Fishes of African floodplains typically comprise populations having short life cycles, relatively few age groups, more generalised feeding behaviour, and which experience spawning success that is strongly influenced by abiotic factors such as hydrological regime of the river, climatic seasonality and habitat characteristics. In addition, African river-floodplain ecosystems do not have appropriate predictive models for estimating yield in these ecosystems. While most predictive models developed to date for floodplain fisheries have taken into account morpho-edaphic factors, they have generally excluded climatic factors as a mega-determinant of the variability of floodplain fisheries. The principle aim of this thesis has been to develop a predictive management model that incorporates data on essential biological aspects of the target species, characteristics of the habitat as well as overall climatic factors, and thus allow for adaptive management of the fisheries in a continuously fluctuating floodplain environment. Lower Shire Floodplain (34 ⁰ 45’ – 35 ⁰ E and 16 ⁰ 00’-17 ⁰ 15’ S) in Malawi, one of the major rivers-floodplains in the Zambezi Basin, was used to test this type of model. The major hypothesis tested in this study was that “the dynamics of the fishery of Lower Shire Floodplain are driven by and adapted to the seasonal, but predictable, hydro-climatic regime of the floodplain”. The specific objectives were: to describe the floodplain’s climate and hydrological pattern; to develop a quantitative characterisation of the major habitats of the Lower Shire Floodplain; to assess the fishery in terms of size, gear utilisation, gear selectivity and yield; to assess the biological parameters of the target species, necessary for the management of the fishery; to analyse how water fluctuation in the floodplain affects the recruitment and life history of the two target species; and to develop of a predictive hydro-climatic model to benefit the management of the fisheries. The climate of the floodplain was characterised by low (765 ± 198 mm) and fairly variable (Cv = 27%) rainfall, which largely occurred between December and January. The area was generally humid (mean RH 68%), with mild to hot (25- 33ºC) and variable monthly mean diurnal (12ºC) temperature. Four quarterly hydro-climatic seasons were identified and comprised: Quarter 1 (Jul-Sep) characterised by hot, dry weather with a low flood regime; Quarter 2 (Oct-Dec) hot, windy, wet weather with low-but-rising flood regime; Quarter 3 (Jan-Mar) hot humid, wet weather with the flood regime at peak; and Quarter 4 (Apr-Jun) humid and cool weather with receding flood regime. The annual hydrograph of the floodplain was represented by four categories of flood regime as: low (Jul-Sept), low-but-rising (Oct-Dec), peak (Jan- Mar), and falling (Apr-Jun). The floodplain experienced a water deficit of 95.1 mm.year⁻¹, and it was hypothesised that ground water recharge maintained water in the floodplain the rest of the year. Three major habitats were identified in the floodplain. The river-floodplain, characterised by deep fast-flowing water, sandy substrate and little emergent vegetation; the permanently connected lagoons, were shallow (≤ 2 m) with sandymud bottom and slow flowing water; and the seasonally connected lagoons had slowflowing stagnant water, with comparatively more emergent and floating vegetation. Physicochemical characteristics of the habitats varied significantly with hydroclimatic seasons (one-way ANOVA, p ≤ 0.05), indicating the influence of flood regime, which joined the habitats in a non-equilibrial spatial distribution. Consequently, during receding and low flood regimes river-floodplain and permanently connected lagoons exhibited similar characteristics while all three habitats had similar characteristics during the rising and peak flood regimes. It was, therefore, concluded that in floodplains, habitats shift horizontally and vertically according to the water level. Gill nets, cast nets, long line, and fish traps accounted for 99% of the total count of gears, and hence considered the major fishing sectors. Two principal species in the floodplain were the catfish Clarias gariepinus and the cichlid Oreochromis mossambicus. Length-at-maximum-selectivity (ø) and length-at-50%-sexual-maturity (Lm₅₀) for each of the two species showed that although they were selected into the four fishing sectors at lengths above maturity, maximum selectivity into the long line occurred at a length before maturity for O. mossambicus. In addition, the width of the gamma selectivity function (σ) indicated that a considerable proportion of juveniles of the two species were also selected. Therefore, it was decided that management for the floodplain be centred on controlling over-fishing and preserving the spawner stock, by banning seine and mosquito nets, and closing river-floodplain and permanently connected lagoons to fishing during the low flood regime. There were significant seasonal variations in CPUE, lowest