An Introduction to the Freshwater Fish Communities of Corcovado National Park, Costa Rica

BRENESIA 21: 47-.66.1983

AN INTRODUCTION TO THE FRESHWATER FISH COMMUNITIES OF CORCOVADO NATIONAL PARK, COSTA RICA

Kirk O. Winemiller Department of Zoology, University of Texas, Austin, Texas, USA 78712

Key Word Index: Corcovado, Costa Rica, fish communities, freshwater ecosystems, trophic relationships

ABSTRACT

The distributional pattern and ecologyof eachspecies of the freshwater fish communities surrounding the Sirena station of CorcovadoPark were investigatedduring the early rainy seasonof 1982.Thirty species were observedor collectedin freshwatersof the park, and an additional 10 specieswere restrictedto intertidal regions.Peripheral division and invading marine fishes were major integral componentsof the freshwater ichthyofauna. These forms were particularly prevalent in the rivers and lower stretchesof smallstreams. Stream width was positively correlated with speciesrichness. Most of the primary and secondary division fishesappeared to be fairly restricted in their use of the range of available food resources.The habitat affinities and approximate abundancesof eachfish specieswas reported,based on visualcensuses. The route of dispersalinto the region by primary and secondarydivision fishes is discussed.Based on comparisonswith earlier studies in CostaRica and Panama,it is hypothesizedthat the marineelements in the park's freshwaterecosystems decline during the dry seasons.Food webs for the QuebradaCamaronal and Rio Claro fish communities were constructed.The major sourcesof primary productivity for the communities are apparently allochthonous, indicating a dependency of the present ichthyofauna on the tropical rainforest as a sourceof energyand nutrients.

The freshwaterecosystems of CorcovadoNational Park in CostaRica are unusualin permitting the study of freshwatertropical fish communitiesin a regionof relatively low freshwaterfish speciesdiversity. Freshwaterfish communitiesin the neotrop- ics are generallycharacterized by an abundanceof primary freshwaterfish species (i.e. membersof families with little or no tolerance to salineconditions) of the

47 subordersCharacoidei and Siluroidei (Lowe-McConnell,1975). By markedcontrast, the freshwaterfish fauna of Corcovadocontains only three speciesof primary freshwater fishes (two char.:oids and one siluroid), and this has been attributed to geographicalfactors inhibiting invasion by these forms onto the Osa Peninsula (Constantzet al., 1981). Secondarydivision freshwater species(i.e., membersof families with limited degreesof toleranceto marinesalinities) are underrepresented at Corcovadoas well, with only three reported (Constantzet aI., 1981) from the more than 25 secondarydivision speciesknown from the Costa Rican mainland (Bussing, 1967). The remainderof the Corcovadofreshwater fish fauna is comprised of peripheral division and marine specieswhich have invadedthe streams, often many kilometers inland. I havechosen to define peripheraldivision speciesas membersof familieswhich areclearly marine in origin, yet they must residein fresh or brackish coastal waters for variable periods of time during their life histories. Peripheralfishes may require freshwaterecosystems for successfulfeeding during juvenile or adult life, or as an environmentfor reproduction and/or larvalfeeding. Marinefishes will be defined asspecies which may residein fresh or br.:kish waters, often for extensiveperiods, yet they are fully capableof completing their entire life cycles in the marineenvironment. Unlike peripheraldivision fishes,many popu- lations of marine fishes resideexclusively in the marine environment,either off- shore or in intertidal zones.The significanceof this new distinction will become apparentin subsequentsections of this report.

Constantz et 8/. (1981) reported a total of 20 fishes from the freshwaters of Corco- vado Park. This represents relatively low speciesrichness for neotropical freshwaters in a region of this size. The freshwater ecosystemsof Corcovado Park provide good opportunities for comparatively-based hypotheses concerning freshwater fish dispersal abilities, historical aspects shaping local species diversity, and tropical freshwater community dynamics. The present study was undertaken to: 1) provide more detailed information on the relative abundances, distributions, and habitat utilization of each freshwater fish species;2) determine the food habits of each fish species; 3) obtain information on the trophic structure of the freshwater ecosystems; and 4) provide a preliminary record of intertidal and estuarine fish species.

Material and Methods

Field sites were chosenalong sectionsof the R(o Claro, Rio Pavo,Quebrada Cama- ronal, Q. Danta, two small streamswhich emptied directly into the ocean,the headwatersof a small, unnamedstream, and one rocky intertidal pool. Additional fish collections and notes were made in intertidal regionsof bedrocksubstrate (A), an oxbow lake near the mouth of the R(o Sirena (B), the Rio Sirenaestuary, and sectionsof streamsbetween the selectedfield sites.Maps were constructed for each field site containing: Profiles of streamdepth, poll or riffle length, streamwidth, substratecomposition, and the location of potential fish refugia (i.e., undercut streambank, roots, logs,boulders, etc.). The length of eachfield site wasdelineated by the natUralboundaries of the pool or riffle under investigation.Consequently, a

4A pool field site in a river was often much longer than a.pool in a smallerstream. Watercurrent velocity was determinedby tossinga35 mm film canister,half-filled with water, into the region of greatestcurrent and timing its passagerate over a designateddistance. At eachfield site, the following data were recorded:1) pH; 2) salinity (ppt); 3) temperature;4) maximum water current velocity; 5) the abundanceof eachfish speciespresent; and 6) the relativeratio of juvenilesto adultsof each fish speciespresent. The presenceof epiphytic algae,zooplankton, crusta- ceans,insolation, and fruiting trees was also noted at each site. Eachfield site was photographedto supplementthe mappedinformation in later analyses. Fisheswere collected by dipnetting, seining,fish traps, angling,and by hand. Habi- tat utilization and estimatesof the relative abundancesof fishes were made by direct observationfrom abovethe water'ssurface and snorkellingbelow the water's surface during normal stream discharges.Special attempts were made to trap or dipnet rareor nocturnalfishes under sectionsof erodedstream bank. Selectedfishes were measuredfor standard length and their stomach morphology and contents were examined.One to six specimensof eachspecies collected were preservedin 10 percent formalin and later depositedin the TexasNatural History Collectionof the TexasMemorial Museum, University of Texas,Austin.

