ABSTRACT

Historical and zoogeographic factors appear to explainthe origin of Neotropical freshwater fish diversity, but proxi-mate factors maintaining such remarkable levels of regionaldiversity, particularly in large floodplain rivers, remainsunknown. Floodplain rivers are characterized by high levels oflandscape and temporal heterogeneity. The littoral zone is com-posed of a mosaic of habitat templets upon which local com-munities are assembled and represents a highly dynamic com-ponent of the landscape due to seasonal water level fluctuationsassociated with the annual flood pulse. Results presented indi-cate littoral species are forced to continually disperse across thelandscape in association with a moving land-water interface

25

Arrington D.A.1,2 Winemiller K.O.1

1 Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, Texas, USA2 Department of Biological Sciences, University of Alabama, Tuscaloosa, USA

ORGANIZATION AND MAINTENANCEOF FISH DIVERSITY IN SHALLOW

WATERS OF TROPICAL FLOODPLAINRIVERS

Key Words: beta diversity, colonization,community assembly, fishes, flood pulse,gamma diversity, spatiotemporal hetero-geneity, Venezuela

and lend support to an implicit theoretical trade offbetween colonization and competition ability amongfishes of tropical floodplain rivers. The continual dis-assembly and reassembly (due to dispersal) of localcommunities across a spatially heterogeneous land-scape should result in low extinction rates (i.e. at theregional level) and could theoretically maintain a near-ly infinite regional species pool. Consequently, wesuggest that the flood pulse paradigm should beexpanded to include a potential mechanistic under-standing of maintenance of high levels of beta andgamma diversity in floodplain rivers. Interactionsamong seasonal hydrology, variability in habitat struc-tural complexity and landscape heterogeneity appearto maintain high aquatic species richness in these low-land rivers. It follows that alteration of seasonal waterlevel fluctuation (e.g. damming) and habitat hetero-geneity (e.g. channelization) should have substantialand negative consequences on the maintenance ofregional biodiversity pools in floodplain rivers.

INTRODUCTION

Tropical floodplain rivers are home to thelargest fraction of freshwater fish biodiversity(Dudgeon 2000; Lundberg 2001) and as such shouldbe a focal point of global conservation efforts.Recently, conservationists have focused their effortson species conservation through identification andconservation of hotspots (i.e. areas with high levels ofendemism) (Meyers et al. 2000). Unfortunately, tropi-cal rivers and associated fish faunas are absent fromthis conservation initiative (Meyers et al. 2000; Brookset al. 2002). Not only have rivers been undervalued inconservation efforts, but also ecological understandingof community and assemblage dynamics in lowlandrivers lags behind other fields of study (e.g. limnologyin temperate lakes). We suggest that better conceptualunderstanding of these systems will lead to more effec-tive conservation and restoration practices.

Recent studies have made great headway inunderstanding large-scale patterns of freshwater fishdiversity. In an insightful review of the source of SouthAmerican freshwater fish fauna richness, Lundberg(2001) identified area and latitudinal gradient as suit-

able explanations for large-scale maintenance ofneotropical riverine freshwater fish diversity, but sug-gested that a historic-zoogeographic perspective isneeded to understand the genesis of this diversity.Lundberg presented evidence that river basins wererepeatedly transformed during periodic geologicalupheavals and changes in basin boundaries and inter-basin connections resulted in sympatric speciationopportunities. Lundberg reasoned that low baselineextinction rates resulted in the present-day richness ofneotropical freshwater fish species.

Although historical and zoogeographic factorsmay explain the origin of neotropical freshwater fishdiversity, there is little understanding of proximate fac-tors (sensu Resetarits and Bernardo 2001) maintainingthis remarkable diversity across the landscape atregional and local scales, particularly in large flood-plain rivers. Fish assemblages in European and NorthAmerican temperate forested lakes are structured by aseries of nested filters operating first at regional andsubsequently at local scales (Tonn 1990; Tonn et al.1990). Stream fish assemblages in the same tworegions (Europe and North America) also exhibitedhierarchical structure with regional zoogeography andlocal habitat templets structuring local fish assem-blages (Matthews 1998; Lamouroux, Poff andAngermeier 2002). Characterization of the relationshipbetween regional and local fauna richness has beenidentified as a useful metric to evaluate the degree thatspecies interactions regulate local community dynam-ics (Cornell and Lawton 1992; Hugueny and Cornell2000). Cornell and Lawton (1992) stated that unsatu-rated assemblages are ubiquitous in nature and as aconsequence regional richness is free of local con-straints, although this logic has been seriously chal-lenged (Shurin et al. 2000). Cornell and Lawton (1992)speculated that the regional species pool regulateslocal richness, which should be a function of landscapeheterogeneity and evolutionary diversification.Empirical data have revealed that contemporary ener-gy availability and habitat heterogeneity successfullypredict the current global distribution of riverine fishdiversity (Guégan, Lek and Oberdorff 1998).Furthermore, fish assemblages in Africa, North

