The portion of the Loémé basin concerned by this study is located in Mayombe, Kouilou Department and extends from the locality of Mvoungouti to that of Bilinga following the railway line of the realignment of the Congo-Ocean Railway (CFCO). The Loémé basin, located south of the Kouilou basin, has a total surface area of approximately 3250 km². The Congolese Mayombe is located between latitude 3˚30' and 4˚50'S and latitude 11˚30' and 13˚12'E. It is located between 40 and 80 km from the Atlantic Ocean with a width of 50 to 80 km. It is an orogeographic and floristic entity, part of which in Congo Brazzaville is preserved by the Dimonika Biosphere Reserve established in March 1988 [16]. Overall, eight stations were prospected in the upper reaches of the Loémé basin, including its different size tributaries (Figure 1). These stations included the Loémé stations in Bilinga (BILO), Bilinga downstream of the bridge (BAPO), Loufouyou in Nemba (NELO), Loémé in Nemba (LONE), Bidounzi in Tsessi (TSBI), Loémé in Tsessi (TSLO), Mvougounti forested banks (MVBF) and Mvougounti grassy banks (MVBH). The stations were chosen based on: 1) their proximity to the various CFCO stations, making them easily of access, and 2) their various environmental aspects, including the habitat type, canopy, riverbank type, and the presence or absence of human activities. The sampling campaigns were spread over two years at the rate of two campaigns per year, one in the dry season and a second one in the rainy season.

Figure 1. Study area and sampling stations.

For each campaign, stations were sampled using a standardized fishing method using four gillnets of 10, 12, 15 and 25 mm mesh size, 30 m long and 1.5 m drop. The nets were set for approximately 14 hours overnight, between 5 p.m. and at 7 a.m., following [17]-[19]. Species were identified and counted (individuals per species) in the field. The Standard Length of each fish was measured to the nearest 0.01 millimeter and the weight to the nearest 0.01 g. Fish whose identifications were uncertain in the field were preserved in 10% formalin for further identification at the ichthyological laboratory of the National Institute for Research in Exact and Natural Sciences (IRSEN). The classification of families was done according to [20] and that of species of the same family, in alphabetical order.

At each station and during every campaign, the water physicochemical variables were recorded. These include the temperature (˚C), conductivity (µS/cm), TDS (mg/l), and pH. The measurements were carried out in-situ using the HANNA and YSI multi-parameter probes. For each measurement, the electrodes are immersed in water and the value of the parameter displayed is read on the screen after stabilization. Turbidity (m) and depth (m) were measured respectively using a Secchi disk and a depth gauge. Water velocity (m/s), distance to the bank (m), canopy height (m), average width (m), and net shading (%) were also recorded. The nature of the substrate was estimated in percentage of sand, gravel, pebbles, mud, dead wood, dead leaves, boulders and large rocks. In addition to these habitat characteristic variables, the presence or absence of agricultural activities in the vicinity of the station, a non-quantifiable variable, was also recorded. A water sample was taken per station. These water samples were each taken using a 0.5 litre plastic bottle at a depth of 20 cm, stored in a cooler and brought back to the laboratory in these same bottles. The dosage of Calcium, Magnesium, Potassium, Ammonium, Aluminum, Phosphates, Sulfates, Chlorides, Carbonates, Nitrates, Nitrites ions and the alkalinity titer was carried out ex-situ at the hydrology laboratory of the National Institute for Research in Exact and Natural Sciences.

Based on the non-significant seasonal differences (p > 0.05) found on more than 70% of the environmental parameters recorded (Table 1), with the student’s t test, using STATISTICA version 8 software, the treatments were carried out without taking into account the seasonal variations. Diversity indices provide more information about community composition than just species richness; they also take into account the relative abundance of different species [21]. Four types of indices, classic in ecology were calculated using PRIMER software, version 5 [22]. These included the Specific richness S corresponding to the simple counting of the number of species present in the sample; the Shannon H' index [23], used in ecology as a measure of specific diversity [24]-[26]. The following expression was used to calculate H':

Table 1. Minimum, maximum, mean and standard deviation values of environmental parameters. Min: minimum, Max: maximum, Avg: mean, SD: standard deviation.

