Use of food resources and resource partitioning among five syntopic species of Hypostomus ( Teleostei : Loricariidae ) in an Atlantic Forest river in southern Brazil

We analyzed the diet and resource partitioning among five syntopic species of Hypostomus Lacépède, 1803 in the Corumbataí River in southeastern Brazil. The gut contents of 352 individuals were assessed and 21 food items were identified and quantified under an optical microscope. The food items found in the gut contents indicate that these suckermouth loricariids are bottom-dwelling detritivorous/periphytivorous catfishes. PERMANOVA and SIMPER analyses indicated variation in the consumption of some resources, and the contribution of periphytic algae was primarily responsible for such dissimilarity. ECOSIM analyses of dietary overlap showed evidence of resource sharing among all species in the dry and rainy periods. This is most likely the result of the predominance of detritus and autochthonous items such as algae and aquatic immature insects in all gut contents. Our data suggest that trophic resources available in the Corumbataí River are explored and partitioned among Hypostomus species, all specialized in surface-grazing foraging behaviour.

ations (winemiller et al. 2008), ontogenetic variations (frehSe et al. 2015), and competition (fauSch et al. 2002).The availability and use of resources play an important role in regulating the biomass, the structure and functioning of the fish community and ecosystem (fauSch et al. 2002(fauSch et al. , winemiller et al. 2008)), and contribute to the maintenance of the high functional and taxonomic diversity of the Neotropical ichthyofauna.
In this study we examined patterns of food utilization of five syntopic species of Hypostomus, which are common and abundant inhabitants of lotic aquatic ecosystems in the Neotropical region.Despite the influence of seasonal changes (Dry or Rainy seasons), we expected that the overlap in the diet of those species would be substantial since they are morphologically similar.Specifically, we attempted to determine whether information on diet and feeding overlap could throw a light on how the available resources are exploited.

MATERIAL AND METHODS
This study was conducted in the Corumbataí River (24°14'S and 51°56'W), in a stretch of approximately 20 km.In this stretch, the river is very homogenous in its general characteristics.There are no artificial and natural barriers (e.g.waterfalls, dams), the average depth is 0.62 m, the average channel width is 26 m, the canopy cover is 25%, the river velocity is 17 m s -1 , and the substrate consists mainly of sand, gravel and rocks.All adjacent regions are surrounded by agriculture and livestock.The Corumbataí River flows into the Ivaí River, a tributary of the left bank of the upper Paraná River, which has a drainage area of 36,899 km 2 and extends for approximately 675 km.It is located in the aquatic ecoregion known as the Upper Paraná, with a reported richness of 310 species and many endemic catfishes (lanGeani et al. 2007).
The climate in our study region is subtropical (caviGlione et al. 2000) with the temperature ranging from 22°C during the warmer months to 18°C during the colder months.This region enjoys a regular amount of evenly distributed rainfall, but during our sampling period the rain volume was one of the lowest recorded in the history of the last 50 years (iaPar 2015).The dry season (March-August 2006) was defined based on the average monthly precipitation, which was 49 mm.The rainy season (September 2006 to February 2007) had an average monthly precipitation of 207 mm.These conditions were used in the analysis of the seasonal variation in the diet of the fish assemblage.
Monthly samples were collected between March 2006 and February 2007, using gillnets placed parallel to the riverbank for 24 hours and a seine net operated for a period of 10 minutes.After capture, all fish were fixed in 4% formaldehyde and after five days they were transferred to 70% alcohol.In the laboratory, individuals were identified, labelled, weighed (total weight, kg), measured (standard length, mm) and dissected for the removal of the gut contents.Voucher specimens of Hypostomus ancistroides Ihering, 1911, H. strigaticeps Regan, 1908, H. commersoni Valen-ciennes, 1836, H. hermanni (Ihering, 1905), and Hypostomus sp. were deposited in the fish collection of the Museu de História Natural Capão da Imbuia (MHNCI 12355-MHNCI 12359).
Food contents were identified to the lowest feasible taxonomical level and quantified using a compound light microscope following a specific methodology based on aranha (1993).After the extraction of the contents of the anterior third of the digestive tract, the material was homogenized and three slides (subsamples) were prepared.The slides were placed on the stage of the microscope and moved to a random location.This random location was defined using the numbering system present on the mechanical stage using one dimension as the abscissa and the other as the ordinate.The field determined by the ordinate and the abscissa (1 × 1 mm) was analysed at 200 and 400 times magnification.A minimum of seven and a maximum of 10 locations per slide were analysed, totalling an average of 25 locations per stomach content.Items were quantified according to their percentage contributions.The frequency of occurrence and percentage contribution methods were used in the diet analysis (hyneS 1950).
Differences in diet composition were investigated through permutational multivariate analysis of variance (PERMANOVA), and similarity percentage analysis (SIMPER) methods performed by the Primer v6 software (clarke & Gorley 2006), using stomachs as samples and seasons (Dry or Rainy) and species as fixed factors.A similarity matrix was constructed using the Bray-Curtis index, using the transformed (logX+1) values of the percentage contribution of food items.This matrix was subjected to a two-way PERMANOVA analysis (anderSon et al. 2008) in order to examine the effects of the factors (seasons and species) on the diet composition.Significant factors were further analyzed using a PERMANOVA pairwise comparison.All tests utilized Bray-Curtis dissimilarity as the metric and were permutated 9,999 times under a reduced model.
Percentage Contribution Analysis (SIMPER) was carried out (clarke & Gorley 2006) in order to estimate the contribution of each variable (food item) to the observed similarity or dissimilarity between samples.It was, therefore, useful in assessing the importance of the different food items to the overall diet, and in identifying the variables that are more important in creating the observed pattern of similarity.
Diet overlaps in each season between all pairs of Hypostomus species were calculated through the Pianka index (Pianka 1973) in the software EcoSim, version 7.72 (Gotelli & entSminGer 2000).This measure of overlap is symmetrical and assumes values ranging from 0 (no prey in common between two species) to 1 (species share the same relative amount of food items), and the overlap is considered high when values exceed 0.60 (sensu wallace 1981).Niche overlaps among Hypostomus species in each season were statistically tested against null models (RA3 algorithm, 1,000 iterations) using the EcoSimR package (Gotelli et al. 2015) with R 3.3.1 statistics software (r develoPment core team 2016), in order to evaluate whether the pattern of niche overlap diverged significantly from a random distribution (absen ce of overlap).