26 Organization and maintenance of fish diversity

America and Western Europe may be either unsaturat-ed i.e. non-interactive (Tonn et al. 1990; Hugueny andPaugy 1995; Griffiths 1997; Oberdorff et al. 1998) orsaturated i.e. interactive (Tonn et al. 1990; Angermeierand Winston 1998). In their comparative study, Tonn etal. (1990) demonstrated that North American assem-blages exhibited an asymptotic relationship betweenlocal and regional richness (i.e. interactive localassemblages), whereas European assemblages did not(i.e. non-interactive local assemblages). Similar totemperate lakes, North American stream fish assem-blages (i.e. Virginia) also showed local community sat-uration (Angermeier and Winston 1998). The likeli-hood that neotropical fish assemblages are saturatedseems high, because regional diversity in these settingsis remarkably high (Jepsen 1997; Arrington andWinemiller 2003) yet alpha diversity is not substantial-ly greater than similar North American samples(Matthews 1998), that have much lower regionaldiversity levels. Though a consensus on this subjecthas not been reached (Cornell and Lawton 1992;Shurin et al. 2000), the possibility exists that localcommunity interactions may regulate regional speciesrichness for fishes in neotropical rivers.

Much attention has focused on spatiotemporaldynamics in lotic environments as a mechanism formaintaining biodiversity (e.g. Schlosser 1987;Townsend 1989; Ward 1989, 1998; Poff and Allan1995; Matthews 1998; Schlosser and Kallemeyn 2000;Oberdorf, Hugueny and Vigneron 2001). Recent workconducted on European floodplain rivers has charac-terized landscape attributes of floodplain rivers asshifting mosaics of habitat features with varying levelsof among-habitat connectivity (Ward, Tockner, Arscotet al. 2002; Amoros and Bornette 2002). The combina-tion of landscape heterogeneity and temporally vari-able among-patch connectivity are common features offloodplain rivers that result in observed patterns andlevels of biodiversity (Ward 1998; Ward et al. 1999;Ward et al. 2001; Tockner et al. 1999; Amoros andBornette 2002; Robinson et al. 2002). The generalsynopsis is that a dynamic landscape composed of amosaic of habitat patches in various successional statesmaintains the high regional diversity levels observed

in floodplain rivers. This should be the case whether ornot local communities are saturated, because effects oflandscape and temporal heterogeneity should over-come competitive exclusion (Levin and Paine 1974;Chesson and Huntley 1997; Hurtt and Pacala 1995).

In this limited review, we hope to address thefollowing questions based on experience with fishassemblages in neotropical floodplain rivers. How areneotropical fishes organized across the river-floodplainlandscape? What factors influence assemblage struc-ture in these local communities? How are such largeregional species pools maintained?

ORGANIZATION OF FISH DIVERSITY INTROPICAL FLOODPLAIN RIVERS

Characterization of patterns of species occur-rences and relative abundance is a major goal in com-munity ecology (Hubbell 2001). A central debateamong community ecologists has been the role ofdeterministic versus stochastic processes often inferredthrough examination of random or non-random pat-terns in assemblage data. Studies of fish assemblagesin temperate streams have demonstrated both random(Grossman, Moyle and Whitaker 1982; Grossman etal. 1998) and non-random patterns (Meffe and Sheldon1990; Jackson, Somers and Harvey 1992; Taylor1996), with results often strongly dependent on thespatial, temporal and numerical scale of the study(Rahel 1990, Angermeier and Winston 1998). Tropicalfloodplain rivers have been studied less frequently andhave yielded mixed results. Goulding, Carvalho andFerreira (1988) concluded that fish assemblages of theRío Negro (Brazil) were random associations ofspecies. More recent studies also support the randomassociation hypothesis (Jepsen 1997; Saint-Paul et al.2000).

A few studies in tropical river systems haveconcluded fish assemblages are structured in a non-random manner. Working in the same system as Jepsen(1997), Arrington (2002) documented non-randomstructure of fish and macroinvertebrate assemblagesamong major habitat types (e.g. sandbank, leaf litter,submerged wood) located in the moving littoral. Fish

