PREDATOR-PREY RELATIONS IN NORTHERN LAKE TANGANYIKA AND FLUCTUATIONS IN THE PELAGIC FISH STOCKS

by

F.C. Roest
FAO Fishery Biologist
Fisheries Department
Rome, Italy

ABSTRACT

The catches of clupeids and their Lates predators from Lake Tanganyika's pelagic waters fluctuate dramatically. In the northern part of the lake, the two major pelagic species, Stolothrissa tanganicae and Lates stappersii show cyclical changes in abundance which are negatively correlated. It is suggested that S. tanganicae, which normally lives in the open lake, migrates to inshore waters to spawn. Juveniles return to the offshore zone and are recruited to the commercial fishery at a length of 55mm. Most of the Limnothrissa miodon population lives permanently in the inshore zone. The spawning period of Stolothrissa coincides with that of its main predator, L. stappersii though the latter species spawns offshore. When juveniles of L. stappersii change from a planktivorous to a piscivorous diet they migrate inshore to feed upon juvenile Stolothrissa and return to the offshore waters at a length of 200–300mm. The combination of simultaneous spawning and differential growth rates for Stolothrissa and L. stappersii results in a two month difference in their recruitment to the industrial fishery and a two year time lag between peak catches of the two species. Recruitment is largely independent of parent stock size and the abundance of clupeids is more closely related to the level of predation than to the annual level of primary productivity.

1. DESCRIPTION OF THE FISHERY

1.1 Introduction

In the northern part of Lake Tanganyika, the pelagic fish stocks are exploited by the industrial, artisanal and traditional light fisheries using respectively, purse seines, lift-nets and scoop-nets. A number of other artisanal gears are used for other species (Enderlein, 1976; Herman, 1984).

The main species exploited are the endemic clupeids Stolothrissa tanganicae and Limnothrissa miodon and their predators, Lates stappersii (formerly Luciolates), Lates mariae, L. angustifrons and L. microlepis. The small species, Stolothrissa and Limnothrissa and juvenile Lates stappersii (<190mm), appear together in the commercial category, ‘ndagala’ (Burundi, Zaire), ‘dagna’ (Tanzania) or ‘kapenta’ (Zambia), whereas Lates stappersii is called ‘mukeke’ (Burundi, Zaire), ‘migebuka’ (Tanzania) and ‘mvolo’ (Zambia). The larger Lates species are known as ‘sangala’. The commercial names ‘ndagala’, ‘mukeke’ and ‘sangala’ will be used in this paper.

1.2 Industrial Fishery

The following description of the history of the industrial fishery in Burundi is largely based on Herman (1984). An industrial fishery on Lake Tanganyika started in the northernmost part of the lake (Bujumbura, Uvira), when Greek fishermen built two fishery vessels of traditional Mediterranean type. Different fishing methods and gears were tested, but the small size of these first boats limited the size of the purse seine nets and obliged the fishermen to exploit only the clupeids (ndagala). During 1956, which was a year in which L. stappersii was very abundant, the number of vessels doubled to four and their size, as well as the dimensions of the nets and the mesh sizes used, increased. In the following years, the number of vessels fishing in the northern part of the lake increased, and by 1958 had increased to 12. As a management measure, this number was held constant until 1960. After the independence of Zaire in 1960, the eight remaining vessels based at Bujumbura were no longer allowed to fish in Zairean waters. No new vessels were allowed to enter the Burundi fleet until 1968. Around 1962, new nets had been introduced which combined small meshes for ndagala in the upper part and larger meshes for predators in the lower part. These nets are believed to have been adopted by all vessels by 1964. A study of the predator-prey relationship, based on industrial fish catches, can thus be made only from 1964 onwards. The numbers of vessels gradually increased to 11 in 1960, 14 in 1970, 20 in 1975 and 23 in late 1976. During 1977, following the recommendation of Turner and Herman (1977), fishing effort was limited to 18 boats. At the end of 1977, five boats left Burundi to fish in Uvira and Kalemie (Zaire). These vessels have since returned, so 23 vessels were again operating in 1980 though their number had decreased to 19 by 1983.

In order to protect the artisanal and traditional fishing zones, prior to 1980 industrial fishing was not allowed within a radius of 15km of Bujumbura nor within 5km of the Burundi shoreline. In addition, since the beginning of the industrial fishery, the total catch has had to be sold at the market in Bujumbura, which considerably aids the collection of statistics.

