as priority areas for B. motebensis (B. motebensis is considered to be the most vulnerable of all three species). The Draai tributary was selected due to the B. motebensis population within the tributary showing isolation from the rest of the tributary populations. In order to conserve B. motebensis from the Southern lineage, the Vanstraatens and Kaaloog tributaries were selected. Reasons for selecting these two specific tributaries within the Southern lineage were that the Vanstraatens tributary had unique alleles (three Evolutionary Significant Units) for B. motebensis and the Kaaloog tributary had high genetic diversity (HD = 0.889, ND2 gene) when compared to the other tributary populations. The Ribbok and Vanstraatens tributaries were selected as priority areas for the conservation of A. uranoscopus. The Ribbok tributary was selected as it showed isolation from the rest of the tributary populations, as seen with the Draai tributary (B. motebensis) and the Vanstraatens tributary was selected to represent the Western lineage as it had the highest diversity for both genes (ND2 and S7). The Ribbok tributary has the highest prioritisation when compared to the Vanstraatens tributary. Chiloglanis pretoriae occurs within the Draai, Vanstraatens, Ribbok and Kaaloog tributaries, therefore by prioritising these tributaries for conservation, C. pretoriae will in turn be conserved.

The introduction, spread and concomitant impacts of non-native species are a global problem. Fish are among the most widely introduced vertebrate groups, with their impacts affecting multiple levels of organisation- from individuals, populations and communities, to entire ecosystems. In South Africa, the largest perceived threat to range-restricted endemic headwater stream fishes is said to be invasion by non-native fishes, however, as is the case elsewhere, invasive impacts are often a case of risk perception rather than actual risk analysis. Two range-restricted headwater species, the Eastern Cape redfin Pseudobarbus afer and the Border barb Barbus trevelyani are redlisted by the International Union for the Conservation of Nature (IUCN) as ‘Endangered’, primarily due to invasion by non-native fishes. To investigate invasions in South Africa, and provide a quantitative estimate of the impact of non-native fishes on the two imperilled endemics, P. afer and B. trevelyani, the overall aims of this thesis were to: (A) Provide a literature review on non-native fish invasions in South Africa; (B) Using two case studies on the headwaters of the perennial Keiskamma and episodic Swartkops River systems, investigate the naturalisation-invasion continuum to provide a holistic view of the invasion process in these variable environments. The specific thesis objectives were: (1) Reviewing current knowledge of invasive impacts of non-native fishes in South Africa; (2) Investigating invasibility of headwater stream environments by non-native fishes; (3) Determining the establishment success of non-native fishes, (4) Assessing the spatial and temporal impacts of invasion; (5) Understanding mechanisms responsible for non-native fish impacts; (6) Investigating the threat of non-native fish invasion on the genetic diversity of two the two headwater fishes, P. afer and B. trevelyani. Results from the literature review of fish invasions (Chapter 1) showed that South Africa has a long history of non-native fish introductions, spanning two and a half centuries. Currently, 55 species have been introduced or translocated. Many of these introduced species have become fully invasive (36%). Their impacts also span multiple levels of biological organisation. There was a general paucity of studies on fish invasions (38 studies), however, of those conducted, reviewed studies placed emphases on invasive impacts (25 studies) and the transport, introduction, establishment and spread stages of the invasion process were largely ignored. The two study systems, the Swartkops and Keiskamma Rivers, were heavily invaded and numbers of introduced species surpassed that of natives (Chapter 2, 3 and 5). Headwater streams had varying invasibility and a number of non-native species were successfully established (Chapter 2, 3, 5 and 6). The remainder of the invasions were casual incursions into headwater streams from source populations in mainstream and impoundment environments which were invasion hotspots. Irrespective of establishment, four predatory invaders (largemouth bass Micropterus salmoides, smallmouth bass M. dolomieu, brown trout Salmo trutta and rainbow trout Oncorhynchus mykiss) impacted heavily on native fish communities (Chapter 3, 4 and 5). Two broad types of invasion were documented, top down invasion by non-native O. mykiss and S. trutta and upstream invasion