RESULTS
A total of 352 individuals of Hypostomus ancistroides, H. strigaticeps, H. commersoni, H. hermanni, and Hypostomus sp. were captured, with a total weight of 19.39 kg.In both seasons there was an overlap in the sizes (SL) of all Hypostomus species.The two most common species (H.commersoni and Hypostomus sp.) comprised 75.8% of all captures and 66.8% of total biomass.No taxon was exclusive to the samples collected in the dry or rainy seasons, and H. commersoni and Hypostomus sp. were the most abundant in both periods (Table 1).
of Coleoptera, Odonata nymphs, rotifers and gastropods were absent from the diet of all species.
SIMPER analysis showed that the similarities among the diets of Hypostomus species' result from the high percentage contribution of detritus, diatoms and algae fragments, and that the dissimilarity between the diets of H. ancistroides and H. commersoni, Hypostomus sp. and H. stigatriceps, and between The results on diet composition of gut contents included 21 different food items (Table 2).In general, the diet of all Hypostomus species was slightly more diverse during the rainy than the dry season.In the dry season, immature aquatic stages  5).
The Ecosim analyses of dietary overlap based on Pianka's index found high values (> 90%) for all pairs of species, indicating substantial similarity in the diets of Hypostomus species in both seasons (Table 6).Comparisons with null (RA3) models indicated that the observed overlap between species in dry (0.96) and rainy (0.99) seasons was greater than expected by chance (0.16, p = 0.001).