Descriptionsof Field Sites

Corcovado park is located on the extreme western edge of the Osa peninsula of Costa Rica's Pacific coast. The park is approximately 41,800 ha and lies at approximately SO30' N latitude and 8'jO30' longitude. Corcovado Park contains extensive areas of primary tropical rainforest plus patches of regenerating second growth, including dense stands of He/iconia, Ca/athes, Trema, Piper, Cecropia and species of other plant genera (Janzen, 1979). The park received 4701 mm of rainfall in 1980 and averaged 364.4 mm over 1979, 1980, 1981 during the period in which the presentstudy was conducted (June 25 to July 28). This period is in the early stageof the major annual rainy seasonof this region.The streamsand riversof the park flow over Cretaceousvolcanic and sedimentaryrocks and, to a lesserdegree, Quaternaryalluvial deposits(Janzen, 1979). The location of eachfield site is providedin Figure 1. Fish speciesnumbers, physi. cal parametersof the water, and stream length, width, and maximum depth are provided in Table 1. The following are verbal descriptionsof the generalhabitat conditions presentat eachfield site. Sites 1 and 3 are shallow, swampypools at the headwatersof smallstreams which ultimately empty into the Rlo Sirena.The poolscontained much emergentvegeta- tion, including Hel/conia sp. Rotting Hel/conia leavesserved as refugia for fishes and substratefor SpyrogyrBsp. and other speciesof filamentousgreen algae. The bottoms of these pools were coveredwith thick laversof vegetativedetritus and

49 mud. At midday, direct sunlight was transmitted through the secondarygrowth forest, producing scatteredpatches of sunlit and shadedwater in eachpool. No ob- servablewater current waspresent at sites 1 and 3. Sites 2, 4, 5, 7, 8, 9, and 11 are pools in the QuebradasDanta and Camaronal.All pools contained regionsof gravel, sand, and leaf litter substrate.Epiphytic blue- greenalgae and diatomswere usuallypresent on rocks in the shallowmarginal areas or deeper regionscontaining observablecurrent in thesepools. Eachpool is bor- dered at one side by an area of eroded (undercut) bank, usually with tree root structuresor bedrock above.A small portion of most pools was exposedto direct sunlight during midday, and thesewere often associatedwith limited algal growth when suitablesubstrates were present. Site 6 is located at the extreme headwatersof the QuebradaCamaronal. This was the last pool observedto contain fish (3 specimensof Agonostomusmont;cola) as one traveled this fork upstream. Stream gradients are much steeperand many moss-and lichen-covered boulders are present in this region of the stream. Leaf litter, bedrock, and gravel comprised the substrate, and very little observable algaewas present at this site. The entire pool remainedshaded by primary forest throughout the day. Site 10 is a shallow riffle section of the O. Camaronalwith a coarsegravel, sand, and leaf litter substrate.Most of the riffle wasexposed to direct sunlight at midday, and this appearedto support thin growths of epiphytic blue-greenalgae and diatoms.

Site 12 is located at the mouth of the a. Camaronalwhere it emptiesinto the Rro Sirenaestuary. Site 12 gradesfrom a deep, narrow streampool, borderedby bedrock, to a shallowslough flowing over rocks,coarse gravel, and finally sandand silt. The entire site is exposedto direct insolation through most of the day and supported diatomson the rock surfaces.

Sites 13 and 14 are located in two separatesmall, shallow streamswhim flow through areasof extensivesecondary forest growth adjacentto the coast. These streamsempty directly into the oceanover sandy beach.The substrateof site 13 consistedof sand and the pool marginswere composedprimarily of grassesand sedges.Site 13 was exposed to insolation throughout most of the day, yet no significant algal growth was observable.Site 14 wasentirely shadedand hasa sand, soil, and leaf litter substrate. Clumps of emergent understory vegetation were scattered throughout site 14. These streams are probably ephemeral,with very restrictedflow during the major dry season.

Sites 15 and 17 are large pools in the Rfo Claro which have bedrock and gravel substrates. Thin growths of blue-green algae and diatoms were present in the shal- so low marginal areasand swift headwatersand tailwaters of each pool. Leaf litter depositswere generally present among the rocks at the pool margins.Major portions of both pools were exposedto direct sunlight at midday. Site 16 is a swift riffle areaof the R(o Claro. The substrateis comprisedof coarse gravel and rocks. Most of this area is exposedto direct sunlight at midday, and a thin layer of algae is present on some rocks, particularly in shallowermarginal regionsof the riffle. Site 18 is located in a channel formed by the presence of a sand bar near the edge of the Rfo Oaro estuary. The substrate here is comprised of sand and fine silt deposits. The area is exposed to direct sunlight throughout the day, yet no significant epiphytic algal growth was detected. Forest litter was commonly deposited along the banks of the channel following floods and high tides.

Site 19 is a large pool of the Rfo Pavo approximately 50 m upstream from the crossing of the park's Casa Marenco trail. The substrate at this site was primarily fine gravel and sand. Unlike the pools of the Rfo Claro (sites 15 and 17) several fallen trees provided cover for the fishes. The entire pool was exposed to sunlight at midday, and very limited growths of epiphytic algae were present in the shal- lower regions. Site 20 is an isolated intertidal pool at low tide, which has a sand and bedrock substrate.Numerous rocks were scatteredthroughout the pool, providing coverfor many small marine fishes. The pool is exposedto direct sunlight throughout the day, and thin layersof epiphytic algaewere presenton many rocks. At high tide, site 20 is inundated by over 3 m of sea water and is exposedto constant wave action.

