Early Dry-Season Community Structure and Habitat Use of Young Fish in Tributaries of the River Sinnamary (French Guiana, South A

Environmental Biology of Fishes 50: 235-256,1997. O 1997 Kluwer Academic Publislzers. Printed in the Netherlands.

Early dry-season community structure and habitat use of young fish in tributaries of the River Sinnamary (French Guiana, South America) before and after hydrodam operation

Dominique Ponton' & Gordon H. Copp' Centre ORSTOM de Cayenne, B.P 165, 97323 Cayenne Cedex, French Guiana, France Department of Environmental Sciences, University of Hertfordshire, College Lane, Hatjïeld, Heits AL10 9AB, UK 1' Received 23.4.1996 Accepted 1.2.1997 ii'"1. Key words: juvenile fish, neotropics, flooded areas, wetlands, river reservoirs, Petit Saut, hydroelectric dams

Synopsis

We examined fish community structure and habitat use at the start of the dry seasons: (1) in 10 tributaries of the River Sinnamary (French Guiana) before and after the start of dam operation, and (2) in 10 upstream tributaries and at 10 littoral sites in the newly-createdreservoir after the start of operation to assess the impact on fish juveniles of a hydroelectric dam built on the river's lower section. After the first year of dam operation, juvenile fish communities downstream of the dam showed an important decrease of the relative abundance of Characiformes, and Perciformes dominated. Principal components analysis revealed a distinct upstream-to- downstream progression in the juvenile fish communities with post-reservoir downstream and reservoir sites representing transitions between the upstream and pre-reservoir downstream sites. Canonical correspondence analysis and electivity indices of fish-habitat associations revealed three relatively distinct groups of sites, corresponding to the downstream, reservoir and upstream taxa. The propohion of juveniles presenting higher-than expected frequencies (Fisher's exact test) towards local environmental variables was higher for taxa more often caught in upstream sites. Inversely, juvenile taxa more frequently observed in downstream and reservoir sites appeared less selective towards local environmental characteristics. In the downstream '. reaches of the river, hydrodam operation is expected to drive the fish community towards a new biologically accommodated state where tolerant species will dominate and sensitive species will be lacking.

Rationale utaries of Guianese rivers (Boujard & Rojas-Beltan 1988, Boujard 1992), however there have been no Unlike many large tropical rivers, whose seasonal published investigations on any aspect of the life hydrological cycles are more predictable (Bayley history of larvae and juveniles. In fact, little is 1988), small tropical rivers such as those in French known of the reproductive styles, early develop- Guiana are subjected to extreme short-term varia- ment, growth, behaviour (Géry 1969,Westby 1988), bility in discharge throughout most of the year and indeed the identity of some fish species in rivers (Westby 1988). Initial investigations have revealed of French Guiana. This lack of information is of par- --_ the important influence of hydrological regime on ticular ,concern with respect to the River Sinnamary, the density of adult fishes in the floodplain and trib- where a hydroelectric dam recently began oper- I N 5'30'

Atlantic Ocean

I I I I I I I 6" -____1.54' 53' 52'

IOlD Dam 5"66'N

R3 R4 R5

I I 1 I I l I

u7 N U8 4'30'

