Kodjoboué lake is located in south-eastern Côte d’Ivoire in the South-Comoé region about 4 km south of the town of Bonoua between longitudes 3˚35'9"W and latitudes 5˚14'11"N. This area is contiguous to the Grand-Bassam wetland and covers an area of 3.404 Km² [20] . This lake communicates with the Comoé river by a thin stream of water. This region has an average altitude of 90 m, its climate is subequatorial with an average annual temperature ranged between 25˚C to 33˚C. The rainfall varies from 1400 to 2500 mm/year and an annual rate of humidity about 80% to 90% [22] . This climate is characterized by four seasons: two dry seasons (mid-July to mid-September and December to March) and two rainy seasons (April to mid-July and mid-September to mid-November) [23] . Four sampling sites were selected in Kodjoboué lake (K1 to K4) (Figure 1). These sites were sampled during eight sampling campaigns from June 2014 to May 2017. Table 1 summarizes the characteristics of these sampling sites.

Aquatic insects of the native vegetation, covering the banks of the lake were

sampled at the sites during the eight sampling campaigns with a kick-net (25 × 20 cm, 500 µm mesh size). At each site 5 net strokes are sampled. A net stroke consists of pulling the net over a distance approximately 1 m. The initial net is followed by two round trips on the surface to capture the dislodged organisms. The samples were then pooled to make one composite sample per site. Samples were fixed on 70% ethanol and stored in labelled polyethylene flasks.

In the laboratory, specimens were sorted and identified to the lowest possible taxonomic level using a stereomicroscope Olympus SZ (40× magnification) and a series of identification keys [24] - [29] . FFGs of aquatic insects were identified based on Cummins et al. (2005) [10] and Tachet et al. (2010) [29] . Temperature, pH, conductivity and dissolved oxygen were measured in situ at each sampling site with a portable multi-parameter (HANNA), water transparency was determined with a Secchi disc. Water was also collected with bottle of 1L at each sampling site for nutrients (phosphorus and nitrate) analysis in the laboratory using Rodier et al. (2009) [30] methods.

Species diversity was calculated using the Shannon-Weaver diversity index and Pielou evenness. Taxonomic richness was rarefied in each site per sampling period. Precisely, rarefied richness was used to avoid any bias related to differences in abundances between samples [31] . Before performing comparison analyses, data normality was checked using Shapiro test. As the biotic and environment data distribution follow non-normal distribution (P < 0.05), the non-parametric test of Kruskal-Wallis was performed to compare data variability between sampling sites. When Kruskal-Wallis test is significant, Mann-Whitney test was used for pairwise comparison. The significance threshold was p = 0.05. Analyses were conducted using the R package. The occurrence percentage (FO) was calculated using the following formula: FO = (Ni/Nts) × 100; with Ni = number of samples containing a given species i, and Nts = total number of samples collected. The FO was used to classify species following (Dajoz, 2000) [32] : FO > 50: very frequent species; 25 < FO ≤ 50: frequent species; FO ≤ 25: rare species.

The FFGs include the calculation of the relative abundance of each FFG, Top-down predator control (ratio of predators to all other FFGs) and the calculation of the balance between autotrophy and heterotrophy (production [P]/respiration [R]). This index was calculated as the ratio of scrapers to (shredders + total collectors):

scr, shr, fc and gc represent the abundances of scrapers, shredders, filterers-collectors and gatherers-collectors respectively. According to Cummins et al. (2005) [10] , P/R > 0.75 indicates autotrophy.

Focused Principal Component Analysis (FPCA) [33] was used to assess relationships between taxa abundance and environmental variables. Focused Principal Component Analysis (FPCA) is a variation of the traditional principal component analysis. It uses the same types of matrix as the PCA but differs from it in that it is centred or focused on a variable xi. Indeed, it allows a graphical representation of the correlations that exist between this variable xi and the other variables. The graph provides access not only to the nature (positive or negative) but also to the significance (P < 0.05) of the correlations between variable xi and the other variables. It is also possible to observe on the graph the correlations between the other variables.

In this study, the FPCA was used to determine the variables that influence the abundance of macroinvertebrate taxa that contribute at least 5% of the total abundance. This analysis was performed using the psy package on the R software.

Table 2 shows the variations of environmental variables measured in the four studied sites.

