Lake Kariba, created in 1958, experienced changes, notably the decline of the Limnothrissa miodon fishery, which have been attributed to climatic change. Air temperatures rose abruptly by 1.1°C between 1980 and 1981, but the temperature of the lake did not follow this pattern. Temperatures at 10 m depth increased by 8% between 1961 and 1971, remained stable until 1984, then declined and by 1992 the temperature was about 7% lower than in 1961. The causes of this are unclear but it followed the hot and dry El Niño droughts of 1982-83 and 1991-92. The lake warmed again by 2007-2011 with temperatures at 10 m being about 10% higher than in 1961, while at 40 m it was 16% higher indicating a faster warming rate in deeper waters. The thermocline fell from 15-20 m in 1968 and 1986 to 20-25 m in 2011 and the temperature gradient decreased by > 50%. The epilimnion became more homogenous, with no evidence of anoxia in the upper 20 m in 2007-08, and the normal monomictic thermal regime may change, thus affecting nutrient circulation and the seasonal abundance of plankton. Both zooplankton and phytoplankton communities have evolved since the lake was created, but these changes occurred before any evidence of warming. The fishery has been declining since 1996 but there is no evidence that climate change is responsible; the number of fishing vessels is presently about three times the recommended level and fishing effort is almost certainly the main cause of the problems. All African lakes support fisheries and it is essential to consider fishing, which changes fish species composition, demographics and abundance, characters that could also be affected by climate change.
Understanding the effects of climate change on lakes is challenging because of the wide range of physical, chemical and biological response variables that influence them, both within the lake itself and in its catchment area (Adrian et al., 2009). Although warming has been detected in a number of African lakes (Ogutu-Ohwayo et al., 2016), there are few examples of its direct impacts on their physical characteristics and biota. Few African lakes have comprehensive, long-term data sets that can be used to infer and perhaps predict the impacts of climatic changes. Since these lakes all support fisheries and climate change may reduce fisheries production there is an urgent need to understand the relationship between climate change and fishing effort.
Lake Kariba, created in 1958, one of the world’s largest man-made lakes (c. 5400 km2), is located on the Zambezi River (Zambia-Zimbabwe). The native fish species are largely restricted to areas < 20 m deep and the sardine Limnothrissa miodon (“Kapenta”) was introduced from Lake Tanganyika to colonise open water. It supported the major fishery on the lake, yielding around 30,000 t per annum at its peak. It is a semi-industrial fishery, requiring considerable capital investment in boats, equipment and shore facilities and was the major driver of growth and infrastructure development around the lake (Bourdillon et al., 1985). The fishery is now experiencing considerable difficulties resulting from declining catches, attributed by some to excess fishing effort (Chali et al., 2014) while others implicated climate change (Magadza, 2011; Ndebele-Murisa, 2011). This paper discusses these aspects, addressing some contentious issues and considering some impacts of climate change on the physical environment and fisheries.
Impacts of climate change on the physical environment
Most African lakes lack long-term temperature records and air temperatures have been used as a proxy for water temperature in assessing climate change, including on Lake Kariba (Chifamba, 2000; Magadza, 2011; Ndebele-Murisa et al., 2011). This may not be a reliable indicator because the relationship between air and water temperatures is not always a simple linear one. In Lake Kariba, for example, water and mean minimum air temperatures were related by:
where TW = water temperature, A0 = temperature in current month, A1 and A2 = temperature one month and two months earlier (Clay, 1976).
Air temperatures at the Kariba weather station increased by about 1.1°C between 1962 and 2008 but not in a steady linear fashion (Figure 1a). The mean temperature from 1962-1980 was 24.3°C, and from 1981-2008 it was 25.4°C, but in each period the trend was insignificant. The abrupt change in 1980-81 coincided with a major shift in the Earth’s biophysical systems (Reid et al., 2016). The response of the lake to these temperature changes was not predictable (Figure 1b). Relative temperatures (1961 = 1.0) in March, when the lake is warmest and fully stratified, at depths of 10 m and 40 m (representative of the epilimnion and hypolimnion) increased from their 1961 level by about 8% at 10 m and 4% at 40 m by 1971. They remained relatively stable until 1984 (about 8% and 3% above 1961 levels) when a marked cooling period began and by 1992 temperatures had fallen by about 7% and 13% of the 1961 level at 10 m and 40 m, respectively.