in shallow waters of tropical floodplain rivers 27

assemblages in these local habitats were maximallystructured during the low-water period and less struc-tured in rising- and falling-water periods.Consequently, juxtaposition of multiple habitat typesand the resulting landscape heterogeneity resulted inhigh levels of observed beta diversity, which substan-tially influenced the estimate of the regional speciespool. Similarly, fish assemblages on the floodplain ofthe Brazillian Amazon were found to be non-randomlystructured among major habitat types (Petry, Bayleyand Markle 2003), though habitats in this study werecharacterized by dominant macrophytes. Others haveshown fish assemblage structure in tropical rivers isinfluenced by water type (Ibarra and Stewart 1989;Cox Fernandes 1999), sample depth (Lundberg et al.1987; Stewart, Ibarra and Barriga-Salazar 2002;Hoeinghaus et al. 2003), seasonally falling water lev-els (Cox Fernandes 1999) and diel period sampled(Arrington and Winemiller 2003). Rodriguez andLewis (1997) found structured assemblage patterns inOrinoco floodplain lakes that were correlated withwater clarity. They inferred predation by alternativepredators in clear or turbid lakes was driving assem-blage structure. As Winemiller (1996) hypothesized,tropical floodplain river fish assemblages appear to bestructured by both stochastic and deterministicprocesses and the magnitude of these processes variesseasonally (Arrington 2002).

MAINTENANCE OF FISH DIVERSITY INTROPICAL FLOODPLAIN RIVERS

We suggest that the flood pulse paradigm beexpanded to include a potential mechanistic under-standing of maintenance of high levels of beta andgamma diversity in floodplain rivers. We hypothesizethat the flood pulse (Junk, Bayley and Sparks 1989),i.e. the annual hydrologic pattern of predictable flood-ing of lateral floodplain habitats in large tropicalrivers, regulates community assembly patterns andregional diversity levels. As conceived by Junk et al.(1989), the flood pulse concept linked riverine produc-tivity to predictable annual patterns of flooding andcharacterized the main channel as a passageway forfish migrations. Although some fish species undoubt-edly use the main channel for migration (Junk et al.

1989; Fernandes 1997; Wei et al. 1997; Duque,Taphorn and Winemiller 1998), many species eitherseasonally (low water) or consistently occupy mainchannel habitats (e.g. deep channel, shifting sand-banks, snag complexes). We shift our focus from themain channel / highway analogy (Junk et al. 1989) tothe moving littoral as a dynamic habitat template. Wedefine the moving littoral as the dynamic land-waterecotone occurring along main channel margins andextending onto the floodplain during high water. Themoving littoral, thus, represents a highly dynamiccomponent (i.e. shallow water) of the landscape com-posed of a mosaic of habitat templets upon which localcommunities are assembled (Southwood 1988;Townsend 1989; Bayley 1995; Arrington 2002; Petryet al. 2003). Furthermore, local habitat templates in themoving littoral may be thought of as being disturbed atsome intermediate level (Connell 1978; Townsend,Scarsbrook and Dolédec 1997; Ward et al. 1999; Sheiland Burslem 2003) due to the seasonally predictablepatterns of drying and wetting in lowland rivers (Junket al. 1989; Arrington and Winemiller 2003).

As discussed above, there is considerabledebate regarding the roles of deterministic and sto-chastic processes regulating local fish assemblages. Anew hypothesis receiving considerable attention isHubbell’s (2001) “neutral theory”, which assumes percapita equivalence of within-trophic-level communitymembers. Community change is assumed to occurthrough stochastic ecological processes and randomspeciation. Application of the neutral theory has result-ed in the generation of multiple testable predictions,which can serve as working null hypotheses in com-munity studies. For example, local communities areexpected to be sub-sets of the larger metacommunity,with species relative abundance equal between the twowhen migration rates into the local community arenon-trivial (Hubbell 2001).

Previous studies have shown the importance ofimmigration rates on the structure (including richness)of local communities (MacArthur and Wilson 1967).Dispersal rates determine the importance of local com-munity regulators in zooplankton assemblages

28 Organization and maintenance of fish diversity

(Jenkins and Buikema 1998) and predominantly limitdiversity in newly formed assemblages (Shurin et al.2000). Temporal variation of fish assemblage structurein a Brazilian estuary has been linked to fish immigra-tion and emigration dynamics (Garcia, Vieira andWinemiller 2001). For temperate streams, Schlosser(1987) offered a conceptual model indicating the gen-eralized importance of immigration and extinctionprocesses for the development of fish assemblageattributes. Using Schlosser’s model as a starting point,Taylor and Warren (2001) showed stream fish immi-gration rates were positively related to stream size andnegatively related to flow variability. They then docu-mented patterns of nestedness in fish assemblagestructure that were positively related to extinction ratesand negatively related to immigration rates. They alsoobserved that colonization and extinction dynamics ofspecies appeared “more or less random” in habitatswith high immigration rates, a result predicted byTownsend’s (1989) patch dynamics concept.