Figure/. Map of the River Sinnamary (French Guiana. South America),the reservoir at Petit Saut as it appeared in AuguatlY91. and the & 1YW study sites (Al to AIO: dilwnstream ]?')?I. D1 to DIO downstream 1994. RI tu R7: luwerreservoil-lYY4. RS to R10: upper rescrvair and UlU: U1 to upstream 1994). Only sampled tributaries are presented. 237 ation in the river’s lower section at Petit Saut (Fig- utaries in November, at the end of the dry season. ure 1).As a consequence, assessment of the dam’s Recent evidence supports this pattern (D. Ponton impact on the success of fish reproduction will be unpublished observations) and suggests that the difficult. dry season in Guianese rivers? similar to that in Species-environment interactions play an impor- other South American rivers (Welcomme 1985) and tant role in the life-history of freshwater fishes, es- to the end of summer in some European rivers (e.g. pecially with respect to reproductive and recruit- Copp 1989, Copp et al. 1994), is a crucial period in ment success. Among the most influential environ- the survival of young of the year riverine fishes. mental factors in tropical river systems is hydrological regime (Bayley 1988). As reproduction in tropical fishes is often set in motion by wet season Study sites, material and methods floods (Welcomme 1979, Lowe-McConnell 1987, Munro 1990, Boujard 1992, Schwassmann 1992, Val The River Sinnamary is approximately 260 km in & Almeida-Val 1995), hydroelectric regulation of length (Figure 1).It drains an area of 6565 km2with the Sinnamary’s seasonal hydrological regime a mean annual precipitation of 3000 mm and has an could have devastating effects on fish populations mean annual discharge of 230m3 sV1, ranging be- in the floodplain tributaries downstream of the tween 120 and 400 m3 d (for a description of the en- dam. tire river system, see Boujard 1992 and Tito de Mo- The movement of young and small fish between rais et al. 1995). The river’s lower course meanders the main channel(s) of large tropical rivers and their through a old flat coastal plain dominated by palms flood plains is well known (Welcomme 1985). How- Mauritia flexuosa and Euterpe oleracea as well as ever, in smaller meander-type rivers such as the ‘moutoucliis’Pterocarpus oJSFicinaZis.In this section, River Sinnamary, adult fish migrate between the tidal flows strongly influence the river’s level both river and its floodplain via the tributaries (Boujard on a daily and a seasonal basis, with the salt waters 1992), which function throughout most, if not all, of of the flood tide pushing up under the river’s fresh the year as feeding, spawning and nursery grounds waters. Upstream, four forest types are associated for fish larvae and juveniles (D. Ponton unpublish- with the river and its tributaries (Granville 1979): ed observations). terra firme arborescent assemblage, permanent The aims of the present investigation on young swamp forest rich in monocotyledons, flooded for- fishes in tributaries of the River Sinnamary were ests dominated by pteridophytes and palm swamps twofold: (1) to compare juvenile community struc- (for a complete list of plant species associated with ture and fish-habitat associations in the lower Sin- the River Sinnamary see Tito de Morais et al. 1995). namary before and after the start of the hydrodam’s During the 1992-1993 rainy season, successive operation; and (2) to develop an empirical model of flow events of short duration were superimposed species-habitat associations in the Sinnamary’s trib- on the increase of the river water level from No- utaries. The beginning of the dry season, Julylh- vember onwards (Figure 2). Maximum values were gust was chosen because it is during this period observed between March and May, when water lev- when the young of most species have reached their els were about 5m above those observed at low juvenile period of development (D. Ponton unpub- flow. Starting on 6 January 1994, impoundment im- lished observations). According to Boujard’s (1992) mediately induced a rapid increase in the reser- proposed model of the annual cycle in Guianese riv- voir’s water level. For technical reasons, the reser- ers, July is the sixth interval, when floodplain ‘fish voir water level was kept at 31 m from the beginning are fat and return to the main river’. However, Bou- of the dry season onwards. During most of the peri- jard et al.’s (1990) study on the River Arataye, upon od needed to fill the reservoir, downstream river which the temporal aspects of his model appear to flow was kept at 100 m3 s-’ except during some short be based (Boujard 1992), indicated that young and hydrological events. As a consequence, the river small fishes still inhabited the floodplain of the trib- level remained low during most of the 1993-1994 tributaries upstream from the reservoir in 1994 and - al downstream I992 I993 10 stations in the littoral zone of the reservoir in 1944 (Figure 1). seven in the mid-to-lower reservoir (lower reservoir) and three at its upstream end (up- 4 per reservoir). The exact locations of sampling sites were choosen at random (i.e. without knowing whether fish were present or not) hut were restricted to places where rotenone sampling had been al downstream 1993 - 1994 found to be the most efficient for sampling young fish. i.e. sites with maximal depth < 1.5 m and water 4 velocity < 20 cm sec-'. Indeed. deeper sites impede an efficient retrieval of sunken fish and greater water velocities require the use of greater quantities of toxicant. which could have potentially disastrous effects on downstream fishes. At each site, water tem- peraturc. pH, oxygen and conductivity were first 4 measured with a ICM 51000 multiparameter ana- lyzer before any disturbance. Then. the extents of the sampling sites were delimited by stop nets (1 nini mesh) carefully set on the bottow in order to avoid fish escapes. Three successive doses of PREDATOXB-d_, (6.6% emulsifiable solution ro- upstream 1993 - 1994 of 41 tenone extracted from Dewis rllipfica by Saphyr, Antibes. France) were mixed in the water in order 2 to reach the effective contact time' (sensu Gilder- hus 1072) of the most resistive juveniles taxa of the community with final concentration never exceed- NDJFMAM] ing 3 mg 1.'