Food resource partitioning among species of Astyanax (Characiformes: Characidae) in the Lower Iguaçu River and tributaries, Brazil

ABSTRACT Resource partitioning allows for interspecific coexistence and is frequently reported for similar species. Here, we predicted the existence of resource partitioning among species of Astyanax that co-occur in the Low Iguaçu River and tributaries in Brazil. A total of 848 stomachs of five species of Astyanax were analyzed. Algae, terrestrial plant and fruit/seed were the most consumed resources. Astyanax bifasciatus and A. dissimilis had predominantly herbivorous diets, A. gymnodontus and A. lacustris were omnivorous, and A. minor was mainly algivorous. Permutational analysis of variance showed the species had different diets, and similarity percentage analysis indicated that fruit/seed and terrestrial plant contributed the most to this differentiation. A paired comparison indicated that the trophic breadth of A. gymnodontus differed from that of other species. The food overlap was low for 55% of Astyanax pairs. These results showed alignment with the niche theory, in which differentiation in the use of food resources facilitates the coexistence of species and minimizes competition. These adjustments to coexistence become relevant in the context of endemic species in a highly isolated basin under intense threat (dams, species introduction, deforestation, and pollution) as is the case for the Iguaçu River basin.


Introduction
Unlike terrestrial species or those with aerial or winddispersed life stages, species confined to freshwater environments generally cannot disperse from one catchment to another, unless corridors of aquatic habitats exist to connect them (Abell et al., 2008). Thus, the connections and disconnections between adjacent river basins have long been recognized as essential for the diversification of freshwater fish in the Americas (Buckup, 2011). In contrast, tectonic movements have in some cases established barriers to dispersal (vicariance) between adjacent river basins, 2 e190028 [2] allowing for further speciation and giving rise to endemic fish fauna (Cox et al., 1980;Ricklefs, Schluter, 1993;Albert et al., 2011). The Iguaçu River, a left-bank tributary of the Paraná River in Brazil, has been isolated from the main Paraná valley for approximately 22 million years by a geological lift, which gave rise to the Iguaçu Waterfalls (about 70 m in height; Maack, 1981). The geographic isolation promoted by this natural barrier has led to the occurrence of many endemic species, which represent more than 50% of the fish species in the Iguaçu River basin (Baumgartner et al., 2012). Threats to the integrity of this fish fauna diversity (dams, species introduction, deforestation, and pollution) (Baumgartner et al., 2012), coupled with the high endemism, were recognized by Abell et al. (2008), who included this basin as one of the 426 global freshwater ecoregions requiring urgent conservation planning efforts.
In this basin, Astyanax Baird, Girard 1854 stands out for its richness (Pavanelli, Oliveira, 2009), with 11 described species, of which 10 are endemic (Baumgartner et al., 2012). The occurrence of a large number of congeneric species living in sympatry raises a central ecological question, which is to understand how morpho-physiologically similar species minimize competition for resources (Chesson, 2000;Kylafis, Loreau, 2011;Juncos et al., 2015). According to the niche theory, when sympatric species overlap in the use of resources along one dimension, they must differentiate in another resource in order to coexist (Hutchinson, 1957). Species differ mainly in the use of habitat, food, and time (Schoener, 1974, Wiens et al., 2010Wang et al., 2015). This differentiation is called resource partitioning (Schoener, 1974) and enables interspecific coexistence (Ross, 1986). Moreover, resource partitioning is widely reported for congeners (Barreto, Aranha, 2006;Neves et al., 2018).
In this context, our study was guided by the following question: Is the coexistence of Astyanax species in the Iguaçu River basin favored by food resource partitioning? We predict that Astyanax species of the Low Iguaçu River consume different resources, presenting low food overlap. Thus, the aims of this study were to i) evaluate the food items used by Astyanax species that co-occur in the Low Iguaçu River; ii) investigate possible differences in the consumption of food resources and in the niche breadth; and iii) infer the degree of overlap among species.

Material and Methods
The Iguaçu River rises in the Serra do Mar and flows through a geological fault throughout Paraná State (Maack, 1981). The hydrographic basin covers the Paraná and the Santa Catarina states in Brazil and the Misiones region in Argentina (Dalla Corte et al., 2015). The study area comprised the Low Iguaçu River and its main tributaries in the stretch between the Salto Caxias dam and the Santo Antônio River mouth (Iguaçu National Park), where 25 sampling sites were established (the first sampling site: 25°35'17.04"S-53°29'56.58"W; the last sampling site: 25°35'17.16"S-53°59'25.20"W; Fig. 1).  Fish were collected monthly from September 2013 to August 2014, using gill nets (2.5-14.0 cm mesh) exposed for 24 h and inspected every six hours, as well as electrofishing in the morning. Caught fish were euthanized with 250 mg/L benzocaine [State University of Western Paraná (UNIOESTE) Ethics Committee-Protocol 62/09] and later fixed in 10% neutralized formaldehyde.
Fish were identified according to Garavello, Sampaio (2010) and Baumgartner et al. (2012), measured (total and standard length), weighed (g), eviscerated, and the stomachs were collected. All the analyzed individuals were considered adults based on macroscopic evaluation of gonadal development. Voucher specimens were deposited in the fish collection of the Zoology Museum/ State University of Londrina (MZUEL) (Tab. 1).
