CATCH TRENDS OF LIMNOTHRISSA MIODON IN LAKE KARIBA

by

Justin M.C. Lupikisha
Department of Fisheries,
Central Fisheries Research Institute,
P.O. Box 350100,
Chilanga,
Zambia.

1. INTRODUCTION

Lake Kariba, with a surface area of approximately 5400 km², is one of the largest man made lakes in the world. It is situated along the Zambezi river. The Lake is shared between Zambia and Zimbabwe (49% and 51% respectively). The 40 indigenous, and riverine fish species are inhabitants of the inshore waters of the lake. These form the basis for the inshore fisheries employing about 2000 fishermen in Zambia and 800 in Zimbabwe. The introduced clupeid or sardine Limnothrissa miodon (Boulenger) from Lake Tanganyika in 1967–1968 (Bell Cross and Bell Cross, 1971) occupies the pelagic waters and now supports the commercial fishery in both countries. The clupeid is by far the most economically important fish species in the lake.

Commercial fishing of the sardines started on the Zimbabwean side in 1973 (Machena and Kautsky, 1991). Initially a variety of fishing methods were tried to determine the best method of catching the sardine. These included the use of explosives, pumping and drifting gill nets but they all failed to yield the intended objectives. The methods of fishing employed in Lake Tanganyika were also tried but failed mainly due to two main reasons. The sardine in Lake Kariba has adopted a different feeding strategy. In Lake Tanganyika feeding is done in surface waters at night, whereas in Lake Kariba the fish generally remain below 10 to 15m. Secondly water transparency in Lake Kariba is much lower than in Lake Tanganyika. These are the factors which influenced gear design and the fishing method on Lake Kariba. The fishing method currently in use involves a vessel in the form of a platform mounted on two cylindrical pontoons, a winch (hydraulic or manual) which operates a lift net of varying diameter (6–9m) and with the depth of 14m. Other accessories include a generator which is a power source and mercury vapour bulbs. Although echo-sounders are used, they are optional. In Zambia the sardine fishery started in 1981, shortly after the liberation war in Zimbabwe.

Some aspects of the growth of sardines in Lake Kariba have been described by Cochrane (1984), while the mortality and potential yield have been investigated by Marshall (1985). A considerable volume of research work into its population dynamics and life history is currently going on under the auspices of the Zambia/Zimbabwe SADC fisheries project at Lake Kariba Fisheries Research Institute (LKRFI, Zimbabwe) and at Sinazongwe (Zambia). This paper considers the development of the sardine fishery on Lake Kariba with a bias towards catch and effort trends in commercial catches on both sides of the lake.

2. METHODS

Data on total catch and effort by year, month and fishing zones were extracted from Open Access database and from fisheries statistics reports of Zambia and Zimbabwe. Using these data, plots were prepared to illustrate the historical development and seasonality patterns. For catch per unit effort (CPUE), attempts were made, where possible, to fit total surplus production models. Detailed illustrations of these models are provided by Sparre et al (1989), Schaefer (1954) and Fox (1970).

Fishing effort in this paper refers to boat-nights. This does not take into account net or rig variation in terms of size, net colour, engine size, presence or absence of fish finding devices, hydraulic winch etc. While it is recognised that these attributes increase catch variability between rigs, it is very difficult to define fishing effort which would take into account the above attributes, especially in the absence of data on the time frame when such attributes were effected.

Efforts were also made to adjust Zambian catch data in light of speculations by commercial fishermen that reported catches represent 50% of actual yield. Finally, both short (Index of absolute variation-Ua) and long (Coefficient of variation -CV) term catch variations were computed.

3. RESULTS

3.1. Catch and effort trends

In Zimbabwe catches of sardines increased from 66 tonnes in 1973 to about 22,000 tonnes in 1990 (Fig.2). In the 1982/83 fishing period there was a sudden decrease to about 9,000 tonnes. This was due to the drought which hit the region during the same period. There after, an upward trend was observed. Similarly, effort increased from 616 boat-nights in 1974 to about 60,000 boat-nights in 1990.

