The Nyangara wetland, also known as Laguna, in the fluvial plain of Rusizi River at the northern end of Lake Tanganyika, harbours a fish fauna which offers food and income for 500 people directly or indirectly associated with fisheries. Fishing focuses mainly on two cichlids (Oreochromis niloticus and Astatotilapia burtoni) and the marbled lungfish (Protopterus aethiopicus).
Studies of population structure were based on length-frequency distributions of fish in catch samples. The size at first maturation was estimated using total length and state of gonad maturity estimated macroscopically. Minimum and maximum sizes in catches ranged from 10 to 100 mm total length in A. burtoni and from 20 to 180 mm in O. niloticus. Sizes at first maturation were 65 mm total length for female and 70 mm for male in A. burtoni and 130 mm in O. niloticus. Immature fish comprised 95 % of the catch. The majority of harvesting thus took place before the onset of maturity. In both species, the dominance of very small fish in the population structure, the reduced maximal size, and the early maturation of fish, indicated high fishing mortality.
Although further studies are needed to properly understand the effects of intensive fishing on fish populations in the Nyangara wetland, the present study suggests that the fishing community would be well-advised to follow the precautionary approach and pay more attention to the numbers and types of fishing gear, and to appropriate mesh sizes to ensure successful spawning of fish. To reduce fishing pressure, alternative sources of protein should be sought, such as pond aquaculture, agriculture and animal husbandry for example.
Overexploitation of fish stocks is notoriously common in marine fisheries worldwide (Jackson et al., 2001; Hilborn et al., 2003). In the inland waters, much less information exists on the status of fish stocks (Allan et al., 2005), but overfishing has been suggested in certain cases, although it has seldom been properly documented. In the small scale inland fisheries of Africa, the present fishing pressures seem to be mostly sustainable (Jul-Larsen et al., 2003); while in southeast Asia high exploitation rates have led to very low annual yields per fisherman (Duncan, 1999).
Overfishing was earlier considered as a major threat to the Nile perch fishery in Lake Victoria (Pitcher and Bundy, 1995). Later reports suggested the situation was less alarming (Mkumbo et al., 2007; Cowx et al., 2003), and a recent thorough review did not find any evidence of overfishing (Kolding et al., 2008a). However, in Lake Tanganyika, symptoms of local overfishing have been observed, particularly in the north (Mölsä et al., 2002; Sarvala et al., 2002; Mulimbwa, 2006).
Fishing is a crucial source of livelihood in the rural communities of tropical developing countries, and therefore knowledge of the effects of the fishery is of utmost importance. Heavy fishing mortality may not only reduce population size, but also change the population structure, improve or impair growth, and depending on the selectivity of the fishery, may lead to earlier maturation (e.g. Miller, 1956; Healey, 1980; Henderson et al., 1983; Amundsen, 1988; Sarvala et al., 1994; Jorgensen et al., 2007).
The Nyangara wetland, known as ‘Laguna’, situated in the fluvial plain of Rusizi River at the north end of Lake Tanganyika (Figure 1), harbours a fish fauna, that directly or indirectly offers food and income for approximately 500 people. The fishing community consists of 46 professional and 62 occasional fishers, and other people are employed in fish processing or restaurants. Fishing focuses mainly on two cichlids (Astatotilapia burtoni (Günther, 1894) and Oreochromis niloticus (Linnaeus, 1758)) and the marbled lungfish (Protopterus aethiopicus (Heckel, 1851)), and fishing pressure is likely very high. Until present, however, no study had been conducted regarding sustainability at the existing level of exploitation.
In the present paper, the population structure and maturation of A. burtoni and O. niloticus in the Nyangara wetland are studied in relation to current fishing pressure. Furthermore, the implications for regional fishery management are discussed.