2. CATCH AND EFFORT DATA

Excluding the initial development phase of both artisanal and industrial fisheries (1950–64) and the period of political unrest and its immediate after-effects (1972–75), the total fish catch landed in Burundi has fluctuated between 10,000 and approximately 25,000t (Fig.1). In the period 1965–71, total catches varied between 12,000 and 20,000t and after 1975, they increased to as much as 24,600t (in 1977) before dropping to the 1981–83 levels of 10,000–12,000t. The industrial catch usually constitutes less than 50% of the total catch.

The industrial catch increased steadily to about 8,000t in 1976, after which it fell back to the 1965–66 levels of 3,000–4,000t. There is, however, reason to believe that since 1977 the industrial catches have been underestimated (Herman, 1984). The annual industrial catches show considerable fluctuations in their species composition (Fig.2). Ndagala usually dominates the catch while mukeke contributes, on average, about 25 percent, but the latter contribution to the catch has varied from 3% (1971) to 62% (1978, 1983). Sangala was initially of great importance but declined after 1961 following about five years of commercial exploitation. This was also recorded by Coulter (1976) in Zambian waters.

As detailed data on catch magnitudes and composition are only available for the industrial fishery, this paper will deal mainly with observations from this fishery.

Catches per unit of effort (tonnes per boat year) for the industrial fishery in Burundi during the period 1956–83, show strong periodic fluctations in the abundance of the principal commercial categories (Fig.3). As explained above (1.2), 1964 (when mixed-mesh purse seines were introduced) is considered to be the first year in which the relative abundance of predator and prey species can be assessed from commercial catches. Ndagala catches per unit effort were high in 1966–68, 1971 and 1973–74 and showed a minor peak in the period 1979–81. A somewhat higher CPUE for ndagala in 1861 could also be considered to be a minor peak. Mukeke CPUE show similar peaks alternating with ndagala peaks in 1963, 1967–69, 1976–78 as well as in 1980. The periodicity of the abundance cycles therefore seem to be in the order of 6–7 years for ndagala and 6–8 years for mukeke. The duration of the ndagala abundance increased from three years in 1966–68 to about 4 years in 1971–74 (considering the whole bimodal graph), while their maximum level in t/boat year decreased from 445 in 1966–68 to about 360 in 1971–74 and 135 in 1979–81 over these periods. At the same time, the duration of the mukeke abundances increased from two years in 1962–63 to four years in 1967–70 and to five years in 1975–79 while their maximum level increased from 135 to 145 and 175t/boat year respectively. The total industrial catches per boat year have declined from about 600t in 1967 to the present low level of about 170t in 1983.

Ndagala is composed of variable percentages of three species: Stolothrissa, Limnothrissa and Lates stappersi (Table 1). Stolothrissa, the main component, can be seen to contribute 41–85% of the total annual ndagala catch. In Burundi this species shows a strongly seasonal pattern of abundance which largely determines the seasonality of the ndagala catch (Fig.4). A maximum CPUE is reached in the period October/November to December/January and an annual minimum in April/May – May/June (lunar months). As the life span of Stolothrissa is about 12 months, the increasing monthly CPUE figures of ndagala can be considered to represent the average biomass increase of the main cohort of this species, and the decline from December/January the biomass decay. The “biological year” of Stolothrissa, thus defined, is used in further analyses. As Stolothrissa is the main prey species in the lake it is to be expected that the biological cycles of the predators will be adapted to the timing of the main events in its life cycle. Figure 4 also shows the average monthly abundance of sangala and mukeke (not on the same scale), which can be seen to correlate negatively with ndagala abundance.

On a monthly basis, the alternating abundance of Stolothrissa and mukeke for the period April/May 1973 to September/October 1980 is illustrated in Fig.5. The significant negative linear correlation (p<0.05) between the annual abundances of the two species is shown in Fig.6. Catches of the other Lates spp. are not considered here because of their low levels since 1962 (see Fig.3).

3. BIOLOGY

3.1 Summary of Relevant Data

Stolothrissa has a life span of approximately one year during which it reaches an average maximum size of 90–100mm (fork length). Adults inhabit the open-lake waters while juveniles live inshore and recruit to the offshore fishery at about 55mm (Coulter, 1970; Roest, 1978). Stolothrissa feeds upon zooplankton, shrimps (Atyidae), calanoid and cyclopoid copepods, chiromonid larvae and pupae, and fish eggs and fry (Marlier, 1957; Matthes, 1968; Chene, 1975). The youngest Stolothrissa appear to be phytoplanktivorous and some fry of 3–7mm total length were observed to have intestines full of phytoplankton. It is not known at what age they change from a diet of phytoplankton to zooplankton (FAO, 1978). Spawning is continuous though the exact timing of the annual spawning peak varies from year to year and also from region to region (January-April in Kigoma, Tanzania (Chapman and Van Well, 1978); March-July and November-February in Zambia (Matthes, 1968; Coulter, 1970; Ellis, 1971); February-July and October-December in Burundi (Mann et al., 1973)). The time lapse between two successive spawning maxima is often 13–14 month, corresponding to the time needed for two successive generations to reach sexual maturity (Roest, 1978). Fifty percent maturity sizes in Zambia were studied by Ellis (1971) who found an average of 70mm FL for males and 75mm for females. The Von Bertalanffy growth equation for the period 1972–76 was lt = 93.8 (1-e^-0.211t) (Roest, 1978).