With S: the total number of species present in the station; n_i: number of species individuals in the sample; N: total number sampled individuals. H' varying from 0 to H ′ max and H ′ max = log S .

Pielou’s R equitability index makes it possible to measure the distribution of individuals within species, independently from specific richness.

Its value varies from 0 (dominance of one of the species) to 1 (equidistributional of individuals in the species);

The Simpson index for measuring diversity. It measures the probability that two randomly selected individuals belong to the same species [27].

n_i: represents the number of individuals of species i in the sample and N the total number of individuals. D varies between 0 (maximum diversity) and 1 (minimum diversity).

To identify the relationships between stations, fish communities and different environmental variables, multivariate analyses were performed [28]. These analyses were used to determine the correlation between species richness and environmental variables that influence environmental conditions, using CANOCO 4.5 software for Windows [29]. Numerical abundance matrices of species and environmental variables were used. Data were normalized before ordination, species abundances were transformed into log(x + 1), while environmental variables were transformed into ln(x + 1), or into arcsin x for percentage data [18][30]. Principal Component Analysis (PCA) was chosen, on the one hand, with a gradient length (3 < LG < 4) for the distribution of stations according to species abundances; on the other hand, with LG < 3 for the distribution of stations according to environmental variables [31]. The Monte Carlo test (999 permutations p < 0.05) was performed with variable selection, to retain those that best explain the variations in the fish species data. Only three variables (9.38%) out of 32 were significant at the 0.05 probability threshold (see Table 1).

Out of one thousand three hundred and fifty-six specimens captured during this work, 24 species distributed in 18 genera, 9 families and 5 orders were identified. Cyprinids were the most diverse with 8 species, followed by mormyrids, alestids, claroteids, and cichlids with 3 species each. The Distichodontidae, Hepsetidae, Clariidae and Malapteruridae were the least represented families with only one species each (Table 2).

Table 2. Species collected, their code and their relative abundances in the stations.

Specific richness (S), Shannon (H'), equitability (R) and Simpson (D) indices were calculated based on the numerical abundances of species per station (see Table 3). The highest value of each index is in bold and the lowest is underlined.

Table 3. Ecological diversity indices. N: number of specimens; S: Specific richness; R: equitability; H’: Shannon; D: Simpson; H ′ max : Shannon maximum.

Figure 2(a) and Figure 2(b) show the distribution of diversity in 8 stations through the diversity indices H', H ′ max and R. From this graph, equitability is high when the gap between H_max and H' is low. Five stations (NELO, BILO, LONE, TSLO and MVBH) present, for species found within each of them, an equitable distribution in abundance, the BAPO and TSBI stations have respectively an averagely balanced and unbalanced distribution.

(a)

(b)

Figure 2. Distribution of diversity in stations across diversity indices. (a): H’ and H ′ max ; (b): R.

The results of the Canonical Redundancy Analysis (RDA) (Figure 3(a) and Figure 3(b)) indicate that the first two axes (respectively 47.1% and 20.3%) express 67.4% of the cumulative variance of the fish data. The Monte Carlo test (with 999 permutations) also showed that the contributions of the first two axes are significant (F = 3.974; p = 0.003). The BAPO and BILO stations are very well represented on axis 1 (47.1%) in its positive part, the TSBI and LONE stations are very well represented on axis 2 (20.3%), respectively on the positive and negative sides. The species Brachyalesteskingsleyae, Bryconalesteslongipinnis, Chrysichthysauratus, Enteromiusholotaenia, Hemichromiselongatus, Hepsetuslineata and Marcuseniusmoorii are highly correlated with each other and are grouped together in the BAPO and BILO stations. Clariascamerunensis and Enteromiuscamptacanthus, which are highly correlated, are present in both the MVBF and MVBH stations and the TSBI station. Enteromiusmartoelli, Garraornata and Paramormyropskingsleyae are also highly correlated and characteristic of the MVBF and MVBH stations. Brycinusmacrolepidotus and Distichodusnotospilus are still correlated but to a lesser extent, the former being more present in the TSLO station and the latter in the LONE and NELO stations (Figure 3(a)).

(a)

(b)

Figure 3. ARD of species, stations and the three environmental variables selected upstream. (a): species-stations biplot; (b): environmental variables-stations biplot.