The use of an Atlantic Forest stream by the catfish Scleromystax barbatus ( Quoy & Gaimard , 1824 ) Uso de riacho costeiro da Mata Atlântica pelo limpa-fundo Scleromystax barbatus ( Quoy & Gaimard , 1824 )

Habitat used by fish is often related with the requirements of the species, according to their morphological characteristics, feeding, and reproductive strategies. We describe the habitat preference, foraging behavior, and camouflage strategy of Scleromystax barbatus (Callichthyidae: Corydoradinae), an endemic catfish from coastal streams in the Atlantic Forest. The movement and foraging behavior of 11 individuals were observed using focalfish method, and the habitat structure availability (i.e., depth, bottom type, and mesohabitat) was measured in a clearwater stream stretch with 40 m length and 2.8 m width (± 1.5 sd) in southeastern Brazil. In addition, six individuals were captured to verify stomach contents. Habitat selectivity of the species was assessed using the Ivlev electivity index that considers the frequency of occurrence of each habitat variable used by fish in relation to its environmental availability. Scleromystax barbatus selected pools and runs with sandy bottom in depths between 21 cm and 40 cm, and avoided gravel in shallower riffles (up to 10 cm). Individuals foraged close to the bottom in a head-down posture, speculating and revolving the substrate composed by sand, fallen leaves, and twigs. Their long snout and small barbels were used to dislodge the food items of the substrate, which were quickly sucked by the mouth. Aquatic immature insects (mainly trichopterans) were predominant in the stomach contents. Three fish species, Deuterodon iguape Eigenmann, 1907, Mimagoniates microlepis Steindachner, 1877, and Geophagus brasiliensis (Quoy & Gaimard, 1824), were firstly registered following S. barbatus during its foraging activity. The brilliant golden body color with darkish spots of S. barbatus strongly resembles the stream bottom, which indicates disruptive camouflage.


Introduction
Habitat used by fish is often related with the requirements of the species according to their feeding (Sabino and Castro, 1990;Aranha et al., 1998), reproductive habits (Aranha et al., 1993), and morphological characteristics (Teixeira, 1989;Braga, 2004;Mazzoni et al., 2011).Moreover, spatial resources (Gorman and Karr, 1978;Sheldon, 1968) and microhabitat availability (Carvalho et al., 2006;Sazima et al., 2006;Zuanon et al., 2006) are also important to comprehend how fishes use habitats and how they are spatially distributed.In coastal streams of the Brazilian Atlantic Forest, studies have revealed fish species' preferences on the use of habitats.For instance, Gymnotidae and Synbranchidae species prefer vegetated sites near stream margins (Teixeira, 1989), while Characidae species are generally found near marginal vegetation with lentic or lotic waters (Sabino and Castro, 1990), and Loricariidae and Crenuchidae species are found at rocky lotic waters (Aranha et al., 1998;Barreto and Aranha, 2005).The knowledge of habitat preferences and their different uses by freshwater streams fishes are of stronger interest to conservation due to the increase of deleterious human activities in such environments (Mazzoni et al., 2011).The genus Scleromystax Günther (Siluriformes, Callichthyidae, Corydoradinae) comprises four endemic species that inhabit small tributaries of Atlantic Forest coastal drainages: S. barbatus (Quoy & Gaimard, 1824), S. macropterus (Regan, 1913), S. prionotos (Nijssen & Isbrücker, 1980), and S. salmacis Britto &Reis, 2005 (Britto andReis, 2005).Scleromystax barbatus is a small air-breathing callichthyid catfish that inhabits clearwaters (Gonçalves and Braga, 2013).Actually, it is subject to commercial interest by the aquarium trade industry (Reis, 2003) and few studies investigated its behavior in backwaters of shallow streams (Aranha et al., 1998;Esteves and Lobón-Cerviá, 2001;Barreto and Aranha, 2005).In the present paper, we studied the use of a clearwater coastal stream by S. barbatus in the Atlantic Forest, southeastern Brazil.We evaluated the species' habitat selectivity, its feeding behavior, and social interaction.

