Following recent changes in the ecosystem of Lake Victoria resulting from the introduction of Nile perch, Lates niloticus, in the 1960s, there is urgent need for information on which to formulate rational exploitation and management of the fisheries resources. This paper presents information on the biomass and distribution of dagaa, Rastrineobola argentea, the second most important commercial species in Lake Victoria. Data were collected during five acoustic surveys conducted between 1999 and 2001, using a Simrad EY500 echo–integrator with a 120 kHz split–beam transducer. Fish samples were collected using a frame trawl, bottom trawl and pelagic trawl, all lined with 5 mm mesh size netting. The mean total biomass of R. argentea in the lake was 476,902 ± 339,308 t at an average fish density of 7.3 ± 5.4 t km−2 in the sampled area. Potential yield was estimated at 581,584 ± 305,377 t (8.5 ± 4.4 t km−2). The majority of the biomass of R. argentea (an average of 68%) was distributed in waters of more than 40 m depth. The biomass of R. argentea increased progressively over the survey period, probably as a result of decreased predation pressure by Nile perch.
Lake Victoria is the largest lake in Africa with a surface area of 68,800 km2 and a mean depth of 40 m and maximum depth 84 m. It is shared between Tanzania (51%), Uganda (43%) and Kenya (6%). The lake is located in a densely populated area with some 30 million people living in its immediate locality (Crean et al., 2002). These people depend on the lake not only for food, but also for its water, which serves domestic, industrial and agricultural purposes, as well as for transport.
Lake Victoria has undergone several changes in the productivity mechanism and in the fish community following environmental changes, overfishing and fish introductions (Ogutu-Ohwayo, 1990; Hecky and Bugenyi, 1992; Bundy and Pitcher, 1995). The fisheries of Lake Victoria are now based on three species: the introduced Nile perch (Lates niloticus L.) and Nile tilapia (Oreochromis niloticus L.) and the indigenous dagaa Rastrineobola argentea (P.). Current estimated annual catch of all species from the lake is between 400,000 to 500,000 t with a value of up to US $ 400 million (Crean et al., 2002). Although R. argentea is an indigenous species in the lake, its biomass increased in the sub-littoral areas in the 1980s following a decline in the haplochromine species (Wanink, 1991). Dagaa developed to become the second most important commercial species in Lake Victoria in the 1980s contributing 10–20% to the total commercial catch by weight (Reynolds et al., 1995). Current estimates in Uganda (70,333 t) (Muhoozi, 2002), Tanzania (167,789 t) (Mkumbo, 2002) and Kenya (66,620 t) (Bwathondi et al., 2001) indicate that it is dominating the commercial catches in the lake.
R. argentea is one of the main prey items of the introduced predator, Nile perch (Mkumbo, 2002), which is a big asset to the fisher communities around the lake and the countries sharing the lake through provision of income, foreign exchange and employment (Reynolds et al., 1995; Crean et al., 2002). R. argentea contributes to the ecological efficiency in the lake through transfer of energy from macroinvertebrates and primary consumers to fish, birds and man (Wanink, 1998). It is also a cheap source of protein especially to the poor and middle-income homes (Odongkara, 2001). In Kenya, catches do not meet demand of the internal market (Jansen et al., 1999), while in Uganda and Tanzania part of the catch is exported. Major markets include Democratic Republic of Congo, Rwanda and Burundi (Odongkara, 2001).
Fishing effort targeting R. argentea is not uniform on the lake nor are catch rates. Uganda has the lowest fishing density (0.08 boats km−2) compared with Tanzania (0.10 boats km−2) and Kenya (0.5 boats km−2). Catch rates by the end of the 1990s were lower in Kenyan waters (164 kg boat−1 day−1) (Othina and Tweddle, 1999) than in Tanzania (221 kg boat−1 day−1) (Mkumbo, 2002) and Uganda waters (205 kg boat−1 day−1) (Muhoozi, 2002). The majority of the boats use mosquito seines and are manually operated using oars (85.6%) (LVFO, 2005) and this limits the distance they can cover. This led to a concentration of fishing in the inshore areas and harvesting excessive amounts of juvenile fish of R. argentea and other species (Ogutu-Ohwayo et al., 1998).
