The Loutété River, a left-bank tributary of the Kouilou-Niari, originates on the edge of the Cataractes Plateau [15] at an elevation of approximately 413 m. It is a river that flows through savanna in the Bouenza department and empties into the Niari River. The Loutété River is one of the many tributaries that feed the Niari River on its left bank. The hydrographic network associated with the Loutété is dendritic in nature, reflecting a natural organization without strict structural control, often observed in areas with homogeneous substrate and moderately undulating terrain. The watercourses of the basin exhibit spatial and temporal variation in their hydrological parameters [16] discharge, width, and depth which are strongly influenced by the seasonal rainfall regime. This dynamic causes some rivers to become locally intermittent during the dry season, thereby exacerbating the water vulnerability of local populations. The Loutété region is subject to a transitional equatorial climate referred to by Aubreville (1449) as “Lower Congolese” or “Sudan-Guinean” [17]. It is characterized by a long, cool dry season, with temperatures ranging between 22˚C and 31˚C. The rainy season lasts 7 to 8 months, from October to May, interrupted by a lull or short dry season from January to February (Figure 1). Average rainfall is 1200 mm, with fluctuations ranging from less than 800 mm to just over 1400 mm [17]. However, the dry season lasts 4 to 5 months (mid-May to mid-October). As for the terrain, it consists of plains, valleys, hills, and mountain ranges. The vegetation of the Niari Valley is characterized by a shrubby savanna surrounded by higher, wetter forested areas (the Chaillu and Mayombe Forests) [18].

To conduct and complete this study, five stations were selected along the Loutété River from downstream to upstream during both the rainy and dry seasons. These are: Loutété at the bridge on the RN1 (LRN1), Loutété village Moukanda at the bridge (LVMP), Loutété at Saint Michel (LSM), Loutété village Ntadi Hubuka (LVNH), and Loutété past the village of Mfouati (LVMF). These stations were selected based on the following criteria: accessibility of the site and water depth. The sampling campaigns spanned two years, with two campaigns per year (April-May 2022, August-September 2022, and April-May 2023, August-September 2023), one during the rainy season and one during the dry season.

Figure 1. Study area and sampling stations.Caption: LRN1 (1); LVMP (2); LSM (3); LVNH (4); LVMF (5).

For each sampling campaign, benthic macroinvertebrates from the Loutété River were captured using a very practical tool, the Surber sampler, consisting of two metal frames welded at right angles. The horizontal frame, placed on the substrate, defined an area of 33 × 33 cm, or 0.109 m². The net, made of sieving silk, had a mesh size of 0.3 mm. Sampling was conducted during four campaigns spanning two years: April-May 2022 and August-September 2022; April-May 2023 and August-September 2023, corresponding respectively to a rainy season and a dry season. Sampling was conducted during the day between 7:00 a.m. and 4:00 p.m. At each station and during each sampling campaign, five overnet sweeps were conducted to collect macroinvertebrates. The taxa from the five samples were combined to form a composite sample representative of that station for the sampling campaign in question.

After sampling, a preliminary visual sorting was performed in the field to remove plant debris and large stones in order to reduce the sample size and prevent damage to the macroinvertebrates during transport. Finally, the remainder of the sample was stored in a plastic bag (Ziplock) containing 10% formalin and a label including: the station name, its geographic coordinates (Elevation, Longitude, Latitude), and the collecting date.

In the laboratory, the sample from each station was rinsed with water on a sieve to remove fine sediments and as much formalin as possible. After rinsing, the contents of the sieve were placed in a Petri dish to separate the macroinvertebrates from the remaining sediments under a Carl Zeiss binocular microscope, with magnification ranging from 80× to 320×. The macroinvertebrates thus sorted were then stored in pill boxes containing 95% ethanol (to prevent any deterioration) pending identification. A label bearing the name of the river, the station, and the geographic coordinates (latitude, longitude, and elevation) was placed in each pill box. Specimen identification was carried out at the Oceanography and Environment Laboratory of the National Institute for Research in Exact and Natural Sciences (IRSEN) using the identification keys proposed by: [19]-[26]. Identification was performed for the most part down to the family level. This level of identification is appropriate for biological assessment programs [27]-[29] to determine the state of integrity of aquatic environments.