The reasons for this cooling are unclear as it occurred after the air temperature transition in 1980-81 and followed the hot and dry conditions of the 1982-83 drought, and continued into the 1991-92 drought, then the worst ever experienced in Zimbabwe (Maphosa, 1994). The flow of the Zambezi was reduced and the lake level fell by 7.0 m between 1980 and 1984, (a 25% decrease in volume), remaining around that level during a prolonged dry period that lasted until 1999. There is a gap in the record until 2008-2011 when the temperature at 10 m was 11% higher than in 1961, but 16% higher at 40 m suggesting that the hypolimnion had warmed more rapidly than the epilimnion.
Lake Kariba is a warm, monomictic lake, with only one overturn and one isothermal period during the year (Begg, 1974). It becomes isothermal in July, begins to stratify in September and by December the surface temperature is about 6°C higher than at 40 m. It is fully mixed and oxygenated in July but oxygen concentrations in the hypolimnion decrease as the lake stratifies again and by March it is anoxic. This cycle drives productivity of the open-water ecosystem and is of crucial importance to the Kapenta fishery. During the stratified period nutrients are retained in the hypolimnion, being released at overturn leading to an increase in phytoplankton and zooplankton in July-August. Kapenta breed from about October to March and reach maturity around the time of the plankton peak, and catches are typically highest in August (Marshall, 1988).
However, changes in stratification of the water column suggest that warming of the lake may be having an impact. Temperature profiles in March were similar in 1968 and 1986 but had changed considerably by 2011 (Figure 2a). Surface temperature increased by 1.0°C between 1986 and 2011 while the temperature at 40 m increased by 3.4°C. The steepest thermal gradient, indicating the position of the thermocline, occurred at 15-20 m in 1968 and 1986, but at 25-30 m in 2011 (Figure 2b). Thermal gradients were similar in 1968 and 1986 (0.64 and 0.58°C m-1, respectively) but in 2011 decreased by >50% to 0.28°C m-1. The oxycline began at the same level as the thermocline and the hypolimnion was anoxic in 1986 (Figure 2c). By 2011, however, deoxygenation was less pronounced and there were indications that dissolved oxygen concentrations in the hypolimnion have become less predictable and relatively disorganised (Mahere et al., 2014). The notion that the water column was becoming less variable and more homogenous, was supported by comparisons of mean temperature and oxygen concentrations in the upper 20 m in 1986-87 and 2007-08 (Figure 3). In 2007-08, the surface temperature had increased by 1.5°C but by 2.6°C at 20 m confirming that deeper waters were warming more rapidly. Changes in oxygen concentrations were even more pronounced; in 2007-08, the difference was only 0.6 mg l-1 over 20 m, compared to 5.1 mg l-1 in 1986-87. A striking feature was the presence of a discontinuity in temperature and, especially, in oxygen between 5 and 10 m in 1986-87 indicating the beginning of a thermocline or oxycline at this depth. This discontinuity was absent in 2007-08.
The possible impact of warming on the Kapenta fishery
This fishery is now experiencing difficulties, as the average catch per unit effort (CPUE) fell from 0.30 t boat-night-1 in 1990 to 0.10 t boat-night-1 in 2011 (Kinadjian et al., 2014). This situation was attributed to overfishing although some Zimbabwean workers have discounted fishing effort and instead implicated climate change as the main cause (Magadza, 2011; Ndebele-Murisa et al., 2011, 2014). These arguments were based on a misinterpretation of changes in stratification and the incorrect assumption that changes in zoo- and phytoplankton were a result of climate change, even though they occurred long before there was any evidence of warming (Marshall, 2012a, 2012b, 2017). Magadza (2011) attributed the decline in fish catches to increased temperature, based on values derived from air temperatures. However, there was no significant correlation between CPUE and epilimnetic water temperatures in July, when the lake is coolest and isothermal (r = -0.455, p > 0.05) or in March (r = -0.572, p > 0.05), when it is warmest and fully stratified (Figure 4a) so this seems unlikely.