Recent work by Arrington, Winemiller andLayman (in review) examined the influence of colo-nization rate and habitat complexity on the dynamicsof local fish assemblages in the Cinaruco River, afloodplain river located in the Venezuelan llanos.Habitat patches of varying complexity and coloniza-tion rates were created within broad main-channelsandbanks and were colonized by fishes and macroin-vertebrates for a period of 21 days. In accordance withisland biogeography theory, Arrington et al. found asignificant positive influence of colonization rate onthe number of species in local habitat patches.Furthermore, they observed more complex habitatscontained significantly more species. Results variedwhen treatment effects were evaluated separately fortwo distinct subsets of the fish assemblage. Richnessof fish taxa with low vagility was positively related tocolonization rate and habitat patch complexity, where-as richness of highly vagile fishes was positively relat-ed to patch complexity but not colonization rate.Presumably, increased colonization ability by vagilespecies swamped the influence of habitat patch isola-

tion. These results suggest local community dynamicsin this neotropical floodplain river are dominated bynear continuous dispersal and colonization of habitatpatches in the moving littoral (sensu “patch dynamicsconcept” Townsend 1989) by adult fishes (Arrington etal. in review). Furthermore, low concordance wasobserved between ranks of species from the meta-com-munity and local habitat patches; thus falsifying one ofHubbell’s (2001) hypotheses (see above). In a parallelexperiment, Arrington et al. (in review) documentedlargely stochastic structure of local assemblages innewly formed habitat patches, but increasing levels ofnon-random organization were observed in patches astime progressed (> 24 days). Taken together, these datasuggest dispersal is most important in structuringassemblages in newly formed patches, whereas theinfluence of local processes on assemblage structureincreases as time progresses.

A considerable body of ecological theory hasbeen developed that indicates tradeoffs in colonizationand competitive abilities should preclude competitive-ly dominant species from occupying all suitable habi-tats in a spatially heterogeneous landscape (i.e. themoving littoral of floodplain rivers) and as a conse-quence competitive subordinates should persist in theregional species pool (Levin and Paine 1974; Hurtt andPacala 1995). The experiments by Arrington et al. lendsupport to an implicit theoretical trade off between col-onization and competition ability in fishes of tropicalfloodplain rivers, where littoral species are forced tocontinually disperse across the landscape in associa-tion with a moving land-water interface. This continu-al disassembly and reassembly (due to dispersal) oflocal communities across a spatially heterogeneouslandscape should result in low extinction rates (i.e. atthe regional level) and could theoretically maintain anearly infinite regional species pool (Hurtt and Pacala1995).

Additional studies on the Cinaruco Riverappear to support such a mechanism in maintaining avery large regional species pool in a tropical floodplainriver (Arrington 2002). Through most of 1999, six

in shallow waters of tropical floodplain rivers 29

habitats were sampled from the moving littoral zone inthe Cinaruco River. These habitats function as habitattemplets, upon which local communities are assembled(Arrington 2002) and their spatial distribution in themain channel and floodplain is a dominant componentof spatial habitat heterogeneity. Each month sevenreplicate diurnal samples were collected using thesame seine (see Arrington 2002 for a description ofmethods). We plotted fish species accumulation curvesfor each habitat independently (Figure 1). In each habi-tat, we observed a continual and positive slope of the

accumulation curve, with observed total assemblagerichness (γ diversity) reaching 50 to 80 fish species perhabitat type. In addition to observed species richnessvalues, we estimated γ diversity for specific littoralzone habitats using a non-parametric estimator basedon observed relative abundance data. The technique,known as abundance-based coverage estimator,assumes information about un-sampled species is heldin the rarest classes of species collected (Chao and Lee1992; Colwell and Coddington 1994; Chazdon et al.1998) and can be computed using freely-available soft-

30 Organization and maintenance of fish diversity

Figure 1. Species accumulation curves for standardized seine samples collected from six habitats located in the moving littoral zoneof the Cinaruco River, Venezuela reveal the diversity of tropical floodplain fish assemblages. Samples were collected through most of1999, excluding the peak-wet season (see Arrington 2002). Number of samples collected per habitat was: river rock 45, river sand 69,river snag 47, lagoon leaf 36, lagoon sand 36 and lagoon snag 31. Each point along the solid line represents an estimate of the total com-munity richness (including taxa not sampled) for specific littoral zone habitats based on the relative abundances of the most rare speciesin our samples (see Colwell 1997).