In Zambia fishing effort increased from 18,874 boat-nights in 1982 to 33,383 boat-nights in 1990. On the other hand, catches increased from 4,136 tonnes in 1982 to 9,113 tonnes in 1986. There after it oscillated around 6,000 tonnes (Fig. 2).

From the single factor ANOVA (Analysis of Variance) there is a significant difference between catches of the two countries (F^[cal] = 29.69 and F^0.5[1,18] = 4.41).

Data collected in Zambia from 15 commercial fishermen (rig owners) in Siavonga, Chipepo and Sinazongwe sectors during the implementation of the new catch and effort data recording system on the lake, indicated that catches submitted to the Department of Fisheries (DoF) represent 53% of the actual catch. This gave a raising factor of 1.89 which was used to adjust catch data (Table 1).

Single factor ANOVA on adjusted (Zambia) and Zimbabwean catch data gave no significant difference at 95% confidence level between the two countries.

3.2. Seasonality

Monthly catches of sardines in Zambia showed a marked seasonal variation with peaks in February-March and July-August periods (Fig. 3a). It can be seen in this figure that the July-August peak is much higher than the former peak. Similar patterns have been reported in Zimbabwe. However, there were no clear patterns in daily catches. Low catches were experienced around December-January and May-June.

3.3. Number of rigs

In Zambia the number of rigs has remained fairly constant over the last decade (Fig. 3b). However, there was a slight increase in the Sinazongwe fishing sector between 1986 and 1990. These are probably not new entrants but fishermen moving from other areas such as Siavonga and Chipepo in search of better fishing grounds. This can be seen by a corresponding decline in the number of rigs in Siavonga and Chipepo fishing sectors.

3.4. Changes in CPUE

The overall picture of CPUE in Zimbabwe shows a continuous decline from 1974 up to 1982. Thereafter, a constant pattern emerged (Fig.4). However, more pronounced upward trends were observed in Chalala and Sengwa basins. It would appear that the two basins have no greater CPUE over the whole lake.

Linear regressions of CPUE on effort in each basin indicated a negative correlation in Kariba and Bumi basins (Table 2). However, Chalala, Sengwa and Binga/Mulubizi basins had a different pattern (Figs. 5 and 6) In these basins positive correlations were observed. This indicates that the CPUE has increased with increased effort.

As shown in Table 2 and Figs. 5 and 6 there is evidence that the use of fish finding devices and hydraulic winches in fishing activities may have had no significant effect on CPUE in Kariba and Bumi basins. This can be seen by a steady decline in CPUE (b and r² values).

In Zambia the CPUE has declined in all areas (Fig. 7). Table 3 shows results of a single factor regression analysis of CPUE on effort. The most important point about Table 3 is that the CPUE in all areas has decreased since the fishery started. This is expected of a fishery where fishing effort is ever increasing. The scattered nature of points on the graphs (also shown by low r² values) may be an indication of wide fluctuation in stocks of the sardines coupled with thefts by vessel operators. Tables 2 and 3 also suggest that perhaps total surplus models can not be fitted to this data except that of Kariba and Bumi basins in Zimbabwe and all sectors in Zambia.

3.5 Total surplus production models

Schaefer (1954) and Fox (1970) models assume decreased CPUE as a function of increased fishing intensity and this relationship is not very clear in this fishery. However, the assumption appear to hold for Kariba and Bumi sectors in Zimbabwe and for the whole of Zambian data. This is especially so for Kariba and Bumi basins where r values were reasonably high (0.6 and 0.5 respectively). Maximum sustainable yield (MSY) and effort associated with it (Fmsy) values of 56,284 tonnes and 100,000 boat-nights respectively were obtained using the Fox model fitted to the combined data of Kariba and Bumi basins.

