This thesis represents a detailed investigation into aspects of the biology, ecology and population dynamics of two endemic ariid species, Galeichthys feliceps and G. ater, off the southeast coast of South Africa. The two species are exploited as a by-catch in the commercial ski-boat fishery off Port Alfred, a fishery dominated by highly fecund sparid and sciaenid species. They collectively constitute approximately 10% of the total annual catch in terms of landed mass G. feliceps outnumber G. ater in the catches by a ratio of 3:1. The investigation was designed to provide the biological data required for stock assessment and to determine optimum management strategies for the two populations. The implications of their K-selected life-history styles for exploitation received particular attention. While the two species were sympatric and had similar depth distributions they were found to be allopatric with respect to their foraging habitats. G. feliceps foraged over sandy and muddy substrata in marine and estuarine environments. G. ater fed only on reef-associated species and did not utilise estuaries. Their feeding-associated morphologies were identical and both species preyed primarily on crustaceans (brachyuran crabs and isopods), echiurids, molluscs and polychaetes. The diet of G. ater was broader in terms of the number of species consumed. The two species are mouth-brooders with low fecundity. G. feliceps and G. ater produced a mean of 49 and 32 eggs each, per annum. The buccal incubation period was determined to be in the region of 140 days for G. feliceps. Embryos hatched after approximately 75-80 days and the young began exogenous feeding thereafter. The young fed intra-buccally on detritus provided by the parent. Adult buccal mucus may also have been used as a food source. Young were released at a total length of ± 55mm. Adult males ceased feeding whilst mouth-brooding. Body musculature, abdominal fat and liver reserves provided energy for basal metabolism and males lost approximately 28% of their body mass during buccal incubation. Females expended less reproductive energy than males. Catches were dominated by mature fish (76% in G. feliceps and 97% in G. ater). Females were significantly more abundant in catches during the spawning and mouth-brooding period. The female to male sex ratios were 1.65:1 and 2.23:1 for G. feliceps and G. ater respectively. Age and growth studies revealed that the two species mature at advanced ages (10 and 9 years for G. feliceps and 9 and 7 years for G. ater males and females respectively). They are long-lived, reaching ages in excess of 18 years in G. feliceps and in excess of 15 years in G. ater. Females live longer than males and grow larger. Yield-per-recruit and spawner biomass-per-recruit analyses demonstrated that G. ater were exploited below FO.1 at a level where spawner biomass-per-recruit was reduced to between 45% and 65% of the unexploited level. The G. ater stock was not adversely affected by current levels of fishing effort. For G. feliceps, both sexes were exploited beyond F₀.₁ where spawner biomass-per-recruit was reduced to between 30% and 22% of the unexploited level. G. feliceps were shown to be sensitive to relatively low levels of exploitation, a phenomenon attributed to their highly Kselected life-history style. Should the species become targeted for in the future, effort restrictions in the form of a closed season during the spawning and mouth-brooding period would prove effective in reducing effort and conserving the population sex ratio.

Aspects of the biology of the sharptooth catfish, Clarias gariepinus (Burchell, 1822) were studied in a clear coastal lake (Lake Sibaya, KwaZulu.). Catfish reach maturity towards the end of the first year (total length 200-250mm) and breed in summer. Gonadal maturation is cyclical, and most fishes are ripe between September. and January e ach year. Maturation stages and fecundity are described. Twenty-two catfish spawning runs were witnessed. There were no large aggregations of catfish before the Summer rains, as reported in some impoundments and rivers, but catfish accumulated on the shallow terrace immediately before a spawning run. Spawning usually took place in flooded marginal areas after heavy rain on dark calm nights. Peak activity occurred between 20h00 and 02h30. Pre-nuptial aggression, courtship and mating are described and illustrated in detail for the first time from field observations. Gonadal products are released while in amplexus and fertilized externally. The eggs have an adhesive disc for attachment to plants, and develop rapidly. Early development and feeding are described. The growth rate of catfish was determined using rings on the pectoral spines, and validated using recent techniques. C.gariepinus in Lake Sibaya grow rapidly and remain in good condition to a length of 500mm, but larger catfish have a poor growth rate and condition, and few exceed 650mm. The modal length and weight of the Sibaya population is intermediate between that of other populations but the number and condition of larger catfish in Lake Sibaya is lower. Other