Study area, materials and methods
Nyangara wetland (29° 09′ 55″E-29° 12′ 10″E and 3° 20′ 14″ S −3° 20′ 42″S, mean depth 1.5 m; circumference 14 km) was previously a part of Lake Tanganyika and has a bottom covered by mud and lush vegetation dominated by Typha domingensis, Phraguila mauritianus, Polygonum pulcherum, Nymphaea nonchatri, Cyperus articulatus, Cyperus papyrus, Cyperus digitatum, Digitaria sp, Leersia sp, Panicum sp, Ceratophyllum demersum and Ipomoea fragrans. The Nyangara water is always green, showing an abundance of phytoplankton; the Chlorophyceae accounting for 43%; Cyanophyceae 31%; Euglenophyceae 24% and Pyrrhophyceae 2% of the total phytoplankton species numbers (Kamalebo, Centre de Recherche en Hydrobiologie, Uvira, D.R. Congo, unpubl.).
The water temperature varies between 24.7 and 28.5°C, dissolved oxygen between 82.2 and 137.7%, conductivity between 457 and 829 microsiemens cm−1, pH between 5.9 and 6.5 and the transparency between 15 and 35 cm (Kamalebo, Centre de Recherche en Hydrobiologie, Uvira, D.R. Congo, unpubl.). The abundant macroscopic vegetation, green water colour and low transparency all indicate the high productivity and hypereutrophic condition of the system.
According to an official survey, the total number of fishing gear recorded in the region around Nyangara was 42 seine nets, 4000 trap nets, 110 gill nets, and 1400 hooks, and in addition there were divers catching fish by hand (Environmental Service, pers. comm.). Actually, however, only 20 seine nets, 352 trap nets and 8 gill nets operated during the investigation period (2003–2004).
Samples were collected at the landing site twice a month from February to December 2003. The sampling started at 7 h 00 with fish from gill nets and trap nets, and the samples from seine nets, hooks and divers were obtained around 9 h 00–10 h 00. One kilogram of fresh fish was sampled randomly from each of three seine nets with mesh sizes of 3, 5 and 10 mm and three gill nets with 15, 20 and 30 mm mesh sizes, along with all fish from trap nets, hooks and divers (who catch fish by hand). The fishing by seine nets is done in the daytime and the gill nets are set in water in the evening and removed in the early morning. Selectivities differ between these gears: seine netting is an active and relatively unselective method also catching stationary fish, while gill-netting is a passive and selective method wholly depending on the movements of the fish. The most selective method was hand-fishing by divers who targeted only the larger fish, thus not taking A. burtoni at all.
In the laboratory, fish were grouped by species and total lengths were measured. Gonads were removed and stages of gonad maturation were estimated by visual inspection using the key of De Kimpe (Kaningini, 1995). The length at first maturity was taken to be the smallest total length at which 50% of the fish had ripe gonads belonging to the 3/5 stage. The percentage of immature fish was determined by the number of individuals that were smaller than the estimated size of maturity.
Size frequency distributions were collated monthly, and the modal length was obtained with the graphical method of Bhattacharya (1967). The effects of fishing were assessed from the structure of the fish stocks, the maximum individual size and the observed length at first maturity.
A total of 12423 specimens of A. burtoni were measured and grouped into 10 mm length classes, and 291 females, 569 males and 266 immature individuals were dissected. A total of 8510 of O. niloticus were measured and grouped into 10 mm length classes, and 466 females, 704 males and 358 immature individuals were dissected.
Most of the catch derived from the seine nets and gill nets, and the fish sizes caught by the gill nets were also effectively sampled by the seine nets. Catches from other gear comprised a minor proportion of the total catch and did not much affect the size distributions, particularly in A. burtoni. Consequently, size frequency data from each type of fishing gear were pooled. In O. niloticus, however, the larger fish caught by the divers were visible in the size distributions, making interpretations more difficult. It was impossible to determine growth parameters, because clear modal progression extending to large fish was not observed in the monthly samples (Figure 2, Figure 5; see below).
A. burtoni started recruiting to the catch at 10 mm, and appeared to be fully recruited at 30 mm (Figure 2). The length frequency distribution showed usually one mode around 30 mm and the numbers of larger fish decreased with size (Figure 2). There were weak modal progressions on two occasions, from April to July and from August to October, but these did not extend to the mature size groups and would imply very low growth rates (10 mm or less per month).
