Environmental Biology of Fishes 34: 29-50, 1992.

@ 1992 Kluwer Academic Publishers. Printed in the Netherlands.

Fish assemblages across a complex, tropical freshwater/marine ecotone

Kirk O. Winemiller & Mitchell A. LeslieDepartment of Zoology & Texas Memorial Museum, The University of Texas, Austin, TX 78712, U.S.A1 Present address: Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge,

TN 37831-6036, U.S.A.

Received 10.8.1990 Accepted 6.3.1991

Key words: Caribbean, Costa Rica, Estuary, Gradient analysis, Ordination, Species diversity.Species relative abundance, Species turnover, Tortuguero

Synopsis

Riverine fish assemblages in the temperate zone generally show strong longitudinal patterns of faunalturnover and increases in species richness with increasing stream order. We examined the composition andstructure of tropical fish assemblages across a complex freshwater/marine ecotone in Tortuguero NationalPark on the Caribbean coast of Central America. Species turnover was high between four characteristichabitats that largely corresponded with a longitudinal gradient of stream order over distances of less than30 km. Suites of common fish species characterized each habitat: creeks, rivers, lagoons, and the sea. Inaddition to the habitat endemics, several species spanned two habitat types, but only three species werecollected in more than two habitats. Multivariate gradient analysis of fish assemblages reflected a gradient ofhabitats that to some extent corresponded to fluvial distances. Due to the unusual configuration of coastallagoons lying parallel to the coast, the ordination gradient showed little correlation with linear distance to thecoast. Environmental variables related to habitat size and salinity showed greatest correspondence with thefish assemblage ordination gradient. Invertebrate-feeding fishes were the predominant trophic group in 15 of16 fish assemblages, and inland creek sites contained a greater proportion of herbivores and omnivores thanother sites. The relative fraction of herbivorous and detritivorous fishes showed a monotonic decline alongthe longitudinal habitat gradient from inland to coast. Patterns of species composition and richness atTortuguero Park appeared to agree well with earlier models of factors influencing temperate zone streamfishes. Headwaters have low aquatic primary productivity and contain small colonizing fish species subject tolarge fluctuations in local densities and intermittent competition. Lagoons contain both large and smallspecies, the latter being restricted largely to shallow edge habitats by predation. Lagoons exhibit more lenticenvironmental conditions, experience relatively fewer periodic disturbances than headwaters, and theirassemblages are inferred to be under relatively greater influence of biotic factors. Fish assemblages of riversand calios (swampy side channels and braids) appear to be under less abiotic control than headwaters andinfluenced less by biotic factors than lagoons.

Introduction management concerns, with a primary goal beingimproved understanding of the complex hierarchyof physical and biotic environmental factors im-pinging on the relative success of fish populations.

Interest in the structure of natural fish communitiesis rooted in both ecological theory and fisheries

30

of species associations can be reduced to the ques-tion of between-habitat (beta) diversity, wherestrong longitudinal patterns of species associationgenerally are replaced by a more complex spatialconfiguration associated with microhabitat hetero-geneity (Gorman & Karr 1978, Angermeier & Karr1983, Meffe & Sheldon 1988, Lyons & Schneiper

1990).This study investigates patterns of species associ-

ation and longitudinal zonation of fishes in Tortu-guero National Park, a region of 18 947 ha on CostaRica's Caribbean coast. The park contains a mosaicof diverse terrestrial environments, ranging fromspeciose lowland tropical rainforest to palm swampsand beach communities consisting of a few colo-nizing species. Aquatic environments span a com-plex, feshwater/marine ecotone and include smallheadwater streams within distances of 0.3 to 13 kmfrom the sea coast, swamps, rivers, estuaries, andocean beach. One hundred-thirty fish species havebeen documented from fresh, brackish, and marinewaters of Tortuguero National Park (Caldwell etal. 1959, Gilbert & Kelso 1971, Bussing 1987,Winemiller 1987). We employ multivariate meth-ods of community gradient analysis and note levelsof turnover in the fish fauna over small spatialscales that are comparable to patterns frequentlyobserved across much larger watersheds. We thenrelate species distributions and assemblage affil-iations to physical and biotic environmental fac-tors.

Methods

Species assemblages can be examined from severaldifferent perspectives, including spatiotemporalpatterns of taxonomic composition, trophic ecol-ogy, and reproductive guilds. Investigations of riv-erine fish assemblages have resulted in general hy-potheses of longitudinal zonation (Kuehne 1962,Sheldon 1968, Bishop 1973, Vannote et al. 1980,Schlosser 1987). Investigators working with fishesand aquatic invertebrates have identified stronggradients ranging between (1) low-diversity, head-water assemblages subject to periodic disturbancefrom high variation in water discharge levels (Fish-er et al. 1982, Meffe & Minckley 1987), and (2)high-diversity assemblages in low-gradient, down-stream reaches where biotic interactions appear toexert a relatively greater influence (Allan 1975,Schlosser 1987). Exceptions to this general patternare usually associated with special geological orhistorical features, including rifting (Balon 1974,Balon & Stewart 1983) and desertification (Smith1981). In estuaries, physicochemical gradients areagain conspicuous and either directly or indirectlyinfluence species assemblages. A number of par-ameters tend to be correlated with salinity andform gradients across the freshwater/marine inter-face, and these in turn influence species distribu-tions and local abundances (Weinstein et al. 1980,Felley 1987, Rozas & Odum 1987, Odum 1988).Some species are directly limited by narrow physio-logical salinity tolerance, whereas other specieshaving broad tolerances may be restricted fromcertain locations by physical or biotic factors thatare correlated with salinity (e.g., nutrients, pro-ductivity, habitat structural heterogeneity).

Multivariate methods for ecological gradientanalysis (Hill & Gauch 1980, Gauch 1982, TerBraak 1986) have intensified interests in patterns offish species association and distribution in relationto riverine gradients (e.g., Matthews & Robison1988, Townsend & Peirson 1988, Ibarra & Stewart1989). Most studies of fish assemblages have fo-cused on patterns observed across large spatialscales corresponding to large watersheds or entirezoogeographic regions (Burton & Odum 1945, Saul1975, Cashner & Brown 1977, Horwitz 1978, Gra-ham & Hastings 1984, Matthews 1986, Ibarra &Stewart 1989). At small spatial scales, examination~

Study region

Tortuguero National Park lies within the Carib-bean coastal lowlands of Costa Rica's Lim6n prov-ince between 10° 20' and 10° 35' lat. N (Fig. 1). Theregion is among the wettest on earth, averagingabout 5 m of rainfall per year with two main rainyseasons from July to August and November toJanuary. Rainfall in excess of? m may occur duringan exceptional year. Generally, no month receivesless than 50 mm, and temperature shows little sea-sonal variation with an annual mean of 23-26° C

31

830 30' W

1

,~~~~~~~~~:::--

;rortuguero Village

CARIBBEAN SEA

Aqua Fria Viejo9

10030' N

~~.Agua Fr(aStation -

-

~iOt'-.

~

.g

\Fig. 1. Map of Tortuguero region. Locations of 16 study sites within the Rio Tortuguero, Rio Sierpe, and Caito California drainages are

indicated by numbers. Dark shading represents Lomas de Sierpe, dashed regions represent low-lying, Raphia palm swamps.

Cano Negro, and Laguna Jalova in Fig. 1). Thecoastal lagoons are actually estuaries molded bythe dynamic forces of river sedimentation andcoastal currents. The Laguna Tortuguero averagesapproximately 7.5 m depth at midchannel betweenTortuguero village (depth 13 m) and its mouth(Nordlie & Kelso 1975). Two openings to the sea lieat, or in close proximity to, the park boundary: apermanent northern outlet at the mouth of theLaguna Tortuguero, and the temporary LagunaJalova outlet to the south. The outlet at the LagunaJalova mouth only opens during periods of excep-tional rainfall. For example, the mouth was openduring the peak summer wet seasons in 1983 and1986, but completely blocked by a sand ridge

(Hirth 1963). Most of the western area of the parkis covered by pristine, lowland tropical rainforest,but at least four other vegetative zones have beenrecognized within the reserve: cativo swamp forest,low marsh woodland, herbaceous marshland, andhalophytic coastal vegetation (Nuhn et al. 1967).Many of the low marsh woodlands are dominatedby the palm Raphia taedigera.

Apart from the Lomas de Sierpe hills in the west(Fig. 1), most of the park is coastal plain lying onlya few meters above sea level. Sediments of theplain are of alluvial origin, formed by a coalescenceof river deltas. A distinctive network of narrowlagoons lies parallel to the coast (corresponding toCaiio Palacio, Laguna Tortuguero, Caiio Servulo,

32

(fishes having almost no salinity tolerance), sec-ondary division (fishes having limited or sometimessubstantial salinity tolerance), or peripheral divi-sion (freshwater representatives from marine fam-ilies). Some ichthyologists question the existenceof a high degree of uniformity within higher taxawith respect to divisional affiliations based on salin-ity tolerance (Bussing personal communication). Arecent test of salinity tolerance involving a smallnumber of Costa Rican species confirmed Myer'sdivisions (Winemiller & Morales 1990). In terms ofspecies richness, the Central American ichthyofau-na is dominated by secondary division freshwaterfishes of the families Poeciliidae and Cichlidae,followed by peripheral division fishes, and primarydivision fishes mostly of the Characidae. The Tor-tuguero region lies within Bussing's (1976) Rio SanJuan fish province with about 55 reported fresh-water species. An additional 75 marine specieshave been collected from marine and estuarineenvironments of the park (Gilbert & Kelso 1971,Winemiller unpublished).