Spawning areas have not been defined, but were generally supposed to be pelagic (Poll, 1953; Matthes, 1968; Coulter, 1970). Pelagic eggs and larvae would be driven inshore by wind and wave action. However, after studying gonads from fish caught inshore, Marlier (1957) believed that there was a prespawning migration towards the shoreline. This observation is supported by Enderlein (1976), who, after sampling the artisanal inshore fish catches in Burundi for over a year, concluded that in the period February-April 1974, the Stolothrissa stock moved close inshore. Ripe-running fish have never been found in the industrial fish catches (Mann et al., 1973).

Lates stappersii is Stolothrissa's most important predator. Most of Stolothrissa's observed rapid horizontal and vertical movements are aimed at avoiding this predator. At night, Stolothrissa concentrate in layers and rise in the water column to follow the plankton movements, before moving deeper again during daylight hours. Sequential nightly catches in the same location show great differences in Stolothrissa abundance, while L. stappersii catches vary much less. Apparently the latter follow the rapid clupeid movements much less efficiently than the pelagic Lates microlepis (FAO, 1978).

Adult L. stappersii are pelagic and prefer the upper water layers. Juveniles hatch in the offshore pelagic zone where they are caught from 20mm upwards. Juveniles are planktivorous (Ellis, 1978) and do not start feeding upon fish until they reach a length of 70mm. At 130mm they are entirely piscivorous and feed almost exclusively on Stolothrissa. Juveniles are unevenly distributed over the lake and are virtually absent from the catches around Kigoma (Chapman and Van Well, 1978b) and in Zambia (Coulter, 1970; Pearce, 1985). However, Chapman et al. (1974) found concentrations in the central part of the lake, 200km south of Kigoma. The mean values for fifty percent maturity lengths are 280mm for males and 300mm for females though the values vary throughout the year (Ellis, 1975).

A large part of the Limnothrissa stock spends its entire life in the inshore areas. Although this species, like Stolothrissa is basically planktivorous, its feeding habits are less specialized (Matthes, 1968). The wider range and regular availability of planktonic food inshore are more suitable for Limnothrissa whereas the food preferences of Stolothrissa give it a competitive advantage in the pelagic zone.

Coulter (1976) has described the biology of and fishery for Lates spp. in Lake Tanganyika. Adult L. microlepis are fully pelagic and inhabit the surface waters. Fluctuations in the catches indicate that this species ranges freely and tends to concentrate where prey is abundant. Juvenile L. microlepis live inshore where they possibly feed on clupeids. Lates mariae and L. angustifrons are benthic predators. After reaching sexual maturity (500–550mm FL), L. mariae depends almost entirely upon Stolothrissa for food. During periods of high clupeid abundance, this species also correlates positively with Stolothrissa as witnessed in Zambia by Coulter (1970) and in the Tanzanian part of Lake Tanganyika by Chapman et al. (1974). From the difference in catch composition between purse seines and gillnets, it appears that only part of the population migrates upwards at night to feed on clupeids. Lates angustifrons is the least important pelagic predator of Stolothrissa. According to Coulter (1976), its abundance near the surface is less subject to seasonal variations than that of other species.

In addition to the above species, the following predominantly inshore species are reported by Matthes (1968) to prey on Stolothrissa:- Bathybates fasciatus (almost exclusively), Alestes rhodopleura, Bathybates spp., Boulengerochromis microlepis and Lamprichthys tanganicus. Barilius moorei, Hydrocynus vittatus, Dinotopterus cunningtoni and Lamprologus spp. are occasional predators especially on the young in the inshore areas. To this latter category should be added the larger species of Mastecembelus. Large catfish (Dinotopterus) are frequently seen in industrial catches in Burundi, especially during the period of high offshore clupeid biomass from September to February (personal observation).

