Introduction

In many African countries, where the economy is strongly dependent upon biological resources, the exploitation of natural resources and their conservation are ambiguous (Kamdem Toham & Teugels, 1999; Economic Commission for Africa, 2001). In fact, conservation of resources and development of countries are mutually dependent. Indeed, unless living resources and in particular aquatic ones are used and managed rationally, natural production faces the risk of collapse (Yapi-Gnaoré et al., 2000). Despite their economic, social and ecological importance, Africa’s freshwater systems are being degraded at an alarming rate (Thieme et al., 2005). The multitude of threats (overfishing, land and forest degradation, pollution from burgeoning populations and industries, introduction of exotic species, dams and water withdrawals) suggests that conservation measures are long overdue.

To ensure a sound management and sustainable exploitation of freshwater biodiversity, it is vital to develop a deeper understanding of the factors and processes that determine aquatic diversity at different spatial scales (Ibanez et al., 2007). The examination of variation in fish assemblage composition according to environmental gradients may help to clarify the role of biotic and abiotic factors for community organization, a major issue in aquatic ecology. Thus, scientific interest is increasingly focusing on the assessment of the relationships between species and the environment (Angermeier & Davideanu, 2004; Tejerina-Garro et al., 2005), yet in this field of research, few studies have been conducted on the tropics’ aquatic systems (Lévêque, 2006).

In Côte d’Ivoire (West Africa), recent years have seen an emphasis on research to identify correlations between structural changes in the fish assemblages and environmental variables (Da Costa et al., 2000; Koné et al., 2003; Kouamélan et al., 2003; Yao et al., 2005; Kouadio et al., 2006; Konan et al., 2006). However, no investigation has been undertaken in the Sassandra River basin to detect patterns in fish communities and their possible environmental basis. Previous surveys conducted on this basin dealt mainly with taxonomic and biological aspects (Teugels et al., 1988; Kouamé et al., 2006), and stock assessment and management (Traoré, 1996).

As part of a project entitled ‘‘Characterization, utilization and conservation of freshwater fish biodiversity in Côte d’Ivoire’’, the present article aims to describe distribution patterns in the Lower Sassandra ichthyofauna and to identify correlations between structural changes in the assemblages and environmental variables.

Material and methods

The Sassandra River basin rises in Bela region in Guinea and enters Atlantic Ocean in Côte d’Ivoire, close to the city of the same name (Fig. 1). It is 840 km long and encompasses an area catchment of 75,000 km2, with a mean annual discharge estimated at 513 m3 s−1 (Traoré, 1996; Lemoalle, 1999). The present study was restricted to the lower course of the basin because an ongoing armed conflict has prevented field work in northern Côte d’Ivoire since September 2002. The studied area includes four sectors: (1) the man-made Lake Buyo (drainage area: 920 km2) originated from the building of a hydroelectric dam on the main course of the Sassandra River in 1981 (Traoré, 1989). It is a deep and open water. Aquatic vegetation is abundant and consists mostly of Eichhornia crassipes and Echinochloa pyramidalis; (2) the main channel with a mean slope estimated at 0.85 m km−1, is bordered by gallery forests. The hydrological regime depends on both discharge from the dam and rainfall; (3) the Lobo tributary is shaded by overhanging canopy of riparian vegetation and (4) the Davo tributary receives a discharge of waste of palm oil from a factory. As a result, the water is characterized by high levels of organic pollution. The climate of the studied areas is typical of equatorial rainforest, comprising two dry (July–September and December–March) and two wet (September–November and April–July) seasons (Girard et al., 1971). Surveys occurred at 22 sampling sites of which 2 (Sa1 and Sa2) are located on the Lake Buyo, 7 (Sa3–Sa9) on the main channel, 6 (Sa10–Sa15) on the Lobo tributary and 7 (Sa16–Sa22) on the Davo tributary (Fig. 1). Each site was sampled at three-month intervals from October 2004 to September 2005. Fishes were collected with a fleet of 17 weighted monofilalement gill nets (bar mesh sizes 8, 10, 12, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 80 and 90 mm), each measuring 30 m long by 2.5 m deep. Nets were set overnight (17–7 H) and during the following day (7–12 H). All fishes collected were identified following Paugy et al. (2003a, b). Each specimen was measured to the nearest mm and weighted to nearest 0.1 g.