Collections

This study uses species' numerical abundancesfrom 16 sites that were intensively sampled from

June-August 1983, February-December 1985, andAugust 1986. We combined our species numericalabundances with primary data reported by Gilbert& Kelso (1971) from earlier collections at many ofthe same sites. Because our collections were morefrequent and intense at southern and inland loca-tions and earlier collections were more frequent atnorthern estuarine and ocean/beach sites, the twodata sets complemented one another in providing adata set portraying regional species distributionsand relative abundances across a broad range ofaquatic habitats. Fishes were collected using seines(6.2 x 1.9m, 4.7mm mesh; 2.5 x 1.9m, 3.2mm

mesh), dipnets (3.2mm mesh), castnets (1.9mdiam., 1 cm mesh), gillnets (25 m, 5 cm mesh; 25 m,10cm mesh), and hook and line. In addition, Gil-bert & Kelso (1971) used rotenone at several of theinland stream sites. At a given site, all aquatichabitats were sampled intensively with the primary

throughout 1985. Nordlie & Kelso (1975) observeda tongue of saline water (8.7-10.5ppt) near thebottom of the Laguna Tortuguero (depths> 5 m)during both dry and wet seasons. Nordlie & Kelsoalso reported high-tide salinities in surface watersof the upper Laguna Tortuguero between 0.07 pptduring the wet season and 0.7 ppt during the dryseason. One of us (KOW) measured surface salin-ities less than or equal to 0.1 ppt in the same areathroughout the year 1985.

The Lomas de Sierpe are drained by smallstreams that join east-flowing tributary streamsthat drain the foothills of the central mountainrange lying 50-60 km to the west. Many of thepark's rivers have shifting sand substrates and ex-hibit multiple channels, or braiding, prior to emp-tying into the network of estuarine lagoons. Someof the old river channels, like Cano Agua FriaViejo, are in excess of 3 m depth in some stretches,and exhibit little or no water current during dryperiods. Salinity measurements never reached0.1 ppt over the course of a year in Cano Agua FriaViejo (Winemiller 1987). Dense mats of floatingaquatic plants (Eichhornia, Hydrocotyl, Salvinia,Azolla) sometimes cover the surface of these smallcallOS during extended dry periods. Temporary for-est pools and small creeks (e.g., Quebrada) liescattered about the eastern lowlands and barrierislands that form partitions between sea and la-goons. These small creeks have negligible gra-dients and drain directly into the lower reaches ofrivers and lagoons. Salinity measurements from theupper reaches of Quebrada and nearby forest poolsnever exceeded 0.1 ppt during 1985.

The Central American freshwater fish fauna iscomprised of three basic groups: (1) a major com-ponent of South American-derived freshwater spe-cies, (2) a very minor component of North Amer-ican-derived species, and (3) marine-derived pe-ripheral species (Myers 1966, Miller 1966, Bussing1976). The last group corresponds to species be-longing to predominantly marine or brackish waterfamilies that are physiologically and ecologicallyadapted to live either all, or a portion, of their livesin freshwater environments (Myers 1938,1966).Myers (1938) classified Central American speciesfrom freshwater families as either primary division

33

phyte composition and relative coverage were re-corded at each site. Habitat measurements weregenerally made between 1200 and 1400 h on the daythe sampling effort was initiated. Complete recordsof sample sites and dates are on file at the TexasNatural History Collection of the Texas MemorialMuseum. Sample sites used in the analysis are asfollows: (1) ocean/beach environment adjacent tothe park and Tortuguero village, (2) the mouth/outlet of the Laguna Tortuguero, (3) Laguna Tor-tuguero between its outlet and national park head-quarters, (4) Rio Tortuguero between Cafio Ben-jamin and its confluence with the Laguna Tortu-guero, (5) Cafio Benjamin, a sluggish, blackwaterwestern tributary of the Rio Tortuguero, (6) sidechannel of the Rio Agua Fria (station 31 in Gilbert& Kelso, 1971), (7) Rio Agua Fria, a stretch within1 km of Agua Fria park station, (8) small west-slope streams draining the Lomas de Sierpe andentering the Rio Agua Fria in the vicinity of AguaFria park station, (9) Cafio Agua Fria Viejo, an oldswampy channel (braid) of the Rio Tortuguero,(10) east-slope streams draining the Lomas deSierpe and entering old channels of the Rio Tortu-guero northeast of the Lomas, (11) forest pool,isolated during the dry season and approximately100 m from Cafio Agua Fria Viejo, (12) Quebrada,a small stream draining the barrier island and lyingon the park's northern boundary approximately20 m north of the national park headquarters, (13)Cafio Servulo, a deep, sluggish blackwater sloughentering the Laguna Tortuguero in the central east-ern park, (14) Laguna Jalova, a temporary inlet atthe park's southeastern border, (15) Cafio Negro, anarrow, winding channel connecting the LagunaTortuguero and Laguna Jalova, (16) Cafio Cali-fornia, a river forming part of the park's southernboundary and flowing into the Laguna Jalova (Fig.

1).

goal being attainment of a sample with all species inrelative proportions approximating those of thelocal assemblage. A given sampling effort lastedfrom approximately 0.5 to 5 days, and the effortwas continued until an hour or more of collectingproduced no additional species for the sample (fur-ther details given in Winemiller 1987, 1990). Obvi-ously, some sites were more efficiently and fre-quently sampled than others, however by summingmany individual samples into annual site totals andconversion of species' numerical counts into frac-tional abundances, our estimates of species' rela-tive abundances and spatial distributions shouldhave minimized most strong systematic biases.Two modifications were made on entries in theGilbert & Kelso (1971) data set. Gilbert & Kelso'ssample of 2689 Pomadasys crocro larvae from theLaguna Tortuguero on August 15, 1963 represent-ed an unusual collection from a single massive ag-gregation of migrating fish larvae. Because thislarge number would have greatly affected the en-tire distribution of species' numerical relativeabundances for Laguna Tortuguero, it was enteredas one individual for the present analysis. Thisadjustment yielded a total of 95 individuals for thesite, which we feel more accurately represented theannual relative abundance of the species within theassemblage context. Also, three species were listedby Gilbert & Kelso as 'many' in the central LagunaTortuguero (Eleotris spp., Gobionellus fasciatus)because they were the predominant fish compo-nents of the tismiche (huge migrating aggregationsof postlarval shrimp and larval fishes). We enteredeach of these as 250 individuals.

At each sample site during the 1983-1986 collec-tions, stream width and depth were measured tothe nearest 1 cm with a tape (at 3-4 points in alateral transect). Water current velocity was esti-mated by timing the passage of a film canister filledhalf-way with water over a measured distance inmidchannel. Dissolved oxygen was estimated byWinkler titration, salinity was measured by meteror calculated from density measurements, pH wasmeasured with reagent grade pH paper, and watercolor was estimated with the Hazen scale (eachperformed 2-6 x per sample date). Qualitativeshoreline characteristics, substrate, aquatic macro-

Data analysis

Fishes were identified following Lopez-Sanchez(1968), Gilbert & Kelso (1971), Villa (1982), Rivas(1986), and Bussing (1987). Specimens were depos-ited in the Florida State Museum, Gainesville (Gil-

34

bert & Kelso 1971), the Natural History Museum ofthe University of Costa Rica, San Jose, and theTexas Memorial Museum, Austin. Species' numer-ical abundance data were combined for sites thatwere sampled repeatedly, and sample totals wereconverted to species' relative abundances. Differ-ences in relative abundances between wet and dryseasons were recorded for Quebrada, Cafio AguaFria Viejo, Laguna Tortuguero, Laguna Tortugue-ro mouth, Cafio Negro, and forest pool. Speciesdiversity at each site was calculated as the index ofShannon & Weaver (1963):

H' = - L Nj/N loglo Nj/N

as well as linear distances to the coast were mea-sured to the nearest 0.1 km on a 1: 50000 map(Servicio de Parques Nacionales, Ministerio deAgricultura y Ganaderia, San Jose, Costa Rica). Inaddition to DCA, a direct gradient analysis wasperformed with de trended canonical correspond-ence analysis (DCCA), a multivariate techniquethat ordinates sites using information from bothspecies abundances and a set of environmental var-iables (Ter Braak 1986, 1988). In addition to fluvialdistance, we placed the following environmentalvariables into discrete categories: average channelwidth, average depth, water current, water color(clear versus tannin-stained), salinity (calculatedfrom density), and substrate. Categories werebased on the measurements or estimates that wererepeated over time at each site, with the exceptionof the lomas creeks which were sampled only dur-ing June 1985. Recorded temperature, dissolvedoxygen, and pH values did not show large between-site or temporal variation and thus were excludedfrom further consideration.

Euclidean distances based on site scores on thefirst two DCA axes were used as estimates of simi-larity in clustering by a hierarchical linkage algo-rithm (Wilkinson 1989). A jackknifing method wasused to estimate the consistency of nodes and ter-minal units of the dendrogram (Ibarra & Stewart1989). Consistency of nodes and site arrangementsin the dendrogram were estimated by comparisonswith a series of dendrograms in which one of the 16sites was removed sequentially from the dgta set.The trophic status of each species was categorizedas either piscivore, invertebrate-feeder, omnivore,or herbivore (the latter includes varying degrees ofdetritivory) based on stomach contents data gath-ered by Winemiller (1987, 1990, unpublished,Winemiller & Pianka 1990).

The Shannon index is a function of two compo-nents, species richness (i.e., length of the frequen-cy distribution of species' relative abundances) andevenness (i.e., degree to which a distribution isplatykurtic), and here we contrast H' with speciesrichness as a simple comparative index of relativeevenness (Tramer 1969, Hurlbert 1971). Turnoverof species between adjacent sites along longitudi-nal gradients was examined by calculating the per-centage of species shared between adjacent sitesfrom downstream to upstream, and by plotting thepercentage of original species remaining from agiven site while moving through sequential samplesalong the upstream gradient.