Sizes at first maturity were 65 and 70 mm for females and males, respectively (Figure 3) and below a length of 40 mm, sex was indeterminable (Figure 3). Accordingly, 95% of the fish caught were immature (Figure 4). The percentage of mature fish increased with size of fish (Figure 3). A sexual dimorphism in colour and size was observed. The males were black and larger than females which were whitish. The ratio male/female in the catch was 1.95:1.
O. niloticus started recruiting at 20 mm total length, appeared to be fully recruited at 30 mm, and the maximum length was 180 mm (Figure 5). Small fish with the modal length of 30 mm were abundant throughout the year (except in the August samples). There was a weakly defined modal progression from February until June (possibly extending to August, although with practically no growth from June to August), implying a growth rate of roughly 20 mm per month (slightly more in the beginning and slowing down at larger sizes). In the latter half of the study year, no modal progression was visible, but a broad peak of larger fish of 80–120 mm total length was present in August-November (Figure 5).
Size at first maturity was 130 mm total length for both males and females, and for fish under 90 mm, sex was indeterminable (Figure 6). The percentage of mature fish increased with size of fish (Figure 6). Even in this species, 95% of fish harvest consisted of immature fish (Figure 7). The ratio of males to females was 1.5:1.
Our paper is the first documentation of the locally important fishery and its effects in this wetland. Although our data base was relatively small, the reported population structures were based on representative catch sampling covering all seasons. Our study area is in the tropical zone where the seasonality of abiotic factors such as air and water temperature is restricted and allows extended spawning periods. Accordingly, there were no clear changes in population structure during this period, making it impossible to derive more detailed life-history information from our data. Variations in the catch per effort could not be used to help in cohort separation or estimating the population strength, because of strong changes in the catchability of fish. In the rainy season with higher water levels, fish were dispersed among the vegetation where fishing was impossible, while in the dry season with lower water levels, fish were forced to leave the vegetation for open water and the catches increased (Head of the local fishermen, pers. comm.). The analysis of length frequency distributions only showed the presence of small fish throughout the year, suggesting that reproduction takes place year-round. This is consistent with what is known of the reproductive activities of the species in captivity (A. burtoni is widely reared by aquarium hobbyists and O. niloticus is an important aquaculture species). In both species, besides the continuous reproduction, the apparently very high fishing pressure was probably another reason preventing us from observing clear cohort succession. A cohort which appeared in one month was immediately harvested and already disappeared next month. Moreover, the fact that the modal progression could not be followed into the mature size groups was most likely due to a higher fishing mortality of the larger fish. In this situation, growth rates derived from the size frequency distributions were likely to be biased towards unrealistically low rates.
Although recruitment of small fish was practically continuous throughout the year, there were weak signs of potential seasonality in reproduction. In A. burtoni, the two pulses of small fish in April and August might indicate peaks of increased reproductive activity, but they are difficult to associate with environmental factors because they represent completely different phases of the seasonal cycle, April being towards the end of the wet season and August in the latter half of the dry season. Fish reproductive activities in the region are known to be generally more dynamic in the first half of the year (Plisnier et al., 1988; Mulimbwa and Shirakihara, 1994). In O. niloticus the right-hand side of the size distributions was confounded because most of the bigger fish were diver-caught, and their proportion of the total catch was seasonally variable. Moreover, the complete lack of small fish in August has to be a sampling artifact; most likely the August samples derived solely from the divers' catch.
A drawback in our analyses is that the fishery information is temporally limited. The extent of the present fishery is known, but there are no long-term data to document possible changes in the fishing pressures.
The Nyangara wetland is connected to a canal derived from River Rusizi, which could be thought to lead to an exchange with other fish populations. Such mixing between different populations would of course affect the possibilities to manage the fish stocks. However, we maintain that the fish populations in the wetland are largely self-sustaining, because of the long distance (5 km) from the river (and there is no connection with Lake Tanganyika). A. burtoni is not found in the main Rusizi River, although it is present in the Kiliba and Mwaba swamps located in the Rusizi plain and at the mouth of the Rusizi River (Nshombo, 2008, unpubl.). The seasonal development of the size frequency distributions also did not indicate any clearly flood-associated changes in A. burtoni. Moreover, the effects of the intensive fishing remain even if there might be some immigration from the surrounding areas – very few fish lived long enough to attain sexual maturity.