ware (Estimate S, Colwell 1997). In some habitats,such as river rock (Figure 1), the estimated size of thespecies pool far exceeds observed values. These localcommunities were consistently composed of relativelyrare species. Thus, it appears that community assemblywithin isolated habitat patches, such as our river rockhabitats, are more dependent upon stochastic coloniza-tion processes (i.e. colonization limitation).Furthermore, these patches often contained speciescharacteristic of adjacent open sandbank habitats,which suggests leaky boundaries may lead to “masseffects” (Townsend 1989) in local patches that are notbiologically saturated. Others have hypothesized thatlow dispersal or connectivity among patches shouldresult in lowered α diversity, but promote β and γdiversity (Hubbell 1997), particularly in floodplainrivers that are characterized by high levels of spa-tiotemporal heterogeneity (Ward et al. 1999; Tockneret al. 1999; Amoros and Bornette 2002). In more con-tiguous (higher connectivity) patches, lower estimatesof total richness may reflect reduced persistence of rare(Hubbell 2001) or competitively inferior (Hurt andPacala 1995) species with higher colonization rates(higher α diversity but lower β diversity; Amoros andBornette 2002). At present we are unable to identifythe mechanism(s) driving the difference betweenobserved and expected diversity (Figure 1). But, ourexperimental results indicate that a combination of col-onization limitation (dispersal) and biotic interactionsresult in low α diversity but high β diversity(Arrington et al. in review). We submit that the annualflood pulse interacts with basin geomorphology andadds temporal heterogeneity to an already spatiallyheterogeneous landscape, both of which are critical inmaintaining high levels of γ diversity observed in low-land tropical rivers (Figure 2).

THREATS TO FISH DIVERSITY IN TROPICALFLOODPLAIN RIVERS

Rivers face a number of anthropogenic threats(Allan and Flecker 1993; Crisman et al. 2003) and dambuilding appears particularly damaging to tropicalriverine biodiversity (Grossman, Dowd and Crawford1990; Goulding, Smith and Mahar 1996; Pringle et al.2000). Large floodplain rivers are characterized by a

remarkable degree of spatiotemporal heterogeneity intheir natural state (Ward et al. 1999, 2001, 2002) andthis heterogeneity is maintained by fluvial dynamicsacting on landscape features. In the Tagliamento River(Italy) corridor, for example, landscape features expe-rienced up to 80 percent turnover in a 3-year period,but features maintained similar relative proportionsacross the landscape (Ward et al. 2001). Constructionof dams for flood control or hydroelectric power gen-eration constrains these fluvial dynamics and canresult in dramatic loss of spatial heterogeneity (Toth etal. 1995; Schmidt et al. 1998). If the interactionbetween the natural rise and fall of flood waters andfloodplain spatial heterogeneity (habitat templates fororganisms) maintains regional diversity levels in trop-ical floodplain rivers (Figure 2), then loss of the floodpulse not only will impact biological production (Junk

in shallow waters of tropical floodplain rivers 31

Figure 2. A conceptual model illustrating the importance ofspatiotemporal heterogeneity in maintaining biological diversity infloodplain rivers. This model is derived from the intermediate dis-turbance hypothesis (Connell 1978; Shiel and Burslem 2003) andthe “patch dynamics concept” (Townsend 1989) and is supportedby fish data collected from the Cinaruco River, Venezuela, anunregulated, tropical lowland river. Anthropogenic alterationssuch as channelization and flow regulation are expected to resultin compromised heterogeneity; direction of impact is indicated bydotted lines.

et al. 1989; Bayley 1995), but impoverish regionalspecies pools (Grossman et al. 1990; Ward et al. 1999).Furthermore, reduction of landscape heterogeneitymay impair the resilience typically observed in thesesystems (Townsend 1989; Meffe and Sheldon 1990;Townsend et al. 1997). Consequently, restorationstrategies for floodplain rivers must emphasize thereturn of hydrologic variability characteristic of thepre-impacted system (e.g. Toth, Arrington and Begue1997) as well as re-establishing among-habitat connec-tivity (Toth et al. 1998).

Interactions among seasonal hydrology, vari-ability in habitat structural complexity and landscapeheterogeneity appear to maintain high aquatic speciesrichness in these lowland rivers. It follows that alter-ation of seasonal water level fluctuation (e.g.damming) and habitat heterogeneity (e.g. channeliza-tion) should have substantial and negative conse-quences on the maintenance of regional biodiversitypools in floodplain rivers. Better ecological under-standing is needed to properly manage and preservebiological diversity in tropical floodplain rivers.

ACKNOWLEDGEMENTS

Our research was funded by the NationalScience Foundation (NSF DEB-0107456), TheNational Geographic Society, L.T. Jordan Institute andthe International Sportfish Fund. This work would nothave been possible without the help and assistance ofJ. Arrington, C. Garcia, J. Garcia, C. Layman, C.Lofgren, C. Marzuola, J. Marzuola, E. Pelaez, A.Stergios, B. Stergios, D. Taphorn and G. Webb. C.Layman and an anonymous reviewer provided helpfulcomments on this manuscript.

LITERATURE CITED

Allan J.D. & Flecker A.S. 1993. Biodiversity conserva-tion in running waters. Bioscience, 43: 32-42.