Observations The number of speciesat eachfreshwater field site wasextremely variable, ranging from one (sites 6, 13, 14) to 15 (site 19). In addition to providing total species numbers,Table 1 summarizesthe number of primary division, secondarydivision, peripheral,and marine fish speciesat each site. No continuous pattern emergedas total speciesrichness was tracked from the headwatersof each river and streamto its mouth. SPSS(Statistical Packagefor the Social Sciences,Nie et al., 1975) was usedto compute a multiple regressionanalysis of total numbersof fish speciesand streamcurrent, maximumdepth, maximumwidth, and the length of eachfield site. The results, summarizedin Table 2, indicate that approximatelyhalf of the total varianceis explainedby differencesin streamwidth. Additional variancewithin the pattern of speciesrichness is explainedby depth, current, and site length, in order; and together these four variablesaccounted for 73 percent of the variance.The ratio of the numbersof peripheraland marine speciesto the numberof freshwater division speciesincreased from 1.0 at the headwatersof the a. Camaronalto 12.0

)1 at its mouth. A simpleregression of peripheraland marineto freshwaterspecies for the sevensites on the a. Camaronalresulted in a positivetrend (r2 = .494), however this wasstatistically nonsignificant (F = 4.873, df = 1, 5, p = .078). The distribution of freshwater,peripheral, and marinedivision fishes in the park doesnot correlate with salinity gradientssince dissolved solids were uniformly low « 1 ppt) throughout the streamsand rivers (Table 1). The exceptionswere the two river estuaries during high tides (Table 1).

The distribution and habitat affinities of each fish species within the park are pre- sented in Table 3. AstysnBX f.cistl/$ was by far the most abundant and widely distributed primary freshwater fish species in the area. Brachyrhsphis rhsbdophora was the most numerous secondary division species and was present at 14 of the 19 freshwater field sites. The distribution of Poeci/is sphenops was much more restricted to regions of filamentous green algal growth. Poeci/iopsis turrubarensis and OxYZygOnectBSdovii were limited to the RIo Sirena estuary and adjacent stream mouths where B. msbdophors was absent.

Among peripheral fishes, Awaous transandeanus, Sicydium pittieri, Gobiomorus maculatus, Eleotris picta, and Agonostomus monticola were distributed far inland in significant numbers. s. pittieri, in particular, may reproduce in freshwaters as evidenced by the numerous juveniles throughout its distribution and the presence of a brilliantly colored territorial male in the middle reaches of the Q. Camaronal. Larger, predatory peripheral division and marine fishes (Centropomus and Lutjanus spp.) were present in significant numbers in the pools of rivers, even far inland (site 19). The abundance of these forms fell off sharply as one traveled toward the heedwaters of the smaller streams and pool volumes declined. Juveniles of Pomada- sys bayanus were commonly encountered in pools of the streams. Several individuals were observed to exhibit intraspecific aggression near regions of cover such as eroded stream bank. A single adult specimen (-1.5 m standard length) of the saw- fish, Pristis sp., was observed at site 19 and one other location of the Rfo Pavo. The most abundant fish speciesof the intertidal zones appearedto Abudefduf an8logus,Hypsoblennius striatus, and 88thygobiusramosus. The assessmentof the abundancesof speciessuch as Gobiesoxsp., Gymnothor8x sp., and Mslacoctenus zonifer was hamperedby the presenceof many deep, undercut bedrock crevices utilized by theseforms. Preliminary information on the food habits of the principal freshwater-dwelling fishesof the park is presentedin Table 4. Epiphytic algaeand smallshrimp of the family Palaemonidaeapparently comprised the major autochthonousbasis for the ichthyomasssustained in the rivers and streams.Major allochthonous sourcesof food input were forest vegetativedetritus, mature fruits and seeds,and terrestrial insects,including ants.Aquatic snailswere very abundantin the R(o Claro,but only minor componentsof the R(o Pavo and streamecosystems. Other foods utilized werecrayfish, aquatic insectlarvae (Ephemeroptera), and microcrustaceans.

S2 A striking characteristicof the streamsat Corcovadois the paucity of autochthonous primary productivity, and the apparent extreme paucity of aquatic insects and zooplankton. Preliminary schemesfor the trophic interactions of the communities of the Q. Camaronal(Fig. 2) and Rro Claro (Fig. 3) were constructed basedon the approximaterelative abundancesand food habits of each species.In both systemsthe major sourceof primary production takesthe form of allochthonous input from the rainforest.Snails and shrimp compriseda much greaterpropor- tion of the "primary aquatic consumer" trophic level in the Rfo Claro than in the Q. Camaronal.Both communitiesare representativeof a "food web" trophic structure, wherein many speciespopulations utilize food resourcesfrom the samebroad categories,or utilize food from two or more trophic levels.The specieswhich comprised the lower trophic levelsof the t~ systemswere quite similar, howeverthe Rro Oaro supported severalmore predatory speciesin the higher trophic levels (Figs.2 and 3). Although the food habits data for Corcovadofishes should be consideredprelimi- nary due to low samplesizes, several species exhibited fairly rigid selection for specific items. Fruits and seedsconstituted the major portion of the diet of A fasc;atus,and to a slightly lesserdegree H. savage;.Several species appeared to be strict algal grazers,including P. sphenops(green algae, esp. Spyrogyrs spp.), P. tu- rrubarens;s(diatoms), and s. p;tt;er; (diatoms and blue-greenalgae). The diets of B. rhsbdophora and A mont;co/a consistedof terrestrial insects,with B. rhsbdophora consumingmostly ants and tiny dipterans. The gut of the single analyzed specimenof Sphoero;dessnnu/atus was p.:ked with many aquatic snailsand shell fragments.Several AMOUS trsnsandeanuswere observedsucking up portions of fine gravel and sand substrateand allowing the inorganicparticles to sift through their opercular oPenings.One cKJultA. trsnsandeanusat site 16 was observedto periodically dash up from the substrateto capture individual particlesof drifting detritus. The gut contentsof two individualsrevealed only fine particlesof detritus.

