Benthic macroinvertebrates in the Kenyan waters of Lake Victoria were sampled on a monthly basis from March 1994 to March 1995. Ekman grab samples were obtained from 10 stations representing all the major ecological zones of the lake. The aim of the study was to estimate the abundance and diversity of benthos.
The highest mean (± standard error) benthic density was 15113.4 ± 3885.3 m−2 at stations 53, which was significantly different from that of all the other stations (73.5 ± 16.8 to 1625.8 ± 275.4 m−2). Oligochaeta species, which are indicators of organic pollution, dominated the benthic abundance at station 53. Benthos in this station were exposed to both organic and inorganic pollutants as evidenced by highest level of nitrates and trace metals contaminants copper, iron, manganese and aluminum in water at the bottom of water column compared to that at all the other stations. The pollutants find their way to these sites from towns of Webuye, Eldoret, and Kakagema in the catchment through runoff and rivers. The elevated benthic abundance at station 53 could be due to organic and inorganic pollutants.
Shannon index values for the stations in the Winam Gulf were more than triple the values estimated for the open waters implying the gulf had better water quality than the main lake. The gulf mean (± standard error) species heterogeneity values ranged from 0.7 ± 0.2 to 1.1 ± 0.1 whereas in open waters (except station 54) the values were 0.07 to 0.2 ± 0.05. This could be attributed to mixing of water throughout the water column in the gulf ensuring high oxygen levels throughout the water column, which favoured diverse benthos to flourish. In contrast, in deep waters (ca. 40 m) of the main lake stratification occurred most time of year resulting in anoxia in the hypolimnion. Thus, only the dipterans chironomids and chaoborids and oligochates that are adapted to withstand low oxygen levels were recorded, albeit in low densities.
Benthic macroinvertebrates may be defined simply as those aquatic bottom dwelling invertebrates retained by a sieve of aperture mesh size 0.5 to 0.6 mm (APHA, 1985; Wiederholm, 1980). They constitute the main food items of most fish species in Lake Victoria especially the juveniles (Crul et al., 1993), therefore forming an important trophic link. According to Mwebaza-Ndawula (1990), benthos enhance primary production and release of nutrients in the lake, thus contributing in the recycling of energy. Benthos are indicators of the status of water quality in their environment and have been used in many parts of the world to access and monitor water pollution (Premazzi and Chiaudani, 1992).
The endemism and extinction of the biodiversity of Lake Victoria has attracted the attention of both the local as well as the international scientific communities in efforts to conserve it (Muli, 1996). Attention has focused on documenting the changes in the diversity of phytoplankton, fish and the physical and chemical properties of the lake water, ignoring the benthos (Muli and Mavuti, 2001). Due to the fore-mentioned bias, historical documentation of benthic diversity are not comprehensive enough to demonstrate changes and trends of benthic diversity over the last five decades, as is the case for the other trophic groups. Studies undertaken when the lake was pristine (1950 and 60s) reported abundance of only a few species. Macdonald (1956) estimated the abundance of chaoborids and chironomids, whereas Brown (1994) documented the species richness of molluscs on the shores of the Kenya waters of lake based on samples collected by dredging in 1971 to 1973. Contemporary studies of Okedi (1990) and Mbahinzireki (1994) were comprehensive and estimated the abundance of most of major benthic taxonomic groups of the northern portion of the lake in Uganda. Estimation of benthic diversity using diversity indices has not been documented. There is paucity of quantitative data on abundance and diversity of benthos especially for the Kenya waters of the main lake and the enclosed Winam Gulf.
According to Davies and Hart (1981), knowledge on the taxonomy of invertebrates is far from complete and identification of a number of less common benthic groups of insects is still inadequate. Numerous species are yet to be described and it is likely that many have become extinct due to pollution before we even had a description of them (Muli, 1996). While carrying out biodiversity studies in the region, complete counts of invertebrates species are impractical at present due to inadequate knowledge and limited financial resources. Measurements of abundance and diversity using diversity indices provide rough estimates, which are useful and practical in monitoring conservation of fauna over long-term periods. According to Odum (1971), for detection and evaluation of pollution using diversity indices one needs to recognize species but not necessarily to identify them by name.