3.2 Further Biological Data from Industrial Catches in Burundi

With the exception of the northernmost part of Burundi (Ruzizi plain), the shoreline of Lake Tanganyika is steep and the water near the shore is often deep. Pelagic species are thus able to make use of this inshore zone and a large part of the Limnothrissa stock lives permanently in these deep inshore regions of the lake. It has been suggested that at least part of the Limnothrissa population moves into the offshore zone at a length of about 100mm when following Stolothrissa which are recruiting to the offshore (>5km) industrial fishing zone (Henderson, 1976). The reported average recruitment length of about 55mm for Stolothrissa (Coulter, 1970; Roest, 1978) was confirmed for the period 1971–80 (from 146 average lengths of lowest monthly Bhattacharya (1967) peaks in samples) and found to be 56.9mm. In view of the rather large standard deviation of 13.6cm, it was decided to consider the monthly recruitment index for Stolothrissa to be the sum of the numbers of individuals of 52, 55 and 58 mm FL caught per boatnight divided by 3. These recruits, with an average length of around 55mm, have an age of about 4 months. The monthly variations in recruitment strength indicate that either spawning intensity or survival of juvenile Stolothrissa are subject to rather large fluctuations within the year (Fig. 7). Spawning periods back-calculated from peak recruitments are shown in Fig. 8. They are variable within the year and, for the period 1973–79, show a regular shift with the interval between two annual peaks often being about 13 months (as already observed by Roest, 1978). The main cohort of 1974 can be traced back to April/May and the following cohorts to May/June in 1975, June/July in 1976, June/ July in 1977 and to August/September in 1979. This implies that the main annual Stolothrissa cohort does not originate from the maximum annual abundance of this species at the end/beginning of the year, but rather from the period of lowest annual adult Stolothrissa abundance in the industrial catches.

The main spawning period of Stolothrissa coincides with that of Lates stappersii (as was also noted by Chapman and Van Well, 1978) which, during this period, shows an increased catchability as it concentrates in the pelagic zone. It would, therfore, be reasonable to suppose that Stolothrissa might have developed a strategy to avoid sharing the same pelagic breeding site with its main predator. The absence of ripe-running Stolothrissa in offshore catches (Mann et al., 1973) suggests that prior to spawning, fish move out of the offshore fishing zone and closer inshore (Marlier, 1957 and Enderlein, 1976). They then spawn in the inshore, pelagic area where water depths are great enough for the normal development of the sinking pelagic eggs (Matthes, 1968). This inshore movement would have the additional advantage that larvae would not have to swim or be transported over large distances when moving inshore where their further development takes place. The relative absence of adult Stolothrissa in the industrial catches during their presumed spawning period may, to some extent, be explained by the fact that Lates stappersii are at that moment much more numerous and, because of their higher market value, always constitute the target species of this fishery. Stolothrissa landings per boat may therefore not be representative of their true abundance as they could either be partially discarded or the fishery might selectively exploit the mukeke stocks.

According to Fig. 9, 190mm is the approximate length at which equal numbers of Lates stappersii occur in the commercial categories ndgala and mukeke. These are sampled separately and, surprisingly, until now no attempt has been made to combine the length frequencies of juveniles and adults of this species. Ellis (1978) treats them separately. Length-converted catch curves (Pauly, 1984), using L∞ = 473mm, K = 0.4 and t = 0 for the individual years 1973/74 to 1979/80 are shown in Fig. 10 and a total curve for the whole period is given in Fig. 11. These figures clearly show the strongly bimodal nature of the catch curves, and indicate that fish in the 10–22 month age group (130–250mm) are less well represented than those in other age groups. As the industrial fishery covers the whole pelagic area of the northern part of the lake, fish belonging to the missing length group are likely to be absent from this zone. Ellis (1978) states that Lates stappersii start preying upon fish from a length of 70mm and are almost exclusively piscivorous at 130mm. Ellis (op.cit.) also indicates that juvenile L. stappersii feed almost entirely on Stolothrissa, thus it is likely that the missing length group of L. stappersii are where its main prey, juvenile Stolothrissa, live. Until their recruitment length of 55–57mm, Stolothrissa inhabit the inshore areas where they hatched or developed and where only artisanal fishing is allowed. Data on the species composition and length composition of catches from inshore areas are unfortunately rare and are mostly limited to the period 1974/75 for Burundi (Enderlein, 1976). Visits to artisanal landing places have indeed confirmed the presence of young L. stappersii of 130–250mm in the catches, although even here they tend to be relatively scarce, being too fast-swimming to be caught in the slow artisanal and traditional gears (lift net, scoop net). L. stappersii spawn and hatch in the pelagic zone, during the period of peak annual catches of this species. Juveniles are caught from 20mm upwards, and Fig. 11 shows that there is probably 100 percent retention from an age of approximately six months (corresponding to 90mm forklength), after which they gradually disappear to the inshore zones. The average instantaneous total mortality coefficient calculated from the total length-converted catch curve (1973–80) is in the order of Z = 1.78 per year. When continued backwards, the line, with negative slope Z, passes almost exactly through the points on the catch curve corresponding to the lengths of 90 and 110mm, indicating that Z is constant from the age of 6 months onwards.