DISCUSSION
All five species of Hypostomus were sampled along the main channel of the Corumbataí River, in the dry and rainy seasons.The relatively high species richness and abundance of representatives of this genus can be explained by the great amount of fast-flowing aquatic environments and rocky substrate, environments that favor the occurrence of armored catfish (weBer 2003), in addition to many illuminated stretches with submerged streamside vegetation, where loricariids display a surface-grazing foraging behaviour (montoya-BurGoS et al. 2003, luJan et al. 2012).
Food items ingested by Hypostomus species in the Corumbataí River indicate that these suckermouth loricariids are bottom-dwelling detritivorous/periphytivorous catfishes.The large and diverse number of food items registered in the diet of Hypostomus species, in our data, with a predominance of detritus and periphyton, was expected (e.g., caSatti 2002, caSatti et al. 2009, cardone et al. 2006, delariva & aGoStinho 2001, Gomiero & BraGa 2008, meSchiatti & arcifa 2009).Detritus is a protein-rich organic material, and there is a large number of micro and macro organisms associated with the degradation processes of this material (fauSch et al. 2002(fauSch et al. , flecker et al. 2002(fauSch et al. , mcintyre et al. 2007)).is a complex community of algae, bacteria, fungi, protozoa, micro crustaceans, organic and inorganic material adhered to, or associated with, the substrate (JacoBSen 2008).
The foraging patterns evidence of resource sharing the dry and rainy periods, as observed values of niche overlap were higher than those expected by chance.The high consumption of detritus and also autochthonous items as and aquatic immature insects seems to the feeding overlap, and probably reflect their high availability in the aquatic environment, allowing the coexistence of these catfishes.The high degree of niche overlap was already registered among de-Table 4. Pair-wise tests based on PERMANOVA analysis, considering the significant factors in the use of food resources.P (perm) Probability.*Indicates the significant results.et al. 2016), and this feeding strategy requires complex anatomical-physiological adaptations in order to exploit detritus (Bowen 1983).The suckermouth-scraping feeding mechanism (Schaefer & lauder 1986)  Variations in resource utilization among Hypostomus species can be also explained by differences in how the species use their foraging habitats.These differences, in part, result from the different attachment mechanisms they display.These attachment mechanisms ensure that the individuals can maintain their position [while feeding] and result in the differential occupation of lotic systems by the various species (Buck & Sazima 1995, Garavello & Garavello 2004, PaGotto et al. 2011).In addition, the differences in the morphology and functional versatility (Schaefer & lauder 1986, luJan & armBruSter 2012) of the ventrally oriented feeding apparatus, which bear teeth that touch the substrate from which algae and other food items are scraped, are also relevant to resource partitioning, since these morphological differences result in different ways of obtaining food (delariva & aGoStinho 2001).
In summary, our data show that trophic resources available in the Corumbataí River are shared among five morphologically similar Hypostomus species, all specialized in surface-grazing foraging behaviour.The food resource partitioning between loricariid species may be interpreted as an adaptive response that results in a greater exploitation of the available resources, and differences in the diet between coexisting similar species are probably associated with morphological peculiarities of the feeding apparatus of each species.
and the long and coiled digestive tract (delariva & aGoStinho 2001) are remarkable functional traits that are essential for the success of loricariids in exploiting detritus and associated items (delariva & aGoStinho 2001, luJan et al. 2012).Our analyses also indicate variation in the consumption of some resources among Hypostomus species and seasons, and the contribution of periphytic algae was primarily responsible for such dissimilarity.Differences in periphyton composition, for example, depend on several environmental characteristics, such as the type of substrate (cattaneo et al. 1997), level of disturbance (BiGGS et al. 1998), hydrodynamics (alGarte et al. 2009), nutrient availability (moulton et al. 2010), light (hill 1996), and biological control by grazing (roSemond et al. 1993), affecting the availability of resource and consequently feeding overlap.In addition, factors such an increase in the sediment by rains increase the accumulation of particulate matter on with a consequent reduction in photosynthetic biomass and in the diversity and density of organisms (e.g.algae, vercellino & Bicudo 2006), with opposite effect the dry season (feliSBerto & rodriGueS 2005, JacoBSen 2008, martinS & fernandeS 2011).

Table 1 .
Number of individuals (N), standard length (SL) variation (range, mean and SD = standard deviation, mm), and total biomass (WG, kg) registered for the loricariids captured from the Corumbataí River in the dry and rainy seasons.

Table 3 .
Results of the two-way crossed PERMANOVA analysis on the use of food resources, considering the factors season and species.(df) Degree of freedom, (SS) Sum of squares, (MS) Mean sum of squares, (Pseudo-F) Statistic, P (perm) Probability.
the diets of Hypostomus sp. and H. ancistroides, H. commersoni, H. hermanni and H. stigatriceps resulted from the variable percentage contributions of filamentous algae, Chlorophyceae, Cyanophyceae, Euglena species, free-living nematodes, insects remains, and testate amoebae.Those items also contributed to the overall dissimilarity between the dry and rainy seasons for H. ancistroides, Hypostomus sp. and H. stigatriceps (Table