The aim of this study was to estimate the abundance and diversity of benthos in Kenya waters of Lake Victoria. This information will be useful in water quality and biodiversity monitoring in future. In addition, it is useful in biodiversity conservation, for instance in distinguishing areas to be protected from those that can be used for human exploitation.
Materials and methods
The Kenyan waters of Lake Victoria where this study was done is shown in Figure 1. Table 1 describes the stations where samples were collected. Sampling sites were selected to represent all the major ecological zones/habitats of the lakes. In 1984, C. Foxall (with M. Litterick and S.N. Njuguna, Lake Basin Development Authority, Kisumu, Kenya) in an unpublished report initially described the ecological zones on the basis of level of organic pollution of the lake waters (Figure 1). Subsequently, Mavuti and Litterick (1991) adapted the zonation to describe zooplankton species composition and diversity of the lake. The zonation is adapted for use in this paper. A detailed description of the study area, which includes hydrological, morphometric and biotic characteristics, can be found in Muli (2003). The rainfall is more or less continuous throughout the year and there is very little distinction between the start of the short rain season from the long rain season. There are three rainfall peaks; the main ones correspond to the long rain (March–May) and short rainy seasons (October–December). The third peak is minor and falls in July–August. About 40% of the annual rainfall occurs during long rain (March–May) and 28% during the short rainy season (October–December; Muli, 2003). The remaining amount falls during the other months of the year. Since rainfall is continuous throughout the year and there is no month without rainfall, there is no definite dry season (Muli, 2003). There is very little variation in the mean monthly temperature, air temperature range is 21.9 to 24.3°C (Ogalo, 1981).
Four replicate samples were taken from 10 stations representing water depths from 2 m to 40 m in all the six ecological zones of the lake (Mavuti and Litterick, 1991). Sampling was quantitative using an Ekman grab with a sampling area of 225 cm2. A pilot survey showed that four grab samples was the optimum number to represent the benthos density adequately (Muli and Mavuti, 2001). The four grab samples collected at a site were combined to form one composite sample. Sampling was carried out on a monthly basis from March 1994 to March 1995, except in December 1994 and February 1995. Organisms were separated from sediment using a series of sieves of progressively finer mesh size (20, 2, 1 and 0.5 mm). In the field, specimens were fixed using formalin (10%). Mollusca were first anaesthetized using menthol crystals. In the laboratory, all specimens were preserved in alcohol (70%) for later identification. Organisms were identified to at least family/genus and abundance of each taxon was recorded. Identification keys of Pennak (1978), Mandahl-Barth (1973, 1988) and Brown (1994) were used. Although Pennak (1978) is written specifically for species of Eastern United States, it is sufficient to identify some of the Lake Victoria insect groups to order/family/genus levels. There are no suitable identification keys for all Lake Victoria insect groups and therefore identification to species level for all species is impractical at the moment (Davies and Hart, 1981; Muli, 2003). Problems of taxonomy should not be a deterrent in biodiversity assessment and biomonitoring studies using diversity indices and biotic indices based on invertebrates, as identification to family/genus level is sufficient (Mason, 1991).
The density of macroinvertebrates (numbers m−2) in each sample was calculated according to Clark et al. (1989):
Three aspects of the taxonomic diversity were estimated: taxa richness, the distribution of individuals among taxa (equitability or evenness) and the heterogeneity, a measure that encompasses the first two diversity measures. The indices used were, respectively, number of species (S), Pielou evenness index: J = H′/H′ max. and Shannon heterogeneity index: H′ = − Σpiln pi (Sheehan, 1984). The species abundance, richness, heterogeneity and evenness for all the collected data were calculated per station per month. Muli and Mavuti (2001) showed that there were no seasonal changes in the composition or benthic density. Therefore, temporal variation of benthos was not considered in this paper.
A one-way Analysis of variance (ANOVA) was used to test whether spatial variations in macroinvertebrates abundance and species diversity measured in various stations were significantly different. The ANOVA tests were based on the monthly values calculated from the composite sample made from four replicate samples for the various stations. All ANOVA tests mentioned above were performed on Log10(X+1)-transformed data. The data was transformed to normalize it. Whenever ANOVA resulted in significant F-values (P < 0.05), a posteriori comparison of means using Student-Newman-Keuls test (SNK) was conducted to determine the location of significant differences. Significance level was 5% for all statistical tests carried out (Zar, 1974).