by M. salmoides and M. dolomieu (Chapter 3 and 5). Their impacts included changes in community structure, extirpation from invaded stream reaches resulting in contracted distribution, and isolation and fragmentation of native fish populations. The impacts of non-native predatory fishes were particularly acute for P. afer and B. trevelyani. Where non-native predatory fish occurred, P. afer and B. trevelyani had been extirpated (Chapter 3 and 5). As a result both native species exhibited contracted distributions (>20% habitat loss due to invasion). Upstream invasion by centrarchids isolated and fragmented P. afer populations into headwater refugia, while top down invasion by salmonids excluded B. trevelyani from invaded, more pristine stream reaches, by forcing the species into degraded unsuitable lower stream reaches. Predation also disrupted population processes such as adult dispersal for P. afer, and centrarchid-invaded zones acted as demographic sinks, where adults dispersing through invaded reaches were rapidly depleted. While the Mandela lineage of P. afer exhibited little within or between drainage genetic structuring, B. trevelyani was >4% divergent between drainages, and up to 2% divergent between streams within the Keiskamma River system (Chapter 7). The distribution of genetic diversity for B. trevelyani also indicated that the loss of diversity was imminent without immediate conservation interventions. This thesis has provided conclusive evidence that native fishes are vulnerable to invasion and that non-native predatory fishes have significant impacts on native fishes in Eastern Cape headwater streams. If management and conservation measures are implemented, the unwanted introduction and spread of non-native fishes may be restricted, allowing native fishes opportunities for recovery.

Despite more than a century of introductions, fish invasions are poorly understood in South Africa. This thesis aims to: (1) provide baseline information on the fish fauna in five inland water bodies (four dams: Theewaterskloof, Clanwilliam, Quaggaskloof, and Voelvlei and a natural lake, Groenvlei) in the Western Cape; (2) determine which species were established and provide a baseline of abundance and size structure for monitoring and, (3) make preliminary recommendations for a fishery. First, criteria for determining establishment were developed by undertaking a full biological assessment of bluegill Lepomis macrochirus, a data-poor species in South Africa. This included the assessment of growth rate using validated estimates of age derived from otoliths in two dams (Clanwilliam Dam: Lt=335.9(1−eo.113(t+1.06)); Groenvlei: Lt=287.2(1-e−0.171(t+0.54)). Length at 50% maturity was similar for males (146 mmFL) and females (147 mmFL). Length frequency analysis combined with data on the age and growth and reproduction demonstrated that established populations of bluegill contained several age cohorts as well as mature and juvenile fishes. A meta-analysis was conducted to: (1) compile a list of initial stocking dates for all alien fishes in each waterbody and (2) use angler tournament data to assess for additional species and persistence over time. Data from the meta-analysis was combined with ichthyological survey data to determine establishment success. These data demonstrated that nine non-native species had established in the five water bodies. Clanwilliam Dam had the largest number (7) of non-native species established, Groenvlei (5) and Theewaterskloof, Brandvlei and Voelvlei the lowest with four species. The state of invasion of each species in the province showed that all nine species are categorised as invasive. This information on establishment was then integrated into a framework for rapidly determining what type of fishery would be most suitable for the 5 waterbodies. Potential yield was estimated using morphoedaphic models, were 260 t/yr for Theewaterskloof Dam, 93 t/yr for Voelvlei, 81 t/yr for Brandvlei, 59 t/yr for Groenvlei and 55 t/yr for Groenvlei. The CPUE for Groenvlei was the highest in gill nets with 1.32 ± 1.94 kg/h, then Voelvlei 1.05 ± 1.34 kg/h followed by Brandvlei (0.84 ± 1.48 kg/h), Clanwilliam (0.71 ± 1.55 kg/h) and lastly Theewaterskloof (0.36 ± 0.41 kg/h). Only Theewaterskloof could produce in excess of 100 t/yr but less than 400 t/yr of fish making it suitable for a small scale fishery. The CPUE values of Theewaterskloof were the lowest and a gill net fishery on this water body may not produce sufficient fish for the fishery. All water bodies had a high recreational angling usage, the opening of a small scale fishery could be detrimental to this industry. Subsistence and recreational anglers should be promoted with the possibility of a long line industry targeting Clarias gariepinus.