studies on C.gariepinus growth are reviewed. Very high and low first year increments obtained previously may be due to methodological errors, but catfish do appear to have variable growth rates. The size reached by C.gariepinus is reviewed. Catfish are equipped morphologically and behaviourally to feed on a wide variety of·prey in different situations, but they usually feed at night on exposed, active benthic organisms. Social hunting facilitates the capture of prey which is too fast or manoevrable for a single predator. Food preferences are scored by three different methods and presented for the total catfish population and for different length groups, seasons, and habitats. Juveniles feed more often and on a greater variety of prey, and their condition is better than that of adults. The proportions of several food items in the diet changed at different lake levels as marginal areas became inundated. In particular, increasing lake levels resulted in a change in relative density of three cichlid prey in the shallow water feeding area of catfish. This situation provided the opportunity to study the effect of changing cichlid density on predation by catfish. The world literature on factors influencing the predator-prey relationship in fish is reviewed and several key factors are identified. The interaction of fish predators and prey in Africa, and particularly Lake Sibaya, is examined in detail, and prey density, modified by accessibility, is found to be an important factor. The effect of cichlid prey density on catfish predation was therefore investigated using field collections and experimental observations. Catfish prey on the cichlid species with greatest relative abundance, with some deviations which can be explained in terms of accessibility. The experimental results also describe feeding periodicity, imitative hunting and the effect of water depth , day-night light intensity changes and predator number on predation efficiency on cichlids. Catfish predation is discussed in terms of recent predation theory, and it is concluded that their ability to switch from one prey to another as prey availability (= density+ accessibility) changes , is an an important aspect of their success as predators. Catfish predate more efficiently on invertebrates, but feed largely on fishes after the first year. This preference for fish prey is probably related to food quality, as fishes contributed over 75% of the energy content of all ingested food, as well as abundant protein and the correct amino acid balance. Finally, I conclude that the role of C.gariepinus as a predator on S.mossambicus is three-fold: (a) to enhance species' fitness by removing enfeebled individuals i.e. to act as a 'cleanser' (b) to restrict the distribution of prey under certain conditions i.e. act as a 'restricter' (c) to dampen fluctuations in cichlid abundance i.e. act as a 'regulator'. The relative importance of these roles may alter at different lake levels in Lake Sibaya in response to changes in the vulnerability of fish prey.

The aim of this study was to evaluate the aquaculture potential of a hybrid between the two largest African silurid catfish, Heterobranchus longifilis (HL) and Clarias gariepinus (CG) . A morphometric, meristic and karyological study showed that the hybrid shared some of the physical characteristics of both C. gariepinus and H. longifilis, while its chromosome complement and fundamental number (2N = 54, FN = 95), was intermediate between that of H. longifilis (2N = 52, FN = 92) and C. gariepinus (2N = 56, FN = 97). The HLxCG cross could therefore be regarded as a true hybrid. For the characters tested, no morphological or karyological differences were apparent between the HLxCG hybrids produced in West Africa and those produced in southern Africa. In southern Africa, the HL♂xCG♀ cross had higher fertilization and hatching rates than pure strain H. longifilis, C. gariepinus or the reverse cross. There was no difference in the survival of the HL♂xCG♀ hybrid larvae and C. gariepinus larvae up to the onset of exogenous feeding. Pure strain H. longifilis juveniles had a faster growth rate than the H. longifilis ♂ x C. gariepinus ♀ juveniles, but the hybrid always grew at a faster rate than C. gariepinus or the reverse cross. The HL♂xCG♀ cross was consequently chosen as the hybrid with the greatest potential for siluroid aquaculture in southern Africa. The HL♂xCG♀ hybrid showed evidence of partial gonadic, gametic and post-zygotic sterility in both sexes. The hybrid was not completely sterile, as it was artificially induced to spawn, and a small number of viable F₂ hybrid and F₁ hybrid x C. gariepinus larvae were produced. However, in the light of its probable reproductive strategy, it is highly unlikely that the hybrid would pose an ecological risk to the southern African region. The H. longifilis ♂ x C. gariepinus ♀ hybrid was compared to C. gariepinus for selected water quality preferences and tolerances. The hybrid had a wider temperature preference (28 °C to 34 °C) than C. gariepinus (28 °C to 30 °C), but appeared to be more dependent on aerial respiration than C. gariepinus. The air-breathing frequency of hybrid fish began to increase at dissolved oxygen concentrations below 3 .8 mg.