Figure 2 depicts the mean density of benthos at the various stations. The estimated mean total benthos density was highly variable between and within the various stations. The range between the smallest and the largest density was 15039.9 m−2. Generally the largest quantities of benthos were obtained in stations 53 and 54 in ecological zone V (Figure 2). In contrast, the lowest density occurred at deep-water stations 32 and 99 in zone VI and inshore station 17 in zone I. The mean (±SE) total benthic density in zone V ranged from 4350.5 ± 2046.7 to 15113.4 ± 3885.3 m−2. At the other stations the density ranged from 73.5± 16.8 to 1625.8± 275.4 m−2. The one-way ANOVA test showed the benthic density in the various stations was significantly different (F 0.05 (1) 9, 84 = 12.769, P < 0.05). Subsequent analysis using the SNK test based on results of one-way ANOVA, showed the benthic density in station 53 was significantly different from that of all the other stations.
The relative abundance of the major benthos taxonomic groups is depicted in Figure 3. There was a clear trend in the distribution of the taxonomic groups. The benthic abundance in the main lake and in Winam gulf was composed mainly of Oligochaeta and Mollusca. Insecta groups contributed a very small portion. There was a distinction in species composition between river influenced inshore stations, which are organically polluted, and those that are unpolluted. In polluted stations (zone I), oligochaetes, chironomids and gastropods dominated, whilst in unpolluted stations (zones II and V) Ephemeroptera and Bivalvia were the most abundant groups. In zone I stations, the commons gastropods were Bellamya unicolor (Olivier) and Melanoides tuberculata (Müller) while Branchiura sowerbyii Beddard and Alma emini (Michaelsen) were the dominant oligochaetes. On the other hand, in zones II and V, Povilla adusta Navás was the most abundant ephemeropteran species followed by Caenis spp., while Sphaerium nyanzae Smith was the dominant bivalve. The only exception among the unpolluted inshore stations was station 53, whose benthic abundance were dominated by Oligochaeta. In both offshore and inshore stations in the gulf (zones III and IV), which are not under direct river influence and are unpolluted, mollusc species were the only dominant benthos. Sphaerium nyanzae Smith, Bellamya unicolor and Melanoides tuberculata were the most abundant molluscs in these two habitats. In the deep offshore waters of the main lake (zone VI), an unpolluted zone, Oligochaeta species, Branchiura sowerbyii and Tubifex spp. were the most abundant benthos. Other groups that formed a substantial population in zone VI were Sphaerium spp. (Bivalvia) and Chaoborus spp. (Diptera).
The results of the three estimates of benthic macroinvertebrates diversity are depicted in Figure 4. The cumulative species richness recorded over whole sampling period generally followed a similar pattern as the species richness index. The species richness values estimated for the gulf stations and inshore main lake were more than double that estimated for the offshore deepwater stations in the main lake. The highest species richness values were recorded in stations 2 and 54 whilst the lowest were recorded in stations 32 and 99. The mean (± SE) species richness (S) recorded was generally low in all the stations sampled. It ranged from 1.3 ± 0.3 to 7.6 ± 0.8. The one-way ANOVA test showed the species richness in stations was very highly significantly different (F0.05(1) 9, 84 = 17.083, P < 0.001). Subsequently, the SNK test could not distinguish which were different.
Species heterogeneity followed similar pattern as species richness. Except for station 54, the estimated Shannon index values for the stations in the gulf were more than triple the values estimated for the open waters. The gulf mean (± SE) species heterogeneity values ranged from 0.7 ± 0.2 to 1.1 ± 0.1 whereas in open waters (except station 54) the values were 0.07 to 0.2 ± 0.05. One-way ANOVA test showed the species heterogeneity in the various stations was very highly significantly different (F0.05(1) 9, 84 = 9.281, P < 0.001). Subsequent analysis using the SNK test, showed that species heterogeneity in stations in open waters (32, 53 and 99) was not different. Similarly heterogeneity in the gulf stations was not significantly different. However, the heterogeneity in the gulf stations was significantly different from that at the open water stations except for station 54.