Discussion With the addition of Oxyzygonectesdov;;, the presentstudy increasesthe number of known freshwaterdivision fishesat Corcovado(Constantzetal., 1981) to seven. To the previous list of 14 fisheswith marine affinities presentin the park (Cons- tantz .:...jl., 1981), this study reports 13 additional peripheraldivision and marine species,plus 10 additional marine speciesfrom intertidal zones.The three pipefish specimens,which were returned to the Texas Natural History Collection, were determined to be Pseudopha//usstarks;. P.wudopha//uselcaptansnse was reported as moderatelyabundant in the a. Camaronalby Constantzet al. (1981), wheretwo of my specimenswere collected. Gob;esox potan;us and RhamdiaMgnsr; were not observedduring the study although both were previouslyreported as rare in the Q. Camaronal.Despite the fact that they were reported in slightly different manners, there appearto be major disparitiesin the approximationsof the abundanceand distribution of many of the fishesbetween 1979 (Constantzet al., 1991) and 1982.

~~ This may be due to differencesin season,since the 1979 observationswere made during the major dry seasonand the 1982 observationsduring the early portion of the major wet season.Differences in collecting techniquesand censusingmethods may havealso contributed to differencesin abundanceestimates.

The ichthyofauna of the park's freshwater ecosystems is a mixture of freshwater, peripheral, and marine division forms. Certainly the composition of these communities is temporally dynamic in nature, and they will ultimately require detailed lon- gitudinal studies before the factors influencing their structure are fully understood. If, as indicated by Table 2, stream volume has a major influence on species richness, then major differences in the number of peripheral and marine division fishes might be expected between wet and dry seasoncensuses. Several investigations in a variety of temperate and tropical lotic systems have also supported the view that stream volume, width, and/or discharge may explain a major portion of the variability observed within fish species diversity patterns (Gorman and Karr, 1978; Smith, 1981; Livingstone et al., 1982). I hypothesize that many more individuals of the marine division species join the ichthyofaunas of small coastal tropical streams, such as those at Corcovado, during wet seasonswhen stream volumes, discharge rates, arxl food resources are higher. Allochthonous food input to the Tortuguero estUary of eastern Costa Rica was found to be much higher during the wet seasons(Nordlie and Kelso, 1975). In a small Panamanian stream, fish foods were more abundant during the wet season (Zaret and Rand, 1971). During the dry seasonsat Corcova- do, many marine fishes may be excluded from freshwaters due to competition with freshwater and peripheral division fishes. In reviewing the distributional patterns of African freshwater fishes, Roberts (1976) noted that, when present, freshwater division fishes tend to exclude more euryhaline forms from inland habitats. Tiny palaemonid shrimp were extremely numerous in the streams and rivers of Corco- vado during the early rainy season, and these should provide a major potential source of food for temporary marine invaders as well as permanent freshwater resi- dents. The view, that primary and secondary division freshwater fishes dominate the ecosystems at Corcovado when stream volumes are lower, is also supported by the observation that freshwater division fishes comprised a greater proportion of the speciesrichness as one traveled away from the mouths of streams. Corcov~o Park lies within the isthmian faunal province of Miller (1966) and Bus- sing (1976). This region of CentralAmerica is by far the richestin primary division freshwaterfish species(Myers, 1966; Miller, 1966; Bussing,1967), with most forms derivedfrom South Americanfaunal elements.Of the three primary division sp«:ies presentat Corcov~o, Astyanax fa.w:;stusand Rhamdiawagner; are reportedto have limited toleral~es to salineconditions (Miller, 1966; Bussing,1976). Both of these speciesexhibit ubiquitous rangesin lower CentralAmerica, and A. fa.w:;stusin particular extends from Argentina to southern Texas on the Atlantic slope (Miller, 1966). It is therefore quite likely that these specieshave utilized a coastal route for dispersalinto this distal region of the Osapeninsula. If streamheadwater cap- tures h~ played a significant role in the dispersalof primary freshwaterfishes into the park, more specieswould be expected.The degreeof salinity toleranceby

~ Hyphessobrycon savagei is unknown. Like the primary freshwater fishes, the secondary division fishes of Corcovado are represented by several species which are ubiquitous in Central America. The presence of these seven primary and secondary division freshwater fishes suggests that these forms have dispersed into the Osa fairly recently (i.e., compared to major dispersals of these families along the mainland Central American slopes) and by way of coastal routes, possibly during periods of unusually high rainfall. The recent arrival of freshwater division fishes into the Corcovado region is further supported by Bussing's (1967) observation that a large number of indigenous species comprise the Costa Rican freshwater fish fauna. The peripheral and marine division fishes of Corcovado are primarily species which are widely distributed along the Pacific coast of Central America. Several species, such as Agonostomus monticola, Centropomus undecimaJis, and Mugil curema are known from both coasts of Central America. Apparently, the presence of Abudef- duf analogus in my collections represents the first report of this species for the Pacific coast of Central America. The preliminary data on the diets of Corcovadofishes indicated that severalfresh. water and peripheraldivision specieshave fairly restrictedfeeding habits (Table4). One marine invader, Sphoeroidesannulatus, may speCializeon snails, using its "beak-like" dentition to crush shells. Most of the marine speciesin freshwaters at Corcovadoare large predatorswhich apparently exploit the numerousshrimp and small fishes. Much more information is neededon the diets of thesefishes before it can be determinedif food resourcesare a limiting factor for marinefish bio. massduring the wet season.The descriptiveinformation on streamhabitat use by the fishesof Corcovado(Table 3) suggeststhat habitat selectioncould be onemeans by which fishesecologically segregate during periodsof depressedresources. This is an emerginggeneralization from recent studiesof tropical stream(Zaret and Rand, 1971; Roberts, 1972; Bishop, 1973; Gorman and Karr, 1978) aswell astemperate freshwaterfish ecology(Winn, 1958;Allen, 1969; Mendelson,1975; Symons, 1976; Werneret al., 1977). The distributional patternsof fisheswithin the park (Table3) probably relatesmost directly to habitat affinities. Consequently,the total species richnessat each site should relate, not only to streamvolume (Table 2), but also to environmentalheterogeneity in termsof habitat diversity at eachsite. For exam- ple, Poecilia sphenopswas only abundant in environmentswhere there was low water current velocitiesand significantgrowth of fil~entous greenalgae. The freshwater ecosystemsof CorcovadoPark appearto be consistent with the physical characteristicsof many tropical rainforest headwater streams (Myers, 1947; Roberts, 1972; Bishop, 1973; lowe-McConnell, 1975). Their waters are apparentlyquite nutrient poor, contain little primary production, areoften entirely shaded,exhibit low thermal variability, and receiveseasonal patterns of rainfall. One major difference is the absenceof high freshwaterfish speciesdiversity at Corcovadodue to the peninsulareffect. The Rio Claro and Rio Sirenaalso lack extensive stretchesof lowland floodplains and extensiveestuaries prior to their dischargeinto the Pacific Ocean. It is this unique combination of physical and biotic similarities and dissimilaritiesto other tropical regionswhich givesthe fresh-