ℓ⁻¹, while C. gariepinus only showed an increase in air-breathing frequency at concentrations below 3.0 mg.ℓ¹. The hybrid was more tolerant of un-ionised ammonia (96-hour LC₅₀ = 9.1 mg.ℓ⁻¹) than C. gariepinus (96-hour LC₅₀ = 6.5 mg.ℓ⁻¹), but their 96-hour LC₅₀ salinity tolerances were similar (10.8-11.0 g.ℓ⁻¹). The H. longifilis ♂ x C. gariepinus ♀ hybrid had a higher fillet yield (43.9 %) than C. gariepinus (38.9 %), but the crude protein content and amino acid profile of the two groups were similar. Catfish are traditionally grown in earthen ponds under semi-intensive conditions in southern Africa, at around 4 kg of fish per cubic meter of water (kg.m⁻³). However, the hybrid could tolerate densities of up to 415 kg of fish per cubic metre of water, if the water was exchanged hourly (kg.m⁻³.hr⁻¹), and the density at which yield was optimised was rounded off to 400 kg.m³.hr⁻¹. The high threshold density and water quality tolerances of the H. longifilis ♂ x C. gariepinus ♀ hybrid indicates that it is ideally suited for highly intensive aquaculture. It was concluded that it would be more productive to farm the HL♂xCG♀ hybrid on an intensive basis in southern Africa, than it would be to farm C. gariepinus in the traditional manner.

Austroglanis gilli and A. barnardi are endemic to the Clanwilliam-Olifants System in the Western Cape, South Africa. The populations of each of these species are considered to be threatened by various anthropogenic activities, which include inappropriate agricultural practice and impacts of alien invasive fish species. The purpose of this thesis was to assess the life-history of these two endangered species in order to contribute to understanding their biology. Such information is vital for the development of strategies for their conservation. Marginal zone and marginal increment analyses from sectioned lapilliar otoliths of both A. gilli and A. barnardi showed a unimodal peak, suggesting a single annulus formation. The oldest specimens of A. gilli and A. barnardi were 12+ and 14+ years, respectively. The growth of A. gilli and A. barnardi was relatively slow and was best described by the von Bertalanffy growth curve as: L, = 131.56(1 - exp(- 0.27(t - 1.18 ))) for male and L, = 113.86(1 - exp(- 0.43(1- 0.74))) for female A. gilli from the Rondegat River. In the Noordhoeks River, growth was L, = 99.67(1 - exp(- 0.53(t - 0.35))) for male and L, = 96.60(1 - exp(- 0.64(t - 0.11 ))) for female A. gilli, and L, =71.02(1-exp(-0.26(1-3.07))) and L, = 69.50(1-exp(-0.36(1 - 1.88))) for male and female A. barnardi, respectively. The average natural mortality for the combined sexes was estimated using catch curve analysis at 0.37 ± 0.12 per year for A. gilli from the Rondegat River, and at 0.71 ± 0.05 and 0.39 ± 0.04 per year for A. gilli and A. barnardi from the Noordhoeks River, respectively. For A. gilli in the Rondegat River, the first maturity was estimated at 3.3 years (97.3 mm SL) for males and 3.1 years (94.3 mm SL) for females. In the Noordhoeks River, the first maturity for A. gilli was estimated at 2.0 years (71.9 mm SL) for males and 1.7 (66.4 mm SL) for females, and for A. barnardi at 2.0 (55.0 nun SL) for males and 2.9 years (58.9 mm SL) for females. Maturity corresponded closely to the asymptotic sizes from the von Bertalanffy curves, suggesting a shift in energy use from somatic growth to gonad development. Macroscopic assessment of the state of gonads, the Ganado-somatic index and histological examinations revealed that both A. gilli and A. barnardi have a single spawning season. The presence of oocytes in different stages of development in each of the ovaries of mature females collected between November and January suggested asynchronous, iteroperous serial spawning. The resorption of yolk was observed from ovaries collected between February and March and no vitellogenic oocyte was visible from ovaries collected between April and August. A similar trend was observed for males, with spermatozoa filling the lumen between November and January and some residual spermatozoa being present in the lumens soon after the breeding season. Spermatocytes and spermatids were dominant in the testes until October. An Index of Relative Importance revealed that A. gilli feeds predominantly on the benthic macroinvertebrate larvae of Ephemeroptera (particularly Baetidae), Diptera (particularly Chironomidae and Simuliidae) and Trichoptera. Austroglanis barnardi feeds predominantly on dipteran larvae (particularly Chironornidae and Simuliidae). Chi square contingency tables showed a significant difference in the dominant prey items of A. gilli and A. barnardi in the Noordhoeks River (X₂= 53.79, d.f. = 4, p > 0.001) and A. gilli between Rondegat and Noordhoeks rivers (x₂ = 34.74, d.f. = 4, p > 0.001). The Spearman's rank correlation test showed no shifts in the diet of A. barnardi from Noordhoeks River and A. gilli from Rondegat River with a change in size and season (p>O.OS). However, there was a shift in the