The correlation coefficients (r²) of Schaefer plots were 0.67 for Kariba and Bumi basins and 0.13 for the Zambian data.

4. DISCUSSION

Advances in fishing technology and crew experience appear to have distorted CPUE trends in the fishery. This is particularly so in Zimbabwe where technological interactions are more pronounced. Consequently, existing catch and effort data of the sardine fishery is of little utility as far as the management of the fishery is concerned. Therefore, it does seem necessary to redefine effort which would take into account differences in catches that are bought about by differences in technology. Attributes which seemingly increase vessel efficiency and bring about huge catch differences between vessels include, size and colour of the net (mainly in Zambia where this is apparent), presence of echo-sounders, use of powered winches, number and power of lights per rig, crew experience and engine power.

An attempt to define effort based on such parameters was made (Chifamba, pers. comm). The results of a single factor ANOVA and multiple regressions, confirmed the effects of these attributes. However, the effect of these factors were difficult to separate since they are interrelated. This study also indicates that there may have been other factors other than those measured which had large influence on catch. This made the exercise to standardise fishing effort very difficult.

It would also appear that in order to define effective fishing effort it is necessary for DoF and LKFRI to carry out experimental fishing to separate the effects of these attributes. This may entail acquisition of experimental rigs by the two institutions. The rigs are very expensive but in the long run they may prove worthwhile.

Alternatively, a different effort measure may be defined. For example time spent looking for targeted shoals of sardines and hauling of a net could be considered. It is anticipated that time would increase with reduced stock sizes. This way, a direct relationship between fishing effort and stock sizes may be enhanced. Number of hauls made in a night is another index that could be used as a measure of effort.

In Zambia, under-reporting is probably the main cause of inconsistencies since technological advances are to a certain extent non-existent. The root causes of fish thefts on the lake should be examined. It is believed the poor marketing system coupled with scarcity of sardine are the main areas of concern. Traders spend long periods of time before they could be afforded chance to buy a bag of sardine. In certain cases traders have to travel long distances moving from one commercial fisherman to another in search of the sardine. This makes offers given by crew members, to sell part of the catch to traders on the lake before landing, very attractive. To counteract this, a centrally located place where all commercial fishermen could take their catches for sale to traders is being suggested. This could be made effective by issuing a certificate of origin of fish by DoF without which a trader would not be allowed to pass through check posts. This way thefts on the lake would be reduced.

Adjusted catch data given in this paper should be looked upon with caution. The raising factor used assumes daily thefts in all fishing companies, which is not often the case. Meaningful adjustments may be made when data on frequency of thefts both in time and space are available.

Seasonality pattern in mean monthly catches of L. miodon observed in this paper appear to be related to nutrient inflow and lake turnover. Recently, Masundire (1992) has shown that zooplankton ( the main food for the sardine) densities exhibit two peaks, one associated with the rainy season (January-March) and the other associated with lake turnover (June-September). He has also shown that in the Kariba basin there was a significant relationship between zooplankton biomass and CPUE of the sardines with a time lag of two months. Marshall (1982) also suggested a two months time lag between food abundance and increases in monthly catches of the sardines.

Catch variations related to the lunar cycle were not very clear. This is not surprising given the fact that fishing activities are closed during periods of full moon.

To the biologist the interesting aspect of catch and effort data is in using it to provide indices of stock abundance from which optimum exploitation levels of the resource could be derived. This forms the basis for fishery management. As indicated in this paper CPUE trends do not seem to be related to the relative abundance of the sardine stocks in the lake. The overall picture shown is that of increased CPUE in the last 10 years. Possible reasons for this situation are given above.

On the other hand, CPUE trends observed in the lake today may be due to the resilience of the sardine stocks. This is common among schooling clupeids and other small pelagics especially when subjected to very high fishing pressure. The fishery remains relatively efficient because a day spent fishing is efficiently directed to the ‘last’ schools. In other words catchability coefficient rises at low levels of biomass. The development of CPUE through time (years) shows no signs of deteriorating stock until the fishery suddenly collapses. Marshall (1987) demonstrated that the sardine's total biomass declined from 248,538 tonnes in 1974 to only 24,634 tonnes in 1985. This may be an indication that the fishery is heading for a disaster.