A comparison of the numbers of fish present according to the catch curve and the total mortality line in Fig. 11 enables an estimation to be made of the percentages of juvenile Lates stappersii of 130–270mm living inshore (Fig. 12). The graph shows that less than 50% of the portion of the L. stappersii population between 140mm and 250mm FL is present in the offshore industrial fishing zone. A second recruitment to the industrial fishery starts from approximately 200mm reaching 50% at 245mm and 100% at 290mm which is the approximate length of 50% maturity (280mm males, 300mm females) in this species (Ellis, 1975). Fish therefore apparently return to the pelagic area to spawn. The sum, divided by 3, of the numbers of individuals of 70mm, 90mm and 110mm (FL) caught per unit of effort was used as a monthly recruitment index for L. stappersii (Fig. 7). According to the growth curve used, these fish have an average age of 0.53 year or 6.3 months or approximately 7 lunar months. The correspondence in timing between the numbers of these recruits and the monthly CPUE of mukeke in the industrial catches in Burundian waters (Fig. 13) seems to confirm that they originate from the spawning concentrations. There does not seem to be a straightforward parent/recruitment relationship.

The average age of L. stappersii at sexual maturity (corresponding to a length of 290mm) is about 28 months. Spawning periods can therefore be expected to shift from year to year. The timing of the months of annual peak abundance of mukeke in the industrial fish catches in Burundi is summarized in Fig. 14 for the period 1956–83. This figure shows that there is a gradual shift in supposed spawning period from December-January in 1962 and 1963 to July-August-September in the period 1976–79. This annual shift closely resembles the one observed for Stolothrissa spawning or juvenile survival shown in Fig. 8.

4. SPECIES INTERACTION

The apparent simultaneous hatching of Stolothrissa and L. stappersii which are observed as recruits respectively four and six months later, probably reflects optimum survival conditions of juveniles of the two species (which have the same planktonic food habits) as much as exactly simultaneous spawning. Stolothrissa juveniles, however, live somewhat closer inshore than those of L. stappersii. The observed shifts in peak abundance of L. stappersii and in the most successful (though not necessarily most intensive) spawning periods of Stolothrissa (as estimated by back calculation), can be explained by their life cycles. L. stappersii takes in excess of two years to reach sexual maturity and Stolothrissa more than six months. The frequent “return” to the period December-May (L. stappersii, Fig. 14) and April-May (Stolothrissa) probably indicates that these are the optimum spawning periods which guarantee maximum juvenile survival in the following months. Both L. stappersii and Stolothrissa have prolonged if not continuous spawning periods, (Poll, 1953) but apart from larval mortality (e.g. wind induced mortality of pelagic larvae), the presence of suitable planktonic food and the numbers of predators (and possibly the parent stock size) determine the “success” of a brood expressed as the recruitment index respectively 4 and 6 months after birth. It is probable that until L. stappersii change over to their piscivorous diet at 130mm, they live under conditions similar to those experienced by Stolothrissa and probably even suffer similar mortality rates. After this “critical” phase, L. stappersii move inshore at the age of 8–10 months and start to prey upon a new generation of Stolothrissa. During the following year, the numbers of Stolothrissa juveniles (<55mm) and possibly some other less important prey species in the inshore habitat, determine the survival rate of L. stappersii.

Typically, the simultaneous hatching or origin of L. stappersii and Stolothrissa results in maximum recruitment of juvenile L. stappersii to the industrial catches two months after the recruitment of Stolothrissa. On an annual basis (Fig. 3), there is a time lag of about two years between ndagala and mukeke peaks in industrial catches as mukeke hatched at the same time as Stolothrissa in a given year return to the pelagic area to spawn for the first time two years later. The 1963 peak of L. stappersii abundance clearly followed the 1961 abundance of ndagala (supposedly mainly composed of Stolothrissa), but all ndagala-sized fish would be suitable prey except for the larger Limnothrissa. Ndagala catches peaked again in 1966–1967–1968, followed by high mukeke abundance in the following year. During the first year of ndagala abundance, both Stolothrissa and L. stappersii find suitable planktonic food and few predators thereby guaranteeing optimum survival rates while, in the second year, inshore prey would be abundantly available. The single ndagala peak in 1971 followed by a low abundance in 1972 led to a very modest increase in mukeke CPUE in 1972–73. In both 1973 and 1974, ndagala (Stolothrissa) abundance was high and Lates stappersii abundance increased in 1975–77. Since 1977, the low CPUE of ndagala are clearly a result of the relatively high abundance of L. stappersii, but the continued high catch rates of the latter, especially the 1980 peak, are difficult to understand from the available data.