. This gradual increase in rotenone con- )ASO centration allowed the progressive collection of the Month least resistant (mostly small Characiformes), then the moderately resistant (mostly Siluriformes),and Figrtw -1. Water dnwnqlrcam from thc liydrodam in levels 1997- finally the most resistantjuvenile fish (mostly Gym- IL?).? and 19~13-1W.in the reservoir and upstream in 19Y3-lW4. from Novemher (beginning of the rainy season1 to October. notiformes) at the water surface before they died Nate the different scales un the vertical axeb. and sank to the bottoni. At each sampling site fish collection was performed by a minimum of 4 per- rainy season. A natural flow regime. including fluc- sons equipped with 1 nim mesh dip nets. All speci- tuations. resumed in the river downstream of the mens were immediately preserved in 90% alcohol. dani as soon as the filling of the reservoir stopped in The fish not surfacing after rotenone application July 1993. Upstream from the reservoir, flow regime were retrieved by stiring the water in crevices and presented a natural pattern with water level differ- under trunks, removing woody debris, and carefully ences between low and high flow reaching 2 ni and checking for dead fish in leaf litter (tutal duration of 1 the highest water levels being observed between sampling 1-2 hours). No attempt was made to de- April and May (Figure 2). toxify rotenone outside the sampling area with po- The fish were collccted from stretches of known tassium permanganate because: ( 1) sampling vol- length in 10 tributary streams downstream from the dam in 1993 (henceforth, the pre-reservoir down- ' Eflectivc contact time refers to the length of time a fish must hc stream) and 1993 (post-reservoir downstream). 10 exp~\cdto rotenrine (CI reaull in dcath. 239 ume was always small compared to surrounding wa- ed data) or different types of traps (11% of species: ters; (2) most fish species outside of the sampling A. Bezançon & D. Ponton unpublished data), scien- site were found to detect and avoid rotenone; (3) tific and aquarium literature (5% of species). When clay, largely represented in our sampling site, is size at first maturity was not available (40% of spe- known to reduce rapidly the toxicity of rotenone cies), length-frequency distributions were estab- (Gilderhus 1982);and (4) above 23 "C the half life of lished and specimens belonging to the first modal this toxicant is less than 1 day (Bettoli & Maceina size group of the population's size distribution were 1996). considered as immature. The number of specimens After fish sampling, eleven other environmental for each taxon and the number of taxa at a given site variables were measured quantitatively (mean wa- were then divided by the site's surface area to deter- ter depth, mean channel width, water velocity), vi- mine relative densities (individuals mu') and the rel- sually estimated (percentage of bottom covered by ative number of taxa per sample (taxa m-'). mud and sand in substrate, by leaf cover on bottom, Comparisons of the mean number of species be- by branches, tree trunks, etc. within site), and deter- tween site and years were performed on the resid- mined by cartographic methods (distance from riv- uals of the linear regression for species richness vs. er, distance from estuary), or assigned by qualita- sampling area (species richness = 0.075 * sampling tive class (transparency, channel shape). For bot- area + 14.484, F = 4.715, df = 39, p = 0.036) by using tom substrate, at least three or four samples were an exact permutation test (Good 1993) with a mod- examined to verify visual estimates. Transparency ified SAS macro (Foster 1995 and personal commu- was judged by eye using visibility of the bottom at a nication). As the number of individuals was not re- depth of about 0.5 m. Distance from estuary was es- lated significantly to sampling area (F = 1.4334, df = timated roughly, as the most recent 1/50 O00 maps 39, p = 0.2386), comparisons of the mean relative available to us date from 1966 (sheets NB-22-VI- abundance of individuals were directly tested by us- II-la,- IC& - 2b; Institut Géographique National, ing the same permutation test. Shannon diversity Paris), and field experience proved them to be index and evenness were also calculated for each largely approximate. section of the river in 1993 and 1994. Fish reproduc- In the laboratory, all the collected specimens tive potential at the various sites was assessed from were transferred to 70% alcohol and identified us- the presence and absence of fish taxa at the various ing keys for adults by Géry (1977), Le Bail et al?, sites (e.g. Copp 1989) using principal components Rojas-Beltran3, Lauzanne (unpublished) and keys analysis, which is appropriate for relatively homo- for young stages by Ponton (unpublished). Stan- geneous data sets with relatively low species turn- dard lengths (SL) were compared to the size at first over (Gauch 1982). maturity in order to separate juveniles from adults. Thirteen environmental variables were retained Sizes at first maturity were obtained from fish spe- for analysis. pH (mean = 5.5, SD = 0.446) and con- cies caught with gill nets (44% of species: L. Lau- ductivity (mean = 26.2, SD = 3.282) varied little and zanne, L. Tito de Morais & B. de Mérona unpublish- were excluded. Retained environmental data were converted to semi-quantitative categories using re- J. ciprocal and natural logarithmic transformations LeBail, P.Y., P. Planquette & Géry. 1983. Clé de détermina- for oxygen content (ppm) and distance from estu- tion des poissons continentaux et côtiers de Guyane. Bull. Liai- son Groupe Rech. Guyane No 6 & 8. INRA, 97310 Kourou, ary, respectively, and then tested for normality (Lil- French Guiana. lefors 1967). Using software by Chessel & Dolédec4 LeBail, P.Y., P. Planquette & J. Géry. 1984. Clé de détermination and Thioulouse (1989), the data on taxa and envi- des poissons continentaux et côtiers de Guyane. Bull. Liaison Groupe Rech. Guyane No 9. INRA, 97310 Kourou, French Guiana, France. Chesse1,D. & S. Dolédec. 1992.ADEVersion 3.7: Hypercard@ Rojas-Beltran, R. 1984. Clé de détermination des poissons con- Stacks and Quick Basic MicroSoftO Program library for the tinentaux et côtiers de Guyane. Bull. Liaison Groupe Rech. Analysis of Environmental Data. 9 fasc. URA CNRS 1451, Uni- Guyane No 7. INRA, 97310 Kourou, French Guiana, France. versité Lyon 1,69622 Villeurbanne cedex, France. 820 pp. .. 1. Irih(c, List of fish tasa. authority. ahhreviated species code. number of juveniles (Nj).tntal numher of individuals (Nt) and range of standard length (SL) in study strechcs from 10 downstream sites in late July -early August 1993 and 30 others in August 1994.

