Rev. Biol. Trop. 46(4): 951-959, 1998 www.ucr.ac.cr Www.ots.ac.cr www.ots.duke.edu

Habita! use and food partitioning of the fishes in a coastal stream of Atlantic Forest, Brazil

J. M. R. Aranha, D. F. Takeuti & T. M. Yoshimura

Depto Zoologia, Setor de Ciencias Biológicas, Universidade Federal do Paraná, CP 19.020, CEP 81531-990, Curitiba, Brasil. Fax: 55 (041)266-2042, E-mail: [email protected]

Received 5-II-1998. Corrected22-VI-1998. Accepted 13-VIII-1998.

Abstrad: We analysed the fish assemblage in the "Mergulhiio"stream (soulhem Brazil) with underwater obser­ vations for habitat use, considering water depth, current velocity, bottom type, shadow froID vegetation cover, distance of stream-edge, and vertical position. Stomach contents or foregut content samples of the most abun­ dant were collected from 26 species (lO families). The fish assemblage occupied the bottom stream. The similarity analysis of spalía! occupation ofspecies grouped four habitat use guilds: A) "Iambaris" (Aslyanaxsp. and Deutetodonlangei), Characidium spp. (C. lane; and C. plerosliclum) and Rineloricaria kronei used thebot­ tom in deep sites and waters with middle current; B) Pimelodella pappenheimi and Corydoras barbalusused the bottom in siles with JOwer cUI'l"ént; C) Mimagoniates microlepis used the surface of the water column; and D) Phalloceros caudimacullUus used shallow sítes and waters without current. Species with few records were analysed descriptively. Diet similarity suggested seventrophic guilds: Microglanis sp. and PiTMIodellapappen­ heimi: omnivorouslca:mivorous gui14; Corydoras barbatus: omnivorous/insectivorous guild; Characidium [ane;: aquatic insectivorous guild, Ilillinly aquatic insects; Mimagoniates microlepis: terrestriál insectivorous guild, mainly te.rrestrial insects; Deuterodon [angei and ASlyanax sp.: omnivorous/herbivorous guild; Rineloricaria kronei, Kronichth.yssubteres, . leucofrenatus and obtusa: herbivÓfous guild; and PhaUoceros caudimaculatus: a!givorous guild. When the guilds were similar, the species were geÍletalists in diet and in habitat use.

Key wol'ds: Habitat IIse, feeding, resource partitioning, AtIantic Forest, coastal stream, Brazil.

Neotropical freshwater fishes are very would reduce the importance of the competi­ diverse and fish assemblagesare quite rich in tive relationships (e.g. Connor & Simberloff this regíon. So the interspecific relationships 1979, Strong 1980, Grossman et al. 1982, are intricate in these communities (Lowe­ Wiens 1977). McConneU 1987). The resource partitioning in a community Sevetal researchers believe that competi­ may be important to the knowledge of the tion would bean important ecological and evo­ nature of these interrelationships (Schoener lutionary factor (e.g.· Schoener 1974, Yant et 1974). The most important dimensions of al. 1984). Nevertheless, others believe that the resource partitioning for fishassemblages are environmental instahility of the abiotic factors the habitat and trophic segregation (Ross 952 REVISTA DE BIOLOGIA TROPICAL

1986), being the habitat segregation most large rock (more than 25 cm); 7, litter and 8, important in freshwater than it is in marine trunkand branehes. The shadow of the vegeta­ environments (Horn 1972). tion cover was recorded by the presence (1) or In Brazil, studies of microhabitat and/or absence (O) of the vegetation itself or branches trophic strueture of freshwater fish assem­ over the point-sample. blages have been increasing during the last few years, but few studies have been eonducted in eoastal streams (e.g. Costa 1987, Teixeira Direction of Observation 1989, Sabino & Castro 1990). In this paper, the microhabitat use and RightEdge trophic strueture in guilds of the fishes were analysed in a eoastal stream of the Atlantic Forest in Paraná State.