Detrended correspondence analysis (DCA)served as an ordination method for identifying as-semblage gradients based on species' relative abun-dances (Hill 1979, Hill & Gauch 1980, Gauch1982). We used the rare-species downweightingoption on raw relative abundances and 26 segmentsfor detrending site scores on the gradient axes. Wealso performed DCA without downweighting andwith detrending by polynomials rather than seg-ments using CANOCO (Ter Braak 1986, 1988).Since very little variation was observed among testsusing the same data set, especially on the first gra-dient axis, we report only the results from DCAwith downweighting and de trending by segments.Correlations between fluvial distance (i.e., riverdistance between site and sea outlet) and the firstDCA axis provided one indirect estimate of gra-dient analysis (Ter Braak 1986). Fluvial distances

Results

Species richness of Tortuguero Park fish assem-blages varied from a low of 6 in the forest pool to ahigh of 80 in the Laguna Tortuguero. High faunalturnovers (i.e., the inverse of speci~s retention),ranging from about 60% to more than 90% re-

35

80

'"

'"";c(Q

<(Jc

~II'II'II'

P II', " 0--1_'" U 1' , , , I '. --O,

60-

40;

20

80-

60-

40,

20-

100-

80-

2"00C/):J0

">Q)'-Q.

E

,g1:JQ)c:

"co"'Q)'-C/)Q)'(3Q)Q.C/)

"6Q)0)co"EQ)uQ)

Co

",n..th "0-_---0

west lomas

10 1 5 20 25 30

b

N

'"'X(U

ct(Ja

13

1\-..4\

200.7

~

~100- ~0 I , i

0 200 400 600 ---DCA axis 1

Fig. 3. Plots of the first two DCA axes based on species relativeabundances for all 16 sites (a), and 15 sites with the isolatedforest pools excluded (b). Lines connect adjacent sites withinthe Rio Tortuguero drainage (stemming from site 1) and CaiioCalifornia drainage (stemming from site 14).

0 5 10 1 5 20 25 30 Rnn

60

40

20

0 2 3 4

Fluvial distance (km)6

Fig. 2. Species turnover (expressed as percentages of speciesshared between adjacent sites moving upstream) in the RIoTortuguero drainage (top), Caiio Servulo drainage (middle),and Caiio California drainage (bottom).

placement of species from the previous down-stream station, were observed in 9 out of 14 transi-tions in the Rio Tortuguero/Laguna Tortuguerodrainage and two of three transitions tested in theLaguna Jalova drainage (Fig. 2, Appendix 1).Some of these distinct faunal transitions occurredover distances as small as a few meters, as betweenthe Laguna Tortuguero and the small barrier islandcreek, Quebrada (20% species retention). Five ofthe 21 species recorded from Quebrada (Gymno-tus, Rhamdia, Rivulus, Synbranchus, Cichlasomaalfari) were never collected from the adjacent la-goon.

The first DCA axis described a longitudinal gra-dient running from low-order tributary streamscontaining relatively few species (low scores) tomore speciose downstream sites and finally theocean (Fig. 3, eigenvalue = 0.84.4). The secondDCA axis (eigenvalue = 0.328) was much moredifficult to interpret, but appeared to order trib-utary stream sites along an east (high scores) towest (low scores) gradient when all sites were in-cluded in the analysis (Fig. 3a). When the isolatedforest pool (site 11) was eliminated from the analy-sis, the ordination of the remaining sites on the firstDCA axis was virtually unchanged (Fig. 3b), andordination of sites on DCA axis 2 yielded no gener-al pattern of interpretation (Caiio Servulo, Negro,and Laguna Jalova with high scores versus CaiioCalifornia and a side channel of the Rio Agua Friawith low scores).

The longitudinal pattern of zonation impliedfrom the first DCA axis can be seen clearly in thespatial pattern of distribution of the most common

I ' , I \I '" I \

I " I \'6 I \

I \I \

\I \I \: b0 east lomas

0 I , forest pool

0 5

35

80

'"

III";c(Q

ctUC

~IIIIII

II IP II I

I " 0--1_1" . 1' , , , I '. --O,

$"USIJ):J0

">Q)'-Cl.

E

.g1:JQ)c:

"CO

"'Q)'-IJ)Q)

'(3Q)Cl.IJ)

"6

Q)OJCO"EQ)u

Q5Co

60 -

40;

20-

80c

60-

40,

20 c

100-

80

,""uth '0._---0

west lomas

10 1 5 20 25 30

b

N

III"j(IU

~(J0

13

't200.

7

--\100- ~ ~

0 I , i0 200 400 600 '---

DCA axis 1

Fig. 3. Plots of the first two DCA axes based on species relative

abundances for all 16 sites (a), and 15 sites with the isolated

forest pools excluded (b). Lines connect adjacent sites within

the Rio Tortuguero drainage (stemming from site 1) and Caiio

California drainage (stemming from site 14).

0 5 10 1 5 20 25 30 ROO

60

40

20

0 2 3 4

Fluvial distance (km)6

Fig. 2. Species turnover (expressed as percentages of speciesshared between adjacent sites moving upstream) in the RIOTortuguero drainage (top), Caiio Servulo drainage (middle),and Caiio California drainage (bottom).

placement of species from the previous down-stream station, were observed in 9 out of 14 transi-tions in the Rio Tortuguero/Laguna Tortuguerodrainage and two of three transitions tested in theLaguna Jalova drainage (Fig. 2, Appendix 1).Some of these distinct faunal transitions occurredover distances as small as a few meters, as betweenthe Laguna Tortuguero and the small barrier islandcreek, Quebrada (20% species retention). Five ofthe 21 species recorded from Quebrada (Gymno-tus, Rhamdia, Rivulus, Synbranchus, Cichlasomaalfari) were never collected from the adjacent la-

goon.

The first DCA axis described a longitudinal gra-dient running from low-order tributary streamscontaining relatively few species (low scores) tomore speciose downstream sites and finally theocean (Fig. 3, eigenvalue = 0.84.4). The secondDCA axis (eigenvalue = 0.328) was much moredifficult to interpret, but appeared to order trib-utary stream sites along an east (high scores) towest (low scores) gradient when all sites were in-cluded in the analysis (Fig. 3a). When the isolatedforest pool (site 11) was eliminated from the analy-sis, the ordination of the remaining sites on the firstDCA axis was virtually unchanged (Fig. 3b), andordination of sites on DCA axis 2 yielded no gener-al pattern of interpretation (Caiio Servulo, Negro,and Laguna Jalova with high scores versus CaiioCalifornia and a side channel of the Rio Agua Friawith low scores).

The longitudinal pattern of zonation impliedfrom the first DCA axis can be seen clearly in thespatial pattern of distribution of the most common

i ' " \I '" I \

I " I \'6 I \

I \I \".1 \: b0 east lomas

0 I , forest pool

0 5

36

Table 1. Common species characteristic of each of the four major faunal zones and ubiquitous species common in two or more adjacentzones (I) of the aquatic environmental gradient in Tortuguero National Park.-Creeks

-Rivers Lagoons

-Ocean

Astyanax fasciatus

Bryconamericus scleropariusGymnotus cylindricusRhamdia guatemalensisRivulus isthmensisPriapichthys annectensCichlasoma septemfasciatum

Astyanaxfasciatus :.Alfaro cultratusPhallichthys amatesCichlasoma alfariCichlasoma nigrofasciatumGobiomorus dormitor . . . . . . . . . .. . .. Gobiomorus dormitorEleotris amblyopsis . . . . . . . . . . . . . . . . Eleotris amblyopsisEleotris pisonis

Atractosteus tropicus

Brycon guatemalensisRoeboides guatemalensis

Brachyrhaphis parisminaCichlasoma doviiCichlasoma nicaraguenseCichlasoma rostratum

Belonesox belizanusPoecilia gilliMelaniris milleriStrongylura timucuCichlasoma centrarchusCichlasoma citrinellumCichlasoma loiselleiCichlasoma maculicauda

Herotilapia multispinosaPomadasys crocro

Lutjanus jocuCentropomus pectinatusOostethus lineatusAchirus lineatus

Citharichthys spilopterusTrinectes paulistanusDormitator maculatusEvorthodus lyricusGobionellus boleosomaGobionellus fasciatusGobiosoma spes

Anchoa lamprotaeniaAnchoviella elongata

Myrophis punctatusHyporhamphus roberti

Centropomus parallelusDiapterus plumieriDiapterus rhombeusEucinostomus melanopterusBairdella ronchus

Micropogonias fumieriMicrodesmus carri

Bathygobius soporatorSphoeroides testudineus

Megalops atlanticusMugil curemaCaranx hipposCaranx latus

Oligoplites palometaCentropomus undecimalis

Harengula jaguanaOdontognathus compressusOpisthonema oglinumPellona harroweriColeotropis blackbumi

Polydactylus virginicusConodon nobilis

Pomadasys corvinaeformisLarimus breviceps

Ophioscion panamensisStellifer colonensis

37

Euclide~n dist~nce 200

Fig. 4. Dendrogram based on average clustering algorithmshowing site groupings based on DCA 1 and 2 scores. Horizon-tal axis is the average Euclidean distance between samples andnodes. Numbers by nodes indicate percent consistencies basedon the jacknifing technique. Seven basic clusters were indenti-fied: a single isolated forest pool, an inland stream group, twoinland river sites, two heterogeneous groupings comprised ofdownstream river sites, and a coastal group.