Order family suhfamily Genuslspecies Authority 'lisa code SL N,/N, Min-max (mm)

Characifornies Hemiodontidae Hemiodantinae ISIIS 11-62 Curimatidae Chilnduntinae Puyo. 1945 CZ SI9 ?I 1-71 Curimatinae Steindachner. lRlO Ch 15115 574s cy 7361741 10-9s Anostomidae Géry. 1960 AV 317 60- loo (Bloch. 1704) Lf 31Q 70-33 GCry. Planquette 6: Leßail. 1991 Lg 111 s7 Eigenmann. 1911 Lr hl17 50-lt;l Steindachner. lqlU 1-P 111 17 LS hlb 40-145 Erythrinidae (Schneider. lSOl) Fe iniin 13-8s (Spix. 1829) HO 112 121-130 (Val. in Cuv. &Val. 1846) Ha 33133 S-167 (ßloch. 17941 H m 94195 iI -1sn Lebiasinidae 1911) Pyrrhulininae (Regan. CC 3nmo 13-39 Günther, IS73 Nb s51so 13-2s Val. in Cuv.. 1846 Pf 13n11s1 13-63 Gasteropelecidae Linnaeus, 175s GS 51s 21 -44 C'haracidae ('S 713 10-12 Characidiinae Fowler. 11114 cf 1151115 Q-41 (Eigenmann. 19OP) MI h li1 43-43 MS 314 16-33 ((;&-y. 1960) Ml c hh/b7 11-11 i'haraoinac (Bloch. 1794) Af 14117 14-1OCl (Schomhurgk. 1S4l) Am 111 54-54 Günther, 1864 CP 1I4 38-131 Cheirodontinar (Ulrey. 1894) Plì1 8741874 9-14 Géry. I960 Ps 13 1411 314 u-76 Stethaprioninae (Val. in Cuv. Sr Val.. 1849) PO 313 31-43 Tetragonoptcrinae (Linnaeus. 1758) Ah 414 30-7s Gery. Planquette & LeBail, 1Wti AL 1511151 lh-71 (ÌCry. Planquette 6: LeBail, 19% AM 1911Q 31-48 BI 43/45 11-1I4 ß1 511153 3-87 B3 91'1 4s-84 BS 13/13 13-17 (Gery, 1?591 Hh 7/12 14/32 (Steindachner. ISS?) H o h75lS 15 11-3 1 (Steindachner, ISS11 Hs 111 33 (Gill, IS581 Ilu 031'J5 I 1-31 b/l' Schult;., 1944 Hy 11-13 (Steindnacher. IXS1) blc 21 1X1O7 1340 Ish41 (Günther. MC 741121 18-87 Mh Gery. 19h6 I711171 Ih-35 (CÌünther. ISb.1) Mo b7(J1788 14-98 ~o Ill 55 Eigrnmann, IR(19 Ph 333140s 13-55 241