MATERIALSAND METHODS 3

The "Mergulhiio" stream is a second order 5m LeftEdge stream Atlantic Forest eoastal stream in Antonina municipality (Paraná S tate, Brazil; F10w water 2S017'S, 48°44'W), with very clear water, which allows underwater observations. In the Fig. l. Location of tbe sample-points in transects for mea­ studied area the contiguous vegetation is eom­ surement of variables of habitats and observation of habi­ posed of trees and shrubs on the left edge and tat use. The grid has ten transects at 5 m intervals. herbaeeous plants on the right. The average depth is one meter, and the average distanee between the edges is 5m. This researehwas eondueted in daylight in The underwater observations were done September and Oetober 1993, and January and with mask and snorkel during 6 min in each Mareh 1994. point-sample, where we noted the number of A grid with ten transeets (adapted from fishes of eaeh species, their distance from the Gorman & Karr 1978; Fig. 1) with 80-130 water's edge and their vertical position. The point-samples was used for measuring four distance from the water's edge was classified habitat variables and for observation of the in three categories: O, nearest to right edge; 1, fishes. The habitat variable measurements mid-stream and 2, nearest to left edge. The were the depth of water, eurrent velocity, bot­ vertical position was categorized as 1 (upper tom type and shadow of vegetation eover. The third of the water column), 2 (middle third), 3 depth was measured in eentimeters and classi­ (lower third) and 4 (bottom). fied in eategories of ten of eentimeters (1= 0- During the underwater observation, the 10 em to 9= 80-90 cm). The current velocity feeding behaviour of the species was noted, was measured with a Pitot tube at two cen­ and grouped using the same foraging tactic timeters under the surface and classified in categories deseribed by Sazima (1986). four categories: 0, imperceptible; 1, slow After the observation, fishes were collect­ (0.00-0.25m.s-l); 2, moderate (0.25-0.50 m.s-I); ed with fishing-nets, casting nets and sieves to and 3, fast (0.50-0.75 m.s-l). The substrate or study the fish assemblage and their dieto bottom type was classified in eight categoríes: The stomach or samples of the foregut 1, sand; 2, gravel; 3, pebble; 4, stone (less than contents of the most abundant collected 10 cm of diameter); 5, rock (10-25 cm); 6, species (when individuals were over ten) were ARANHA el al.: Habitat use aud food partitioningof the fishesin a coastal stream 953

examined, under estereoscopic microscope The habitat variables that prevails in the and optica! microscope, to assign the feeding "Mergulhao" stream were the sand-type sub­ guilds by analysis by the occurrence methods, strate(category 1), current velocity ti11 0.5 m.s-1 which the percent composition of each food (categories 1 and 2) and depths between 40 and item in the total number of the items found 60 cm (categories 5 and 6). (Hyslop 1980). The significant differences in the habitat use, by species and by assemblage, in each Data Analysis: Chi-square goodness-of­ environmental variable category and íts mainly fit-test was used to compare frequency distrib­ occurrence is showed in Table 2. The habitat utions of habitat use in relation to the avail­ use by the fish assemblage was significant to ability of each resource. Calculations of vertical position and the tendency to occupy the expected frequencies were based on the envi­ bottom was corroborated. "Lambaris" tended to ronmental features and the sample sizes were occupy the bottom of deep stretches with mod­ standardized to a total of 100, as suggested in erate and fast current velocity. Mimagoniates Gorman (1988). For this purpose, we used the microlepis preferred siles in the left edge with data for each species and for all species togeth­ litter and trunklbranch substrate type, water ero AH species analysis provided an assessment surface and shaded areas. Characidium spp. of how the assemblage used the available preferred water with current more than 0.25 resources. mJs in deeper sites and occupied the bottom in The similarity in the habitat use by species unshaded areas. Pimelodella pappenheimi was estimated by the Morisita's measure occupied the stream bottom in areaswith sand, (Smith & Zaret 1982). A cluster analysis with gravel and litter substratetypes, water with cur­ UPGMA method (Romesburg 1990) was rent untilI 0.25 m.s-l in unshaded areas with applied using the similarity data. The similari­ moderate depths. Corydoras barbatus occupied ty of the percent composition of each food item almost exclusively the stream bottom mainly was ca1culated using the Morisita's measure the right edge, in litter and trunklbranch sub­ modified by Horn (Smith & Zaret 1982) and strate type and waters whose current was lower the results were grouped by UPGMA method than 0.25 m.s-l. Rineloncaria kromei occurred (Romesburg 1990) to determine the feeding in the right edge in unshaded deeper areas with guilds. sand substrate type and water currents moder­ ate and fast. Phalloceros caudimaculatus occu­ pied shallow and shaded areas with litter sub­ RESULTS strate type, and waters with slow current in the right edge of the stream. We collected 26 species belonging to 10 It was not possible to do the x2 test in the families in the "Mergulhao" stream (Table 1). vertical position variable for Characidium The underwater observations totalized 14 spp., Corydoras barbatus and Rineloricaria hours and 2462 observation records of kronei because these species used almost Mimagoniates microlepis, Phalloceros caudi­ exclusively the stream bottom. maculatus, Corydoras barbatus, Pimelodella The similarity analysis of spatial occupa­ pappenheimi, Rineloncaria kronei, Astyanax tion of species allowed us to determine four sp., Deuterodon langei, Characidium lanei and groups: "lambaris", Characidium spp. and C. pterostictum. The accurate specific differen­ Rineloncaria kronei (group A), used the bot­ tiation of Astyanax sp. and Deuterodon langei tom in deep sites and waters with current more was impossible to do during the observations than 0.25 m.s-l; Pimelodella pappenheimi and and they were treated together and designated Corydoras barbatus (group B), occurred the here by "lambaris". Likewise, C. lanei and C. bottom in sites with lower current (until 0.25 pterostictum were treated as Characidium spp. m.s-l) and litter substrate type; Phalloceros 954 REVISTA DE BIOLOGIA TROPICAL