Tortuguero and the lower Rio Agua Fria (includingQuebrada) plus Cafio California, (2) Cafio Servu-10, Cafio Negro, Laguna Jalova, and the Rio AguaFria, (3) the two Lomas creeks, and (4) the forestpool. The Rio Agua Fria assemblage's affiliationwith southern coastal systems arises from similarloadings on DCA axis 2 (Fig. 3). Although sub-stantial differences between the two assemblageswere apparent, Rio Agua Fria and Cafio Negroshared several species in similar relative propor-tions (e.g., Brycon guaternalensis, Roeboides gua-temalensis, Cichlasoma dovii, C. rostraturn).

Site scores on the first DCA axis showed weaknegative correlations with linear and fluvial dis-tances from the coast (Fig. 5). Species diversity wasnot significantly correlated with DCA axis 1, andspecies richness showed a weak positive correla-tion. Species richness showed a general declinewith increasing distance from the sea within the Rio

Tortuguero/Laguna Tortuguero drainage (Fig. 6).Laguna Tortuguero contained the greatest numberof species (N = 80), followed by the swampy side-channel, Cafio Agua Fria Viejo (N = 58) locatedapproximately 8 km upstream. Species diversity(H') showed no consistent trend and relatively lit-tle variation (range = 0.68-0.89) over the course of

the flow between the lomas headwaters and theocean. Species richness and diversity were uncorre-

fish species (Table 1). A large number of specieswere characteristic of only one type of aquatic hab-itat in the region. For example the characid, Bryco-namericus scleroparius, and the poeciliid, Priap-ichthys annectens, were only found in the smallstreams of the Lomas de Sierpe, and the killifish,Rivulus isthmensis, was only taken in flooded for-ests, isolated forest pools, and the upper reaches ofQuebrada (also forested). Similarly, the characid,Brycon guatemalensis, and the poeciliid, Brachy-rhaphis parismina, were common in rivers but ex-ceedingly rare or absent from feeder creeks andlagoons. A large number of species were foundonly in the marine environment (Table 1, Appen-dix 1), whereas another group of essentially marinespecies were almost always encountered in the es-sentially freshwater (with brackish layer beneath)lagoons (e.g., Eucinostomus melanopterus, Sphoe-roides testudineus). Another large group is com-prised of species that were common in two differenthabitats, but generally not more than two. Exam-ples include Alfaro cultratus inhabiting creek-poolsand river-edge habitats, and the pike livebearer,Belonesox belizanus, common along shorelines ofboth rivers and lagoons. Only three of the 130species collected from the region could be consid-ered common components of the fish assemblagesof more than two general habitat types. The pri-mary division fish, Astyanx fasciatus was abundantin nearly all aquatic environments of the park, withthe exclusion of lagoon outlets and the marine envi-ronment. The sleepers, Eleotris amblyopsis andGobiomorus dormitor were taken in small trib-utary streams, rivers, callOS, and lagoon littoralzones. Eleotris amblyopsis, was thought to be thefirst or second most important larval fish compo-nent of the tismiche (huge aggregations of postlar-val palaemonid shrimp and larval fishes) enteringthe Tortuguero lagoon (Gilbert & Kelso 1971,Winemiller unpublished).

Clustering based on similarities of scores on thefirst two DCA axes produced large separationsbetween fish assemblages of the ocean, LagunaTortuguero outlet, and the other 14 sites (Fig. 4).Within the freshwater/estuarine cluster, four sub-groups are identified at distance of 63 DCA units:(1) sites running along a gradient between Laguna

38

5 10 15

Linear distance from coast (km)

~

(/)

';<m

«uc

c:0

Q)...0u

cn

0 5 10 15 20 25 30

Fluvial distance from coast (km)

Fig. 5. Plots and correlations of site scores on the first DCA axis with site distance from coast, fluvial distance, species diversity, andspecies richness.

(e.g., Caiio Servulo). In general, clear headwatersgradually gave rise to increased levels of stainingfrom leached organic compounds in the downstreamdirection. Fast water currents (> 0.35 ms-l) wereonly observed on the ocean coast and in riversduring periods of high rainfall. The transition be.tween freshwater and saline water is apparentlyrather abrupt near the surface of the water column(i.e., lagoon salinity measurements never exceed-ed 1.0ppt over 10 months in 1985) whereas mostmixing probably occurs near lagoon mouths. Bot-tom salinities in coastal lagoons exhibit a muchmore complex pattern of variation. Salinities atbottom depths ranged between 8.7 and 10.5 pptduring both wet and dry seasons over a distance of5 km in the lower Laguna Tortuguero (Nord lie &Kelso 1975). During the dry season, lagoon bottomsalinities ranged from 8.7 to 10.3 between low andhigh tides.

lated within the Rio Tortuguero drainage (r =0.09).

Plots of percent species remaining at successiveupstream sites starting from three initial down-stream sites in the Rio Tortuguero drainage(ocean, mouth, lagoon) show a rapid initial turn-over of species within the first 10 km (Fig. 7). Be-yond 10 km, the rate of species deletions and re-placements falls off sharply until all of the originalocean and lagoon outlet species are replaced in thecreeks on the western slope of the lomas. Only onelagoon species, Gobiomorus dormitor, was foundboth in the lagoon and small creeks on the westernslope of the lomas.

Environmental characteristics of the 16 samplesites varied widely, often over very short distancesbetween adjacent habitats (Table 2). Substrateswere generally sandy with vegetative litter and siltcollecting in forest pools and deep, sluggish canos

39

InInQ)~

""

.~

InQ)'uQ)Co

In

Fig. 6. Relationship between species richness and diversity plot.ted by fluvial distance in the Rio Tortuguero drainage (longitu-dinal course from mouth to western lomas streams).

0 10 20 30

Fluvial distance (km)

Fig. 7. Percentages of fish species from the initial assemblage

remaining in comparisons of successive longitudinal samples

moving upstream from the ocean/beach (a), Laguna Tortuguero

mouth (b), and Laguna Tortuguero (c).

chus were more abundant in Laguna Tortugueroduring the dry season.

Relative percentages of species assigned to fourtrophic categories (herbivore/detritivore, omni-vore, invertebrate feeder, piscivore) exhibited in-tersite variation (Fig. 8). Contingency table analy-sis for frequencies of species in four trophic cate-gories in 16 sites was nonsignificant (X2 = 39.3,45 df, P = 0.71). Invertebrate feeders were the

most species rich group at 15 of the 16 sites. Theomnivorous cichlid, Cichlasoma nigrofasciatum,was a numerically-dominant species at Calio Servu-10, a blackwater site with low fish density and spe-cies richness (N = 14). When fish assemblages

were grouped by basic habitat (4 creeks,S rivers,Slagoons, 1 ocean), significant associations emergedfor both the number of individuals collected andnumber of species collected within the varioustrophic categories (Table 4). Mean relative per-centages of individuals in different trophic cate-gories was significantly associated with habitat(X2= 81.5, 9df, p< 0.0001). Sampling in a down-stream progression, herbivorous fishes droppedout of assemblages, from a high of 30% in creeks toa low of 5% in the sea. Omnivores were not adominant group in any habitat and were entirelyabsent from the sea. Much of the decline in thesetwo groups was associated with a reciprocal in-crease in invertebrate-feeding fishes in lagoon andocean environments. Mean relative percentages ofspecies in different trophic categories showed the

Canonical correspondence analysis (DCCA) re-sulted in the same ordination of sites along the firstgradient axis (eigenvalue = 0.782) with coastalsites having low scores and headwaters having highscores. A Monte Carlo test of statistical signif-icance employing 99 null randomizations (TerBraak 1988) resulted in p < 0.025 for the first axiswith an overall trace at p < 0.05. Of the 7 physicalvariables listed in Table 2 (water type = salinity,clarity), channel width (- 185), depth (134), andsalinity (144) had high loadings on the first gradientaxis (fluvial distance - 63, current - 79, clarity- 62, substrate 79). The second axis (eigenvalue =0.323) was also influenced primarily by channelwidth (- 91) and depth (83) plus substrate (109).Loadings of the other four environmental variableson axis 2 ranged between - 77 (clarity) and 2 (flu-

vial distance).When data were broken down by season, several

common species in Laguna Tortuguero showedsubstantial seasonal shifts in their relative abun-dances (Table 3). Melaniris milleri, Eucinostomusmelanopterus, and Eleotris amblyopsis had thegreatest relative abundances during the wet sea-son. Wet season increases in the numbers of E.amblyopsis in the Laguna Tortuguero were duelargely to downstream rafting in floating mats ofaquatic vegetation torn free from swampy up-stream locations by fast water. M. milleri had high-er levels of reproduction during the wet season atTortuguero (Winemiller unpublished). Marinespecies Centropomus pectinatus and Bairdiella ron-

?:"~Q)>

"C

'"Q)"QQ)C-oo

40

same trends as individual percentages (X2 = 43.3,9 df, P < 0.0001). The absolute number of differentspecies revealed no consistent pattern betweentrophic categories and habitats (X2 = 16.46, 9 df,P = 0.058).

