Wadi Tafna, which is located in northwestern Algeria, originates from the Tlemcen Mountains (Tell Atlas) and flows 170 km to the north and empties into the Mediterranean Sea. Its catchment area is about 7245 km² (Figure 1).

Tafna catchment mainly includes two geological structures: Tlemcen Mountains in the south, containing dolomitic terrains rich in carbonates of the Upper Jurassic, and the plain of Remchi in the north, consisting of clay-sandstone deposits of the Miocene. Valley bottoms are covered with alluvial Quaternary deposits [15] . The regional climate is Mediterranean with mild and wet winters; it is semi-arid over the entire catchment.

The upstream lands are highly cracked and karst-like, which gives them high permeability and a significant flow of groundwater that emerges at the surface as springs [16] . This provides the upstream wadis with some hydrological stability. In contrast, the downstream basin receives irregular and mostly seasonal rainfall, which results in a series of floods and low water periods in wadis at medium and low altitudes [17] . In the immediate vicinity of the upstream sites, the land is partly used for limited market gardening. Downstream, crop growing predominates in the plain, alternating with vegetable crops. These are irrigated by pumped water that affects the flow of streams.

Sampling was carried out at seven sites along wadi Tafna, one site in wadi Khemis, another in wadi Chouly, and two other sites in wadi Isser, the main tributary of wadi Tafna (Figure 1). To achieve representative sampling of the catchment, the 11 sites were randomly drawn from upstream to downstream. Three classes of sampling sites were selected (1) sites of high Tafna, from 0.5 to 11.5 km from the source and located between 1090 and 650 m altitude, (2) sites of middle Tafna, from 50 to 54.5 km, at altitudes between 303 and 220 m, and (3) sites of low Tafna at more than 75 km with an altitude between 83 and 45 m.

Benthic samples were taken using a Surber sampler (mesh net 300 µm), considering four main physical factors: current velocity, type of substrate, and aquatic and riparian vegetation. Measurements of the instantaneous stream velocity, using the weighted float technique [6] , allowed five classes of currents velocity to be identified (VF: very fast, F: fast, M: medium, S: slow, VS: very slow). The substrate nature (Table 1) was determined from the erosion facies (coarse grains: rocks, stones and pebbles) or from sedimentation (fine grains: sand, silt, and mud). Three types of substrates were defined as eroded, deposited and heterogeneous (fine and coarse elements). Aquatic vegetation included all living organic materials in the form of dense herbs, submerged and emergent spermaphytes, and bryophytes and algae. The riparian vegetation was assessed through plant stratification (herbaceous, shrubby and arborous). Thus, 143 samples were collected at 11 sampling sites in 2009 (March to November). The sampling effort was the same for all sites, at the rate of two outlets per month, but the number of samples per site was linked to the specific spatial heterogeneity.

The black flies were counted and determined morphologically to species level, using the analytic key and descriptions of Simuliidae of Morocco [18] .

R = rocks, P = cobble, G = gravel, S = sand, Si = silt, M = mud, C = clear, T = turbid, Ps = pluristratified, Sh = shrublands, Hb = herbaceous, A = agriculture, Gz = grazing.

The spatio-temporal distribution of black flies was first studied using a factorial analysis (AFC) from the data matrix, taking into account species abundances only in prolific samples containing at least two individuals. An analysis of inertia was then performed to assess the proportion of variability explained by each of the five variables: five variables: station, month of sampling, stream velocity, substrate nature and class of distance from the source. A Monte Carlo Permutation Test was performed to test the significance of this variability. The dot plot was used to illustrate the distribution of species abundance for each site. These statistical analyses were performed using Minitab 16.1 statistical software.

The structure of black fly communities for each site was determined by calculating diversity indices: specific richness S, Margalef index (Dmg = S − 1/ln (N)), Shannon index (H′ = −Σpilnpi), and Piélou index (J′ = H′/Hmax). Hierarchical Cluster Analysis (HCA), performed from the matrix of samples for black flies and the studied physical characteristics, was used to measure the degree of similarity, according to the euclidean distances and aggregation, using Ward’s method (moment of second order).

Ten species were recorded among more than 7208 specimens (Table 2). All belong to the genus Simulium, represented by five subgenera, namely Eusimulium, Nevermannia, Simulium, Trichodagmia, and Wilhelmia. The most frequent species in our samples have a wide Palaearctic distribution. These species include S. velutinum, S. ornatum (complex), S. trifasciatum and S. pseudequinum. Simulium intermedium is widespread in Europe and also in northwest Africa.

Some less abundant and less frequent species have either a Mediterranean distribution (S. bezzii and S. galloprovinciale) or an Ibero-Maghreb distribution with limited extension (S. sergenti and S. quadrifila). Simulium ruficorne has an Ethiopian distribution.

Of all samples (143), 70% did not contain black flies; 38.5% of these were taken from areas with slow current and a deposited substrate consisting of fine sediments (sand, silt, and mud), 26.6% from fine substrate colonized by aquatic vegetation forming dense herbaceous growth, and 4.9% from a bottom covered with algae. These habitats were predominant at sites K.BB and T.SB. Although located in high and medium Tafna, these sites are upstream of dams (Beni Bahdel and Hammam Boughrara, respectively).