Species equitability generally followed similar pattern as species richness and heterogeneity. The estimated Evenness index values for the stations in the gulf were higher than the values estimated for the open waters. The gulf mean (± SE) species equitability values ranged from 1.1 ± 0.2 to 1.4 ± 0.1 whereas in open waters the values were 0.3 ± 0.1 to 0.8 ± 0.1.
One-way ANOVA test showed the species equitability in the various stations was very highly significantly different (F0.05(1) 9, 84 = 4.842, P < 0.001). Subsequent analysis using the SNK test could not distinguish which stations were significantly different from the others.
Figure 5 shows relative species richness based on number of species recorded over the whole sampling period. The relative species richness showed similar pattern as the relative abundance. There was a distinction in species richness between river influenced inshore stations, which are organically polluted, and those that are unpolluted. In polluted stations (zone I), Mollusca and Diptera dominated in species richness, whilst in unpolluted stations (zones II and V) in addition, the Ephemeroptera formed a substantial portion of the species. The only exception among the unpolluted inshore stations was station 53, whose benthic species richness was dominated by molluscs and dipterans. In both offshore and inshore stations in the gulf (zones III and IV), which not under direct river influence and are unpolluted, Mollusca and Diptera were the most speciose groups. In deep offshore waters of the main lake, (in zone VI, stations 32 and 99), where it is unpolluted, the community was dominated by Oligochaeta and Diptera with few other species present.
There are no documented data on abundance and diversity of benthos for the Kenya portion of Lake Victoria. Likewise there were no diversity values from other parts of the lake to compare with those for the Kenya waters. The mean density values reported for the Kenya waters were comparable to those reported for the Uganda portion of the lake (Mbahinzireki, 1994). Burgis et al. (1988) reported dominance of benthic fauna by Oligochaeta and Mollusca based on data collected in 1984. High densities and dominance of Alma emini (Michaelsen) and Pila ovata nyanzae (Smith) as reported by Burgis et al. (1988) were not encountered in the Kenyan stations. Molluscs dominated in the gulf while Tubificidae dominated in the main lake. However, in the main lake at station 54, molluscs were the most abundant group. The oligochaete species Alma emini, Branchiura sowerbyii and Tubifex spp. that dominated are known to be able to withstand very low oxygen concentration (Rzóska, 1976; Premazzi and Chiaudani, 1992). Tubificidae species such as Branchiura sowerbyii and Tubifex spp. are known worldwide to be indicators of high organic pollution in water bodies. On the other hand, molluscs of the family Ancylidae and insects of the orders Ephemeroptera and Plecoptera are known worldwide to indicate that a water body is not organically polluted. The presence of only these tubificids species in very high densities at a site in a water body in the absence of ancylids, ephemeropterans and plecopterans, is an indicator the site might be organically polluted. The Biotic indices such the Belgian Biotic Index (BBI), Biological Monitoring Working Party-score (BMWP) and Trent Biotic index (TBI) use presence and absence of these ‘indicator species’ at sampling site in the assessment of organic pollution (Mason, 1991). Organic pollution is characterized by very low oxygen concentration, high biological oxygen demand, high ammonia and nitrate concentrations. Since the lake has abundant and diverse benthos, dominated by non-insects it can be inferred that it is eutrophic (Premazzi and Chiaundani, 1992).