5. DISCUSSION

The timing of the annual primary production peak is dependent upon the seasonal enrichment of the epilimnion through the mechanisms of upwelling in the Zambian region of the lake (Coulter, 1968), temporary destratification (after cessation of the wind in the northern end of the lake) and internal waves. Although the same S. -S.E. wind blowing along the whole lake in the dry season is directly or indirectly responsible for these phenomena, considerable differences in timing of ndagala abundance patterns occur in Burundi, Zambia and Zaire (Fig.15, data from Herman, 1984). The temporal variation in the abundance of Stolothrissa (due to variations in food abundance and predator numbers) in different parts of the lake, coupled with their short life span indicates that a number of different stock units occur in the lake. Coulter (1970) considered the clupeid population of the southwest arm of Lake Tanganyika to be a separate unit. Available information seems to indicate that there are only a few Lates stappersii nurseries around the lake. Most of the shores are very steep and L. stappersii may find the optimal conditions for the postulated inshore phase of its life (when it preys upon juvenile Stolothrissa) only in certain areas (e.g. in Burundi but not in Zambia). When adult, L. stappersii leave these areas and occupy the pelagic areas to prey on offshore Stolothrissa. Depending on the timing and distance from the spawning or inshore growing areas, the distribution of L. stappersii may be very uneven throughout the lake. The annual concentration of L. stappersii in the northern part of the lake may thus be partially composed of fish from other areas, possibly including Kigoma, although Henderson (1976) suggests low levels of mixing of the adult populations.

In view of the importance of the predator-prey relationship in determining the abundance of clupeid prey, the relation between annual primary production and abundance of pelagic clupeids is likely to be less close than suggested by Coulter (1981) and Hecky et al. (1981). Numbers of recruits of L. stappersii appearing at the age of two years in the catches have been dependent on two years of conditions favourable to ndagala development. After the first year, their survival is dependent on factors affecting larval mortality, abundance of planktonic food as well as predation, fishing pressure and possibly, to a lesser extent, parent stock size. During the second year they depend largely on the abundance of suitable prey which, in its turn, depends on the prey parent stock, predators, wind and plankton conditions.

Constructing a model that would take all or most of these factors into account would require a considerable volume of limnological and fisheries data, but would have no predictive value. It would, however, be interesting to survey the inshore part of the stocks of both Stolothrissa (<55mm) and L. stappersii (130 – 250). For the latter, industrial fishing gears could be used though a quantitative study of Stolothrissa would require smaller meshed gear. In the absence of such data, the logarithmic multiple regression model linking prey-recruitment strength to prey parent biomass and predator biomass (Hopkins et al, 1982) was applied. This method has been successfully used by its authors for calculating recruitment of penaeid shrimps, Indian halibut and false trevally in the presence of different demersal predator densities. Table 2 summarizes the data used in this analysis. The following equations were obtained:

ln R = 5.63 + 0.34 ln B Parents + 0.24 ln B Preds_I

ln R = 5.88 + 0.34 ln B Parents + 0.20 ln B Preds_II,

where R = Recruitment index Stolothrissa in numbers (prey)

B Parents = Relative biomass of Stolothrissa prey parents, based on CPUE

B Preds_I = Relative biomass of the predators mukeke, sangala plus all Limnothrissa caught in the industrial catches

B Preds_II = Relative biomass of the predators mukeke, sangala plus Limnothrissa equal to or over 100mm length

Distinction between different sizes of Limnothrissa did not change the correlation coefficient (R² = 0.67). Both equations are significant at p<0.05. The significance of the relationship found with this crude model indicates its usefulness for an understanding of the underlying mechanisms, but the model remains of limited use because of its non-predictive nature.

Limnothrisssa survival rates inshore are probably higher than those of Stolothrissa offshore as they are less vulnerable to adult L. stappersii predation. The Limnothrissa population is therefore less subject to fluctuations. In Zambia, Limnothrissa miodon sometimes equals Stolothrissa in the catch (Coulter, 1970; Pearce, 1985), but this has never occurred in Burundi (see Table 1) and Limnothrissa did not take advantage of the failure of Stolothrissa in 1978/79 (see Fig.5).