Study area
The study was conducted in Boguçá stream (24°28'54.71''Sand 47°13'02.05''W)at Juréia-Itatins reserve, an undisturbed protected area in pristine conditions of Atlantic F orest in São Paulo state, southeastern Brazil.The climate is subtropical humid without a pronounced dry season.The hotter and rainier season occurs from October to April and the lesser pronounced wet season occurs from May to September.The average annual rainfall and temperature are 2,277 mm and 21.4 ºC, respectively (Marques and Duleba, 2004).The main river of Juréia-Itatins reserve is Una do Prelado, a lowland and medium-sized river (100 m length) which runs parallel to the seashore and surrounds the Juréia mountain range (Por, 1986(Por, , 2004)).Boguçá is a 2 nd order clearwater stream (considering a 1:50.000scale map), tributary of the Una do Prelado River.In a previous survey (February 8 th , 2010) using electrofishing gear, we recorded, in decreasing order of abundance, 12 fish species at the studied site: Mimagoniates microlepis (Steindachner, 1877) (DZSJRP 13236, 13238, 13242-45, 13249-50, 13253, 13255, 13258).
The use of an Atlantic Forest stream by the catfi sh Scleromystax barbatus (Quoy & Gaimard, 1824)

Sampling procedures and data analysis
Data collection was conducted on February 24-25 th , 2011 in a stretch of 40 m length and 2.8 m width (± 1.5 sd) in the Boguçá stream.We used the focal-fish method (see Lehner, 1998) with observers positioned outside the water at the stream borders.The clearwaters of Boguçá stream provide sufficient transparency to observe fish behavior.We did not conduct underwater observations due to shallowness of the stream.The behavior of 11 individuals of S. barbatus was observed with naked eye on diurnal periods (from 09 a.m. to 03 p.m.).Individuals were also photographed and video-taped after observations.Each individual was observed interruptedly during one-hour (whenever possible), and the following variables were recorded in samplings conducted every five minutes: fish activity (foraging, moving without forage, or inactive), mesohabitat and bottom type used (see sampling description of the stream habitat structure below), fish position in the water column (surface, midwater, or bottom), and its social interaction (grouped or solitary).
At the end of observations, six individuals of S. barbatus were collected and immediately fixed in 10% formalin.This number was arbitrarily defined to avoid impact on local population, since two visual censuses conducted at the same time by three observers revealed 33 ± 5 individuals on the stream stretch.Caught specimens were measured to obtain the standard length (SL, in mm).Stomach contents were analyzed to verify fish's diet.Food items were identified up to the lowest taxonomic level possible under a stereomicroscope, according to Borror and Delong (1969).
The stream habitat structure was evaluated in 40 linear transects.For each one meter of each transect, we registered the following variables: stream depth (cm), predominant bot-tom type (clay: < 0.05 mm, sand: 0.05-2 mm, gravel: 2-10 mm, and plant debris), and mesohabitat type (pool, run, and riffle).These measures were realized at the end of the fish observations to avoid the habitat structure disturbance.The selectivity of S. barbatus on habitat variables, such as mesohabitat, bottom type, and position on water column, was evaluated by Ivlev electivity index (E), according to: E = (r i -P i )/ (r i + P i ), where r i is the frequency of occurrence of each habitat variable i used by fishes, and P i is the frequency of occurrence of each habitat variable i available on the stream stretch.This index varies from -1 (non-selective) to +1 (selective), and electivity is absent when E = 0 (Ivlev, 1961).

General behavior and habitat use
We observed 11 individuals of S. barbatus in 615 min (55.9 ± 12.6 min per individual), which resulted in 134 samplings.The species was more frequently registered in an active way (N = 105 or 78.4% of samplings), than in an inactive one (N = 29 or 21.6 % of samplings).When active, S. barbatus was foraging (N = 99 or 76.7% of samplings), and rarely moved without foraging on the stream (N = 5 or 3.9%).Active and inactive individuals were recorded both solitary (N = 96 or 70.1% of samplings) and grouped with conspecific (2-6 individuals, mean = 2.5 ± 0.9 sd; N = 41 or 29.9% of samplings).Agonistic behavior was rarely recorded; only one individual demonstrated agonistic behavior with a conspecific in one sampling.Scleromystax barbatus specimens were most frequent at the stream bottom (N = 126 or 94% of samplings) and were rarely recorded in the middle of the water column (N = 8 or 6% of samplings).Scleromystax barbatus individuals were registered using almost all avail-able habitats (Table 1), but this species preferred depths between 11 cm and 50 cm, clay and sandy bottoms, pools and runs.Depths up to 10 cm, riffles, gravel, and plant debris bottom were avoided (Table 1).
The brilliant golden body color with darkish spots of S. barbatus resembles the stream bottom coloration (Figure 1).This color resemblance allied to a stationary position of individuals in threatening situations, i.e. when we approached them with an object (piece of branch), indicate disruptive camouflage.