O. niloticus is originally an African cichlid which has been spread by humans all over the warm belt. It has great economic significance and is widely used in aquaculture. Exploitation levels of the natural or introduced populations differ among countries, and the differing fishing practices lead to different maximal lengths and sizes of maturation (Table 1). Greater lengths are recorded where fish stocks are under- or moderately exploited, and earlier maturation occurs when the fish stocks are subject to a heavy fishing pressure, as demonstrated in Lake Victoria (Ojuok et al., 2007; Njiru et al., 2008). Besides heavy fishing, abiotic conditions may, however, also affect the growth rate and size at maturity (Kolding et al., 2008b). Oxygen stress in the extremely productive environment might have contributed to the reduced maximum size and size at maturity of O. niloticus in Nyangara wetland and possibly also in other small and shallow waterbodies (Kolding et al., 2008b). In our case, however, the high fishing pressure was likely the main factor, as O. niloticus catches consisted mainly of immature fish. The tremendous flexibility of this species in growth and size at maturity is certainly one factor enabling its success in a broad range of environments, including the extreme conditions of the Nyangara wetland.
A. burtoni is an endemic species to lagoons, swamps and rivers in Lake Tanganyika area and elsewhere in East Africa (Coulter, 1991) and its biology is poorly known. Indicative for high levels of exploitation in the Nyangara wetland, length at recruitment to the catch was far below the size of sexual maturity, and the maximum length of fish in the population was reduced compared to the maximum lengths reported historically in Lake Tanganyika and in its tributary, the Lobozi river (Table 2).
More detailed fish biology studies are needed to understand the effect of intensive fishing on fish populations in the Nyangara wetland. For instance, estimates of gonad maturation should be carried out by making use of histological research techniques to accurately follow the different stages of spermatogenesis and ovogenesis.
The present fishery in Nyangara wetland seems to have a strong impact on population structures of both fish species studied here (O. niloticus and A. burtoni). In both species, most of the catch consists of small, immature fish. Their length frequency distibutions are skewed towards young and small sizes; sexual maturation takes place at a small size. Data from a longer period of time is needed however, to determine whether the catch structure, size distributions and maturation sizes are changing with time. Fishing pressure is indeed very high, as evidenced by the extreme dominance of juvenile fish in the catch, but high juvenile dominance alone does not make a fishery unsustainable. Even in certain boreal lakes such high juvenile dominance in a pelagic fishery has been sustainable in the long-term (Sarvala et al., 1998).
Because of the apparently very high potential productivity of the Nyangara wetland, the fishing tolerance of the fish populations can be very high as well, as in many other small-scale fisheries in Africa (Jul-Larsen et al., 2003). Long-term monitoring of the Nyangara fish populations is thus important. In the meanwhile, applying a precautionary approach to prevent further deterioration of fish stocks, we must pay more attention to the numbers and types of fishing gears and to appropriate mesh sizes that are used in the spawning areas. People should also be encouraged to diversify their diets. In order to facilitate this, the community should develop programs promoting aquaculture (in ponds), agriculture and animal husbandry (cattle, poultry and pigs) in the region.
We wish to express sincere thanks to Dr. Hannu Mölsä of Kuopio University (Finland), Prof. Eric Odada of Nairobi University (Kenya) and Dr. Steffen Pauls of Chicago Field Museum (United States) for their scientific remarks and revision of this manuscript. For practical help in the laboratory of Hydrobiological Research Centre (C.R.H./Uvira, D. R. Congo), we thank the General Director Dr. Nshombo Muderhwa, and the technicians Mr. Bahane Byeragi, Kisago Saleh and Watuna Igudji. The first author was financially supported by grants from the MacArthur Foundation, through Prof. John Bates of Chicago Field Museum, the Head of ‘Programme Biodiversité des Ecosystèmes Aquatiques et Terrestre du Rift Albertin (P BEATRA)’ and the International Society of Limnology (SIL), which are gratefully acknowledged.