TABLE 1

Fish species collected and!or observed in the "MergulhJo" stream during the study. and the codes used in the resulls of microhabitat or diet

Species Microhabitat codes Diet codes

Family Erythrinidae Hoplias malabaricus (Bloch. 1794) Family Characidae Characidium lanei Travassos, 1967 Cha Ch! Characidiumplerostictum Gomes, 1947 Cha Oligosarcus hepsetus (Cuvier, 1817) Mimagoniates microlepis (Steindachner,1876) Mim Mim Astyanaxsp, Ast Ast Deulerodon langei Travassos. 1957 Deu Deu Family Pirnelodidae Microglanis sp,n * Mig Mig Heptapteros mustelinus VaIenciennes, 1840 Pimelodella pappenheimi Ahl, 1928 Pim Pim Rhamdia que len (Quoy & Gairnard, 1824) Family Trichornycteridae Homodiaetus sp.n * Family Callichthyidae Corydoras barbatus (Quoy & Gairnard, 1824) Cor Cor Family Hemipsilichthys sp.n * subte res Ribeiro, 1908 Kro Kro Hisonotus leurofrenatus (Ribeiro, 1908) His His (Ribeiro, 1911) Pse Pse Schizolecis guntheri Britski & Garavelo, 1984 Sch Sch Rineloricaria kronei (Ribeiro, 1911) Rhi Rhi Ancistrus sp. Family Gyrnnotidae Gymnotus pantherinus Steindachner, 1908 Family Poeciliidae Phalloceros caudimaclllatus (Hensel, 1868) Pha Pha Farnily Cichlidae Cichlasoma facetum (Jenyns, 1842) Crenicichla sp.n * Geophagus brasiliensis (Quoy & Gairnard, 1824) Family Synbranchidae Synbranchlls marmoratus Bloch, 1795

"*,, new species (hat are being described.

caudimaculatus (group C) occupied shallow Kroníchthys subtáes and Microglanis sp. sites and waters with little current; and were not seen during underwater observations, Mimagoniates microlepis (group D), the sur­ However, during the samplings, they always face of water column, occupied microhabitats fonned by branches, Species with few records were not leaves and other accumulated materials in fallen analysed statistícally and their results were branches in the water, which suggests their just descriptive, preference for these microhabitats, Geophagus brasiliensis occurred in the Furthermore, there were few individuals, in left edge of the stream, upon litter and trunks the right margin of the stream in the middle of and branches, in shaded areas, the grass, Hoplias malabaricus, Gymnotus ARANHA et al.: Habita! use and food partitioning of the ftshes in a eoastal stream 955

TABLE2

The significant differences in fhe IUlbitat use, byspecies and by assembliige, in each environmental variable category, compared with the,environmental availability, and its mainly occurrence

' Fish assemblage "Lainbaris" Categories x2 Mainly occurrence x2 Mainly occurrence Depth * Deepsites (9) Substrate Curren! Fáster (2, 3) Shadow Edgedistance Vertical position " Bottom(4) * Bottom (4)

Mimagoniates microlepis Characidium spp. Categories x2 Mainlyoccurrence x2 Mainly occurrence Depth * Deepsiles (7, 8, 9) Substrate Litter (7); trul1k and branches"('8) Currenl * Faster 2,( 3) Shadow *' Preserice (l) * Al,Isence (2) Edge distance lO Leftedge (2) .. No occuped die leftedge (2) Vertical position " Upper (1) n/p Bottom (4)

PimáHiella pappenheimi éG/ydotasbarba tus Categories x2 Maillly oceurrence x2 Main1y océurrence .. Depth Moderate deep (5, 6, 7) .. Moderate (5, 6) aná deep (S, 9) " " Substrate Sand (1), grave! (2), Ií!ter (7) Liner (7); trul'lk andbranches' (8) Curren! *. SlóW (l) '" Slow (1) Shadow " Ausence (2) Edge distance .. Rigl1tedge (O) Vertical position .. Bortoro (3,4) n/p Bottoro (4)

