Assessing disruption of longitudinal connectivity on macroinvertebrate assemblages in a semiarid lowland river Avaliando a interrupção da conectividade longitudinal em assembleias de macroinvertebrados em um rio semi-árido

Aim: Our aim in this study was evaluate the effects of flow regulation for irrigation on the macroinvertebrate assemblages in a semiarid river. Methods: We sampled two reaches in Dulce River; one placed upstream a weir that diverts flow into a network of irrigation channels and the other downstream that weir, in the assessment of the fluvial discontinuity. We assess the differences among reaches and sites, environmental variables, invertebrate density, richness and Shannon-Wiener index applying non-parametric analyses of variance Kruskal Wallis. The similarity percentage analysis (SIMPER) was used to identify which species contributed to the dissimilarities on macroinvertebrate assemblage structure. Canonical Correspondence Analysis (CCA) was performed with the total set of samples to explore macroinvertebrate distribution in reaches and associations of the assemblages with habitat variables. Results: The density, richness and Shannon index values did not show differences between the reaches located upstream and downstream. Beta diversity (Whittaker) was 0.72 among upstream sites, 0.56 among downstream sites and higher species turnover (0.73) was obtained between both reaches. The Canonical Correspondence Analysis explained 46.71% of the variance differentiating upstream sites explained by higher values of organic matter of bottom sediments and discharge, high density of Nais communis, Bothrioneurum americanum, Pelomus, Stephensoniana trivandrana, Pristina  menoni, P. jenkinae, P.longidentata, P. americana, Dero obtusa, Endotribelos, Heleobia and Turbellaria. The downstream sites were associated to coarser substratum and higher density of Lopescladius, Polypedilum, Cricotopus, Thienamaniella, Cryptochironomus, Baetidae, Nematoda and Corbicula fluminea. Conclusions: The low-flow disturbance had effects on the composition of the benthic invertebrate assemblages, but attributes (such as density and richness) showed a lower variability probably because of taxa replacement.

Although dams construction is one of the most widespread human impacts and the negative effects of large reservoirs are well documented, there are few studies on the effects of small dams (e.g.<10 m height) (Sharma et al., 2005;Ambers, 2007;Principe 2010, Fencl et al., 2015).The understanding of this phenomenon and its consequences for the maintenance of ecosystem functions and associated services is needed to guide management decisions on water resources and river restoration projects (Arthington et al., 2006(Arthington et al., , 2010)).
We hypothesized that the flow reduction resulting from the subtraction of water to the irrigation system modifies the composition and structure of macroinvertebrate assemblages dowsntream.

Study area
The study was performed in the Dulce River (Santiago del Estero Province) of Salí-Dulce endorheic basin, with headwaters in the mountainous regions from the northwest, then flow through the plain into the Mar Chiquita lake in central region of Argentina.
The Dulce River is located in Chaco Seco Ecoregion (Burkart et al., 1999;Brown & Pacheco, 2005).The Chaco climate is distinguished by its strong seasonality, with summer maxima of up to 49 °C.The rainfall declines from over 1000 mm/year in the east to less than 500 mm/year in the west,

Introduction
Streams and rivers are intensely modified due to different hydrological alteration (dam construction and water diversion) that may have large impacts on invertebrate community composition (Rosenberg et al., 2000).The river regulation alters natural flow regime (frequency, magnitude and duration) with effects on the physical-chemical characteristics, the morphology of the river, and the heterogeneity of habitats.These alterations have significant effects on the distribution and structure of macroinvertebrate assemblages and affect the natural connectivity of lotic ecosystem (Poff et al., 1997;Bunn & Arthington, 2002;Suen & Eheart, 2006).
Human disruption of hydrological linkages, such as the longitudinal connectivity (Ward & Stanford, 1983) and lateral connectivity (Amoros & Roux, 1988) through the dam or reservoir built impact on ecological process.The flow regulation generate spatial variations (among others) in stream bed topography, local flow regime, changes in the sedimentation rates, nutrient dynamics, organic matter input and decomposition patterns and productivity (Ward & Stanford, 1983;Dewson et al., 2007b;Tupinambás et al., 2015).
Palavras-chave: regulação do fluxo; barragem de desvio; ecologia de rios; diversidade macroinvertebrados; rio Dulce.with a dry season in the winter and spring and a rainy season in the summer (Pennington et al., 2000).
The Dulce River is regulated since 1966 from headwaters by Río Hondo dam with an annual medium discharge of 130 m 3 .s - .The peak flows pulse start in November/December reaching the maximum flow in February/March, and the low water level occurs since April to October (Gallego, 2012) .
The water use of the Dulce River occurs 50 km downstream at Los Quiroga town, where is located a weir, that by raising water level diverts flow into a network of irrigation channels (Figure 1).From 60 to 80 m 3 .s - are diverting during practically the whole year, with the exception of June when the floodgates are closed in order to carry out their cleaning (Figure 2).