diet of A. gilli from the Noordhoeks River which could suggest a distinct patchiness of benthic macroinvertebrates between the riffle feeding areas used by juveniles and other biotopes used by adults. The occurrence, in stomach contents, of other prey items from a wide variety of taxa and the presence of allochthonous material from the terrestrial environment could suggest an opportunistic feeding guild for both Austroglanis species. The life-history traits of A. gilli and A. barnardi, which are charaterized by slow growth, long life span and low relative fecundity, indicate that both species are relatively precocial and K-selected. The population of a precocial species is relatively stable and if population numbers were to be greatly reduced, they would require a long time to rebuild. An urgent conservation intervention is therefore recommended for Austroglanis spp. so as to maintain the diversity of populations within these species. The creation of protected river reserves and raising public conservation awareness may minimise activities that result in altered river hydrology and the destruction of complex benthic habitats.

Invasive species are of particular concern as they have the potential to alter community structure and food web relationships within their invaded habitats. African sharptooth catfish Clarias gariepinus, a generalist predator, was introduced through an inter-basin water transfer scheme into the Great Fish and Sundays Rivers, Eastern Cape, South Africa, where it threatens the native riverine biota. This thesis assessed its impact from a trophic perspective. Patterns in catfish distribution and abundance revealed an upstream to downstream gradient that was associated with spatial distribution of most species within the mainstream, and a mainstream to tributary gradient that was associated with the spatial distribution of native minnows. The catfish was predicted to occur widely within the mainstem habitats and to decrease progressively along the mainstrem to tributary gradient with the physico-chemical environment being a good proxy for predicting both its occurrence and abundance. The results suggest the catfish proliferated within mainstem habitats where invasion resistance was possibly reduced due to alteration of flow. Population dynamics and size structuring of two native cyprinid minnows Pseudobarbus afer and Barbus anoplus, threatened by catfish, were examined within uninvaded headwater streams in relation to their proximate physical habitats. Their habitats were characterised by seasonal variation in physico-chemical conditions and a spatial variation in substrata compositions. No evidence of differences was found between seasons for density and capture probability for either species. The population size and density for P. afer was found to increase with increasing proportion of boulders. In comparison, B. anoplus population size and probability of capture increased with increasing proportion of bedrock and bank vegetation, respectively. Size structuring was explained predominantly by seasonality and habitat variables for P. afer and B. anoplus, respectively. Stable isotope ratios of carbon and nitrogen were used to compare the spatial variation in both the community-wide and catfish-specific niches and to estimate catfish prey sources from different habitats within the invaded systems. Aquatic community and catfish niches were statistically different among localities, suggesting that each locality had a distinct community-wide trophic structure. Dispersion metrics indicated no evidence of differences in the clustering among individuals, but provided evidence of differences in path trajectories for the comparisons of catfish populations that suggested dietary plasticity within different localities. Dietary studies revealed both ontogenetic shift and omnivory that suggested that catfish may exhibit less pronounced top-down effects within its invaded habitats. Manipulative experiments were used to test the response of benthic macroinvertebrates within two rivers that were differentially impacted by catfish as a presstype disturbance. Macroinvertebrates were non-responsive to catfish presence within a system where catfish had previously been established. In contrast, excluding catfish in this system indicated a response that suggested the importance of refuge within invaded habitats and the possible recovery pattern of certain macroinvertebrate taxa. By comparison, introduction of catfish within previously uninvaded localities provided evidence of direct catfish impact through elimination of conspicuous taxa. Acoustic telemetry was used to investigate catfish movement patterns within an invaded lentic habitat and provided evidence that habitat utilisation was non-random. The shallow and structured river mouth habitat, which was most utilised, was probably the most ideal for its breeding and feeding. This inferred potential overlap with native species and suggested the risk of predation and competitive interference. Catfish also exhibited both nocturnal and diurnal activity patterns that were probably related to feeding.