Amoros C. & Bornette G. 2002. Connectivity and bio-complexity in waterbodies of riverine flood-plains. Freshwater Biology, 47: 761-776.

Angermeier P.L. & Winston M.R. 1998. Local vs. region-al influences on local diversity in stream fishcommunities of Virginia. Ecology, 79: 911-927.

Arrington D.A. 2002. Evaluation of the relationshipbetween habitat structure, community struc-ture, and community assembly in a neotropicalblackwater river. Texas A&M University,Texas, College Station. (Doctoral dissertation)

Arrington D.A. & Winemiller K.O. 2003. Dielchangeover in sandbank fish assemblages in aneotropical floodplain river. Journal of FishBiology, 63: 1-18.

Arrington D.A., Winemiller K.O. & Layman C.A. InReview community assembly at the patchscale in a speciose tropical river. Oecologia.

Bayley P.B. 1995. Understanding large river-floodplainecosystems. Bioscience, 45: 153-158

Brooks T.M., Mittermeier R.A., Mittermeier C.G., daFonseca G.A., Rylands B., Konstant A.B.,Flick W. R., Pilgrim R., Oldfield J., Magin G.& Hilton-Taylor C. 2002. Habitat loss andextinction in the hotspots of biodiversity.Conservation Biology, 16: 909-923.

Chao A. & Lee S.M. 1992. Estimating the number ofclasses via sample coverage. Journal of theAmerican Statistical Association, 87: 210-217.

32 Organization and maintenance of fish diversity

Chazdon R.L., Colwell R.K., Denslow J.S. & GuariguataM.R. 1998. Statistical methods for estimatingspecies richness of woody regeneration in pri-mary and secondary rain forests of NE CostaRiva. In F. Dallmeier & J.A. Comiskey eds.Forest biodiversity research, monitoring andmodeling: Conceptual background and oldworld case studies. Paris, ParthenonPublishing. pp. 285-309.

Chesson P. & Huntly N. 1997. The roles of harsh andfluctuating conditions in the dynamics of eco-logical communities. American Naturalist,150: 519-553.

Colwell R.H. 1997. Estimate S: Statistical estimation ofspecies richness and shared species from sam-ples. Version 5. User’s Guide and applicationpublished. (Available at:http://viceroy.eeb.uconn.edu/estimates.)

Colwell R.H. & Coddington J.A. 1994. Estimating terres-trial biodiversity through extrapolation.Philosophical Transactions of the RoyalSociety of London Series B, 345:101-118.

Connell J.H. 1978. Diversity in tropical rain forests andcoral reefs. Science, 199: 1302-1310.

Cornell H.V. & Lawton J.H. 1992. Species interactions,local and regional processes and limits to therichness of ecological communities: A theoret-ical perspective. Journal of Animal Ecology,61: 1-12.

Cox Fernandes C. 1999. Detrended canonical correspon-dence analysis (DCCA) of electric fish assem-blages in the Amazon. In A.L. Val & V.M.F.Almeida-Val eds. Biology of Tropical Fishes.Manaus, INPA. pp. 21-39.

Crisman T.L., Chapman L.J., Chapman C.A. & KaufmanL.S. eds. 2003. Conservation, ecology, andmanagement of African fresh waters.Gainesville, Fl., University Press of Florida.

Dudgeon D. 2000. Large-scale hydrological changes intropical Asia: Prospects for riverine biodiversi-ty. Bioscience, 50: 793-806.

Duque A.B., Taphorn D.C. & Winemiller K.O. 1998.Ecology of the coporo, Prochilodus mariaeCharaciformes, Prochilodontidae, and status ofannual migrations in western Venezuela.Environmental Biology of Fishes, 53: 33-46.

Fernandes C.C. 1997. Lateral migration of fishes inAmazon floodplains. Ecology of FreshwaterFish, 6: 36-44.

Garcia A.M., Vieira J.P. & Winemiller K.O. 2001.Dynamics of the shallow-water fish assem-blage of the Patos Lagoon estuary Brazil dur-ing cold and warm ENSO episodes. Journal ofFish Biology, 59: 1218-1238.

Goulding M., Carvalho M.L. & Ferreira E.G. 1988. RíoNegro: Rich life in poor water. The Hague:SPB Academic Publishing.

Goulding M., Smith N.J.H. & Mahar D.J. 1996. Floods offortune: Ecology and economy along theAmazon. New York, Columbia UniversityPress.

Griffiths D. 1997. Local and regional species richness inNorth American lacustrine fish. Journal ofAnimal Ecology, 66: 49-56

Grossman G. D., Dowd J.F. & Crawford M. 1990.Assemblage stability in stream fishes: Areview. Environmental Management, 14: 661-671.