Diet and feeding behavior
The six individuals collected ranged from 37 to 58 mm SL.All specimens were females with full stomachs.Aquatic immature insects of autochthonous origin, mainly trichopterans, were predominant in the stomach contents (Table 2).Sand grains occurred in all stomachs, and unicellular algae (Bacillariophyta and Chlorophyceae) were also registered.Scleromystax barbatus was observed foraging close to the bottom speculating the substrate (N = 95 or 96.9% of samplings), or picking up food items in drift (N = 3 or 3.1% of samplings).Individuals assumed a head-down position at stream bottom supported by the ossified spines of the pectoral fins.
During substrate revolving (mainly sandy bottom with plant debris -composed by fallen leaves and twigs), they used the long snout and barbels assuming the head-down posture when food items were quickly sucked by the mouth.Substrate clouds frequently resulted from foraging movements.

Discussion
Freshwater organisms may present strong preferences for local habitat variables (Lamouroux et al., 1999) as verified to macroinvertebrates (Orth and Maughan, 1983), and fish (Grossman and Freeman, 1987;Robertson and Crook, 1999).Depth and stream flow (Buck and Sazima, 1995), temperature, dissolved oxygen, turbidity and conductivity (Fialho et al., 2008), and bottom type (Mazzoni et al., 2011) are environmental variables often related with habitat preferences by fish.In this study, S. barbatus selected pools and runs with sandy bottom and avoided gravels mostly in shallower riffles.The only one study about S. barbatus reported their habitat preference for lentic sites, however with predominance of plant debris in the bottom (Aranha et al., 1998).
Our results indicated an active diurnal foraging of S. barbatus in addition to the usually crepuscular and/or nocturnal habits registered among Callichthyidae (Paxton, 1997).Most individuals of S. barbatus were registered foraging in an active way, i.e. hunting by speculation (sensu Curio, 1976) or using grubber excavation while moving (sensu Sazima, 1986), without signals of territoriality provoked by agonistic encounters.Scleromystax barbatus fed mainly on aquatic immature insects, mostly trichopterans buried in the sand.The presence of sand grains and unicellular algae in S. barbatus stomach contents is a consequence of the indirect ingestion of these items during foraging (Casatti et al., 2009).Grubber excavation while moving also creates possibilities for S. barbatus to act as a nuclear species to diurnal species, such as Deuterodon iguape, M. microlepis, and G. brasiliensis.When foraging in the bottom, S. barbatus induces substrate clouds and the follower species took the advantage to feed on exposed items.Callichthyidae are potential species to perform the nuclear role due to their bottom-forage habit, however, as far as we know, no other study recorded S. barbatus as nuclear species in the Atlantic Forest.In a unique record to date, Sazima (1986) found Astyanax bimaculatus following Corydoras polystictus.
Camouflage associated with disruptive colors is one of the most displayed defense mechanism by bottom species (Carvalho et al., 2006).
Regarding the resemblance of the golden-and-black body color of S. barbatus with the color of the most available bottoms in the stream stretch, i.e. sand and plant debris, we suggest that this species strongly relies on disruptive camouflage during its foraging activity.In risky situations, individuals ceased the foraging activity and became inactive, 'mixing' their body with the stream bottom.Fishes returned to forage when disturbing conditions ceased.According to Sazima et al. (2006), disruptive camouflage is related to defense against visually oriented predators, mainly piscivorous fishes.Further, cryptic and sand-color patterns in fishes are advantageous for evasion from predators (Zuanon et al., 2006).The advance of deforestation in the Brazilian Atlantic rainforest may cause drastic negative effects to the endemic fish fauna (Nogueira et al., 2010).We expect that our results concerning habitat preferences and behavior of S. barbatus in a pristine area of the Atlantic Forest serve as comparative data for future conservationist studies and management activities.

Figure 1 .
Figure 1.Disruptive camouflage of Scleromystax barbatus in a clearwater coastal stream of Brazilian Atlantic rainforest.The photo was taken during the study in their natural habitat, without flash.The arrow indicates an inactive individual of S. barbatus.

Table 1 .
Relative frequency (%) of depth, bottom type, and available mesohabitat categories and mesohabitat categories used by Scleromystax barbatus, indicating species selectivity according to Ivlev electivity index (E).

Table 2 .
Abundance (N) and frequency of occurrence (FO%) of food items recorded in stomach contents of Scleromystax barbatus (37 mm -58 mm SL) from Boguçá stream, southeastern Brazil.