Fig. 1
figure 1

Map of the Lower Sassandra River Basin with indication of the sampling sites (Sa1–Sa22)

Full size image

A total of 19 variables were used to describe environmental conditions at each sampling site. Six physical and chemical features were considered: pH (measured with a pH-metre WTW-pH 330), temperature of water (measured in °C with a thermometer built into the pH-metre), dissolved oxygen (measured in mg l−1 with an oxymeter WTW DIGI 330), water transparency (measured to the nearest cm with a Secchi disc), conductivity and total dissolved solids (measured in μS cm−1 and mg l−1, respectively, using a conductimeter WTW-LF 340). All these variables were measured at 7 H, 12 H and 17 H, and at 50 cm depth, except for water transparency. The habitat variables included are: current speed (measured in m s−1 using a propeller-driven current metre Valeport BF 008), depth (m), width (m), mean canopy closure and height (visually estimated and expressed in % and to the nearest m), aquatic plants (measured as % of stream surface overlain by this shelter category) and substrate type (mud, sand, clay, gravel, rock, mixed clay-mud and leaves-wood, measured as % of stream bottom surface overlain by each substrate type). These 19 environmental variables were used to interpret the variation in species at each sampling site following Kamdem Toham & Teugels (1997, 1998).

Cluster analysis was applied to identify grouping of sampling sites with similar fish assemblages. This analysis is a multivariate technique that sorts objects (such as sampling units) into groups or clusters based upon their overall resemblance to one another (Ludwig & Reynolds, 1988). It was performed with the software Statistica version 6.0, using Euclidean distances and Ward’s method (Pielou, 1984). Associations between fish assemblages and environmental variables were examined, using Canonical Correspondence Analysis (CCA), an ordination technique designed for direct analysis of the relationships between multivariate ecological data sets (Ter Braak & Smilauer, 2002). Significance tests for the general model relating assemblage structure to environmental variables were based on Monte Carlo permutation tests (1000 permutations, P < 0.05). Environmental variables and fish data were log10 (x + 1) transformed prior analysis. Only species captured at least in five sampling sites were included in the analysis, because rare species have a minor influence on results of multivariate statistics (Gauch, 1982). The ordination was performed using the computer program Canoco version 4.5.

Results

Composition of the fish fauna

A total of 64 fish species, from 43 genera and 26 families, were collected in the sampling sites (Table 1). Eleven marine and/or brackish water species (Elops lacerta, Caranx hyppos, Lutjanus goreensis, Pomadasys jubelini, Sarotherodon galilaeus, S. melanotheron, T. guineensis, Tylochromis jentinki, Liza falcipinnis, Awaous lateristriga and Eleotris vittata), two introduced species (Heterotis niloticus and Oreochromis niloticus), one hybrid (T. guineensis × T. zillii) and one endemic species (Synodontis koensis) were reported. Cichlids was the most diverse group, represented by 12 species and amounting to 19% of the total number of fish species, followed by Cyprinids (13%), Mormyrids and Alestiids (11% each). The other families were poorly represented with less than four species.

Table 1 Fish distribution in the Lower Sassandra River Basin
Full size table

Patterns of fish species assemblages

Hierarchical cluster analysis (Fig. 2) allowed to discriminate the sampling sites located on the tributaries (sites Sa10–Sa22: cluster A) from the others (sites Sa1–Sa9: cluster B). Within the cluster A, the subgroup A1 gathered the sites of the Lobo tributary (sites Sa10–Sa15) and the subgroup A2 consisted of those prospected on the Davo tributary (sites Sa16–Sa22). The cluster B included the sampling sites of the Lake Buyo (sites Sa1 and Sa2: subgroup B2) and those located on the main channel (sites Sa3–Sa9: subgroup B1).