Rineloricaria kronei Phalloceros c�udimacuiatus Categories x2 Mainly OCCllrrence x2 Main!"occurreoce Depth * Deep siles (7, 8, 9) .. Shallow siles (3, 4) Substrate .. Occurred: sand (1» * Occutred: litter (7) Current * Faster (2, 3) .. Very Slow (0, 1) Shadow * Ausence (2) * Presence (l) Edgedistance * Ríght edge (O) .. Right edge (O) .. Vertical position n/p BOlloro (4) Bottoro (4)

"*,, denotes distributions signiflcantly different ffomexpected (a = 0.01). "n/p"denotes thal it was notpossible to use chi-square test.

pantherinus and Synbranchus mannoratus pre­ Homodiaetus sp. was collected in a small ferred places next to the riparian vegetation. area ofrapids ofilie stream. Hisonotus leucofrenatl,ls, Pseudotothyris obtusa and Schizolecis guntheri had few Féeding Behavior: The foraging tactic of records during the observations. Nevertheless, Mimagoniates micro/epis was considered as they occupied the bottom, in depths varying "smface picker", picking up floating organ­ from 40.5 to 88.0 cm and preferred the edges isms in the surface; Characidium spp. were of the stream. considered as "sit-and-wait predators", seden­ Crenicichla sp. and Oligosarcus hepsetus tary fishes that ambush prey. Corydoras bar­ were observed in deep microhabitats in the batus and Rineloricaria kronei were consid­ middle of the water column: ered as "grubers", excavating the substrate 958 REVISTADE BIOLOGJATROPICAL

and the characteristic "shallow environment" Herbivorous guild occupied the bottolÍl can have distin�tmagnitude. arid the edge of the stream. except K.subleres P. caudimat;ulatus preferred to occupy the (which used a microhabitat sim.ilar to shallow edges"of the stream almost without Microglanis sp.) ana Rini!loricaria kronei ' ' current. These results corifirm the description (which occupiéd the bottom in deep sites in made by Teixeira (1989), Sabino & Castro unshaded areas). - (1990) and Aranha & Carama�(lhi(in' prep.) P. caudimacúliitus was a specialhlt in the about the distribution of this sp�cies in coastal habitat use and food. It constituted a spatial streams in Rio Grande do Sul, Sao PauIo and and a trophic guild ,?yi tself. Rio de Janeiro, respectiVe}:y. '" These results did lÍot elucidate too theoret­ The micr.ohabitats occupied by "lam­ ical question if strean'I fish' assemblages are baris", Characidium spp. and Pimélodella organized by deterministic processeS'such as pappenheimi were similar to those indicated competition. However, (he fish assertlblage of by Sabino & Castro (1990) respectively for the "Mergulhao" stream does not have a ran­ Deuterodon iguape, Characidium sp. and dom pattem of the habitat use, ID opposition to Rhamdella minuta. These pairs of species the random organizatibn who'se sóme authors belonging to the sarne subfamilies, and may believe in (e. g. Matthews 1982). execute similar functipDs as a reflex of the We conclude that the differences'in the common evolutionaryhistory of these groups. spatial andtropbié.&uUds in the"Mergulhao" Loricariidae an� Callichthyiq� ¡studied stream makepossible tire segreption of these fishes occupied rnainly the bottom of the resources� Ir the comPetition is an important stream and the edge with vegetation. These ecologic factor in this stteam. t'be spatiál and results agree with those frorn Teixeira (1989) food partitioning stiouJda uanmaee the·coexis­ for sorne loricariids and fo,r Corydoras palea" tence of toose species iD lhC "Mergulhiio" tus, and with .those from·Aranha et al. (1993) stream. for two others Corydoras species. Probably. this preferential microhabitat is characteristic · . in these groups. The study óf st�h contents indicated high frequency of algáe "mostin oíthe species. The auth6rs thartk'S. Mátcia de Menezes Probably it is due to the availability of this who belped us dútingsamptita,g; MárciaS. de .' resource in the "Mergulhao" stream: Menezes ánd Vtt¡íma S� Uiedafor readillg thé The collected nuinber: of'Hoplias mal­ manuscripl; Carlos S. LüCéJ'laahd the gtaffof abaricus and Oligosa7J;us hepsetus was quite ichthyologistsof the PUCRsfór identifying srnall and the stomach contents of these several fjSh· speciés; Jtdio· C. Garavelo species were not sttidied.· In tbe literature. (UFSCar) for iden:tifying ; these species have been appóillted as piscivo­ Francisco Langeani (llNESP/Slo losé do Rio rous (e.g. Costa 1987. Teixeira 1989). We sup­ Preto) for identifying Rinelbricaria ánd the pose that these species. together with CNPq for tbe' scholarships to the junior Synbranchus matmoratus, constitute a pisciv­ authors. orous guild in the "Mergulhiio" stream. The comparative analysis of the trophic and habitat use guilds showed that the group­ REFERENcEs ing of species was different, suggesting resource partitioning by the species. Aranha, J.M.R., E.P. Carafuaschi & U. Carámásóbi. 1993. Ocupa\;oo espacial, aliínéi)�1ee ép0Ca reprodutiva D. langei and Astyanax sp. weregeneralist de duali espécies de. COtyd01MS (Siluroi dei, Ca ­ in the food preference (OIhnivorouslherbivo­ lÍichthyidae)coexistentes l'IO·riQ �bari(Botucatu, ' js guild) and in the habitat use too .. , 'SP). Rev: bras: hoollÓ: 453-466. ARANHA et al.: Habitat use and food partitioning of the fishes in a cO¡istal stream 959