For effective management of R. argentea in Lake Victoria, there is need for more scientific knowledge. This knowledge is currently lacking or limited. Previous lake-wide fishery surveys conducted on Lake Victoria never targeted R. argentea, nor was this species targeted by regular bottom trawl surveys conducted by individual research institutions around the lake. The only available biomass estimates of R. argentea in Lake Victoria are from indirect methods (Moreau et al., 1993; Pitcher et al., 1996; Villanueva and Moreau, 2002). Collection of commercial catch data of R. argentea by the fisheries staff is difficult because landings are at night. Available reports show scanty information in Uganda and Tanzania (Bwathondi et al., 2001). Inadequate information on the stock of R. argentea creates uncertainty on how to exploit the stock on a sustainable basis without affecting the ecosystem of the lake. Commercial catches of R. argentea from Kenyan waters have levelled off while those in Uganda and Tanzania waters are still increasing (Njiru et al., 2002, Muhoozi, 2002; Mkumbo, 2002). Furthermore, R. argentea stocks are infested by the parasite Ligula intestinalis (Linnaeus) which appears to interfere with the species' reproductive capacity (Marshall, 2001), while the stock may also be susceptible to the resurgence of haplochromines (Wanink and Witte, 2000).
The objectives of this paper are to provide information for a better understanding of the spatial and temporal changes of the biomass of R. argentea and factors that influence them; and to formulate options for the management of R. argentea in Lake Victoria.
Material and methods
Five surveys were conducted between August 1999 and September 2001 using a Simrad EY500 echo-integrator connected to a 120 kHz split beam transducer (9° beamwidth). The transducer was hull mounted on the RV Victoria Explorer, which is 16.5 m long. The acoustic system was calibrated before and after the survey, using a 23 mm diameter copper sphere of target strength −40.4 dB for the first four surveys and a 33.2 mm tungsten carbide sphere of target strength −40.6 dB for the latter.
Acoustic and fish data from sample trawling were collected along pre-designed cruise tracks covering both inshore and offshore waters (Figure 1). Factors considered in designing the tracks included: recording should take place during daytime when fish are better segregated into identifiable categories, the speed of the boat (10 knots), time required for two to three trawl stations a day, and the need for the vessel to go to a sheltered bay or jetty at the end of the day. During the first survey of August 1999, a stratified design was used which provided more coverage in the inshore areas where most of the fish were expected to be. For the subsequent four surveys, two extra days were added to allow more coverage in offshore areas as they had never been surveyed before.
Sampling of echo traces was done with a 3.5 m × 3.5 m frame trawl and a bottom trawl with a vertical opening of 3.5 m during the first four surveys. Catches of the frame trawl consisted of mainly small fish (R. argentea, haplochromines and juvenile Nile perch). A pelagic trawl with a vertical opening of 10 m was used during the fifth survey to overcome this limitation in order to attempt to catch larger specimens of the various species. The cod ends of all the trawls had a 25 mm mesh net, but were lined with 5 mm mosquito netting to ensure retention of the small fish. Operating depth of the frame trawl and pelagic trawl was monitored using a Furuno CR 24 netsonde. Fishing duration ranged between 10 and 30 minutes, depending on the concentration of echo traces, at a speed of about 3 knots. After each trawl haul, the catch was sampled to determine the species and size composition of the fish.
Acoustic data were collected mostly during the day, when there was some separation between traces of different types and not during the night when all traces would mix (at night it would also become impossible to discriminate between plankton and fish). Echoes from the entire water column were integrated in 10 m layers from 5 m down to 0.5 m from the bottom during the first two surveys and was later changed to start from 3 m from the surface for the last three surveys. Integrator tables showing the Nautical Area Scattering Coefficient (NASC) were printed as hard copies of echograms, with all relevant information related to the position.
Schools of R. argentea were found separated from other traces, distributed in contact with the bottom or in the whole water column (Figure 2). Echo energy of R. argentea when mixed with other species was obtained using proportions derived from trawl catches. To convert acoustic echo energy allocated to R. argentea to biomass, a target strength function which takes the form of TS = 20 Log10 (L)-b20 was required (Foote, 1987). A b20 value of −72 dB was used (Tumwebaze, 2003). Biomass calculation methods were applied as described by Tumwebaze (2003).