Fig. 2
figure 2

Cluster dendrogram summarizing similarity among sampling sites based on their fish assemblages. A1 (Sa10–Sa15) = Lobo tributary sites, A2 (Sa16–Sa22) = Davo tributary sites, B1 (Sa3–Sa9) = main channel sites and B2 (Sa1–Sa2) = Lake Buyo sites

Full size image

The Lake Buyo ichthyofauna (n = 33) was distinguishable from those of other catchment areas by a lower proportion of Mormyrid species and the absence of most of marine and/or brackish water species (Table 1). The main channel exhibited the greater species richness (n = 57) and the higher number of marine and/or brackish water species (n = 11). Within the Lobo tributary (n = 49), fish assemblages were mostly characterized by the presence of more Barbus species than found in the other parts of the basin. On the other hand, the sampling sites of the Davo tributary (n = 35) were distinguished by a low proportion of both Barbus and cichlid species.

Quantitative aspects of fish distribution

Fish community comparisons based on the relative abundance of fish families showed that in the Lake Buyo, the dominant family was Cichlidae (40% of the total number of fishes sampled in this catchment area), followed by Claroteidae (15%), Alestidae (12%) and Cyprinidae (11%). In the main channel, Alestiids were the most abundant (34%), followed by Mormyrids (18%) and Clupeids (12%). In the Lobo tributary, the dominant families were Mormyridae (30%), Schilbeidae (22%) and Alestidae (21%). In the Davo tributary, Alestidae were the most abundant (28%), followed by Claroteidae (17%), Schilbeidae (15%) and Mormyridae (8%).

Patterns of species richness

The fluctuations in the species richness within each catchment area were summarized in Fig. 3. In the main channel (Fig. 3a) and the Lobo tributary (Fig. 3b), the number of species increased progressively from upstream to downstream sampling sites. In contrast, an irregular distribution of fishes was observed in the Davo tributary (Fig. 3c), where the species richness varied from 16 species (Sa16) to 22 species (Sa17), before it decreased greatly, reaching the minimum value (n = 7 species) at the sampling site Sa19. Then, the number of species increased up to 19 at the sampling site Sa22.

Fig. 3
figure 3

Spatial evolution of the species richness within the Lower Sassandra River Basin. (a) main channel, (b) Lobo tributary, (c) Davo tributary, and * = Palm industry

Full size image

Patterns of species composition in relation to environmental variables

The CCA showed the overall relationships between species distribution and environmental variables (Fig. 4). Among the 19 environmental variables initially included in the analysis, only seven were significantly (P < 0.05) related to assemblage structure and accounted for 56.78% of the variance explained by all the original variables: canopy closure (9.74%), leaves-wood (9.63%), aquatic plants (8.75%), temperature (8.98%), width (7.05%), total dissolved solids (6.61%) and depth (6.02%). Among the ordination axes extracted from the analysis, the first two axes (with, respectively, eigenvalue λ1 = 0.57 and eigenvalue λ2 = 0.32), explained 54.4% of the cumulative variance in the species data. Monte Carlo permutation attested that both axes were significant (P < 0.05). Therefore our interpretation will be limited to these two axes which displayed four groups of sampling sites:

Fig. 4
figure 4

Canonical correspondence analysis (CCA) plots showing species (a) and sites (b) distribution in relation to significant environmental variables. See Table 1 for species codes. A1 (Sa10–Sa15) = Lobo tributary sites, A2 (Sa16–Sa22) = Davo tributary sites, B1 (Sa3–Sa9) = main channel sites and B2 (Sa1–Sa2) = Lake Buyo sites

Full size image

Group A1 was representative of the Lobo tributary sampling sites (Sa10–Sa15) having a maximal canopy closure. The fish species correlated to this variable were Barbus ablabes, B. macrops, Chromidotilapia guntheri, Hemichromis fasciatus, Marcusenius furcidens, M. senegalensis, M. ussheri, Mormyrops anguilloides, Mormyrus rume, Petrocephalus bovei, Pollimyrus isidori, Schilbe intermedius and S. mandibularis, which were negatively correlated with axis 1 and positively correlated with axis 2.