Discussion

Fish assemblages of the Tortuguero region exhibita strong pattern of longitudinal zonation over afluvial distance of only 28 km between creeksdraining the Lomas de Sierpe and the sea. Thislongitudinal zonation is similar to patterns observ-ed in temperate zone coastal faunas (Felley 1987,Rozas & Odum 1987, Odum 1988) and is accompa-nied by a series of species replacements and dele-tions in successive downstream locations. Yet, thesite clusterings that combined two drainages in-dicated that fluvial zonation does not always have ahigh correspondence with spatial coordinates. Thecomplex mosaic of waterways within TortugueroPark resulted in poor correlation between zonationof fish assemblages and minimum linear distance tothe sea coast. Streams on the western slope of the

Lomas de Sierpe flow west before joining the RioAgua Fria, which in turn flows north to join thenortheast-bound Rio Tortuguero. Streams drain-ing the eastern slope of the lomas flow either northto join the Rio Tortuguero drainage, east to palmswamps associated with Cafio Servulo, or south tothe Rio Sierpe. The unusual arrangement of la-goons, parallel and in close proximity to the coast,is a dominant feature of the region, and one whichgreatly reduces correlations between linear dis-tance and fluvial distance. In addition, the coastallagoons receive effluent waters from numerous in-dependent tributaries that vary greatly in size andenvironmental conditions. For example, the Lagu-na Tortuguero/Laguna Jalova network has six ma-jor western tributaries plus an unknown number ofsmall forest creeks entering from the west and fromthe barrier island to the east. Due to the high andfairly evenly distributed annual rainfall, the smallcreeks, even those under tidal influence, are essen-tially freshwater environments throughout the

year.At least six of the common species from creek

assemblages appeared to be largely restricted tocreek environments (Table 1). Quebrada is sep-

Table 2. Some physical features of the principal sample sites during the wet season. Water current was classified as slow (S, 0-0.10 mS-I), moderate (M, 0.10-0.35 m S-I), or fast (F, > 0.35 m S-I). Water type was classified as clear (C) or tanin-stained (T) and saline (S,salinity 10-35 ppt), brackish (B, 1-10ppt), or fresh (F, < 1 ppt). Substrate type was classified as fine silt/mud (M), sand (S), or vegetativelitter (L).

Site Fluvial distance Ave. width

(m) (m)

Ave. depth

(m)

Current Substrate typeWater type

FFMISF/MMISMF/MSINMMISNSSSSM

C,

B,

B,

B,

B,

T/(

C,

C,

T,

C,

C,

C,

T,

T,

T,

T,

SSS/LS/LSlUMS/LS/LS/LS/LSUMS/L/MUMSS/LSIL

-110300404.0

207.01.7

26.31.56.02.5

501005060

-6.0

10.02.01.01.51.10.31,50.50.40.4

10.06.02.02.0

OceanMouth Laguna Tortuguero

Laguna TortugueroRio TortugueroCaiio BenjaminRio Agua Fria ArmRio Agua FriaWest Lomas CreeksCaiio Agua Fria ViejoEast Lomas CreeksForest PoolQuebradaCaiio ServuloLaguna JalovaCaiio NegroCaiio California

00.13.59.5

10.518.724.528.211.716.011.85.5

25.70.11.95.6

SSIBBIPFF:, FFFFFFFFBBIPBIP

41

freshwaters upstream from lagoons and droppedout altogether in the smaller rivers and streams(e.g., Centropomus spp., Eucinostomus spp., Ca-ranx spp., Mugil curema). Similarly, most if not allof the species from freshwater families cannot tol-erate salinities at levels encountered at the lagoonIt)outh and sea coast. Yet, a number of freshwaterspecies were firmly established components of theupper and middle reaches of coastal lagoonsthroughout the year. The characid, Astyanax fas-ciatus, was a dominant species at a large number ofsites, including coastal lagoons. A. fasciatus is ageneralized omnivore (Winemiller 1982, 1987,1990) with a very wide geographic distribution. Asthe species is currently defined, A. fasciatus rangesthroughout much of the neotropics from northernArgentina to central Mexico (Miller 1966). Where-as A. fasciatus was the most numerically abundantspecies within the region (comprising approximate-ly 10% of the total regional fish sample), two pe-ripheral freshwater species from marine-affiliatedfamilies, Eleotris amblyopsis and Gobiomorus dor-milar, showed the widest distributions along thelongitudinal stream gradient. Gobiomorus and sev-eral marine species (Carcharinus leucas, Pristis pe-rotteti, Tarpon atlanticus) have established stocksas far inland as Lake Nicaragua (Villa 1982, Buss-

ing 1987).

arated from the coast by only 200-300m (lineardistance), but its faunal composition was a blend ofspecies characteristic of inland creeks and small,edge-dwelling lagoon and river species. Hyphes-sobrycon, Gymnotus, Rhamdia, and Rivu/us ex-hibited large gaps in regional distribution betweenlomas streams and Quebrada. The Rio Tortugueroand lagoon environments apparently inhibit inva-sion by these and other inland species for reasonsthat are currently unknown. Salinity alone does notrestrict freshwater fishes from invading the la-goons, because the upper water column layer isgenerally < 1 ppt and is essentially freshwater«O.lppt) during the 7-8 wettest months. Oppor-tunities for dispersal from inland to coastal creekswould abound during wet seasons in the form ofdrifting rafts of Hydrocoty/ and other floating aq-uatic macrophytes.

Inland penetration and establishment of manymarine fish species is clearly a consequence of limi-tations in physiological mechanisms for dealingwith a hypotonic environment. Sixty-two percentof fishes comprising the ocean assemblage werenever found beyond the mouth of the lagoon, and86% were never encountered further inland thanthe coastal lagoons. Several essentially marine spe-cies with obvious capabilities for adjusting to thefreshwater environment were extremely rare in

Table 3. Fish species showing seasonal shifts in relative abundance in Laguna Tortuguero in the region of the national park headquarters

during 1985.

Species Wet months Dry months

No. sampled Relative abundance No. sampled Relative abundance

Melaniris milleri

Centropomus pectinatusEucinostomus melanopterusBairdella ronchusEleotris amblyopsis

Primary division speciesPrimary div. individualsSecondary div. speciesSecondary div. individualsPeripheral & marine speciesPeripheral & marine indiv.Total speciesTotal individuals

8481419

76

213107537

589

(0.122)(0.118)(0.059)(0.013)(0.110)

(0.04)(0.02)(0.20)(0.11)(0.76)(0.87)

34330

9146

4

(0.044)(0.429)(0.012)(0.190)(0.005)

(0.02)(0.02)(0.20)(0.06)(0.78)(0.92)

179

4236

71049

67946

769

42

14 1513

Sample site

Fig. 8. Distributions of percentages of total individuals collectedfrom four trophic categories in 14 Tortuguero Park fish as-semblages. Numbers above bars are total number of speciescollected in each category at each site (fractional values resultedfrom species exhibiting major ontogenetic feeding shifts thatreceived half assignment to two different trophic categories).

cies encountered in greater abundances in drainag-es to the west of Tortuguero Park (Bussing 1987,Winemiller unpublished). Similarly, Cichlasomanicaraguense, C. septemfasciatum, and C. tubawere much more abundant in forested regions tothe west of the park. Several related taxa have verysimilar ecologies and seemed to replace one an-other in different habitats. As stream size increas-es, Alfaro cultratus replaces Priapichthys annectensas the dominant insectivorous poeciliid. In some ofthe smallest forest streams, Priapichthys may bejoined by a second insectivorous poeciliid, Brach-yrhaphis holdrigei. In rivers and callOS, Alfaro in-habits quieter and shallower waters than another

surface-dwelling insectivore, Brachyrhaphis paris-mina. The algae-grazing poeciliid Phallichthysamates was more abundant in smaller streams, andthe larger Poecilia gilli was more prevalent in la-goons. The two species coexisted in large numbersin intermediate environments such as river back-waters and callOS. Cichlasoma tuba is a large detri-tivorous/algivorous cichlid of clearwater rivers,such as the Rio Agua Fria, Rio Sierpe, upper CanoCalifornia, and higher gradient rivers lying to thewest. Cichlasoma tuba is entirely replaced by thedeeper-bodied C. maculicauda in the lower reachesof rivers, callOS, and coastal lagoons.

These observations lead us to the general ques-tion of whether regional patterns of species turn-over reflect fishes' physiological responses to theirphysicochemical environment, or response to thebiological forces of predation and competition?For example, Matthews (1987) found good corre-spondence between the responsiveness of NorthAmerican cyprinids to harsh conditions in labora-tory tests and their distributions in the wild. Threeout of seven physical environmental variables thatwe included in DCCA had high loadings on the firstordination axis. Among these, the physiologicaleffects of salinity level are obviously of major im-portance in structuring fish assemblages across thelower reaches of the freshwater/marine ecotone(i.e., from deep waters of lagoons to the ocean).Salinity was probably not a major factor influen-cing fish assemblages further inland. The other twodominant variables in DCCA, stream width anddepth, are components of habitat size. Of course

Zonation of Tortuguero fish assemblages corre-sponded to four basic habitats: creeks, rivers, la-goons, and the sea. Even though creek habitatsexhibit an irregular spatial arrangement within theregion's mosaic of winding, low-gradient watercourses, fairly distinctive fish assemblages can bemapped over a general longitudinal pattern of flu-vial habitat zonation that in large part correspond-ed to the first axis from DCA. Lomas creeks con-tained three species that were encountered in noother region of the park. Apparently, the Lomas deSierpe constitute the eastern-most limit of suitablehabitat for Bryconamericus scleroparius, Priapich-thys annectens, and Brachyrhaphis holdridgei, spe-

43

many variables, both measured and unmeasured,might be expected to covary to some extent withhabitat size within a fluvial gradient. Yet, the mostspecies rich assemblages were not necessarily asso-ciated with the largest habitats. Laguna Tortugue-ro yielded a richer ichthyofauna (80 species) thanthe ocean/beach (42 species). Some of this differ-ence could have been associated with lower effi-ciency in sampling the deeper and more turbulentwaters of the coast. Cafio Agua Fria Viejo had thesecond highest number of species (N = 58) butonly ranked eighth in size of habitat (Table 2).Species richness in this swampy channel is probablyinfluenced by a habitat edge effect (i.e., 'mass ef-fect' of Shmida & Wilson 1985) that fosters ele-ments associated with both river and lagoon envi-ronments. In effect, callOS and backwaters repre-sent an additional kind of ecotone involving a tran-sition between (1) small clearwater riversdominated by freshwater division fishes and (2) thepredominantly lentic environment of lagoons con-taining many marine faunal elements. Several fishspecies common in either rivers or lagoons, but notboth (Table 1), coexisted in Cafio Agua Fria Viejo.In addition, cafio and lagoon habitats contained thegreatest structural heterogeneity in the form of

diverse aquatic vegetation, submerged trees andlimbs, coarse detritus, and leaf packs. Species di-versity of fishes and other vertebrates is often posi-tively correlated with habitat structural heteroge-neity (Gorman & Karr 1978).