The presence of these dams results in slow flow, homogenization of facies, and sedimentation of fine elements, eliminating coarse microhabitats favorable for black flies. Inertia analysis showed that the highest inertia percentage concerns the variable station (68.8%) with a significant difference, according to the Permutation Test

(p = 0.019), revealing that distributions would primarily be determined by the factor station; p remained insignificant for the other four variables analyzed.

At all the sites, the Margalef index values were less than 2 and the index H' fluctuated between 1.97 and 0.97, indicating low diversity. However, the values of the index J′, always greater than 0.5, indicated the presence of regular and balanced communities (Table 3).

The communities of upper Tafna sites (T.GB, T.AG, T.AZ, and CH.Y) displayed the greatest specific richness due to certain typical species. Simulium galloprovinciale was collected exclusively in wadi Chouly (CH.Y) and S. quadrifila at site T.AG (Figure 2). Upstream of Beni Bahdel dam, the two sites T.AG and T.AZ have S. sergenti. However, S. bezzii is abundant in samples from three upstream sites (T.GB, T.AG and CH.Y). Moreover, the lowland sites have mostly species such as S. velutinum, S. trifasciatum, and S. pseudequinum. These species, the most abundant and ubiquitous taxa, are constant over the entire Tafna catchment (Figure 2).

The 42 prolific samples tested with respect to the main physical characteristics are shown in a dendrogram, resulting from hierarchical cluster analysis (HCA) (Figure 3). Twelve groups were found, each corresponding to a

DS = class of distance from the source, N = sample size, S = specific richness, D.mg = Margalef index, H′ = Shannon index, J′ = eveness.

microhabitat identified by the stream-substrate complex. The number of samples per microhabitat ranged from 1 to 7. Two large sets could be distinguished: (1) four microhabitats (MH 9-10-11-12) with strong currents on eroded substrate, including 17 samples (40.5%), (2) eight microhabitats, including 22 samples (52.4%) with low current and heterogeneous or deposited substrates, and three samples (7.1%) with fast currents and heterogeneous substrates. Table 4 summarizes the characteristics of the 12 microhabitats, their occurrence (%) and their location per station. The microhabitat MH11, with fast current and eroded substrate without vegetation, was distinguished by a higher importance (16.7%). Microhabitats MH2 and MH3 were the least representative; they were characterized by an average current and eroded or deposited substrates. The nine remaining microhabitats had a similar number of samples (3 or 4). The distribution of these microhabitats per site (Table 4) indicated a clear between-station effect. Therefore, microhabitat MH11 was found at four sites, and remained fairly representative. The other microhabitats were found at 2 or 3 sites in upper (T.GB, T.AG, T.AZ) and middle (T.HB, T.RE, T.PC) Tafna and rarely in its tributaries (CH.Y, I.RE, I.OM). Microhabitat MH2, with an average current and eroded substrate, was represented by a single sample taken at station T.GB, downstream of the source of wadi Tafna.

Analysis of simuliid distributions in the 12 defined microhabitats (Table 5) revealed that species richness, as well as abundance, varied from one microhabitat to another. On an eroded substrate and fast current (MH11), the maximum richness (eight species) was found, with high abundances, exceeding 1450 individuals for the S. ornatum (complex) and 900 individuals for S. trifasciatum. With the same specific richness, abundance was substantially lower for microhabitat MH6, which had heterogeneous substrate and slow current. However, the abundance decreased slightly (up to 78 individuals for S. pseudequinum) on eroded substrates covered with aquatic vegetation with a fast current (MH10). In contrast, microhabitats with deposited substrate had few individuals, especially microhabitat MH8 where the current was slow.

The most common species S. velutinum had less than 20 individuals. Intermediate values of species richness were observed in the other microhabitats, displaying abundances higher than 100 individuals for S. intermedium and S. trifasciatum in microhabitats MH4 and MH6 with heterogeneous substrates.

Species abundance differed according to the microhabitats (Table 5). Simulium velutinum was the most plastic and widespread, occuppying all types of microhabitats. It was followed by S. pseudequinum, S. trifasciatum

and S. intermedium, which were encountered in up to 10 microhabitats. These four species colonized habitats with fast and slow currents and different substrates.

Simulium bezzii was found in four microhabitats, with a significantly higher defined by the current-substrate complex. (SE = substrate eroded; SH = heterogeneous substrate; SD = substrate deposited) were characterized by fast current and eroded substrate, with or without vegetation.

The S. ornatum complex was found in three microhabitats; its presence in microhabitat MH11 was represented by more than 1450 individuals. Simulium sergenti accounted for only 18 individuals, of which 13 were in microhabitat MH11. A similar but smaller number of S. galloprovinciale was noted in microhabitats MH10 and MH11, each with a strong current and eroded substrate. The lowest abundance was found for S. quadrifila with only eight individuals distributed among microhabitats MH6 and MH10.