It will be understood from the fluctuating abundance of ndagala and mukeke that the application of surplus production models (e.g. by Turner and Herman, 1977 and Herman, 1978) is not very meaningful and only leads to an average potential which offers few possibilities for management. Turner and Herman (op. cit.) tried to apply a global Schaefer model to the combined industrial, artisanal and traditional catches after converting small-scale fishing units to equivalents of purse seiners. However, they noticed that there were great changes in relative efficiency of the different units over the period 1967–76, which made them consider the presence of two distinct fish stocks, one inshore and one offshore to which separate models were applied. An increase in the efficiency of artisanal gears over the last eight years is also mentioned by Pearce (1985). Such increases in efficiency may be partly due to changes in the composition of the available fish stocks. It is well known, at least in Burundi, that the artisanal fishermen catch relatively few predators with their slow gears. In spite of the great abundance of Lates stappersii since 1975, offshore surface gillnetting has never come past the test fishing stage (Chapman et al., 1974). The introduction of faster inshore gears would allow the artisanal fishermen to exploit the inshore L. stappersii stock of 1–2 year old fish which are at present virtually unexploited. A reduction of their numbers would undoubtedly lead to a greater survival of Stolothrissa juveniles which would also be beneficial to the small-scale fishery. However, the absence of juvenile L. stappersii in Kigoma, as well as in other areas of the lake, could mean that the northern lake area constitutes a “nursery” for this species. Such a fishery would therefore need to be closely managed. Roest (1978) estimates the inshore area in the north to be 115,000ha or some 28% of the total area.

Past management measures introduced for the industrial fishery (see section 1.2) have all been meaningless in the biological sense. In as far as such measures were meant to protect the artisanal fishery and to avoid overfishing the offshore stocks, it should be realized that only half of the northern zone is intensively fished and that recruitment levels are largely independent of parent stock size. Economic constraints are likely to limit industrial fishery development before recruitment overfishing occurs. Turner and Herman (1977) discuss the possibility that intensive fishing in Burundi waters may attract fish from neighbouring areas. Because present conditions around the lake are not favourable for rapid development of a lakewide industrial fishery (Herman, 1978) and because of the complex interrelationship between L. stappersii and Stolothrissa at all stages of their life histories, Coulter's (1976) forecast of a fishery dominated permanently by clupeids is not likely to come about in the foreseeable future.

6. REFERENCES

Bhattacharya, C.G., 1967 A simple method of resolution of a distribution into Gaussian components Biometrics, 23:115–35

Chapman, D.W. and P. Van Well, 1978 Growth and mortality of Stolothrissa tanganicae. Trans.Am.Fish.Soc., 107:26–35

Chapman, D.W. and P. Van Well, 1978b Observations on the biology of Luciolates stappersii in Lake Tanganyika (Tanzania). Trans.Am.Fish.Soc., 107:567–73

Chapman, D.W., J. Bayona and C. Ellis, 1974 Analysis of test fishing and limnological sampling in the Tanzanian waters of Lake Tanganyika in October 1974. Lake Tanganyika fishery research and development project. Working paper No. 12. Rome, FAO, URT/71/012/12:27p.

Chené, G., 1976 Etude des problèmes relatifs aux fluctuations piscicoles du lac Tanganyika. Liège, Université de Liège, Faculté des Sciences, 108p.

Coulter, G.W., 1968 Hydrological processes and primary production in Lake Tanganyika. Proc.Conf.Great Lakes Res., 11:609–26

Coulter, G.W., 1970 Population changes within a group of fish species in Lake Tanganyika following their exploitation. J.Fish.Biol., 2:329–53

Coulter, G.W., 1976 The biology of Lates species (Nile perch) in Lake Tanganyika, and the status of the pelagic fishery for Lates species and Luciolates stappersii (Blgr.). J.Fish.Biol., 9:235–59

Coulter, G.W., 1981 Biomass, production and potential yield of the Lake Tanganyika pelagic fish community. Trans.Am.Fish.Soc., 110:325–35

Ellis, C.M.A., 1971 The size at maturity and breeding seasons of sardines in southern Lake Tanganyika. Afr.J.Trop. Hydrobiol.Fish., 1:59–66

Ellis, C.M.A., 1975 Biology and predatory impact of Luciolates stappersii. Unpublished manuscript, Service des Peches, Bujumbura

Ellis, C.M.A., 1978 Biology of Luciolates stappersii in Lake Tanganyika (Burundi). Trans.Am.Fish.Soc., 107(4):557–66

Enderlein, H.O., 1976 Biological sampling survey No. 1 of the traditional and artisanal fisheries. Lake Tanganyika (Burundi). A report prepared for the Fishery Survey and Development Project. Rome, FAO, FI:DP/BDI/508/6:17p.