Table 1. Continued.

Order family subfamily Genuslspecies Authority Taxa code SL NjN, Min-max (”)

Siluriformes Auchenipteridae Tatia intermedia (Steindnacher, 1876) Ti 18/18 14-45 Pimelodidae Piinelodella cristata (Müller & Troschel, 1848) Pc 42/46 28-120 1840) 12/14 70-106 Pimelodella gracilis (Val. in Cuv. &Val., pg . Pinielodus ornatus (Kner, 1857) Pt 212 130-135 Pseudopi~nelodusraninus (Valenciennes, 1840) Pr 13/13 11-95 Pseudopinielodus zungaro (Humboldt, 1833) Pz 111 28 1824) 717 72-195 Rhamdia quelen (Quoy & Gaimard, Rq Cetopsidae Hemicetopsis sp. Hc 111 17 Aspredinidae , Bunocephalinae Bunocephalus coracoideus Cope, 1874 Bc 213 34-77 1909) 313 54-63 Trichomycteridae Triclioniycterus guianense (Eigenmann, Tg Callichthyidae Corydoras octocirrus (Nijssen, 1970) Co 616 . 43-61 Hoplosternum thoracatuni (Val. in Cuv. &Val., 1840) Ht 313 25-108 I‘ Loricariidae Ancistrus aff. hoplogenys (Günther, 1864) Ah 8/10 24-82 Harttia surinamensis Boeseman, 1971 Hr Oll 101 Lasiaiicistrus niger (Norman, 1926) Ln 214 62-112 Gymnotiformes Sternopygidae Eigenmannia virescens (Valenciennes, 1847) Ev 18/19 11-147 Sternopygus niacrurus (Bloch & Schneider, 1801) Sm 16126 311300 Hypopomidae Bracliypapomus beebei (Schultz, 1944) Bb 761105 25-196 Hypoponius artedi (Kaup. 1856) HA 4113 43-257 Gymno tidae Gyinizotus anguillaris Hoedeman, 1962 Ga 78/78 36-248 Gynznotus carapo Linnaeus, 1758 Gc 82/82 42-335 17/17 15-34 Gyinnotus spp. (unidentif. juv.) GY Cyprinodontiformes Aplocheilidae Rivulus agilae Hoedeman, 1954 Ra 51/66 9-31 Rivulus igneus Huber, 1991 Ri 12/34 11-50 Rivulus xiphidius Huber, 1979 Rx 1471147 9-25 Poeciliidae 1894) 6-21 Poeciliinae Poecilia parae (Eigenmann, PP 66/66 Poecilia sp. Pe 414 13-14 Synbranchiformes 1795 515 101-150 Synbranchidae Syiibranclius mamioratus Bloch, SY Perciformes Nandidae Nandinae Polycentrus schoniburgkii Müller & Troschel, 1848 Pk 44/51 6/56 Ciclilidae Cleitliracara niaronii (Steindachner, 1882) Cm 38/45 17-67 Ciclilasorna biinaculatuni (Linnaeus, 1758) Cb 111 103 Creiiicichla saxatilis Linnaeus, 1758) cs 2531259 91175 1905) 1404l1418 7-120 Krobia guianensis (Regan, Kg Nannacara anomala Regan, 1905 Na 3381465 7-62 Satanoperca aff. leucosticta (Müller & Troschel, 1848) s1 79/87 12-158 Eleotridae Dormitator niacrophtlialinus Puyo, 1944 Dm 313 19-21 Eleatris anzblyopsis (Cope, 1870) Ea 83911273 9-71 I unidentified fish other than larvae 313 9-13 unidentified fish larvae 313 5 eigen values loo downstream 1993