Grossman G.D., Moyle P.B. & Whitaker J.O. 1982.Stochasticity in structural and functional char-acteristics of an Indiana stream fish assem-blage: A test of community theory. AmericanNaturalist, 120: 423-454.

in shallow waters of tropical floodplain rivers 33

Grossman G.D., Ratajczak R.E. Jr., Crawford M. &Freeman M.C. 1998. Assemblage organizationin stream fishes: Effects of environmental vari-ation and interspecific interactions. EcologicalMonographs, 68: 395-420.

Guégan J., Lek S. & Oberdorff T. 1998. Energy availabil-ity and habitat heterogeneity predict globalriverine fish diversity. Nature, 391: 382-384.

Hoeinghaus D.J., Layman C.A., Arrington D.A. &Winemiller K.O. 2003. Spatiotemporal varia-tion in fish assemblage structure in tropicalfloodplain creeks. Environmental Biology ofFishes 49: 421-433.

Hubbell S.P. 1997. A unified theory of biogeography andrelative species abundance and its applicationto tropical rain forests and coral reefs. CoralReefs 16: Suppl., S9-S21.

Hubbell S.P. 2001. The Unified Neutral Theory ofBiodiversity and Biogeography. Princeton,NJ., Princeton University Press.

Hugueny, B. & Cornell H.V. 2000. Predicting the rela-tionship between local and regional speciesrichness from a patch occupancy dynamicsmodel. Journal of Animal Ecology, 69:194-200.

Hugueny, B. & Paugy D. 1995. Unsaturated fish commu-nities in African rivers. American Naturalist,146: 162-169.

Hurtt G.C. & Pacala S.W. 1995. The consequences ofrecruitment limitations: Reconciling chance,history and competitive differences betweenplants. Journal of Theoretical Biology, 176: 1-12.

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

Jackson D.A., Sommers K.M. & Harvey H.H. 1992. Nullmodels and fish communities: Evidence ofnonrandom patterns. American Naturalist,139: 930-951.

Jenkins D.G. & Buikema Jr. A.L. 1998. Do similar com-munities develop in similar sites? A test withzooplankton structure and function. EcologicalMonographs, 68: 421-433.

Jepsen D.B. 1997. Fish species diversity in sand bankhabitats of a Neotropical river. EnvironmentalBiology of Fishes, 49: 449-460.

Junk W.J., Bayley P.B. & Sparks R.E. 1989. The floodpulse concept in river-floodplain systems. InD.P. Dodge ed. Proceedings of the internation-al large rivers symposium. Can. J. Fish. Aquat.Sci. Spec. Publ., 106: 110-127.

Lamouroux N., Poff N.L. & Angermeier P.L. 2002.Intercontinental convergence of stream fishcommunity traits along geomorphic andhydraulic gradients. Ecology, 83: 1792-1807.

Levin S.A. & Paine R.T. 1974. Disturbance, patch forma-tion, and community structure. Proceedings ofthe National Academy of Sciences USA 71,2744-2747.

Lundberg J. 2001. Freshwater riches of the Amazon.Natural History, 110: 36-43.

Lundberg J.G., Lewis Jr. W.M., Saunders J.F. III & Mago-Leccia F. 1987. A major food web componentin the Orinoco River Channel: Evidence fromplanktivorous electric fishes. Science, 237: 81-83.

MacArthur R.H. & Wilson E.O. 1967. The Theory ofisland biogeography. Princeton, NJ., PrincetonUniversity Press.

Matthews W.J. 1998. Patterns in Freshwater FishEcology. New York, Chapman and Hall.

34 Organization and maintenance of fish diversity

Meffe G.K. & Sheldon A.L. 1990. Post-defaunationrecovery of fish assemblages in southeasternblackwater streams. Ecology, 71: 657-667.

Meyers N., Mittermeier R.A., Mittermeier C. G., daFonseca G.A.B. & Kent J. 2000. Biodiversityhotspots for conservation priorities. Nature,403: 853-858.

Oberdorff T., Hugueny B., Compin A. & Belkessam D.1998. Non-interactive fish communities in thecoastal streams of North-western France.Journal of Animal Ecology, 67: 472-484.

Oberdorff T., Hugueny B. & Vigneron T. 2001. Is assem-blage variability related to environmental vari-ability? An answer for riverine fish. Oikos, 93:419-428.

Petry P., Bayley P.B. & Markle D.F. 2003. Relationshipsbetween fish assemblages, macrophytes andenvironmental gradients in the Amazon Riverfloodplain. Journal of Fish Biology, 63.

Poff N.L. & Allan J.D. 1995. Functional organization ofstream fish assemblages in relation to hydro-logical variability. Ecology, 76: 606-627.

Pringle C.M., Freeman M.C. & Freeman B.J. 2000.Regional effects of hydrologic alterations onriverine macrobiota in the New World:Tropical-temperate comparisons. Bioscience,50: 807-823.