Sampling
Two reaches were selected in Dulce River, one was placed upstream the weir at Los Quiroga and the other downstream for the assessment of the fluvial discontinuity.Four sites were sampled in each reach (S1-S4 upstream and S5-S8 downstream) in October 1 st , 2014 (Figure 1).On each site, the center and the right bank were sampled.Three samples of bottom sediments for assessing benthic invertebrates and one sample for granulometric and bottom organic matter analyses were also collected in each sampling site (center and right bank) with a Rigosha grab of 100 cm 2 .The benthic samples were sieved (200 µm mesh size) and fixed in 5% formaldehyde solution in the field.
In each sampling site, temperature (standard thermometer), conductivity and pH (with Hanna  checker), dissolved oxygen (Winkler method), BOD 5 (incubated 20 °C), transparency with Secchi disc, depth and current velocity (with floats) were measured in the center.
The granulometric composition of each sample (%) was determined using a screening method according to Wentworth scale (1932).The organic content of the sediment samples was determined using ash-free dry-weight method (550 °C for 3 h); the difference between the initial weight of sample and weight after ashing gave the percentage of the organic sediment content.

Data analysis
To assess the differences among reaches and sites, environmental variables, invertebrate density, richness and Shannon-Wiener index were subjected to non-parametric analyses of variance (Kruskal-Wallis test, p<0.05).The analyses were run using InfoStat software (version 2016, Di Rienzo et al. 2016).Beta diversity (spatial species turnover) among the sites upstream and downstream the weir, and between the two reaches was calculated according to Whittaker (1972).The similarity percentage analysis (SIMPER) was used to identify which species contributed to the dissimilarities (Bray-Curtis matrixes) on macroinvertebrate assemblage structure.
Canonical Correspondence Analysis (CCA) (Ter Braak, 1986) was performed with the total set of samples to explore macroinvertebrate distribution in reaches and associations of the assemblages with habitat variables.Redundant variables were removed using a Spearman Rank Correlation comparison method (threshold value: r ≥ 0•6 or r ≤0•6), CCA and SIMPER were performed using the Past Software (version 3.14, Hammer et al., 2015).Only taxa with density >100 ind.m -2 were selected.Abundance data were log10 (x+1) transformed and a Monte Carlo permutation test was performed for determining the significance of eigenvalues derived from the CCA.

Environmental variables
The temperature, conductivity, transparency, depth, discharge and coarse sand percentage presented differences between reaches upstream and downstream according to Kruskal-Wallis results (Table 1).The lowest dissolved oxygen values (3.24 mg.l -1 ) and the transparency (15 cm) were obtained in S8, where the BOD (18.7 mg.l -1 ) and conductivity (794 µS.cm -1 ) reach the maximum value.
The bed substratum was characterized by scarce percentages of gravel with higher values only in the banks of S4 (45%) and S6 (40%) and mainly composed of medium and coarser sands substratum.Finer substrate were dominant in the reach upstream the dam, bank of S2 (65.52%) and center S1 (46.67%),where the highest organic matter values was obtained (1.39% in S2 and 1.26% in S1).

Macroinvertebrate assemblage structure
Density, richness and Shannon index did not show significant differences between the upstream and downstream reaches (KW test, p=0.07, p= 0.47 and p=0.87).
It was found that the upstream and downstream sites differed in the macroinvertebrate    The Ephemeroptera Caenidae and Baetidae and Corbicula fluminea MÜLLER, 1774 (Bivalvia) were mainly associated with downstream sites.Ceratopogonidae was represented in both reaches.Crustacea Copepoda and Ostracoda were more abundant in the downstream reach.
The CCA grouped benthic samples mainly by reaches (Figure 4).The first two ordination axes explained 46.71% of species-environment relation (Eigenvalues: axis 1: 0.41; axis 2: 0.26) and Monte Carlo permutation test showed that the difference between axes was significant (p =0.001).
Most samples from upstream sites were located on the negative quadrant of the first axis where the most important environmental variables in the ordination were bottom organic matter, dissolved oxygen and discharge and the invertebrates with the highest density were Pristina menoni, P. jenkinae, P.longidentata, P. americana, P. sima, Stephensoniana trivandrana, Slavina sp., Aulodrilus piguetti, Dero obtusa, Nais communis, Bothrioneurum americanum, Pelomus, and Turbellaria.On the other hand samples collected in downstream sites were placed on the positive quadrant (Figure 4), associated to coarser substratum (gravel, coarse and medium sand) and higher density of Lopescladius, Riethia, Apedilum, Caladomya, Dajlmabatista, Tanytarsus, Harnischia, Baetidae, Nematoda and Corbicula fluminea (Figure 4).