0.4 12.9 1- -+68 O 1.8 1.5 I

downstream 1994

o N o a -1 $ "1 upper reservoir 'I 994

O 2 -2 -1 1 3 PC 1 2.6 O I 0.9 2.1 51 Fiww.?. Principal compilnents analysis of the JO sites hy 68 species data matrix in ahsenseipresence, with (a\ eigcn values and (11) PC1-hy-PC2 plot. The first two xieh account Tor 29% o1 the variability. The upstream (LI), reservoir (R),downstream IWî (A) Lind downstream 1W (DI trihutaries are indicated.

ronmental variables were then subjected to canonical correspondence analysis (CCA ), a method of direct gradient analysis that provides an empirically based method of generating both predictive (Ter Fi~uw4. Relative ahundance of Characiformes. Siluritommes, Gymnc~tif~~rnies.Cv~rincidantiT1,rnie~ and Percilormes juveniles Braak 1986) and descriptive (Chessel et al. 19S7) in thc pre-reservoir (1903) downstream anti post-reservoir models of species-habitat associations. CCA is also (lcW)downstrcam tributaries. the Itwcr and uppel- reservoir a powerful alternative to niultivariate analysis of sites and ths upstream (1W)tributaries. variance in the analysis of data from 'before-aftcr- control-impact" studies with and without replica- on in the group of samples having a given category tion of the impacted site (Ter Braak & Verdonschot of environmental variable and the frequency of that 1995). taxon in all samples (non-null). The samples-by-variables niatris was cross-tab- ulated with that for samples-by-tasa (converted to absence/presence. l/O. tasa occurring at less than Results * three sites being eliminated) to determine the frequencies of occurrencc and variables-tasa associ- A total of 13 309 specimens. representing 86 tasa 13 ations (Fisher's exact tust. Good 19931. and to gen- (Sf)distinct spccies) from fanlilies and 6 orders erate environmental profiles of hahitat use fur each were collccted at the 40 study sites (Table 1). Only taxon. The habitat selectivity profiles were calculat- Htrrrritr ,~iiriiz~iiizc~i.sis( Siluriformes, Loricaridac ed as the difference between the frequency ofa tas- was represented strictly by adults. The 11 994juve- 243

niles were dominated by Characiformes and Perci- upstream-to-downstream progression (Figure 3) in formes, which accounted for 69.2% and 25% of the the juvenile fish assemblages of the sites, with post- individuals, respectively. Moenkliausia collettii, reservoir downstream tributaries and the reservoir Pseudopristella siniulata, Pristella maxillaris, Cy- sites representing transitions in the upstream and phocharax sp. and Hernigrarnnzus ocellifer were the pre-reservoir downstream sites. Exceptions in this most abundant Characiformes. Krobia guimemis pattern were the two sites just downstream of the and Eleotris ariiblyopsis dominated the Perci- dam (Ag, Alo) and the three sites at the upstream formes. end of the reservoir (R8-10), which had juvenile Principal components analysis revealed a distinct communitiesresembling those of the upstream trib-

Table 2. Number (n) and estimated densities (ED, n m”) of individuals, richness (S) and relative richness (RR, S m”) number of taxa, Shannon diversity index and eveness for the different river’s sections and year. With SE = standard error.

~ Downstream 1993 Downstream 1994 Lower reservoir Upper reservoir Upstream 1994

n ED n ED n ED n ED n ED

~ ~ 423 7.83 78 1.O4 47 0.42 130 2.17 270 4.50 411 20.55 197 2.19 55 2.20 399 19.95 489 4.53 114 2.48 549 14.64 185 7.40 271 5.42 565 9.42 1703 17.74 ’ 269 2.69 23 0.92 756 10.08 276 4.60 221 5.53 63 2.52 389 11.11 232 2.90 84 0.93 843 15.33 548 15.22 153 4.08 188 2.85 61 2.44 357 4.41 18 1.20 48 1.00 185 4.63 148 1.97 199 1.99 211 11.72 123 2.73 25 0.56 667 47.64