Rahel F.J. 1990. The hierarchical nature of communitypersistence: A problem of scale. AmericanNaturalist, 136: 328-344.

Resetarits W.J., & Bernardo J. eds. 2001. ExperimentalEcology: Issues and perspectives. New York,Oxford University Press.

Robinson C., Tockner K. & Ward J.V. 2002. The fauna ofdynamic riverine landscapes. FreshwaterBiology, 47: 661-677.

Rodríguez M.A. & Lewis W.M. 1997. Structure of fishassemblages along environmental gradients in flood-plain lakes of the Orinoco River. EcologicalMonographs, 67: 109-128.

Saint-Paul U., Zuanon J., Villacorta Correa M.A., GarcíaM., Noemi Fabré N., Berger U. & Junk W.J.2000. Fish communities in central Amazonianwhite- and blackwater floodplains.Environmental Biology of Fishes, 57: 235-250.

Schlosser I.J. 1987. A conceptual framework for fishcommunities in small warmwater streams. InW.J. Matthews and D.C. Heins eds.Community and evolutionary ecology of NorthAmerican stream fishes. Norman, OK, USA,University of Oklahoma Press. pp. 17-24.

Schlosser I.J. & Kallemeyn L.W. 2000. Spatial variationin fish assemblages across a beaver-influencedsuccessional landscape. Ecology, 81: 1371-1382.

Schmidt J.C., Webb R.H., Valdez R.A., Marzolf R. &Stevens L.E. 1998. Science and values in riverrestoration in the Grand Canyon. Bioscience,48: 735-747.

Sheil D. & Burslem D.F.R.P. 2003. Disturbing hypothe-ses in tropical forests. Trends in Ecology andEvolution, 18: 18-26.

Shurin J.B., Havel J.E., Leibold M.A. & Pinel-Alloul B.2000. Local and regional zooplankton speciesrichness: A scale-independent test for satura-tion. Ecology, 81: 3062-3073

Southwood T.R.E. 1988. Tactics, strategies and templets.Oikos, 52: 3-18.

Stewart D.J., Ibarra M. & Barriga-Salazar R. 2002.Comparison of deep-river and adjacent sandy-beach fish assemblages in the Napo RiverBasin, Eastern Ecuador. Copeia, 2002: 333-343.

in shallow waters of tropical floodplain rivers 35

Taylor C.M. 1996. Abundance and distribution within aguild of benthic stream fishes: Local processesand regional patterns. Freshwater Biology, 36:385-396.

Taylor C.M. & Warren M.L. 2001. Dynamics in speciescomposition of stream fish assemblages:Environmental variability and nested subsets.Ecology, 82: 2320-2330.

Tockner K., Schiemer F., Baumgartner C., Kum G.,Weigand E., Zweimüller I. & Ward J.V. 1999.The Danube restoration project: Species diver-sity patterns across connectivity gradients inthe floodplain system. Regulated Rivers:Research & Management, 15: 245-258.

Tonn W.M. 1990. Climate change and fish communities:a conceptual approach. Transactions of theAmerican Fisheries Society, 119: 337-352.

Tonn W.M., Magnuson J.J., Rask M. & Toivonen J. 1990.Intercontinental comparison of small-lake fishassemblages: The balance between local andregional processes. American Naturalist, 136:345-375.

Toth L.A., Arrington D.A. & Begue G. 1997. Headwatersrestoration and reestablishment of natural flowregimes: Kissimmee River of Florida. In J.E.Williams, C.A. Wood & M.P. Dombeck eds.Watershed restoration: Principles and prac-tices. Bethesda, MY., American FisheriesSociety. pp. 425-444.

Toth L.A., Arrington D.A., Brady M.A. & Muszick D.A.1995. Conceptual evaluation of factors poten-tially affecting restoration of habitat structurewithin the channelized Kissimmee Riverecosystem. Restoration Ecology, 3: 160-180.

Toth L.A., Melvin S., Arrington D.A. & Chaimberlain J.1998. Hydrologic manipulations of the chan-nelized Kissimmee River. Bioscience, 48: 757-764.

Townsend C.R. 1989. The patch dynamics concept ofstream community ecology. Journal of theNorth American Benthological Society, 8: 36-50.

Townsend C.R., Scarsbrook M.R. & Dolédec S. 1997.The intermediate disturbance hypothesis, refu-gia, and biodiversity in streams. Limnologyand Oceanography, 42: 938-949.

Ward J.V. 1989. The four-dimensional nature of loticecosystems. Journal of the North AmericanBenthological Society, 8: 2-8.

Ward J.V. 1998. Riverine landscapes: Biodiversity pat-terns, disturbance regimes, and aquatic conser-vation. Biological Conservation, 83: 269-278.,47: 517-539.

36