~~ water ecosystemsof CorcovadoPark a tremendouspotential for comparativetests of the factors governingthe evolution of tropical streamfish communities.For ex- ample, I would predict that the large numers of palaemonidshrimp, supported largely by terrestrially-derivedvegetative detritus, could support many more freshwater fishes in higher trophic levelsduring the rainy season,if they were intro- duced. The reproductivebiology of the peripheraldivision fish speciesof the park remain virtually unknown. Thesefishes are very important componentsof many freshwatercommunities, and additional life history information could contribute much to clarifying the oPerationaldefinitions of fish divisions basedon marine dispersiveabilities and ecology.

Acknowledgments I would like to thank Frank Pezoldfor his extensiveefforts in verifying and deter- mining the identification of speciescollected at CorcovadoPark. Clark Hubbsand LawrenceGilbert provided important logistical assistancewhich otherwisewould havemade the study impossible.Thomas Grohman and LawrenceGilbert aided in the collection of severalfish species.Frank Pezold, LawrenceGilbert, and John Rawlinsprovided important criticismsand suggestionsduring the preparationof the manuscript. I would especiallylike to thank the efforts of the CostaRica National Park Servicefor their provision of a collecting permit and hospitality during the study. The data was compiled during participation in the U. T. Zoology 384L coursein tropical ecology.

Resumen 5e investigaron las comunidadesde pecesde aguadulce en el ParqueNacional Corcovadodel 25 junio al 28 de julio. 5e observaron0 recolectarontreinta especies de pecesde aguadulce, asi como diez especiesque seencuentran solamente en zo- nascon influenciade las mareas. Los pecesde lis divisionesde aQuadulce mas abundantesy extensamentedistribuidos son Asryanax fasciatusy Brachyrhaphisrhabdophora. De 105peces de la divisi6n periferica, Sicydiumpittieri, Awaous trangndeanus,Gobiomorus maculatus y Eleotris picta fueron 105mas extensamentedistribuidos en lis aQuasdulces de la regi6n. Variasespecies de la divisi6n marina tambien se encontraroncon fre. cuencia en aQuadulce, entre otras, Mugil curema, Pomadasysbayanus, Lutjanus colorado, L no~mfa.:iatus, y Spheroidesannulatus. Estas especies se encontraron en abundanciasobretodo en 105rfos y las regionesbails de 105arroyos. EI unico ejemplarde pel sierra (Pristis sp.) observado,se encontr6 en el rfo Pavocasi a kilo. metro y medio de la desembocadura. Examinamoslos contenidosde los est6m~os de varios ejemplaresde las especies dulceaculcolasmas comunes. Los alimentos terrestresles eran muy importantes; '6 entre otros, pepitas,frutas, insectos(especialmente hormigas) y material organico. Ahora bien, la mayor parte de la comida aut6ctonaen 105regrmenes de 105peces fue algas,camarones y pecespequenos. La rflayoria de 105peces de lasdivisiones de aQuadulce parecentener habitosde alimentaci6nparticulares. . La particularidaddel habitat en el arroyo de cads especiesugiere la posibilidadde que estefactor puedaser un metodo adicionalde segregaci6necol6gica.

Tomando en cuenta la informaci6n preliminar de que dispon(amos sabre los habitos de alimentaci6n, se construyeron redes de alimentaci6n para las comunidades de Quebrada Camaronal y r(o Claro. Se encontr6 que el ancho del arroyo se correlacio- na positivamente con la riqueza de especiesde peces. Con base en los estudios preli- minares hechos en Costa Rica y Panama,se ofrece la hip6tesis de que los elementos marinas en las aQuasdulces del parque bajan en nCAmerodurante las estaciones secas. Tambien que par 10 menos cinco de los pecesde las divisiones de ague dulce emigra- ran hacia el interior de la regi6n par rutas costeras.

Literature Cited

Allen, K. R. 1969. Distinctive aspects of the ecology of stream fishes: A review. J. Fish. Res. Bd. Canada26: 1429-1438. Bishop, J. E. 1973. Limnology of a sm.1 Malavan river, Sungai Gombek, Monogr~h. Bioi. 22. Th. Hague: W. Junk. 435 pp.

Bussing. W. A. 1967. New species Md new records of Costa Rican freshwater fishes with a ten- tetive list of ~ecies. Rev. Bioi. Trap. 14: 206-249. Bussing, W. A. 1976. Geographic. distribution of the San JU81 ichthyofauna of Centr. Amer- ica with remarks on its origin 81d ecology. p.s 167-176 In T. B. Thorson, ed. Investiga- tions of the ichthyofauna of Nicaraguan lakes. Lincoln: Univ. Nebresk.. ConstMtz, G. D., W. A. Bussing. and W. G. Saul. 1981. Freshwater fishes of Corcovado Nation- . Park, Costa Rica. Proc. Acad. Nat. Sci. Phil.. 133: 16-19. GormM, O. T. and J. R. Karr. 1978. Habitet structure and stream fish communities. Ecology 69:607-616.