Shannon diversity index values were generally low indicating a stressed environment or physically controlled by a few individuals (Odum, 1971). Considering the dominance of non-insect types, the high density of tubicifidae and low diversity indices, it can be inferred that the lake is eutrophic. The benthic fauna distribution and abundance is consistent with recent past and current findings on the increasing eutrophication (Hecky, 1993) and deoxygenation in Lake Victoria (Hecky et al., 1994). Consistently low oxygen concentrations (< 4 mg l−1) in the hypolimnion covering large areas in stations 4, 32 and 99 have been reported (Hecky et al., 1994) throughout since the beginning of last decade. Elsewhere, low diversities have been reported in areas with stress from rigorous physical environment and pollution (Boesch, 1972) or in areas with extremely high temperature, hydrostatic pressure, salinities and low oxygen concentration (Paul, 1975). The Winam Gulf had generally higher species diversity than the main lake, implying the gulf had better water quality than the main lake. This could be attributed to mixing of water throughout the in the gulf water column thus increasing the oxygen concentration at the bottom. The gulf is sufficiently isolated from the main lake and behaves limnologically independently. In the main lake, in deep offshore waters (ca. 40 m) especially in stations 32 and 99, non-mixing of water throughout the water column has been reported before (Hecky et al., 1994). During the study period, complete mixing of water occurred only once during the month of September. As a result, the dissolved oxygen concentration at the lake bottom during this study was 0.3 to 2.29 and 2.6 to 3.1 mg l−1 at stations 32 and 99, respectively (Gichuki, 1995). Thus in deep water stations, low oxygen concentration could be one of the major influences on the structure and low diversity of the benthic community.
Stations 2 and 54 had the highest diversity and abundance of ephemeropteran species. This could be interpreted to mean these two stations had the best water quality or the most diverse habitat compared to all the other stations. This could further imply the Rivers Sondu-Miriu and Yala, which drain into these stations, had better quality compared to the Rivers Kasat and Nzoia. The elevated abundance of tubificids and absence of ancylids at the stations near the mouths of Rivers Kasat and Nzoia is probably an indicator of contaminants from paper mills and agro-based industries. The towns of Kakamega, Webuye and Eldoret add organic pollution to the lake via River Nzoia as their sewage treatment plants have not been working efficiently for the more than a decade (Muli, 1996). Similar situations prevail in the towns of Kisumu and Homa Bay. Untreated organic effluents from Kisumu town find their way to the lake through the River Kisat. In Homa Bay town, raw sewage is discharged directly to the lake from sewage lagoons, which are less than fifty metres from the lake. Organic pollution is known worldwide to change benthic species composition, abundance and diversity (Mason, 1991). The species diversity decreases substantially but the abundance of one or two species that are tolerant to the effluent increases tremendously. This seems to be the case for stations near the mouths of Rivers Kasat and Nzoia.
Species diversity at station 53 near Nzoia river mouth was much less than at station 54, both in zone V while density of oligochaetes was greatest at station 53 for all stations. The difference could be due to the high organic and inorganic contaminants, which benthos are exposed to at station 53 compared to station 54 as described below. River Yala is not exposed to large amounts of organic effluents, as the major towns in its catchment have no agro-chemical industries to discharge organic effluents into the river. In addition, the Yala swamp complex (Lake Kanyaboli, Lake Namboyo, Lake Sare and Yala swamp itself) ensures clean water enters Lake Victoria through the Yala River mouth. The river traverses through the extensive Yala swamp (> 17,500 ha) in diverse deltoid water channels whose mouths open into Lake Sare (Abila, 1998). The Yala swamp filters the turbid river water before releasing it into several minor lakes within the swamp that include Lake Namboyo and Lake Sare. These lakes are natural ‘sediment tanks’ and Lake Sare is the final one before the water is released into Lake Victoria near station 54. On the contrast, the Nzoia River mouth is an estuary and since it lacks an extensive swamp, most of the water which enters the lake through its mouth is not filtered. Thus, the water that enters Lake Victoria through River Yala at station 54 is less turbid compared to what enters the lake through River Nzoia at station 53.
Gichuki (1995) and Mwamburi and Oloo (1997) estimated values of some of the organic and inorganic contaminants in the lake, while Mwamburi (2003) estimated values of inorganic contaminants in the afferent rivers of the lake. Their values were based on analysis of water and sediment samples collected concurrently with the benthic samples analyzed for this study. The levels of the estimated contaminants gave an insight on principal environmental variables determining the benthic distribution pattern in the lake. According to Gichuki (1995), water variables temperature, pH, conductivity, alkalinity, and hardness at the lake-bottom waters at the various stations were not substantially different, except for turbidity and nitrates. Nitrate-N recorded at station 53 was 189.5 μg N l−1, which was nine times compared to average value 22.2 μ g N l−1 (range, 7.1–47.5) recorded at all the other stations sampled. On the other hand, turbidity at station 53 was 385.6 NTU, which was five times compared to average value 78.6 NTU (range, 10.9–176.8) recorded at all the other stations sampled.