Stomach contents were analyzed using optical and stereo microscopes. The items were quantified according to the volumetric method (volume percentage of each item in relation to the total volume of the stomach contents) (Hyslop, 1980). The volume of each food item was determined using graduated test tubes; for volumes lower than 0.1 mL, a counting chamber was used (Hellawell, Abel, 1971). Food items were identified according to Bicudo, Bicudo (1970) for algae and McCafferty (1981), Borror, Delong (1988), and Mugnai et al. (2010) for invertebrates.
To identify possible differences among the diets of Astyanax species, a permutational analysis of variance (PERMANOVA, Bray-Curtis index) with 9999 permutations was applied to the volume matrix of food items. The data were transformed using the square root to reduce asymmetry and satisfy the premise of homogeneity. When significant differences were detected, pairwise tests using the pseudo t-statistic were applied to determine which species showed different diets. The food items that most contributed to distinguish the species pairs were verified using similarity percentage (SIMPER) analysis (Clarke, 1993).
Interspecific differences in diet breadth were evaluated by permutational multivariate analysis of dispersion (PERMDISP) (Anderson, 2006). This analysis was used to measure the distance from the centroid of a group defined a priori, in this case the species, and is calculated by principal co-ordinates analysis (PCoA). Calculation of the centroid of the group was performed using Bray-Curtis dissimilarity index values to compare the average dissimilarity among the n individual observations within the group. In this case, the mean distances of the samples in relation to the group mean (centroid) correspond to the niche breadth. In order to test the null hypothesis that the niche breadth did not differ among species, F-statistics were calculated using the Monte Carlo method with 9999 randomizations. The pairwise comparison by Tukey's honestly significant difference method was used to indicate differences in the interspecific variability of individuals of each species.
The diet overlap per sample was calculated based on the volume matrix of food items for each pair of Astyanax species at each site and month of collection using the Pianka index (Pianka, 1973) described by the following equation: where, Ojk = Pianka niche overlap between species j and k; Pij = proportion of food resource i in the diet of species j; Pik = proportion of food resource i in the diet of species k; n = total number of food resources.
The Pianka index ranges from 0 to 1 and indicates low (0.00-0.39), intermediate (0.40-0.60), or high overlap (0.60-1.00), where high overlap values may be indicative of competition or resource partitioning (modified from Grossman, 1986;Novakowski et al., 2008). The overlap index values were transformed to a percentage, ranging from 0 to 100.
We identified 40 food items consumed by the species that, according to their origin, were classified as allochthonous, autochthonous, and undetermined (Tab. 1). The food items were grouped into 10 categories: terrestrial plant, fruit/seed, terrestrial insect, terrestrial invertebrate, algae, aquatic plant, aquatic insect, fish, aquatic invertebrate, and detritus (Tab. 1).
Algae, terrestrial plant, fruit/seed, and detritus were the most consumed resources. Astyanax bifasciatus and A. gymnodontus foraged on 82.5% and 72.5% (33 and 29 items) of the 40 identified items, respectively. Astyanax lacustris and A. minor ingested 24 different food resources, and A. dissimilis consumed approximately 60% of the listed food items (23 out of 40) (Tab. 1).
Items of allochthonous and autochthonous origin appeared in different proportions in the diets of the fish species (Fig. 2). Allochthonous resources (fruit/seed) were predominant in the diet of A. lacustris (Tab. 1). On the other hand, autochthonous resources (algae) dominated the diet of A. minor (Fig. 2). The other species consumed similar proportions of both sources (Fig. 2)  Astyanax gymnodontus and A. lacustris exhibited an omnivorous diet in which fruit/seed was the main resources, but the species complemented this with different foods (detritus, fish, and algae for the former and detritus and terrestrial insect for the latter; Tab. 1). The other species showed an herbivorous diet. Astyanax bifasciatus predominantly consumed terrestrial plant and algae. Astyanax dissimilis principally consumed algae and fruit/seed while A. minor mainly consumed algae (Tab. 1).
The PERMANOVA (pseudo-F = 12.73; p < 0.01) indicated differences among the diets of the species pairs, except for the pair A. dissimilis × A. minor that showed similar diet preferences (pseudo-t statistic; p > 0.05) (Tab. 2). The observed dissimilarity, verified through SIMPER, indicated terrestrial plant and fruit/seed as the main food items that distinguished A. lacustris from the other species. In addition, the Rhodophyceae Compsopogon caeruleus was responsible for the dissimilarity in the diets of A. bifasciatus, A. dissimilis,A. gymnodontus,and A. minor (Tab. 2 Astyanax species showed average centroid distance values above 0.6 (Tab. 1; Fig. 3), indicating a wide trophic niche breadth. Astyanax gymnodontus presented the largest niche (0.65) (Tab.1; Fig. 3) significantly different from that of other species (PERMDISP, F = 3.78 p <0.01) (Tab. 3).