Group A2 comprised the sampling sites of the Davo tributary (Sa16–Sa22). This area was distinguished by fish assemblage including Brycinus imberi, B. longipinnis, B. nurse, Chrysichthys maurus, C. nigrodigitatus, Clarias anguillaris, Hepsetus odoe, Heterobranchus isopterus, H. longifilis, Parachanna obscura, S. bastiani, S. koensis, S. punctifer and Raiamas senegalensis with negative scores and associated to a high mean % of leaves-wood in the substrate.

Group B1 represented the sampling sites of the main channel (Sa3–Sa9). This area was distinguished by fish assemblage including Alestes baremoze, B. macrolepidotus, Distichodus rostratus, Hydrocynus forskalii, Labeo coubie, Lates niloticus, Papyrocranus afer, S. galilaeus and S. melanotheron which were positively correlated with axis 1 and negatively correlated with axis 2. The main channel was characterized by low canopy closure and high temperature and total dissolved solids.

Group B2 included the sampling sites of the Lake Buyo (Sa1 and Sa2). The main fish species found in this area were H. bimaculatus, L. parvus, O. niloticus, Pellonula leonensis, T. guineensis and T. zillii with positive scores. Sites in the Lake were characterized by aquatic plants, and wider and deeper habitats.

Discussion

During the present survey, a total of 53 freshwater fish species were identified in the Sassandra River Basin, versus 72 species previously recorded by Teugels et al. (1988). The differences between these data were mainly related to the lack of monitoring both upstream and in the small tributaries. Nevertheless, five species were reported from the Sassandra River for the first time: P. afer (Notopteridae), Micralestes elongatus (Alestidae), Parailia pellucida (Schilbeidae), Tilapia busumana and T. mariae (Cichlidae), suggesting that the number of species known from this basin may be far from complete.

In general, the damming of a river generates numerous changes in aquatic habitats through the transformation of the ecosystem from the fluviatile (lotic environment) to the lacustrine (lentic environment), the modifications in physical and chemical features of water, the habitat fragmentation and the alteration of both the natural hydrology and the hydrographic integrity (Jackson et al., 1988; World Commission on Dams, 2000). Additionally, dams may disrupt fish migrations, affecting the ability of fish to move upstream and onto the floodplains to spawn. It is thus to be expected that an impoundment would have a profound effect upon the composition of the original fish fauna of the basin. In the present study, it appeared that within the Lake Buyo, the Cichlidae made up the greatest part of the catches. The predominance of the Cichlids in lacustrine fish faunas has been observed in several other man-made lakes, including Faé (Koné et al., 2003) in Côte d’Ivoire, and Kainji (Ita, 1984) in Nigeria. These findings underlined the adaptability of this fish family to lacustrine systems. One of the factors favourable to the expansion of the Cichlids in these lentic biotopes is the high production of plankton (Laë & Lévêque, 1999; Ouattara et al., 2001) on which they feed. Another point of similarity in the reservoirs ichthyofauna is the low proportion of Mormyrids, probably due to the inundating of the riverine habitat of these benthic-feeding fishes to anoxic depths. It is a pity that the scarcity of data on the fish distribution before the closure of the dam precludes accurate analysis of the evolution of fish fauna diversity in the Lake Buyo.

Our results revealed a general increase in species richness from upstream to downstream areas in the main channel and the Lobo tributary. This increase is generally attributed to a downstream increase in available habitat, habitat size, habitat heterogeneity and a decrease in environmental fluctuations (Grenouillet et al., 2004; Tejerina-Garro et al., 2005). The irregular fish distribution observed within the Davo tributary is mostly due to the activities of the palm oils industry located at the sampling site 19. The major concern is the discharge of the waste water and waste of palm oils into the stream. In addition, the water is extracted for cooling, washing or as a solvent as part of the production process. The degradation of the forest along the river for creating palm tree farms and pest control in these exploitations also constitute a threat for the aquatic biodiversity. As a result, the fish assemblage within the Davo tributary exhibited a low proportion of Mormyrids (8%) while the Siluriformes (Claroteidae, Schilbeidae and Mochokidae) are relatively abundant (41%). Indeed, Mormyrids are known to be very sensitive to water quality or habitat degradation (Kouamélan et al., 2005). Conversely, the Siluriformes are more tolerant to poor environmental conditions. Furthermore, during the fieldwork, local fishermen reported that they have suspended their activities at and around the sampling sites Sa19 and Sa20 because of the catches decline.