Janzen, D. H. 1979. U. of Santa Rosa National Park and CorcO/ado National Park, Costa Rica, by biologists. Unpublished manu~ript, Univ. Penn. 47 pp. Livingstone, D. A., M. Rowland, and P. A. Bailey. 1982. On tha size of African riverine fish f.,nas. Amer. Zool. 22:361-369. Low- McConnell, R. H. 1975. Fish communities in tropic8 freshwet8r. London: Lon~en. 337 pp. Mendelson, J. 1975. Feeding relationships among speciesof Notropi6 (Pisces: Cyprinidae) in a Wisconsin stream. Eool. Monogr. 45: 199-230.

S7 Miller, R. R. 1956. Geogr~hic. distribution of Centr. Americ., freshweter fishes. Copeia 1966: 773-802. Myers, G. S. 1938. Freshwater fishes .,d West Indian zoogeogr.,hy. Smithson. Rpt. 1937: 339-364. Myers, G. S. 1947. The Amazon and its fishes, Part 4. The fish in its environment. Aqu... J. 18:8-19. Myers, G. S. 1966. Derivation of the freshwater fish fauna of Centr. America. Copeia 1966: 766-773. Nordlie, F. G. and D. P. Kelso. 1975. Trophic relationships in a tropical estUary. Rev. Bioi. Trop. 23: 77-99. Nie, N. H., C. H. Hull, J. G. Jenkins, K. Steinbrenner,.,d D. H. Bent. 1976. StatistM:. Package for the Soci. Sciences.New York: McGraw-Hili. 675 pp. Roberts, T. R. 1972. Ecology of fishes in the Am.on .,d Congo besins. Bull. Mus. Comp. Zool., Harvard 143: 117-147. Roberts, T. R. 1976. Geogr~hic distribution of Afrlc., freshwater fishes. Zool. J. Linn. Soc. 57: 249-319. Smith, G. R. 1981. Effects of habitat size on species richness and ~ult body sizes of desert fishes. Pages 125-171 In R. J. Nai~and D. L SoItz, eds. Fishes in North American deserts. New York: John Wiley & Sons, Symons, P. E. K. 1976. Behavior and growth of juvenile Atl.,tlc salmon (SaI/1X)saI.-).,d three competitors at two stream velocities. J. Fish. Res. Bd. Canada33: 1766-1773. Werner, E. E., D. J. Hall, D. R. Laughlin, D. J. Wagner, L. A. Wilsrn.,n, and F. C. Funk. 1977. Habitat partitioning in a freshwater fish community. J. Fish. Res. Bd. Cenada34: 360-370. Winn, H. E. 1968. Comparative reproductive behavior and ecology of fourtMn speciesof darters (Pisces- Percidael. Ecol. Monogr. 28: 155-191. Zaret, T. M. and A. S. Rand 1971. Competition in tropical stream fishes: Support for the corn. petitive exclusion principle. Ecology 52: 336-342.

~8 ~ - - co N ~ N N r-. - ID 0 0 ~r-. II) 1D"a) OlD .~ ~~~.~~~~N~~N~N~dd~ :c -- --- Nit) -IDMM- 8~ ca - 'C ca > - - ~ oJ oJ ::; > OON~Nm~~~MM~-O~~-~ 010' I I I "Ai .N~~oiN.r.:~MIrir.:NIriO ~~ lliro." .~ ai M--~ M .. ,~ ca M ~ 'a - - ., "'i ..I ..I :J -s ~ .. ~~~g~gg~~~2~~~~~~8...... ',,; ...... N ',,;,,; ~~ .,u - ~ 1 e ... ca > M ~ ~ GI 1-lmMloIOMUJt")~~mCD""";:- I I I -; ... ~l\, E ~ ~ O;~ ~0;~t'!0;~~0;r"-;:::0!~ -0 E C ca ca ~ ca --- - 0- 3 i .J~~ ~ ca ~ OOONN~~N~~~~~~OOON I -0: -0: -0: I -0: -0 ,W ~~~~~~~~~~~~~~~~~~ Qlr--r-- r-- .. M NNNNNNNNNNNNNNNNNN NNN N ~ > .2 ~ :; 0- ~~oo 000000 ~~oooo 010 10 . I ...... I ...... roo.. roo r.: I I > E wGw'" ,..,..,..,..,..,.. GG,..,.., c ca 0 ~ . .. . .c ~ .-. -- --- ..J ..JZ ..JZZ 8 0 "...~~ ~-~ ~';:'~~"ai"~~ -or .- M vvvvvvvvvvvvvvvvvv v~ f! -0 C ~ e ca OOOO-O--NONmoo~.~~ In. -' t..I ., - - E ~ 'a C c ca M ., --O.~-~~~N~~OO~M~N ""0 ot'ff,) I 'y: ., -0.B . 0- M '"¥ ~.. ~ M ~ ""~""O-N -0 O""N =0 .-"6. ""---O-O u .c 0 ; .; II ~"'- J >'~:I: --ON-ONNNONOOO-O-O NO 0"'"0 ~c- E e" ~ -01el:-o c ~ ~ .~N~~-~OOM-~--.~~~ 10. ..h~ a.B'~ I------.. - - e-~ ""0 - !..=- E ca II ~ . ~ >- c . &-..J ::a -NM.~~~~OO-NM.~~~~ 010 1-0- ca ~~~~~~~~~ -N ~m!a I- in ; . .