The diet overlap was low for 55% of the Astyanax pairs, especially those composed of A. lacustris, as well as the pairs A. bifasciatus × A. gymnodontus, A. gymnodontus × A. minor, and A. dissimilis × A. minor (Fig. 4).

Discussion
Our findings corroborated our prediction. Astyanax species consumed a variety of resources, exhibiting trophic segregation in co-occurrence. The observed consumption of foods of different trophic levels and origins (autochthonous and allochthonous) revealed a more generalist and opportunistic diet for A. gymnodontus and A. lacustris, while the other species had herbivorous dietary preferences. Astyanax is known for its opportunistic feeding habits (Lobón-Cerviá, Bennemann, 2000) and the use of different feeding strategies (Abelha et al., 2001(Abelha et al., , 2006Mazzoni et al., 2010) that enable the species to forage at almost all trophic levels (Abelha et al., 2006). Most of the previous dietary information about Astyanax species from Iguaçu River is from pre-and post-impoundment studies in which the trophic plasticity of this fish group is revealed by their opportunistic feeding behavior in response to environmental changes (see Cassemiro et al., 2002Cassemiro et al., , 2005Loureiro-Crippa, Hahn, 2006;Delariva et al., 2013;Mise et al., 2013).
This peculiarity explains the wide array of food items of different origins (autochthonous and allochthonous) explored by the studied species and corroborates their wide trophic niche. The riparian forest in the sample sites, especially inside the Iguaçu National Park (almost pristine), may play an essential role in the abundance of these distinct food resources. This vegetation is a source of allochthonous food, such as terrestrial plant, fruit/seed and terrestrial insect, besides being a substrate represented by trunks, branches, and other structures that fall into the river bed and are colonized by algae and aquatic invertebrates, increasing the availability of autochthonous food resources to fish fauna.
The expressive consumption of fruit/seed and/ or terrestrial plant by A. bifasciatus, A. gymnodontus, and A. lacustris and algae by A. dissimilis and A. minor emphasized the preference toward foods of plant origin. Consistent with our results, most Astyanax species were previously described as herbivorous: A. bifasciatus, A. dissimilis (Delariva et al., 2013;Mise et al., 2013), and A. gymnodontus in the pre-impoundment period of Salto Caxias Reservoir (Cassemiro et al., 2005;Delariva et al., 2013) and A. lacustris (Cassemiro et al., 2002(Cassemiro et al., , 2005Loureiro-Crippa, Hahn, 2006), whose preference for fruit/ seed was also reported by Andrian et al. (2001) and Silva et al. (2014). Additionally, A. minor was described as detritivorous in the Segredo Reservoir (Mise et al., 2013) and as herbivorous in lotic environments in the region of Salto Caxias Reservoir, showing a diet mainly composed of algae and plants (Delariva et al., 2013).
Although the species fed predominantly on terrestrial plant, fruit/seed and algae, the proportion of these main resources and those that were complementary (e.g., aquatic plant, terrestrial insect, detritus, and fish) varied among the different Astyanax species. These results highlight the ability of the species to differentiate their diets despite the usual tendency of Astyanax populations to adopt a generalist diet. The low values of food overlap may indicate trophic segregation among species (Corrêa et al., 2011). This is an expected result for similar species since the niche theory (Hutchinson, 1957) predicts that coexistence among species depends on a single difference or limiting similarity in order to minimize competition. Under these conditions, resource partitioning becomes an important mechanism for fish species segregation (Zaret, Rand, 1971;Edlund, Magnhagem, 1980), which is apparently operating among the populations of A. bifasciatus, A. gymnodontus, A. lacustris, A. dissimilis, and A. minor in the Low Iguaçu River. The similarity found in the diet of the pair A. minor × A. dissimilis was the only counterpoint to this result, but it does not discard ecological separation between these species since we investigated only one niche dimension. Usually congeneric species inhabiting the same area avoid competition through three possible ecological differences: they exploit different habitats or microhabitats, they consume different foods, or they are active at different times (Pianka, 2000). Another point to consider is that the similarity in the diet was mainly influenced by the high consumption of the alga Compsopogon caeruleus. Although the availability of food resources was not measured in the sampled sites, there must be high algal abundance in the sites owing to the rocky consistency of most of the sampled lotic stretches of the Iguaçu River and its tributaries, which may provide substrates for algal colonization, thus, reducing food limitations for A. minor × A. dissimilis.
Despite the low values, the overlap of terrestrial plant, fruit/seed and Compsopogon caeruleus reinforced the importance of these food items in maintaining the studied Astyanax populations. This information gains relevance considering the landscape threats over the Iguaçu basin (Baumgartner et al., 2012;Kliemann et al., 2018), especially by dam construction and removal or degradation of the riparian vegetation. The loss of habitat heterogeneity in the conversion of lotic environments to reservoirs (Agostinho et al., 2007) and the limitation of allochthonous resources can be decisive for the interactions among species, with increasing competition and possible displacement of species with low fitness.
In conclusion, the wide range of food resources consumed by the evaluated Astyanax species demonstrates the potential of these species to exploit all environmental compartments. However, differences in the proportions and main resources consumed by each species indicate that trophic segregation promotes the co-occurrence of Astyanax species in the Low Iguaçu River.