Connor, E.F. & D.S. Sjmberloff: 1979. The assembly of Matthews, WJ. 1982. SmaU fish community structure in species communities: change or cpmpetition? Eco­ Ozark streams: Structured assembly patterns or ran­ logy 60: 1132 - 1140. dom abundance of species? Amer. MidL Nat. 107: 42-54. Costa, WJ.E.M. 1984. Peixes fluviais do sistema lagunar de Maricá, Río de Jan¡;iro, Brasil. Atlantica, Rio Romesburg, H.C. 1990. Cluster Analysis for Researches. Grande 7: 65-72. Robert E. Krieger, 334p.

Costa, W.JEM. 1987. Fecqing habitats of a fish commu­ Ross, S.T. 1986. Resource partitioning in fish assembla­ nity in a tropical coastal stream,. Rio Mato Grosso, ges: a review of fieldstudies. Copeia 1986: 352-388. Brazil. Stud. Neotrop. Fauna Environ. 22: 145-153. Sabino, J. & R. M. C. Castro. 1990. Alimenta\(ao, período Gorman, O.T. 1988.The dynamicsof habitat use in a guild de atividade e distribui�ao espacial dos peixes de um of Ozark minnows. Ecal. M(lllOg. 58: 1-18. riacho da Floresta Atlantica (Sudeste do Brasil). Rev. brasiL BioL 50: 23-36. Gorman, O.T: & J.R. Karr. 1978. Habitat strncture and stream fish comrnunities. Ecology 59: 507-515. Sazima, 1. 1986. Similarities in feeding behaviour between sorne marine and freshwater fishes in two tropical Grossman, G.D. & A.'dc Sostoa. 1994. Microhabitat use communities. J. Fish BioL 29: 53-65. by fish in the lower Río Matarraña, Spain, 1984- 1987. Ecol. Freshw.·Fish 3: 123-136. Schoener, TW. 1974. Resource partitioning in ecological communities. Science 185: 27-38. Grossman, G.D., P.B. Moyle & J.O. Whitaker Jr. 1982. Stochascity in structural and functional characteris­ Smith, E.P. & TM. Zare!. 1982. Bias in estimating niche tics of an Indiana stream fish assemblage: A test of overlap. Ecology 63: 1248-1253.

cornrnunity theory. Am .. Na!. 120: 423-454. Strong, D.R. 1980. NulJ hypothesis in ecology. Synthese Hom, M.H. 1972. The amoun! ofspace available for ma­ 43: 271-285. rine and freshwater fishes . Ain. Nat. 70: 1295-1297. Teixeira, R.L. 1989. Aspectos da ecologia de alguns pei­ Hyslop, EJ. 1980. Stomach contents analysis - a review xes do arroio Bom Jardim, Triunfo-RS. Rev. brasil. of methods and their application. 1. Fish Siol. 17: BioL 49: 183-192. 411-429. Yant, P.R., J.R. Karr & P.L. Angermeier. 1984. Stochasti­ Krebs, CJ. 1989. EcoJogical Methodology. Harper & city in stream fish communities: an altemative inter­ Row, New York. 654p. pretation. Am Nat. 124: 573-582.

Lowe-McConnell, R.H. 1987. Ecological Studies in Tro­ Wiens, J.A. 1977. On competition and variable enviro­ pical Fish Communities. Cambridge University, ments. Am. Sci. 65: 590-597. Cambridge, XIII+382p.