FAO, 1978 Lake Tanganyika Fishery Research and Development Project, Tanzania. Fishery biology and stock assessment. Technical report 1. Report prepared for the Government of Tanzania by FAO of the UN acting as executing agency for UNDP, based on the work of D.W. Chapman, H. Rufli, P. Van Well and F. Roest. Rome, FAO, FI:DP/URT/71/01:37p.

Hecky, R.E. 1981 et al., Relationship between primary production and fish production in Lake Tanganyika. Trans.Am.Fish. Soc., 110:336–45

Henderson, H.F., 1976 A. Notes on Luciolates based on a study of length frequency diagrams from the ringnet fisheries in Lake Tanganyika. B. On the large size of Limnothrissa in the catches of the ringnet fishery in Tanzania. Lake Tanganyika fishery research and development project. Working paper No. 29. Rome, FAO, FI:DP/URT/71/012/29:10p.

Herman, C., 1978 La peche dans le lac Tanganyika: situation actuelle et perspectives de developpement. CIFA Tech.Pap., 5:16–41

Herman, C., 1984 Les peche sur le lac Tanganyika. Essais de synthese des donnees statistiques du Burundi. MS, FAO Rome, 19p. + tables

Hopkins, D.K. 1982 et al., An alternative to predator-prey ratios in predicting recuitment. Meeresforsch., 29:125–35

Mann, M.J. 1973 et al., A preliminary report on fish biology and stock assessment in Lake Tanganyika (Burundi). A report prepared for the Burundi fishery research on Lake Tanganyika project. Working paper 5. Rome, FAO,FI:DP/BDI/73/020/5:58p.

Marlier, G., 1957 Le Ndagala, poisson pélagique du lac Tanganyika. Bull.Agric.Congo Belge, 48(2):409–22

Matthes, H., 1968 Preliminary investigations into the biology of the Lake Tanganyika Clupeidae. Fish.Res.Bull.Zambia, 4:39–45

Pauly, D., 1984 Some simple methods for the assessment of tropical fish stocks. FAO Fish,Tech.Pap., 234:52p.

Pearce, M.J., 1985 Some effects of Lates spp. on pelagic and demersal fish in Zambian waters of Lake Tanganyika. MS, 18p.

Poll, M., 1953 Exploration hydrobiologique du Lac Tanganika (1946–47). Poissons non Cichlidae. Result.Sci.Explor. Hydrobiol.Lac Tanganyika, 3(5a):251p.

Roest, F.C., 1978 Stolothrissa tanganicae: population dynamics, biomass, evolution and life history in the Burundi waters of Lake Tanganyika. CIFA Tech.Pap., 5:42–62

Turner, J.L. and C. Herman, 1977 Status of the pelagic fisheries of Lake Tanganyika (Burundi) based on catch and effort data. A report prepared for the Burundi fishery research on Lake Tanganyika project. Rome, FAO, FI:DP/BDI/73/020/14:7p.

TABLE 1 Percentage Composition of the Ndagala Catch, Burundi

TABLE 2

Basic Data used in the Application of the Model of Hopkins et al. (1982)
N = numbers, B = relative biomass derived from CPUE

Figure 1: Total recorded fish catch from Burundian waters from 1950 to 1984

Figure 2: Percentage composition of industrial fish catch, Burundi

Figure 3: Catches per unit of effort of the industrial fishery in Burundi from 1956 to 1984

Figure 4: Average monthly abundance of ndagala, sangala and mukeke from industrial fisheries data 1971–1983

Figure 5: CPUE per lunar month of Stolothrissa and Lates stappersii (mukeke)

Figure 6: Correlation between the annual abundances of Stolothrissa and mukeke

Figure 7: Monthly recruitment indexes of Stolothrissa and L. stappersii

Figure 8: Stolothrissa. Apparent months of hatching of major cohorts recruiting at 55 mm

Figure 9: L. stappersii, percentages in “ndagala” and “mukeke” (1971–1980)

Figure 10: Length converted catch curves of Lates stappersii, 1973/74 – 1979/80

Figure 11: Length-converted catch curve for Lates stappersii 1973/74 – 1979/80

Figure 12: Estimation of the proportions of L. stappersii of different lengths in the offshore pelagic area of Lake Tanganyika

Figure 13: Catch per unit effort of mukeke and recruitment strength (/sortie 70+90+110 mm) of Lates stappersii seven lunar months later

Figure 14: Months of peak abundance of L. stappersii in industrial fish catches, Burundi, 1956–1983

Figure 15: Timing of abundance cycle of ndagala in A Burundi

B Zambia

C Zaire