Discussion
The longitudinal connectivity in fluvial system disrupted by dam construction of different size, including weirs, alter the physical and chemical conditions downstream, influencing habitat availability, which will be reflected in changes of macroinvertebrates structure (Poff & Hart, 2002;Tupinambás et al., 2015).In coincidence with Dewson et al. (2007a) the water abstraction decreased water velocity and depth downstream of the weir in the fluvial system of the Río Dulce.Increase of sedimentation rate and finer sediments deposition upstream are often a consequence of flow regulation in small impoundments and substrate size increases below low dams (Fencl et al., 2015).We obtained similar results with finer substratum and higher bottom organic matter in the upstream reach and coarser substratum downstream.However, changes in size substratum and depth around dams depend on many local factors, such as, spatial patterns of channel morphology and sedimentation can decrease because suspended sediment levels in the water decrease because of the presence of aquatic plants that accumulate sediment (Bond 2004;Skalak et al., 2009;Csiki & Rhoads, 2014).Despite it was expected a decrease in the number of taxa downstream of the weir, we did not found significant differences in richness between upstream and downstream reaches.Nevertheless, the taxonomic composition was different with more richness and density of Insecta downstream   2.
Acta Limnologica Brasiliensia, 2017, vol.29, e19 compared to upstream, mainly taxa of short life cycle preferring coarse substratum and low organic matter deposition.Some authors have reported changes in benthic taxonomic composition due to alteration of fluvial habitats by effect of large dams (Gumiero & Salmoiraghi, 1998;Growns & Growns, 2001;Lessard & Hayes, 2003;Dewson et al., 2007b;Poff & Zimmerman, 2010) but others found a relatively minor impact of small dams on the river biota in downstream reaches (Sharma et al., 2005;Ambers, 2007;Principe, 2010).The higher density found downstream (but not significantly different) could be associated with flow regime variability, with the low or reduced flow benefiting the periphyton development and benthic algal biomass in gravel-bed rivers (Suren et al., 2003;Biggs et al., 2005).Therefore, increases in food resources can cause invertebrate densities to increase.However, investigators have found both positive and negative relationships between low flows, substratum particle size and density (Rabeni, et al., 2005;Dewson et al., 2007a;Eedy & Giberson 2007).
Our results are similar to previous studies reported by Maroneze et al. (2011) where density and richness increase when the flows is reduced with dominance of Chironomidae and decrease of Oligochaeta.The Oligochaeta species replacement in the Dulce River upstream and Chironomidae downstream is a result of the disturbance that occurs for the abstraction and water deviation.The CCA results showed that habitat type was the most important source of variation of faunal composition, being substrate composition and discharge the environmental variables affecting the ordination.Upstream was characterized by deeply habitat, finer substratum and high organic matter and Oligochaeta showed a close association with this reach.Downstream reach was characterized by coarse -medium sands, low variation in flow, high green algal mat biomass and Chironomidae and Ephemeroptera were associated with this reach.These habitats produced by hiydrogeomorphologic changes in the river system may be recolonized by organisms of short life cycle (e.g.Diptera), as it has already been reported (Maitland, 1990;Erskine et al., 1999).
The colonization of new habitats depend on the organism dispersion capacity, as passive drift decreases in response to low water velocities during periods of low flow (Bilton et al., 2001;Cañedo-Argüelles et al., 2015;Tonkin et al., 2016), but many studies have shown that active drift increases during periods of low flow (Poff & Ward, 1991).In addition, Walters & Post (2011) concluded that the behavioral responses and the relative mobility of aquatic insects are important in understanding the impacts of flow reduction, as they found decreased biomass, shifts in density of aquatic insect families with significantly increase densities of more mobile insects.
The flow regulation is a dominant factor organizing stream communities and water diversion constitutes barriers that disrupt connectivity of lotic ecosystems and may generate discontinuities in the structure of biological communities.This study finds support for this in that the low-flow disturbance had effects on the composition on the benthic invertebrate assemblages, but also attributes (such as density and richness) showed a lower variability probably because of taxa replacement.

Figure 2 .
Figure 2. Hydrogram corresponding to the period 2014/2015 based on average monthly flows in the Río Hondo dam and the estimated downstream of the weir at Los Quiroga town.

Figure 3 .
Figure 3. Average density (ind.m -2 ) of macroinvertebrates sampled in each site and the mean discharge (m 3 .s - ) corresponding to the previous month to the sampling date.The bars indicate the standard deviation.

Figure 4 .
Figure 4. Triplot of Canonical Correspondence Analysis representing associations between macroinvertebrate densities and environmental variables.Square represent upstream sites, fill square: right bank sites, empty square: center sites.Circles represent downstream sites, fill circles: right bank sites, empty circles: center sites.Sites do not appear reported 0 organism.OM: organic matter; Q: discharge; DO: dissolved oxygen; Gr: gravel; Cs: coarse sand; Ms: medium sands; Te: temperature.Inside elypsis: Oligochaeta taxa and inside rectangles: Insecta taxa.Taxa acronyms in Table2.

Table 1 .
Environmental variables analyzed, Mean and Standard Deviation.

Table 2 .
Average of taxa density (ind.m -2 ) ± standard deviation in upstream and downstream reach.

Table 3 .
Macroinvertebrates contributing to dissimilarity (SIMPER analysis) between upstream and downstream reaches of the weir, and among the sites in both reaches.