Compared with physicochemical parameters,the relative influence of biotic interactions on thestructure of Tortuguero fish assemblages is some-what more difficult to infer. However, additionalecological information concerning two of the as-semblages indicates that biotic interactions prob-ably assume a greater relative role in structuringmore species rich assemblages. Winemiller (1991)performed an ecomorphological analysis of thecore species from the Quebrada and Calio AguaFria fish assemblages, plus eight other freshwaterfish assemblages from three continents. The totalrange of estimated morphological space wasgreater in the more species-rich Agua Fria Viejoassemblage compared with Quebrada fishes, andthe average distance between species was not sig-nificantly smaller in more diverse systems. Thesepatterns suggest greater levels of ecological special-ization in the Agua Fria Viejo assemblage in associ-ation with either an expanded resource base, spe-cies packing, or a combination of both. In addition,

Table 4. Mean percentages of individual fishes and species in four trophic categories collected from four creek sites, five river sites, five

lagoon sites, and ocean beach sites in the Tortuguero region.

Mean percentages by number of individuals collected

Herbivores Omnivores Invert.-feeders Piscivores Total no. indiv

30%26%10%5%

4,549

15%

2ZO/o

7%

0%

3,405

39%39%60%85%

9,684

16%13%23%10%

3,376

CreeksRivers

LagoonsOcean

Total no. indiv. 20,914

Mean percentages by number of species collected

Herbivores Omnivores Invert.-feeders Piscivores Total no. spp.

CreeksRiversLagoonsOcean

Total no. spp.

14%10%9%2%

7

21%13%9%0%

11

51%55%51%60%

74.5

14%22%31%38%

37.5

37669442

130

2,54512,1515,485

733

44

Winemiller & Pianka (1990) employed a null modelto statistically examine two forms of trophic orga-nization in the Quebrada and Cafio Agua Fria Vie-jo assemblages. Both assemblages showed signif-icant community-wide resource segregation duringboth wet and dry seasons. Significant guild struc-ture was observed among Cafio Agua Fria Viejofishes during both seasons and among Quebradafishes during the wet season. Non-random guildstructure and resource segregation, in particular,strongly imply a major effect of biotic interactionson community structure (Lawlor 1980, Winemiller& Pianka 1990).

Relative proportions of both individuals and spe-cies in the four trophic categories reflected a majorlongitudinal gradient in the utilization of benthicdetrital and periphyton resources. Percentages ofherbivore-detritivores declined montonically alonga headwater to ocean gradient. In contrast tostream herbivores, which generally are dominatedby insects plus a significant number of algae-scrap-ing fish species in the neotropics (Power 1984,Winemiller 1990), phytoplanktivorous microcrus-tacea assume dominance in the role of primaryconsumer in lentic freshwater and marine envi"-ronments. Fruits and seeds are additional re-sources available to freshwater herbivores and om-nivores that are essentially unavailable in the sea.The sharp decline in the number of individual her-bivorous and omnivorous fishes at sea was associ-ated with a large increase in the relative fraction ofinvertebrate feeders, mostly in the form of zoo-planktivores. While the relative proportion of pis-civores was actually smaller in the sea than fresh-water habitats, the fraction of species that werepiscivorous was greater. Marine piscivores were forthe most part large species (e.g., Tarpon, Thunnus,Sphyraena), and the disparity between numericaland specific proportions could be attributed tosampling error associated with the high mobilityand patchy distributions of these species in thenearshore marine environment. Even so, data fortrophic categories clearly reflect major differencesin the resource base in different Tortuguero hab-itats. Most primary consumption is by zooplanktonin the nearshore marine environment, and thegreat importance of this resource for fishes can be

seen in the dominance of the marine assemblage byinvertebrate-feeders.

Patterns observed in the Tortuguero fish faunaand their relations to characteristics of the region'saquatic environments generally agree with recentconceptual models of factors influencing longitudi-nal stream succession reviewed recently by Van-note et al. (1980), Fisher (1983), Schlosser (1982,1987), and Odum (1988). The few fish species in-habiting small feeder streams and forest poolswere, for the most part, small species capable ofrapid colonization of periodically disturbed hab-itats. Viviparity and rapid maturation make poeci-Ii ids particularly efficient colonizers. The cyprino-dontid killifish, Rivulus isthmensis, may be themost effective colonizing species at Tortuguero.Rivulus exhibited continuous reproduction, smallclutches, and very rapid maturation, traits associ-ated with a high intrinsic rate of population growthand colonizing ability in fishes (Winemiller 1989).These disturbances occur in the form of seasonaldessication of ephemeral, forest pool habitats, andshort-lived but turbulent flash floods in the highergradient streams of the lomas. Generalist insecti-vores and herbivores/detritivores predominate inthese small creek assemblages. The rainforest ca-nopy is generally quite dense, blocking much of thesunlight. As a consequence, a very large fraction ofthe total energy consumed by fishes and aquaticinvertebrates is of terrestrial origin in these smallforest streams. Ants and other terrestrial insectswere a principal fish food resource in forest streamsand pools. Although competition among fishes forlimited space and food resources is probably im-portant in these small oligotrophic systems, it isprobably intermittent due to periodic disturbancesfollowed by episodes of recolonization.

Downstream, aquatic environments becomelarger and relatively more stable in terms of thedisruptive effects of increased discharge rates dur-ing the rainy season. A large portion of the aquaticvegetation mats in canos are swept out during theperiods of highest rainfall. Dense rafts of aquaticplants, some as large as 50 m2, were commonlyobserved floating down the estuary toward the seaduring summer wet seasons. These rafts transportinvertebrates and fishes, especially eleotrids, from

45

tions to this general spatial segregation by smallfishes are seen in schooling midwater planktivores(i.e., Melaniris, Anchoa, Anchoviella) and crypticand burrowing benthic fishes (e.g., Citharichthys,Microdesmus). The lotic/lentic ecotone embodiedin the cafios seems to harbor high numbers of spe-cies from adjacent creek and river habitats by pro-viding excellent structural refugia for small fishes,and stable deepwater habitat for large species suchas gar, Atractosteus tropicus, and tarpon snook,Centropomus pectinatus. A blend of Schlosser's(1987) conceptual framework for the role of preda-tion, competition, and physical disturbance in tem-perate zone stream fishes and Odum's (1988) gen-eralizations about abiotic factors influencing coast-al marsh communities appears to explain many ofthe patterns produced by tropical fish assemblagesacross the complex freshwater/marine ecotone at

Tortuguero.

Acknowledgements

We thank the Servicio de Parques Nacionales deCosta Rica for its support of the field phases of thisstudy. We are especially grateful to F. Cortez, H.Haug, E. Chamorro, and J. Martinez. In addition,we thank E. Urbina, M. Betancourt, C. Martinez,R. Carlson, P. Lahanas, and J. Zabowski for pro-viding assistance during field collections. The Ser-vicio de Parques Nacionales de Costa Rica provid-ed a scientific collecting permit, and the Depart-mento de Vida Silvestre, Ministerio de Agriculturay Ganaderia provided an export permit for biolog-ical specimens. We thank E.R. Pianka for readingand commenting on an earlier version of the manu-script. Funding for field work was provided by theTinker Foundation and NSF grant BSR-8408379.

inland locations to the estuary. Although floodscertainly alter the structural features and primaryproduction base of rivers and canos (Winemiller1990), their effects on these inland fish communi-ties are probably small compared with headwaters.In these low-gradient regions, flood waters tend tobe much more evenly distributed over both timeand space than floods of higher gradient streams.Many of the smaller species of rivers and canosmoved into the flooded forest during the wet sea-son. A number of these fishes exhibited a seasonalburst of reproduction, leaving hundreds of off-spring in the flooded forest where predators wereencountered at extremely low densities (e.g., As-tyanax, Rhamdia, Poecilia, Cichlasoma nigrofas-ciatum, Herotilapia multispinosa). To some extent,seasonal reproduction by fishes influenced assem-blage composition in lagoons as well. Large aggre-gations of croakers, Micropogonias furnieri, withripe gonads were caught from the Laguna Tortu-guero by villagers during the late November of1985. A dense aggregation of juvenile croakers,Pomadasys crocro, was encountered in LagunaTortuguero in June 1963 (Gilbert & Kelso 1971).As noted previously, at least some of the seasonalshifts in relative species' abundances in Table 3were due to reproductive behavior. Several preda-tory marine species appeared to enter freshwatersprimarily for reproduction. KOW collected tar-pon, Tarpon atlanticus, and snook, Centropomusundecimalis, with ripe gonads at the mouth of theLaguna Tortuguero during irregular intervalsthroughout the year. These species feed while inthe estuary and undoubtedly influence communitystructure. In addition, the periodic arrival of thetismiche (huge aggregations of postlarval shrimpand larval fishes originating from the sea) intocoastal lagoons must have a major effect on com-munity trophic interactions.