An estimate of potential yield of R. argentea was determined using Cadima's formula (Sparre et al., 1989) as:
The total yield used was 304,742 t (4.5 t km−2), 167,789 t from Tanzania (Mkumbo, 2002), 70,333 t from Uganda (Muhoozi, 2002) and 66,620 t from Kenya (Bwathondi et al., 2001). The natural mortality used was 1.8, which was an average from 2.3 from Wandera, (1992), 1.8 from Wandera and Wanink, (1995) and 1.4 from Manyala et al. (1992).
There was significant variation in the biomass of R. argentea between surveys (Kruskal Wallis H test; P < 0.005) (Table 1). The mean biomass for the five surveys was 476,902 ± 339,308 t (95%CL) with an average density of 7.3 ± 5.4 t km−2 (95%CL). The biomass values showed seasonal variation with the average biomass in January/February being 110% higher than that in August/September period. On average, 68% of the biomass was found in waters more than 40 m deep.
The biomass of R. argentea showed consistency in the distribution pattern with preference for areas in the eastern part of the lake (Figure 3). The biomass was highest during the survey of January/February 2001. There was an increase in the biomass during the assessment period (1999–2001) with some areas where dagaa was absent in the first three surveys having a biomass exceeding 20,000 t in the last two surveys in 2001 (Figure 3).
Potential yield of R. argentea
The model estimates the MSY of R. argentea in Lake Victoria to be 581,584 ± 305,377 t (8.5 ± 4.4 t km−2). The estimated yield of R. argentea using an approximate model based on primary production and using Lake Kariba as the base line lake falls within this range (451,578 t, 6.64 t km−2) (Pitcher et al., 1996).
Survey design and methodology
This paper described the first acoustic assessment of R. argentea in Lake Victoria. During the assessment, improvements were made regarding the survey design and sampling of echo traces during the earlier surveys which could have introduced bias. Another source of error could have resulted from recording part of the long transects at night when fish were distributed close to the surface and mixed with Caridina and other macro-invertebrates.
In general, R. argentea seemed to be well suited for surveying by the acoustic method. Being a schooling species, it produces strong echoes that can be well detected by acoustic methods. The schools of R. argentea are small and are not sufficiently dense to cause an ‘overshadowing effect’, a case where the sound energy from the acoustic system cannot pass through the dense aggregations but instead treats them as the lake bottom leading to underestimation of the biomass (Appenzeller and Leggett, 1992). In most cases, R. argentea had a low degree of mixing with other species and was distributed away from the bottom. Of major concern, however, was its close distribution to the surface at times and its vessel-avoidance behaviour.
Comparison with previous estimates of R. argentea
There are no other biomass density estimates of R. argentea obtained from surveys with which to compare directly the results from this study. However, comparisons can be made with estimates obtained from indirect methods. Density estimates of R. argentea for the Kenyan part of the lake by Moreau et al. (1993) using ECOPATH, a multispecies trophic model, were 7.7 t km−2(for 1971-1972) and 7.9 t km−2 (for 1985–1986). Moreau (1995) reported that inputs for the model were obtained from the extensively exploited shallow part of the lake, the Nyanza Gulf and may not apply to the whole lake. Other biomass density estimates of R. argentea by Pitcher et al., (1996) using an approximate model based on primary production in Lake Kariba, ranged from 3.6–15.2 t km−2 (95% limits) with an average of 8.7 t km−2. The density estimates obtained in this study are within the range of estimates from the non-survey methods. The models seem to provide reasonable overall estimates although they may not indicate spatial differences.
Spatial and temporal variation in the biomass of R. argentea
The biomass of R. argentea showed an increasing trend over time but with marked seasonal variations. The increase in the biomass of R. argentea started in the 1980s following the establishment of the introduced Nile perch and the decrease in haplochromines (Wanink, 1991) and it still seems to be increasing. Possible reasons for further increases in the biomass of R. argentea include decrease in the biomass of the predator Nile perch as indicated by a decrease in commercial catches in the Kenyan (Njiru et al., 2002) and Tanzanian parts of Lake Victoria (Mkumbo et al., 2002). Increases in Caridina, which has become a major component of the diet of Nile perch (Mkumbo et al., 2002) and recovery of haplochromines (Witte et al., 2000) may also have released predation pressure on R. argentea and contributed to its increase. Caridina is an important food resource of R. argentea (Budeba, TAFIRI, Dar-es-Salaam, personal communication), which also contributed to the increase of R. argentea biomass.