59 Table 2. Resultsof a stepwisemultiple regressionanalysis of fish speciesnumbers and measuresof streamsize. *

Contribution Overall Overall Variable Partial F Step to R2 R2 F Significance

The regressionequation is y (number of species) = 1,.780 + .420Xl (current) 6.320X2 (depth) - .476x] (width) + .228x4 (length),

60 0 """. .. .c- e . ~ . I . .. t: .E .;0 c ". I 5 - ~. c .. '2 0 E~- . ci ~ ~ e I§ ,_! .E & 4 I~I ~ :;~ . ... ou- U oln ~ it'?:: ~ t ~ ~i 'g 0 Q. .. - .c 0 ~ j j ! £ i£ I ~ e .to ~ - .. .~.."- ~... --- - & " -.. 0 J -..0" -~! ."- ~. .E g E & .. ~i :1 ~£. i -1 i r~ ~: c "~r I '0 ~1; !e ~ tUg" .iE h :- ~ j .: -;:. Ii -1=' -of . °E "~ ~. ..0 I"-;? _- c ~ "o-E ~oi "0 2! .2r.~ ~ o.i~ .E .. 2 ..~ OU Q.8 .'.!".. . c.. U 0 Q, "." - OU 0 ~ c-.. i~' f i i: I -.8 ! ..8 ~.. 0 i ~ .. .&e ~..I.i ~f.s I L-e- -0. 0-.. I 1 -:11 i"!-ES.!; e e ~ "0.. " - m-.c i _i. oiI~.E.~~~c . -. 8 -.0'"~J ! c c ~.t:.. ~E.. ..-~ e1 0 0 -050': .,.~!..;l0- e ,~ ! -s 0. -S E ...... '-='" ." 1: 0 ~ .. " 5 j e e~e..eo-.lilQ.e~ 0 ~:§.E ~o't . ~.iEi=I~- M 'aI'O'O'tia ~:Bt2:~i~j~~i~~~f .c="Sc~" . i&ii :JD<~,.In~ 1 ~~~e~!~&~~£~!~~~J t :"i~ ~ ~ i ~ ~ ~ g 0 i 0 0 ,CM o-~ ~ t. ~ ~ ~ ; 8 8 8 ii.i. ~ ~ i ~ ~ ~ i ~ i ~ ~ ~ ~ -5. . .E -i§~ c - ccc- ccc< - ~0( 0(0(- c .i-8 0 ~ ~ . wi G' , ~ ~ ,.: 1')' , ~ m ~~ ~ . , , ~,.. GI. ... ~ N ,.. . " ""'" ~ , ~ ~ ~ c ~ ~ , , ~ . ~ ";, ~ ~ ~~111"; =.. :~' 0 '" .J ~ ~ a. ,: ~, ... ~ G~, . .-$ >- ~ 0 ~ Ii>' ~ ,..~ .~ ~ . ~~ I a i ~- ,a-. ~ 0" ... ~- - I ,: ~.. 1 ,..8!. ~. ~~, . I ... ~ ~ ~-a ... aa ~,.. ,C'" ..I a... c. ~ ~~~ ~'. ~ ;:;-, in;- -i ;. ..:~. ~ i ~ "', "", ~ i ~ ~. -.. C'!.~ Iq. In .. ,..GIII"'1ft,~ ,~-, 1ft-. ,..,~. -c ~=, N 1~ ~. ~ N'", . in N 11.~cq, N . ~, '.. III~ . ~ ~ Ii~ ~~... ~~~ ~ 811 "'-N~~ ...in ... ';inm~ca" 3' ! c ~ '0.'" ' w c~ .! ~ .~ .! ; .. Jf '2 .~ .~ ., i -=: .~ - --v .0.. .s '. u.. .~ ~ .~:t ;-, ~.. ~ ~ ii: ~ c @ " ~ ~ ~ 0 1 Ii .~ .-;i:l 0 c! a -8 II j -5 I ~~.I~ .'~ .2... 8~ :~ .. i .g M.- .~ ~... -;1 ~ ~ ~ §C ~z .- ~ O.~ i a t ~~! .~ ~ ~ 0:2 I ~ a ! !"';: ~ -J..2 i ~1 ~ "S ! co.! -e ow... ~ ~ -t c-iM --e i Q. 8. a ~ s,~'1 ~ ~~ .. ~ -. ;-i I ] ~ ~ ~ ~ I~ ~ ~ ~~ et ~ --'i~ ~ .Q g I ~ ~} .~ to ~ '" 0 ~ ~8 A."( ~ ! II g A.1Jj 'C( ~ ~ .f ~ &.J

61 ...... »».. cc nI ~I '" ~ 0 - icCCC 00 C .'r. 1 - u.: ..0000 --- i 1 -~o'i u i c~ -e -C -C ~. ~u ~ u~~~~~ rr~ c ..u ! -.c0 - 1 -.. - ~C ~! ~.c.cr.~ --~.. 1 ! J'te!i 8 8 .. .. ~~ ! ~~~gj~~~!.. II '~=Ii"Ic- ; .. -~ -~ 0 . tc . .r.r.r..c ~ a.. ~ 0 ~ ~ I -.;~~~; .0..~!8 8.£;;;;~~~5.;~~~~~.o--- 1 'i .a~~~ ; ...... ~~~~ a=~ ~~g i»~ I .. ..~ I C. ~.& ~ I . 0.0 ".0.0 .o.o"aaaa..~~u '0 C 1 ~~ 0"11"~00~ :i -~~_..c~ ~.! ~o-..ci -. oo.~0.-. ,~~~ ,"" - ... p ~ .,.. C u -r ., ., ., ~ C 8 oW ~ ~ ~ I ~ ~ ~ ~ ~ £ ~ ., ., 8 ! - '.e~- C E E E - u ~uu~ u~ ~EE" ~ :i. e>ou" -"0 0 0 k~~O eoo~ ee"""~e £~c c '0 k."~~".."~...;..-~ .. .. "e~"&.." """~"",,uUHR"o~~~~ .c-- ~ s. ~ £~ § a ~~ ! ~ ! ! ; ~,; ~! !!! ~! ! ! ~ ~ ~~ ! f ! j ; I 0- ., r a I i"O e'! ! ~ ! .. !] ! ~ E E ! e' E E 9 E E E ~ i ~ ~ e'! ! E ~ e' -s S_I~"'.2~~!I~I~~1~2R8~~r88~!!t!~~~~.8 > oJ~~~20c~"~"~u~..oo~000=000.."0~~0~0 m..~IR~mD~a~a~~~~mm~mmmummmoooom~~m~m i ! e. ~ ~ ~ 8 ~i -' i ~ I I I I .8 I I I I I !~~~ 8 0 0 0 :: 8"0 . i ~ ~ G G ~ 0 ~~ 8 --'ao .! 1 s.. . c - EC-I' IICC-CCCCCEI II i 'w ! . ~ '0 0(' Gi .-:. . . ~ . ~ ~> I/)~GGG= ~ ~ ~. ..~ G~ ;'' ~~ .aiai,.:. . "'w~ ~ , _ ~..G> ~ ~. ~ ~. ~ ... ~ c a ~ ~ ~ ~ ~~ m '. ~. ~. ~ ~. ;G.! I :?~ . . '0 ~ w ~. in ~ ~ ... G G i G. i .C ..i ~ Ccl ~i:'i ._~ ~ w) . ~, . ~.~~. ! ~. ~.. ~ ! :: i G c in CCCC! i.~C ~~. ~, ~: ~ ~.~c~!~.3 e:'OC N ~ N ~ N ~ ~ ~ - GO .. NON ~ ~ 1..~ ~ N 0 N - ,- -. _~8 ~1n"1n ~I~I/)~I/)~I/)G ~N ~CCCCN~1/)1/)0~~NCC ~Q.~ . --s~0.- i >co1.- '0 - . .cM ~ ~.. ! M. d .~ ti I 14- tai - ~ ~ .! I ~. ~.. . . - -I ~c C ~ ~ ~ e ~ ~l ~~~~ ii.Ei~~ A "00 .2 ~ ~ I )~o c:; .= . .! i t i! ~M.- ~~r!~ -!.. . I c -:.§ .!. i .. !II,~ il'i ! I ~ 2 i ~ .J ~ ~~ i l J~:J ~ox 0 ~'" ..~ .. .0 I ~ 9 9 ~ i"8 .~,~~a II ~ ~ .E ~ i i i ~ .~~ ia.~.t l ~~ie>.~"tj~~e~,<~i1:t¥o I~ .. ~ e '~. .~ '00 -g <.5< ~'6 o~i-g:l~.!~"1'oQ.g.~ .~ . . ~ :i ce ~ ~ ~ ~ t~ ~~l~~~~~~_9~~~~ ~ .