Temperature ranged from 25.9˚C (K1) to 32˚C (K2). Conductivity varied between 14 µS/cm and 37.1 µS/cm (K1). The dissolved oxygen variation was situated between 0.1 mg/L and 11.3 mg/L (K1). The water transparency of sites values varied from 10 cm to 218 cm (K1). Concerning the pH, it varied from 5.19 (K1) to 8.01 (K4). Regarding the phosphate and nitrate, their values were low, varied from 0.015 mg/L (K2) to 0.31 mg/L (K3) and from 0.1 mg/L (K2 and K4) to 2.65 mg/L (K2 and K3) respectively. Analysis showed no significant differences of these parameters between sites (Kruskal-Wallis, P > 0.05).

A total of 5562 aquatic insect individuals from six orders, 32 families and 68 taxa have been recorded. The order Coleoptera was the most diversified with 19 taxa and eight families. The order of the Diptera was as follows with 15 taxa and

seven families. The least diversified order was the order of Trichoptera with 03 taxa and two families. Families of Dytiscidae and Chironomidae dominated abundances with respectively 33.5% and 26%. The K1 site of Lake Kodjoboué recorded the highest number of aquatic insects (77.68%) followed by the site K2 with 10.48% of the insects harvested. The lowest number of insects was recorded in K4 (5.44%) (Table 4).

Table 3 shows the percentages of the very common, frequent and rare taxa at the four sites. The percentages of very frequent taxa vary between 06.25% (K4) to 12.96% (K1). Those of frequent taxa varied between 21.87% (K4) to 38.88% (K1 and K3). Regarding rare taxa, they are the most numerous at all sites with percentages ranging from 48.14% (K1) to 71.87% (K4). Polypedilum deletum (Diptera) was common to the four sites with a very high occurrence (FO ≥ 50).

The qualitative data on sampling at the lake sites considered are summarized in Table 5. Site K1, with 54 taxa, has the highest taxonomic richness. It is followed by the site K3 with 36 taxa. Site K2, with 32 taxa, has the lowest taxonomic richness of all the stations considered. Of the taxa identified, only 13 are common to all four sites.

The different sites are compared with each other based on their taxonomic composition (Table 5). Sites K1 and K4 contain more common taxa (26), site pairs (K1 - K3) and (K1 - K3) contain fewer common taxa with 18 taxa each. It should also be noted that site K3 contains more specific taxa (04), while sites K1, K2 and K4 each contain only one specific taxon (Figure 2).

The lowest values of the three diversity metrics, Shannon-Weaver index, Pielou’s Evenness Index and rarefied richness were recorded at site K4. However, the highest values were observed at the site K3 for these three metrics. Of these three metrics, only rarefied richness does not show significant variation between the different sites (P > 0.05; Mann-Whitney) (Figure 3).

Five major FFGs were identified in this study: predators (pre), scrapers (scr), shredders (shr), gatherers-collectors (gc), and filter-collectors (fc). The analysis

***Very frequent (%OF > 50); **frequent (25 < %OF ≤ 50);*rare (%OF ≤ 25). (MA) = taxa representing at least 5% of the total abundance.

of the functional composition of the assemblage revealed that the predators were the most abundant (Figure 4) with a proportion of 53.74% of the insects harvested. Gatherers-collectors ranked second with 42.05% of the total insect population. A proportion of 2.84% was observed for the shredders, while filter-collectors and scrapers had the lowest proportions respectively 0.91% and 0.44% of the total abundance.

At the spatial scales (study sites), the sites K2 and K3 had a high proportion of predators compared to the collector-gatherers, while the K1 and K4 sites recorded more gatherers (Figure 5). No scraper was recorded in sites K2, K3 and K4 during this study.

The value of the autotrophic heterotrophy index is less than 0.75 at all four sites. For the top-down predator control, the values are all greater than 0.2 on all our sites (Table 6).

Focused Principal Component Analysis (FPCA) was performed using environmental parameters with a focus on the most abundant taxa (at least 5% of total abundance). Thus, the FPCA states that Hydrovatus sp.is significantly and negatively influenced (P < 0.05) by dissolved oxygen (O₂). Also, this insect is positively correlated with parameters such as conductivity (CND), nitrate (NO₃), temperature (T) and transparency (Transp) but negatively correlated with pH and phosphate (PO₄). Polypedilum deletum is significantly and positively correlated with nitrate (P < 0.05). This taxon is negatively correlated with temperature, pH and dissolved oxygen but positively correlated with other parameters. Cloeon sp. is significantly correlated with dissolved oxygen and pH (P < 0.05). This Ephemeroptera is positively correlated with transparency, but negatively correlated with other parameters. As for Esolus sp., it is negatively correlated with pH and temperature but positively correlated with conductivity, nitrate, phosphate and transparency. Esolus sp. is significantly correlated with dissolved oxygen (P < 0.05) (Figure 6).