Spatial distribution from the five surveys showed that R. argentea had a lake wide distribution and was present virtually throughout the surveyed area. The highest concentration of R. argentea in Lake Victoria seems to be associated with the deepest areas of the lake which form an arc along the eastern side of the lake (Talling, 1966). Scholz et al. (1990) found the thickest sediment deposits in Lake Victoria correlated with the deepest areas of the lake. High concentrations of R. argentea in the area with the thickest sediments may be related to availability of food through increased primary and secondary production following the mixing period of May to August (Talling, 1966).
Major spatial and temporal variations in the biomass estimates and distribution of R. argentea make it difficult to separate seasonal changes from possible declines in the stocks which make assessment and management difficult. The stock of R. argentea showed lower abundance in shallower waters in January/February than in the August/September period, suggesting changes in the occurrence of fish on the fishing grounds. These variations have several implications for the maintenance of catch, price of the fish, quality of the fish, especially with local processing methods (sun drying), and uncertainty about the state of the stock. Short lived species are known for large variations partly contributed by environmental factors (Hilborn and Walters, 1992). Larvae of the small pelagic fish in African lakes, e.g. Engraulicyprus sardella (Lake Malawi) and Limnothrissa miodon (Lake Tanganyika), feed on phytoplankton (Nitzschia spp.) before shifting to zooplankton (Degnbol, 1982), and variation in the productivity of the lakes seems to have an influence on the survival of the juveniles and the recruitment success (Tweddle and Lewis, 1990.). Factors which determine recruitment success and spatial variation in the distribution of the R. argentea stock are not well understood and need to be investigated.
Management of R. argentea
The greater biomass of R. argentea in deeper waters suggests that a larger catch would be obtained if the fishery shifted to deeper waters than where it is currently being operated. Muhoozi (2002) found that powered boats which fish in offshore waters landed about double the catch of manually propelled boats which fished inshore, although the gears used were the same (mosquito seine nets of 5 mm mesh size). Accessibility to higher concentrations of R. argentea stocks in deeper waters is limited by current fishing technology since the majority of the fisherfolk using mosquito seines have boats propelled by oars (85.6%) (LVFO, 2005). There is a need to extend fishing to deeper waters to increase catches. This will also reduce by-catch of juvenile fish, Nile perch, Nile tilapia (Oreochromis niloticus) and haplochromines in commercial catches (Ogutu-Ohwayo et al., 1998).
The yield of R. argentea in Lake Victoria for the year 2000 (304,745 t (4.5 t km−2) calculated from Mkumbo (2002), Muhoozi (2002) and Bwathondi et al. (2001) falls within the 95% CL of the estimated potential yield. This calls for caution in further increase in the fishing effort especially with the resurgence of haplochromines since the biomass of R. argentea in the sublittoral areas increased following the decline of haplochromines (Wanink, 1991; Wanink and Witte, 2000). The decline of R. argentea may not be detected quickly as the two species groups are caught together and are not reported separately. Regular Catch Assessment Surveys (CAS) need to be conducted to provide information necessary for monitoring the performance of the stock. There is need for continuation of acoustic surveys for monitoring as this is the most suitable way to track population changes timely. For protection of the fishery, there is need to control fishing effort, identify and protect nursery areas, institute close seasons and prohibit use of small mesh sizes less than 5 mm.
In conclusion, the ecosystem of Lake Victoria is currently very dynamic and management measures need to change in line with ecosystem change. The sustainability of the R. argentea fishery will depend on the ability to shift the fishery to offshore areas, to control fishing effort, to monitor the changes and to transform management options into actions.
The study was supported by the European Development Fund and the Governments of Kenya, Uganda and Tanzania through the Lake Victoria Fisheries Research project (ACP-327). We acknowledge support and advice from the staff of the fisheries research institutions in Uganda (FIRRI), Kenya (KMFRI) and Tanzania (TAFIRI), and consultants from FRS Marine Laboratory in Scotland.