During each campaign and at each sampling station, physicochemical variables were recorded. Physicochemical water parameters such as pH, temperature, total dissolved solids (TDS), and conductivity were measured in situ using an AMSTRAT multi-parameter probe to characterize the habitat of benthic macroinvertebrates. Water depth (in centimeters, measured using a 10-meter tape), current velocity (estimated at the surface by timing a float released between two fixed points 10 m apart, then converted to m/s), shading type (%), and substrate type were also taken into account. The following substrate categories were identified: sand, gravel, pebbles, mud, dead leaves, small stones, rock, and bedrock. In addition to these habitat-specific variables, two non-quantifiable variables related to the site were considered: the presence or absence of agricultural activities and dwellings in the vicinity of the site.

Based on the non-significant seasonal differences (P > 0.05) found in more than 94.74% of the environmental variables recorded (Table 1), as determined by Student’s t-test using STATISTICA version 8 software, the analyses were conducted without accounting for the seasonal effect. This analysis, which did not account for seasonal effects, provided a comprehensive and synthetic estimate of the taxonomic richness and diversity of the environment. Two approaches were used: one 1) based on synthetic indices commonly used in ecology (specific richness, “S”, the Shannon index “H”, Pielou’s evenness index “R”, Simpson’s diversity index “D”, the Hill index, and Frequency of Occurrence and Constancy (FO%),which in this study apply to taxa; and the other 2) based on multivariate analyses (Principal Component Analysis “PCA” and Canonical Redundancy Analysis “CRA”). The simultaneous use of these two approaches is justified by the fact that studying biodiversity through synthetic indices amounts to studying an ecosystem through indicators that reflect the state of diversity of its biocoenosis [30]-[33].

Specific richness (S): Corresponds to the number of taxa present in the environment;

Shannon diversity index H’ [34]: The Shannon diversity index is used to assess a population within a biotope [35]. It is calculated using the following formula:

where P_i = n_i/N; H’: is the diversity index (in bits); P_i: is the relative frequency of individuals in taxon i; n_i: is the number of individuals in taxon i; N: is the total number of individuals across all taxa; S: the total number of taxa present.

Pielou’s Evenness Index R [36]

Pielou’s R index measures the distribution of individuals within taxa, regardless of taxonomic richness.

R = H’/Hmax where H’max = ln(S) (2)

H’: is the diversity; Hmax: is the maximum diversity expressed in terms of taxonomic richness

Its value ranges from 0 (dominance of a single taxon) to 1 (equal distribution of individuals across taxa).

Simpson’s diversity index D [37]

The Simpson index measures the probability that two individuals selected at random from a collection of S taxa and N individuals belong to the same taxon. This index is a widely used measure of dominance, weighted according to the abundance of the most common taxa and with low sensitivity to sample size [38]. It is given by the following formula:

The Hill diversity index:

Hill = (N2)/(N1) = (1/ʎ)/e^H’ (4)

Its value ranges from 0 (strong dominance of one of the taxa and low evenness) to 1 (even distribution of individuals across taxa and high evenness).

Frequency of Occurrence and Constancy (FO%):

The Frequency of Occurrence (FO) is the ratio, expressed as a percentage, of the number of samples (Pa) from a station where the taxon is present to the total number of samples (P).

FO (%) = Pa/P × 100 (5)

Three groups can thus be defined: the first comprises “very common” taxa with F ≥ 50%; the second group consists of “common” taxa with 25% ≤ F < 50%; and rare taxa form the third group with F < 25%. A fourth group is also defined: this consists of absent individuals with F = 0% [39].