Values of absolute variation (Ua) given in Fig. 2 illustrate an existence of stable fisheries on both sides of the lake and that fishermen faced minimum short term catch variation. This is a scenario for stimulated investment into fishing effort and it is therefore, not surprising to see that effort has ever been increasing. However, for a long time, the sardine fishery in Zimbabwe was very unstable (coefficient of variation CV = 72%). This is due to the droughts which had hit the region from time to time.

5. REFERENCES

Bell Cross, G. and Bell Cross, B. 1971. Introduction of Limnothrissa miodon and Limnocaridina tanganicae into Lake Kariba. Fish. Res. Bull. Zamb. 5: 207 – 214.

Cochrane, K.L. 1984. The influence of food availability, breeding seasons and growth rate on commercial catches of Limnothrissa miodon (Boulenger) in Lake Kariba. J. Fish. Biol. 24: 623–635

Fox, W.W. 1970. An exponential yield model for optimising exploited fish populations Trans. Am. Fish. Soc. 99: 80 – 88.

Machena, C. and Kautsky, N. 1991. Fisheries development on Lake Kariba (Paper presented at the symposium on Lake Tanganyika. May 6–11 1991, University of Kuopio, Finland).

Marshall, B.E. 1982. The influence of river floe on pelagic sardine catches in Lake Kariba. J. Fish. Biol. 20: 465 – 470.

Marshall, B.E. 1985. A study of the population dynamics, production and potential yield of the sardine Limnothrissa (Boulenger) in Lake Kariba. Ph.D Thesis. Rhodes Univ. S. Africa.

Marshall, B.E. 1987. Growth and mortality of the introduced Lake Tanganyika clupeid, Limnothrissa miodon, in Lake Kariba. J. Fish. Biol. 31: 603–615.

Masundire, H.M. 1992. Bioeconomics and production of zooplankton and its relevance to the pelagic fishery in Lake Kariba. D. Phil. Thesis University of Zimbabwe.

Schaefer, M.B. 1954. Some aspects of the dynamics of populations important to the management of commercial marine fisheries. Bull. Inter-Amer. Trop. Tuna Commission 1, 27–56.

Sparre, P., E. Ursin and S.C. Venema 1989. Introduction to tropical fish stock assessment Part 1. Manual. FAO. Fish. Tech. Pap. No. 306/1. Rome. FAO. 337p.

Table 1. Reported and adjusted annual catches of Limnothrissa miodon in Lake Kariba from 1974 (Zimbabwe) and 1982 (Zambia) to 1990.

Note: - Dash (-) means data not available

Table 2. Regression parameters of CPUE on effort by basin from 1974–1990, Zimbabwe

Notes: * highly significant and
* * very highly significant

Table 3. Regression analysis of CPUE on effort in Siavonga, Chipepo and Sinazongwe from 1982–1990, Zambia.

Fig. 1 Sardine fishing areas on Lake Kariba.

Fig. 2 Catch and effort trends in Lake Kariba

Fig. 3 (a) Monthly catches of Limnothrissa miodon in Lake Kariba.

(b) Number of licensed rigs on the Zambian side of Lake Kariba, 1985–1990.

Fig. 4 Changes of catch per unit effort in the various sectors of the Zimbabwean Side of Lake Kariba, 1974–1990.

Fig. 5 The relationship between catch per unit effort and effort in some basins of Lake Kariba, Zimbabwe, 1974–1990.

Fig. 6 (a) Catch per unit effort against effort in Binga/Mulibizi basin.

(b) Catch per unit effort against effort in Kariba/Bumi basin.

Fig. 7 The relationship between catch per unit effort and effort on the Zambian side of Lake Kariba.