River and lagoon assemblages contained morespecies of much greater average size and showinggreater levels of ecological and morphological spe-cializations compared with headwater faunas.Small fishes within these ecosystems are restrictedprimarily to shallow littoral zone habitats, presum-ably in response to predation on a proximate andperhaps even a coevolutionary time scale. Excep-

References cited

Allan, J.D. 1975. The distributional ecology and diversity ofbenthic insects in Cement Creek, Colorado. Ecology 56:1040-1053.

Angermeier, P.L. & J.R. Karr. 1983. Fish communities alongenvironmental gradients in a system of trQpical streams. Env.Bioi. Fish. 9: 117-135.

46

Horwitz, R,J, 1978. Temporal variability patterns and the distri-butional patterns of stream fishes. Ecol. Monog. 48: 307-321.

Hurlbert, S.H. 1971. The nonconcept of species diversity: acritique and alternative parameters. Ecology 52: 577-586.

Ibarra, M. & D.J. Stewart. 1989. Longitudinal zonation ofsandy beach fishes in the Napo River basin, eastern Ecuador.Copeia 1989: 364-381.

Kuehne, R.A. 1962. A classification of streams, illustrated byfish distribution in an eastern Kentucky creek. Ecology 43:608-614.

Lawlor, L.R. 1980. Structure and stability in natural and ran-domly constructed competitive communities. Amer. Nat.116: 394-408.

Lopez Sanchez, M.I. 1968. Clave para los peces de las aguascontinentales de Costa Rica (Key for the freshwater fishes ofCosta Rica). Fac. Cienc. Letras, Universidad de Costa Rica,San Jose. 31 pp. (in Spanish).

Lyons, J.L. & D.W. Schneider. 1990. Factors influencing fishdistribution and diversity in a small coastal river in southwest-ern Costa Rica. Hydrobiologia 203: 1-14.

Matthews, W.J. 1986. Fish faunal 'breaks' and stream order inthe eastern and central United States. Env. Bioi. Fish. 17:81-92.

Matthews, W.J. 1987. Physiological tolerance and selectivity ofstream fishes as related to their geographic ranges and localdistributions. pp. 111-120. In: W.J. Matthews & D.C. Heins(ed.) Community and Evolutionary Ecology of North Amer-ican Stream Fishes, University of Oklahoma Press, Norman.

Matthews, W.J. & H.W. Robison. 1988. The distribution offishes in Arkansas: a multivariate analysis. Copeia 1988: 358-374.

Meffe, G.K. & W.L. Minckley. 1987. Persistence and stabilityof fish and invertebrate assemblages in a repeatedly disturbedSonoran Desert stream. Amer. Midi. Nat. 117: 177-191.

Meffe, G.K. & A.L. Sheldon. 1988. The influence of habitatstructure on fish assemblage composition in southeasternblackwater streams. Amer. Midi. Nat. 120: 225-240.

Miller, R. R. 1966. Geographical distribution of Central Amer-ican freshwater fishes. Copeia 1966: 773-802.

Myers, G.S. 1938. Freshwater fishes and West Indian zoogeo-graphy. Ann. Rept. Smithsonian Inst. 1937: 339-364.

Myers, G.S. 1966. Derivation of the freshwater fish fauna ofCentral America. Copeia 1966: 766-773.

Nordlie, F.G. 1981. Feeding and reproductive biology of ele-otrid fishes in a tropical estuary. J. Fish. Bioi. 18: 97-110.

Nordlie, F.G. & D.P. Kelso. 1975. Trophic relationships in atropical estuary. Rev. Bioi. Trop. 23: 77-99.

Nuhn, H., R. Perez, et al. 1967. Estudio geografico regional dela zona Atlantica norte de Costa Rica (Regional geographicstudy of the north Atlantic zone of Costa Rica). Instituto deTierras y Colonizacion, San Jose. 360pp. (in Spanish).

Odum, W.E. 1988. Comparative ecology of tidal freshwater andsalt marshes. Ann. Rev. Ecol. Syst. 19: 147-176.

Power, M.E. 1984. Habitat quality and the distribution of algae-grazing catfish in a Panamanian stream. J. Anim. Ecol. 53:357-374.

Balon, E.K. 1974. Fishes from the edge of Victoria Falls: demiseof a physical barrier for downstream invasions. Copeia 1974:

643-660.Balon, E.K. & D.J. Stewart. 1983. Fish assemblages in a river

with unusual gradient (Luongo, Africa-Zaire system), reflec-tions on river zonation, and description of another new spe-cies. Env. Bioi. Fish. 9: 225-252.

Bishop, J .E. 1973. Limnology of a small Malayan river SungaiGombak. Monographiae Biologicae 22, Dr W. Junk Publish-ers, The Hague, 485 pp.

Burton, J.W. & E.P. Odum. 1945. The distribution of streamfishes in the vicinity of mountain Lake, Virginia. Ecology 26:

182-194.Bussing, W.A. 1976. Geographic distribution of the San Juan

ichthyofauna of Central America with remarks on its originand ecology. pp. 157-175. In: T.B. Thorson (ed.) Investiga-tions of the Ichthyofauna of Nicaraguan lakes, Univ~rsity ofNebraska, Lincoln.

Bussing, W.A. 1987. Peces de las Aguas Continentales de CostaRica (Freshwater fishes of Costa Rica). Editorial de la Uni-versidad de Costa Rica, San Jose. 271 pp. (in Spanish).

Caldwell, D.K., L.H. Ogren & L. Giovannoli.1959. Systematicand ecological notes on some fishes collected in the vicinity ofTortuguero, Caribbean coast of Costa Rica. Rev. Bioi. Trop.7: 7-33.

Cashner, R.C. & J.D. Brown. 1977. Longitudinal distributionof fishes of the Buffalo River in northwestern Arkansas.Tulane Studies in Zool. Bot. 19: 37-46.

Felley, J.D. 1987. Nekton assemblages of three tributaries tothe Calcasieu estuary, Louisiana. Estuaries 10: 321-329.

Fisher, S.G. 1983. Succession in streams. pp. 7-27. In: J.R.Barnes & G.W. Marshall (ed.) Stream Ecology, Applicationand Testing of General Ecological Theory, Plenum Press,New York.

Fisher, S.G., L.J. Gray, N.B. Grimm & D.E. Bush. 1982.Temporal succession in a desert stream ecosystem followingflash flooding. Ecol. Monogr. 52: 93-110.

Gauch, H.G. 1982. Multivariate analysis in community ecology.Cambridge University Press, Cambridge. 298 pp.

Gilbert, C.R. & D.P. Kelso. 1971. Fishes of the Tortugueroarea, Caribbean Costa Rica. Bull. Florida State Mus., Bioi.

Sci. 16: 1-54.Gorman, O. T. & J.R. Karr. 1978. Habitat structure and stream

fish communities. Ecology 59: 507-515.Graham,J.H. & R.W. Hastings. 1984. Distributional patterns of

sunfishes on the New Jersey coastal plain. Env. Bioi. Fish. 10:

137-148.Hill, M.O. 1979. DECORANA: A FORTRAN program for

detrended correspondence analysis and reciprocal averaging.Section of Ecology and Systematics, Cornell, University, Ith-

aca. 52 pp.Hill, M.O. & H.G. Gauch. 1980. Detrended correspondence

analysis: an improved ordination technique. Vegetatio 42:

47-58.Hirth, H.F. 1963. The ecology of two lizards on a tropical beach.

Ecol. Monogr. 33: 83-112.

47

Rivas, L.R. 1986. Systematic review of the perciform fishes ofthe genus Centropomus. Copeia 1986: 579-611.

Rozas, L.P. & W.E. Odum. 1987. Use of tidal freshwatermarshes by fishes and macrofaunal crustaceans along a marshstream-order gradient. Estuaries 10: 36-43.

Saul, W.G. 1975. An ecological study of fishes at a site in UpperAmazonian Ecuador. Proc. Acad. Nat. Sci. Philadelphia 127:93-134.

Schlosser, 1.1. 1982. Fish community structure and functionalong two habitat gradients in a headwater stream. Ecol.Monogr. 52: 395-414.

Schlosser, 1.1.1987. A conceptual framework for fish communi-ties in small warmwater streams. pp. 17-24. In: W.l. Mat-thews & D.C. Heins (ed.) Community and EvolutionaryEcology of North American Stream Fishes, University ofOklahoma Press, Norman.

Shannon, C.E. & W. Weaver. 1963. The mathematical theory ofcommunication. Univ. of Illinois Press, Urbana. 117pp.

Sheldon, A.L. 1968. Species diversity and longitudinal succes-sion in stream fishes. Ecology 49: 193-198.

Shmida, A. & M.V. Wilson. 1985. Biological determinants ofspecies diversity. 1. Biogeogr. 12: 1-20.

Smith, G.R. 1981. Effects of habitat size on species richness andadult body sizes of desert fishes. pp. 125-171. In: R.I. Naiman& D.L. Soltz (ed.) Fishes in North American Deserts, lohnWiley & Sons, New York.

Ter Braak, C.l.F. 1986. Canonical correspondence analysis: anew eigenvector technique for multivariate direct gradientanalysis. Ecology 67: 1167-1179.

TerBraak, C.l.F. 1988. CANOCO-a FORTRAN program forcanonical community ordination by correspondence analysis,principal components analysis and redundancy analysis (ver-sion 2.1). Report LWA-88-02, Agricultural MathematicsGroup, Wageningen. 95 pp.

Townsend, C.R. & G. Peirson. 1988. Fish community structurein lowland drainage channels. 1. Fish. Bioi. 32: 283-295.