To identify the relationships between the stations, the benthic macroinvertebrate communities, and various environmental variables, multivariate analyses were conducted [40]. They were used to determine the correlation between taxonomic richness and environmental variables that influence habitat conditions, using CANOCO 4.5 software for Windows [41]. Matrices, taxon abundance data, and environmental variables were used. The data were normalized prior to ordination (ACCD); taxon abundances were transformed as log(x + 1), while environmental variables were transformed as ln(x + 1) or arcsin√x for percentage data [42]-[45]. Principal Component Analysis (PCA) was chosen, on the one hand with a gradient length (3 < LG < 4) for the distribution of stations based on taxon abundances; on the other hand with LG < 3 for the distribution of sites based on environmental To identify the relationships between the stations, the benthic macroinvertebrate communities, and various environmental variables, multivariate analyses were conducted [40]. They were used to determine the correlation between taxonomic richness and environmental variables that influence habitat conditions, using CANOCO 4.5 software for Windows [41]. Matrices, taxon abundance data, and environmental variables were used. The data were normalized prior to ordination (ACCD); taxon abundances were transformed as log(x + 1), while environmental variables were transformed as ln(x + 1) or arcsin√x for percentage data [42]-[45]. Principal Component Analysis (PCA) was chosen, on the one hand with a gradient length (3 < LG < 4) for the distribution of stations based on taxon abundances; on the other hand with LG < 3 for the distribution of sites based on environmental variables [46]. The Monte Carlo test (999 permutations, P < 0.05) was performed with variable selection to identify those that best explain the variations in the benthic macroinvertebrate data. Only one variable (5.26%) out of 19 was significant at the 0.05 probability threshold.

Table 1. Comparison of environmental variables at stations along the Loutété River. Significant differences (P < 0.05) are shown in red. DS: Dry season; RS: Rainy season; P: Probability.

Note: 0 = Absent; 1 = Present.

Sampling of macroinvertebrates in the Loutété River yielded 15,336 specimens of macroinvertebrates belonging to 4 classes, 12 orders, and 36 families. Among the orders recorded, Diptera, Hemiptera, Ephemeroptera, Coleoptera, Odonata, and Trichoptera proved to be the most diverse. At the family level, the Hydropsychidae was the most abundant in terms of individuals. The taxonomic composition and frequency of occurrence of benthic macroinvertebrates at the various sampling stations are presented in Table 2.

Table 2. Taxonomic composition and frequency of occurrence of benthic macroinvertebrates at the study sites.

−: F = 0% (taxa absent); +: F ≠ 0% and less than 25% (rare taxa); ++: 25% ≤ F < 50% (common taxa); +++: F ≥ 50% (very common taxa).

Taxonomic richness (S), the Shannon index (H’), the evenness index (R), the Simpson index (D), and the Hill index were calculated based on the numerical abundances of taxa at each site. The highest value for each index is shown in bold, and the lowest is underlined (Table 3).

Table 3. Diversity indices at the sites. S: Taxonomic richness (S); N: total abundance; R: evenness; H’: Shannon; Hmax; D: Simpson.

The results of the Principal Component Analysis (Figure 2) indicate that the first two axes (41.5% and 21.4%, respectively) account for 62.9% of the total variance in the dataset. This PCA generally highlights the formation of three distinct major groups. The first group consists solely of the LVMP station, located on the negative side of the most significant axis, axis 1 (41.5%), which is strongly correlated with depth, bedrock, and velocity. At the opposite end of this same axis lies the second group, consisting of the LVMF station, which is isolated on the positive side of the axis and correlated with canopy closure, dead leaves, and rocks. The third group, on the other hand, comprises the LSM, LRN1, and LVNH stations, which cluster around the origin. However, the LRN1 and LVNH stations are located on the negative side of axis 2, while the LSM station is positioned almost at the zero point of this axis, indicating neutrality with respect to this gradient. This group is generally associated with conductivity, TDS, temperature, pH, gravel, mud, width, small stones, sand, dead wood, and pebbles.

Figure 2. PCA of stations grouped according to environmental variables in the Loutété.

Figure 3. Community composition of benthic macroinvertebrate populations in the Loutété.

The study of correlations between taxa, environmental variables, and all sampling stations in the Loutété River, conducted using Canonical Redundancy Analysis (Figure 3), indicates that the first two axes (45.5% and 22.5%, respectively) account for 68% of the cumulative variance in the taxon data. The correlations between taxa and environmental variables for the two axes are high (0.95 and 0.98, respectively), and the Monte Carlo permutation test (999 permutations) shows that the two axes are significant (F = 2.13; P = 0.04). These results demonstrate that the distribution of taxa, as observed in this analysis, is not random, but rather depends on environmental variables. CRA with “forward selection” identified two environmental variables: sand (Sab) and width (Lar), which proved significant at the 0.05 probability threshold. Although these variables are significant, the majority of taxa are concentrated on the side opposite this gradient.