62 Table 4. Resultsof the stomachcontents analysesfor the principal fish speciesin- habitingthe freshwatersof CorcovadoNational Park.

St.nd.rd Species Stomach cont8nts Alimenury c_1 No. length (mml

A$tyanax faa:istu$ Fruits. seeds.insects Short, sc-like 13 38-95 HyphB$$Obrycon savagei Sm., seeds.ants Short. sac-like 7 35-43 Poscilis sphenops - ,ill; Green alg. Long. coiled 3 58-90 (esp. Spyrogyra spp.) Brschyrhsphi$ rhabdophora Ants. small insects Short, sac-like 7' 18-46 Posciliopsi$ turrubsrensi$ Epiphytic .gee Long. coiled 3 20-28 Oxyzygonectes dovii Ants. sm.1 insects.epi- Short. sac-like 2 48-80 phytic .g.. detritus Sicydium pitt;.,i ."',- Epiphytic .g. Long. coiled 2 56-71 Awaau$ transandsanU$ Detritus Long. coiled 2 74-86 Gobiomoru$ msculetu$ Sm.1 fish. shrimp. Short, sac-like 5 46-168 Ephemeroptera nymphs E/eotris picts Larval fish. shrimp. Short. sac-like 7 38-200 crayfish. snails. seeds, fruits. detritus. insects Agonostomus monticols Insects Short, sac-like 2 81-123 Mugil CUrBmS Detritus, algae Long, coiled 3 25-88 Pomstlssys bBysnus Shrimp, crayfish Short, sac-like 4 133-188 "., Csntropomusundecims/is Fish. shrimp(~) Short, sac-like 7 50.85 SphoBl'oidsssnnulstus Snails S8:-like, long. coiled l' 130 LutiBnus colorado Shrimp. fish(7) Short, sac-like 3 132-166 LutiBnus novemfssciBtus Fish, shrimp(?) Short, sac-like 2 132-150 LutiBnus srgsntiventris Shrimp. fish(?) Short, s-=-like 2 66-140 Pseudophsllus starks; Microcrustaceans(?) Short, sac-like 2 92.110 Eucinostomus sp. Microcrustaceans(7) ? 1 12 Bsthygobius sndrsi Shrimp Short. sac-like 1 15-63

63 "", -. ,

SIRENA ParQue Nacional Corcovado

~ ' .~ M.'...

'. '-'." .:

Fig. 1. Map of the area surrounding the Sirena station of Corcovado Park, Costa Rica and the locations of field sites.

.. c""--"""-~-~~. "-"-'0-= = 'f' .

! i

! II) - C II) II) 0 C ~II) I - (J tU .-C I ~..(J >- (J~ - I ~.c tU .-tU.. .-.c 0)( tU E -II) ;:; I (J.~ 0. ~ C ~ ~ .!! 2 II) I II) ~ E ~ tU .- - (J ~ = tU ->- .- .c >- 2 "C .u 0 tU ~ tU .. 0. - .. >- ~ .. Q) I cO".. .c >- II) ~ .~ o.~ I :I: 00( (/) Q. ~

I Terrestrial Fruits I Diatoms ~ & I & 01 Vegetation Seeds I Blue-green iO algae c (detritus) I 3: I .. I (!) Allochthonous I Autochthonous I

Fig. 2. Block diagram representing the trophic relationships within a typical community of the Quebrada Camaronal. The sizes of the blocks represent the approxi. mate relative abundances of each component in terms of biomass, based on visual censuses.

65 . l. novemfasciatus

I !!. ( 0) u I .,Q) -U .E I .,Q) Ii r .E .- I 0) ::. ~ ., I 0 ::Q) I IU~ Q) 1-- <

? 7

Fruits I Diatoms Terrestrial & I & V . Seeds egetatlon I Blue- green algae I (detrjws) I I I Allochthonous I. Autochth.onous I

Fig. 3. Block diagram representing the trophic relationships within a typical community of the RIo Claro.

66 ~ . .