According to the CCA, the main environmental factors that determined the patterns of fish distribution and assemblage structure were width, aquatic plants, depth, temperature, total dissolved solids, leaves-wood and canopy closure. Surveys undertaken in other basins of Côte d’Ivoire such as the Boubo (Kouamélan et al., 2003), Comoé (Yao et al., 2005) and Mé (Kouadio et al., 2006) have also pointed to leaves-wood, canopy closure, width and temperature as major determinants of fish structure.

The importance of watershed size as a determinant of species diversity and distribution is largely documented for tropical rivers (Hugueny, 1990; Edds, 1993) as well as for temperate ones (Maret, 1997; Taylor et al., 2004). According to Angermeier & Schlosser (1989), three factors can explain the relation between numbers of fish species and stream size. An increase in stream size (1) results in an increase in diversity of habitat types, (2) provides a larger area that can support more individuals and that enables the pool of available species to be sampled more completely and (3) results in different rates of immigration and extinction.

In the Sassandra River, the watershed covered by the invasive plants is equivalent to more than 40% of the whole of the water. The impact of weed infestation should be discussed from two angles. From the point of view of the fish community, we hypothesize with Jackson et al. (1988) that the advent of macrophytes in a man-made lake is beneficial, as the vegetation improved the habitat by increasing food availability, spawning facilities and shelter from predators. Furthermore, the plants take up a large quantity of nutrients from the water, so that these are locked up within the dam basin rather than being lost by downstream discharge. Later, these nutrients are released to the food cycle when the plants themselves die and decay. From the point of view of riparian human populations, weed infestation is a problem, resulting in mounting economic, social and ecological costs. It disrupts navigation and fishing activities and is a health hazard as it offers a suitable habitat for the snail vectors of Schistosomiasis (N’Goran et al., 1997). For example, in the Sassandra River Basin, following the invasion of the macrophytes, the fish capture decreased from 142 tons in 1990 to 32 tons in 1994 (Etien & Arfi, 1996).

Several surveys have highlighted the contribution of depth in shaping of fish communities (Tunesi et al., 2006; Chandra & Terry, 2007). The effect of depth on assemblage structure is probably mediated by fish feeding habits, size and tolerance levels to anoxic conditions.

The role of both instream and outstream covers in the physical, chemical and biological processes in rivers has been stressed by several studies. Dead leaves and woody debris influence habitat characteristics such as depth and current (Angermeier & Karr, 1984). They also ensure food availability for fishes and protection from current and predators. The overhanging vegetation plays a key role in the determination of temperature, the primary production, the physico-chemistry, the hydrological regime and the trophic guild distribution within watershed (Welcomme & Mérona, 1988).

Conclusion

From a conservation standpoint, these results highlighted the need to prioritize conservation actions for the Sassandra basin. This conservation goal is all the more crucial since this basin shelters an endemic species, S. koensis. Specially, the minimization of the impacts of forestry operation, agricultural practices, industrial activities and damming are required. Stream width, stream depth, temperature, total dissolved solids and both instream (presence of wood and leaves, and aquatic plants) and bankside (canopy closure) cover appear to be the most influential factors in shaping species assemblages, and should be taken into consideration when defining aquatic habitat types for conservation planning and integrated river basin management. Furthermore, the report of several species for the first time in the basin, added to the current lack of adequate policing of the ecosystems in the north of Côte d’Ivoire because of civil war, requires us to update the ichthyofauna biodiversity as soon as possible.