Mwamburi and Oloo (1997) reported that total concentrations of Cu, Fe, Mn and Al in the lake-bottom waters at station 53 were substantially elevated compared to those recorded at all the other stations. The concentration of various inorganic contaminants at station 53 were: i) Cu, 0.03 mg l−1, which was three times compared to values 0.002–0.006 mg l−1 recorded at all the other stations; ii) Fe, 9.64 mg l−1, which was two times compared to values of 0.4 to 2.4 mg l−1 recorded at all the other stations; iii) Mn, 0.39 mg l−1, which was two times compared to values 0.06-0.16 mg l−1 recorded at all the other stations; and iv) Al, 6.9 mg l−1, which was one and half times compared to values 0.1–2.4 mg l−1 recorded at all the other stations. Mwamburi and Oloo (1997) attributed the high Cu concentration both in lake-bottom water and sediment at station 54 to modification of natural Cu along the Nzoia River by Pan paper mill factory effluents discharged upstream. Stations on river mouths and shallow areas (2–3 m) in the lake generally showed higher bottom sediment trace metal concentration than offshore deeper areas. Relatively higher Pb and Zn values were estimated for lake-bottom sediments near urban sites (zone I station 10; zone II station 9) compared to other areas in the lake. At station 10, Pb concentration was 122.7 μg g−1 while at all the other stations it ranged from 13.6 to 95.6 μg g−1; Zn was 136.4 μ g g−1 while at the other stations it ranged from 31.8 to 113.6 μ g g−1. Mwamburi and Oloo (1997) attributed high Pb and Zn contamination to effluents from small-scale industries in Kisumu and Homa Bay towns which find their way to the lake through municipal wastes and surface water runoffs. Leaded fuels, marine paints and associated port activities also contribute to metal contamination in the lake. Mwamburi (2003) estimated concentrations of Fe, Cu, Mn, Pb, Zn, Cd, Cr and Al in river-bottom sediments of Rivers Kasat, Kibos, Nyando, Sondu, Awach, Yala and Nzoia. According to Mwamburi (2003), River Kasat sediments contained the highest mean levels of Al (7.77 × 103 μg g−1), Cr (125 μg g−1), Mn (2.10 × 103 μg g−1) and Zn (130 μg g−1) than any other river system. The river was considered polluted with respect to sediment Mn, Cr and Zn content, which were comparatively higher than unpolluted sediments and geochemical background. The high inorganic contaminants in Rivers Nzoia and Kasat find their way into lake near stations 10 and 53. The elevated benthic abundance and low species diversity at station 53 could be partly due to the high level of organic and inorganic contaminants at the station. The high level of the trace metals could be toxic to ephemeropteran, odonatans and plecopterans, which are good indicators of good water quality. Hence in their dearth in stations 10 and 53. Tubificids would then flourish due to less competition and the availability of high levels of nutrients and trace metal elements (Mason, 1991; Premazzi and Chiaundani, 1992). The association between oligochaetes and dipterans Chaoborus sp. and chironomid spp. and high levels of trace metal contaminants has been reported in northern western Lake Victoria, Uganda by Mothersill et al. (1980). They used factor analysis which showed common occurrence of unindentified oligochaetes with Fe, Co and Zn-rich sediments in the areas of organically poor sediments and insects Chaoborus sp. and chironomid spp. with sediments with relatively higher amounts of Zn and Cu.
Although multivariate analysis was beyond the scope of this paper, it should be done to determine the principal environmental variables that determine the benthic community distribution patterns.
This work was carried under the auspices of the Kenya-Belgium joint project on freshwater ecology. I acknowledge the directors of the project Prof. J. Symoens and E. Okemwa for granting me the opportunity to carry out this study. Zablon Awuonda and John Onyango assisted in sampling and sample processing. The map was drawn by Sarah Khanani and John Gichuki. I am grateful to Prof. Niels De Pauw, Dr. Henry Lung'ayia, Dr. Albert Getabu and Messrs Mutune Masai and James Ogari for their advice and support during the study period. Finally, I would like to thank the station coordinator, Baringo Field Station, Mr. Casianas Olilo for granting me time off to write this paper.