in the low flood, and highest during the peak flood regime. Sectioned otoliths were used to determine the age and growth of the two principal species in the floodplain. Marginal zone analysis revealed that annulus formation in all the species occurred during the period of low air and water temperatures, high evaporation and receding water levels, hence a high rate of desiccation and negative water budget. For O. mossambicus, maximum age reached was 6 years while in C. gariepinus it was 9 years. The 3-parameter von Bertalanffy growth model adequately described growth as lt = 177.6(1-e⁻·⁴⁴⁽t ⁺ ¹·⁴³⁾) mm TL for O. mossambicus and lt = 502.9(1-e⁻°·³¹⁽t ⁺ ¹·⁹²⁾) mm TL for C. gariepinus. Female O. mossambicus reached 50%-sexual-maturity at 109 mm SL, while males matured at 105 mm, and both male and female C. gariepinus reached Lm₅₀ at 249 mm SL. The breeding seasons of both O. mossambicus and Clarias gariepinus were between September and March, with modal peaks in January and November, respectively. The annual total mortality rates (Z) were 0.62 ± 0.18 yr⁻¹ for O. mossambicus and 0.93 ± 0.47 yr⁻¹ for C. gariepinus. The mean empirical estimates of natural mortality (M) were 0.46 yr⁻¹ for O. mossambicus and 0.50 year⁻¹ for C. gariepinus, and fishing mortality (F) was calculated as 0.16 yr⁻¹ for O. mossambicus and 0.43 yr⁻¹ for C. gariepinus. The overall exploitation level (Z/K) was 1.41 for O. mossambicus and 3.01 for C. gariepinus. Given that the Z/K ratio was >1, it was asserted that both O. mossambicus and C. gariepinus were mortality-dominated and fairly heavily exploited. However, potential for sustainable exploitation existed since both species showed signs of resilience due short longevity and high rate of natural mortality. Given the limitation of simple exponential models of fish mortality and growth under the situation of seasonal fluctuation of water levels, a simple predictive hydroclimatic- fisheries model was developed. The model predicted the life-history and production parameters fairly accurately (0.53 ≤ r² ≥ 0.98, p ≤ 0.05), and showed that environmental and biological events in the floodplain significantly (0.43 ≤ r² ≥ 0.91, p ≤ 0.05) followed the periodic function of time (day-of-the-year), hence, strongly seasonal. In addition, the flood-pulse preceded all the major biological events, with predicted phase lags established at 55.6º for peak flood, 157.2º, 260.1º, 334.6º and 341.4º for condition factor for O. mossambicus, recruitment, spawning period for O. mossambicus and C. gariepinus, respectively. Therefore, it was recommended that water obstruction on the Shire River must ensure sufficient water flow during the peak flood and spawning period to allow flooding and inundation of the floodplain in order to create habitat favourable for spawning and feeding as well as improve recruitment.

The aim of the study was to investigate local community contributions to the existing Malawi College of Fisheries curriculum with a focus on Chambo fishery. Chambo fish (Oreochromis species) is the most favoured fish in Lake Malawi. Chambo catches have declined over the years. Responding to the declining catches, the Fisheries Department is engaged in a number of management options to address the issues. Most of these management options are governed by scientific recommendations and do not consider the socio-economic situation of the people who are dependent on fishery. This approach to fishery has influenced the Malawi College of Fisheries curriculum. The existing curriculum is product-centred, developed by a consultant. During the development of the curriculum, there was little consultation with the lecturers and no consultation with the local communities who are using the resource. The fishing communities have been fishing for a long time and have acquired knowledge, skills and experience worth investigating for its potential role in improving the existing curriculum. Through the use of interviews, focus group discussions and workshops with local communities local knowledge was identified for inclusion into the curriculum. A review of the existing curriculum revealed that it has primarily technical focus, grounded in the protection, control and management of the fish stocks, while the local knowledge has a practical focus based on existing practices and requiring an understanding of the sources of the issues. There is also more emphasis on a historical perspective and the context in which fishing practices take place at the moment. It was evident from the study that local communities have much of knowledge, skills and experience gained over the years of fishing , and if properly utilised, it can improve the MCF curriculum. I therefore recommend in this study that the curriculum be reviewed in order to integrate and draw on the local knowledge through a deliberative and participatory process between the local communities and the government so that it addresses the needs of the local communities and improves the training of the extension workers.