Tramer, E.J. 1969. Bird species diversity: components of Shan-non's formula. Ecology 50: 927-929.

Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell &C.E. Cushing. 1980. The river continuum concept. Can. J.Fish. Aquat. Sci. 37: 130-137.

Villa, J. 1982. Peces Nicaraguenses de Agua Dulce (Nicaraguanfreshwater fishes). Banco de America, Managua/ 253 pp. (in

Spanish).Weinstein, M.P., S.L. Weiss & M.F. Walters. 1980. Multiple

determinants of community structure in shallow marsh hab-itats, Cape fear River Estuary, North Carolina. USA. Mar.Bioi. 58: 227-243.

Wilkinson, L. 1989. Systat: the system for statistics. Systat, Inc.,Evanston. 638pp.

Winemiller, K.O. 1982. An introduction to the freshwater fishcommunities of Corcovado National Park, Costa Rica. Bre-nesia 21: 47-66.

Winemiller, K.O.1987. Tests ofecomorphological andcommu-nity level convergence among neotropical fish assemblages.Ph.D. Dissertation, University of Texas, Austin. 370pp.

Winemiller, K.O. 1989. Patterns of variation in life historyamong South American fishes in seasonal environments. Oe-cologia 81: 225-241.

Winemiller, K.O. 1990. Spatial and temporal variation in trop-ical fish trophic networks. Ecol. Monogr. 60: 331-367.

Winemiller, K.O. 1991. Ecomorphological diversification inlowland feshwater fish assemblages from five biotic regions.

Ecology (in press).Winemiller, K.O. &N.E. Morales. 1990. Comunidadesdepec-

es del Parque Nacional Corcovado luego del cese de lasactividades mineras (Fish communities of Corcovado Nation-al Park following the cessation of mining activities). Brenesia31: 75-91. (in Spanish).

Winemiller, K.O. & E.R. Pianka. 1990. Organization in naturalassemblages of desert lizards and tropical fishes. Ecol. Mono-gr. 60: 27-55.

48

Appendix 1.

Sample sizes and trophic classifications for fishes collected from 16 sites in Tortuguero National Park, Costa Rica. Abundance data arepooled samples from Gilbert & Kelso (1971) and Winemiller (1987 and unpublished). Sample sites are as follows: I-Ocean or beach, 2-Mouth of Laguna Tortuguero, 3- Laguna Tortuguero, 4- Rio Tortuguero, 5-Caiio Benjamin, 6-Arm of Rio AguaFria, 7 - Rio AguaFria, 8- Creeks on west slope of Lomas, 9-Caiio Agua Fria Viejo, 10-Creeks on east slope of Lomas, ll-Forest pool near Caiio AguaFria Viejo, 12 - Quebrada, 13 - Caiio Servulo, 14 - Laguna Jalova, 15 - Caiio Negro, 16 - Caiio California and tributaries. Trophiccategories are: H - herbivore, I - invertebrate feeder, P - piscivore, 0 - omnivore.

49

Appendix 1. (Continued).

0 0 0

0

-0

0

-0

0

-0

0

00

0

0

00

20 0 23 0-0

-

0

0

-

0

0

01

-0

0

00

0

0

012

0 0 0

00

0

07

0 270 80

04

10

244

01

00

00

00

00

00

5 0 18

796

0 0

20

0 0 0 0 0 0 0

00

0

00

0

20

0 0

00

00

00

02

60

613

02

00

00

00

220

100117

00

00

0

00

00

10

10

0 00 0

1 0

0 200 0

4411

00

00

00

00

0

00

00

00

00

0

214

0

7

0 0 0 0 0 0

530

000

0 0 0 0

10

008

00

000

00

00

3797

10

000

40

00

14

10

00

35

01

00

00

40

1010

129 370 00 0

0 034 1

190 14

000

000

000

011

467

31

0

000

00

0

000

0272

29

20

0

0

50

00

00

20

00

0

0

00

0

0

00

00 H

1

0

0

0

0 0 0 0

0

0

0

000

0

0

000

0

3 0 0

000

0

000

0 0 0 0

000

p

ppp

541

010

3

000

000

000

000

0000000

0000

000

0000000

06

184

000

0000000

0000

0p0

0000000

0000

0

000

000

31265

27000

0001000

0000000

0000000

0000000

0000000

0000000

0101000

2600000

236310

0

070

0100000

0000

0

ppppPII

Pppp

p

PP/IP/I

III

-00

32976

0000

316

36915

1

0000

0

0100

0

000

001

0

00001

0100

0

000

0000

0

000

0009

0

0010

0

0090

0

052

000

0

0 0 0

000

0 0

000

000

000

154351

210

204

9490

00

95

1

000

003

0

1480

00

122

000

000

000

000

000

000

0

70

002

0

00004

001

0

00002

000

0

00000

004

001

0

01002

Priapichthys annectensFamily CyprinodontidaeRivulus isthmensisFamily BelonidaeStrongylura marinaStrongylura timucuFamily ExocoetidaeHyporhamphusroberti hildebrandiFamily BothidaeCitharichthys spilopterusCitharichthys uhleriFamily SoleidaeAchirus lineatusTrinectes paulistanusFamily CynoglossidaeSymphurus plagiusaFamily SyngnathidaeOostethus lineatusPseudophallus mindiiFamily AtherinidaeColeotropis blackburniMelaniris chagresiMelaniris milleriFamily MugilidaeAgonostomus monticolaMugil curemaFamily PolynemidaePolydactylus virginicusFamily SphyraenidaeSphyraena guachanchoFamily ScombridaeScomberomorus cavallaScomberomorus maculatusThunnus sp.Family CarangidaeCaranx hipposCaranx latusChloroscombrus chrysurusOligoplites palometaSelene setapinnisTrachinotus carolinusTrachinotus goodeiFamily CentropomidaeCentropomus ensiferusCentropomus parallelusCentropomus pectinatusCentropomus undecimalisFamily SerranidaeEpinephelus sp.Family LutjanidaeLutjanus griseusLutjanus jocuLutjanus sp.Family PomadasyidaeConodon nobilisPomadasys corvinaeformisPomadasys crocroFamily SparidaeArchosargus probatocephalusFamily GerreidaeDiapterus olisthostomusDiapterus plumieriDiapterus rhombeusEucinostomus argenteusEucinostomus melanopterus

0 0 0

02403

0 0 0 0

05616

0

0000

15

00000

41755

1157

00002

00000

00000

00000

00000

01302

103011

000

20

50

Appendix 1. (Continued)

Species-Sites

-2

-3 4 5 6 7 8 9 10 11

-12 13 14 15 16 Code

0 0000

1910

0110

19

000

0

00000000000

30000000000

00000000000

00000000000

00000000000

00000000000

00000000000

0000()000000

00000000000

00000000000

10000000000

10000000000

00000000000

0518

79260000130

-

PIP1>

2760502

14127

00000

0000000000000

00000000

0007000814711

3434584

15952

45110

13211704

1000

1800000700

012000

0 22 00 0 30 0 40 0 07 8 110 9 30 0 00 0 01 86 320 0 40 0 00 0 00 16 0

0 9 23 526 30 90 00 25 10 0 00 0 0

03

101010030000

1765990

1110

18410

62

5010

7000141000

3000

3900000900

026

1641

944284

0040002

47

18060083000

-IIPIHPI0

0000

-

H0

0P/IP/IPP/I

000000

099000

363752535676

038

041200

001100

000

2511689278168

00

221300

250600

162000

400

00

0 70 00 00 00 100 00 0

0191217

32813

226

00108O.0

0 000

1542

23

0000000

0000000

60

840

14229

1000000

0000000

.0

0570000

0000200

0020500

0000000

-IHT

004

-2

Family SciaenidaeBairdiella ronchusBairdiella sanctaeluciaeCynoscion jamaicensisLarimus brevicepsMenticirrhus americanusMenticirrhus littoralisMicropogonias furnieriOphioscion costaricensisOphioscion panamensisStellifer colonensisUmbrina broussonetiFamily CichlidaeCichlasoma alfariCichlasoma centrarchusCichlasoma citrinellumCichlasoma doviiCichlasoma loiselleiCichlasoma maculicaudaCichlasoma managuenseCichlasoma nicaraguenseCichlasoma nigrofasciatumCichlasoma rostratumCichlasoma septemfasciatumCichlasoma tubaHerotilapia multispinosaFamily EleotridaeDormitator maculatusEleotris amblyopsisEleotris pisonisGobiomorus dormitorGuavina guavinaLeptophilypnus j1uviatilisFamily GobiidaeA waous tajasicaBathygobius soporatorEvorthodus lyricusGobionellus boleosomaGobionellus fasciatusGobionellus pseudofasciatusGobiosoma spesFamily EcheneidaeEcheneis naucratesFamily DactyloscopidaeDactylagnus peratikosFamily MicrodesmidaeMicrodesmus carriFamily BatrachoididaeBatrachoides gilbertiBatrachoides surinamensisPorichthys pauciradiatusPorichthys plectrodonFamily TetraodontidaeSphoeroides testudineusFamily UranoscopidaeAstroscopus y-graecum

-0 0

0 4 0 0 0 0 0 06 0 0

0

0

0 0 0

0

0 0 P/I

1 6

0 34

0

~

0 0 0 0 0 0 0 0 0

0

0

0

0000

0 p

0

0000

0 0 0 0 0 0 0 0

0000

0

0000

0011

0000

5100

0000

0000

0000

0000

0000

0000

0

0000

0000

0000

P/IP/IP/IPIT

0 0 139 0 0 0 0 0

0

0

0

0

0

4 5 0 0

0 0 0 0 0 0 0 0 0 0 0 0

220

130

1802

225300

02222700