It has long been known that Kapenta catches were influenced by river flow (Marshal, 1982, 1988; Chifamba, 2000). As a short-lived species it responded to year-to-year variations in the nutrient supply brought in by the Zambezi and other rivers to replenish those lost through the outflow. There was a significant correlation between the flow of the Zambezi river and CPUE until 1996 but the relationship broke down abruptly after that (Figure 4b). This is almost certainly a result of a great increase in fishing capacity. In 1996, the Zambia-Zimbabwe SADC Fisheries Project recommended that fishing effort should be limited to a total of 500 fishing vessels, with 270 being allocated to Zimbabwe and 230 to Zambia, based on each country’s share of the lake; 56% in Zimbabwe and 44% in Zambia (Paulet, 2014). In 1999, the two countries signed a protocol on the management of the shared fisheries resources on Lake Kariba and the transboundary waters of the Zambezi River, which incorporated these recommendations (Mhlanga and Mhlanga, 2014). Unfortunately, these were ignored and there are now around 1400 fishing vessels on the lake (Figure 4c). This is nearly three times the recommended effort and suggests that excessive fishing capacity is the primary driver of the declining CPUE in the fishery.
The most obvious impact of climate change in Lake Kariba is the relatively recent increase in its temperature, and the most immediate consequences could include a continued weakening of the thermocline and rapid warming of deeper waters. Similar changes were reported from Lake Victoria (Sitoki et al., 2010; Marshall et al., 2013) suggesting that other relatively shallow tropical lakes may behave in a similar fashion. If the observed trends continue, it is possible that the homogenisation of the upper 20 m could extend to deeper layers thus bringing about partial or even complete destratification. It is unclear how this would affect the lake but artificial destratification has been used to reduce anoxia, prevent the accumulation of nutrients, control blooms of blue-green algae and promote the growth of green algae (Visser et al., 2016). If similar processes occurred in a large lake, they could have significant impacts on nutrient cycling, plankton populations and productivity.
Nearly all African lakes already support fisheries that provide a livelihood for local communities, and fishing is now the principal cause of fish mortality in these lakes. It changes the species composition, demography and life history characteristics of exploited stocks, and there is evidence that such changes degrade and destabilise ecosystems (Kuparinen et al., 2016). It is essential, therefore, that fishing is included in any model used to assess the impacts of climate change. Natugonza et al. (2015), for example, noted that Nile Tilapia Oreochromis niloticus in Lake Wamala, Uganda, exhibited a typical r-selected reproductive strategy by remaining small, maturing earlier and exploiting a wider range of food items, which they attributed to adverse conditions caused by climate change. Yet such characteristics are also a result of intensive fishing as shown, for instance, by Nile Tilapia in lakes George and Victoria (Lowe-McConnell, 1982 ; Njiru et al. 2007). The failure to include fishing in the assessment of climate change on Lake Kariba meant that the outcome did not fully represent the situation on the lake and this probably applies to fisheries elsewhere.
Fishery management is a challenging process, especially in Africa where most fisheries are at a small-scale, subsistence level. Apart from providing food security and some income, they may be crucial to survival when, for example, crops fail in countries that lack social welfare safety nets (Marshall, 2016). Fishery managers are therefore expected to maintain sustainable fisheries in the face of an increasing demand for fish and a growing number of people resorting to fishing for survival. In some cases, it may be necessary to reduce fishing effort. Politicians and fishery managers may be reluctant to enforce such measures, and attributing problems to climate change would enable them to avoid the issue of effort reduction. Such is the case on Lake Kariba where there seems to be an obvious need to reduce effort, either by removing boats from the fishery or by management actions, such as a closed season when fish are breeding.
This paper is based on my experiences at the Lake Kariba Fisheries Research Institute and the University of Zimbabwe, 1978-2006. It was presented at the 9th Great Lakes of the World Conference (GLOW 9) held in Kisumu, Kenya, in August 2019. I am grateful to the organisers of that conference, Prof J. Njiru and Prof M. Munawar, for the invitation and to Prof Tumi Tomasson of the United Nations University Fisheries Research Programme for funding my attendance. Jeppe Kolding and Martin van der Knaap drew my attention to the possible impacts of the 1982-83 El Niño on the temperature of the lake.