Mean 360.10 6.61 185.80 3.34 182.43 4.46 266.67 9.18 443.70 12.33 SE 154.7 2.18 48.0 1.34 11.84 2.00 77.68 5.47 61.17 4.10

Downstream 1993 Downstream 1994 Lower reservoir Upper reservoir Upstream 1994 S S RR S RR S RR S RR RR

19 0.35 16 0.2 10 0.1 20 0.3 28 0.5 30 1.50 21 0.2 10 0.4 18 0.9 31 0.3 25 0.54 15 0.4 7 0.3 23 0.5 24 0.4 25 0.26 15 0.2 5 0.2 25 0.3 26 0.43 22 0.6 8 0.3 20 0.6 21 0.26 16 0.2 17 0.3 24 0.7 16 0.43 21 0.3 13 0.5 22 0.3 9 0.60 16 0.3 22 0.6 20 0.27 22 0.2 15 0.8 18 0.40 8 0.2 20 1.4

~ ~ Mean 20.9 0.50 17.20 0.28 10.00 0.30 20.33 0.56 23.10 0.58 t SE 1.9 0.12 1.39 0.04 1.51 0.05 1.45 0.17 1.41 0.11

Downstream 1993 Downstream 1994 Lower reservoir Upper reservoir Upstream 1994

H’ E H’ E H’ E H’ E H’ E

3.681 0.659 3.853 0.679 1.996 0.420 3.644 0.694 3.650 0.647 utaries. The pre-reservoir downstream tributary er for the pre-reservoir downstream tributaries, the AS distinguished itself by a higher proportion of upper reservoir and the upstream tributaries than Perciformes, as in the lower part of the reservoir post-reservoir downstream and lower reservoir (Figures 3,4). sites (Table 3). It indicates that independently of In 1993, the juvenile fish communities down- sampling size the numbers offish species are lower stream nf the dam were dominated by Characi- than expected in the post-reservoir downstream formes (>S3% of the individuals), with Perci- and lower reservoir sites. formes accounting for only 13.9% (Figure 4). In The most abundant taxa also varied between riv- 1994. the same sites showed a significant (Fisher es- er sections and years (Figure 5).Downstream ofthe act test, p < O.nO1) decrease of the relative ahun- dam, the five most abundant juvenile taxa sampled dance of Characiformes (35.9%) when compared to in 1993 were ibí'oerikhairsia collettii, Pristellri nirixil- the relative abundance of Perciformes (55.3%).Af- Itiris, Pseiríloprisrrllri siiniilatn, Hrniigrai~iinirsocel- ter dam operation, communities in the lower reser- li,frr and Eleotris ainblyopsis. In 1994. juveniles of voir contained a very high proportion of Perci- the Perciformes Eleotridae E. ai?iblyopsis re- formes. whereas those in the upper reservoir resem- mained dominant in downstream tributaries. fol- bled more those of upstream reaches. Juveniles of lowed in abundance by one Lebiasinidae. Copella 4 Siluriformes, Gymnotifnrmes and Cyprinodonti- carseiwiiieiisis, and three Cichlidae. Na~~~inenrt~ formes never exceeded 9% of the sample regardless tiiioinnla, Crenicichln saxatilis, and Krobin giiiaiieii- of the river section. Assuming that the character of sis. Young K. gzriaiierisi,~dominated in the lower res- the upstream tributariesin 1994is typical of the trib- ervoir along with Coprlln carswwirieiisis, Sritmio- utaries of the entire Sinnamary system prior to dam percrr sp. aff. lericosticta, Heriiigrriiiirriiis ocrllifrr construction, operation of the hydrodam seems to and Hoplias iiinlabriricirs: these taxa were also have effected a decrease in the relative reproduc- amongst the most abundant taxa in the upper reser- tive success of Characiformes in the downstream voir with n/loerikhorisiacollettii, kl. oligolepis, Cien- tributaries (Figures 3,4). icichltr sax-ntilisand Cyphocliarr~xspp. The five most Juvenile species richness and density varied abundant juveniles caught in the upstream tributar- among sites (Table 3).In 1994. the mean numbers of ies are all Characiformes (ilfoeizklinirsirrchi:\wrgj~- individuals per unit of sampling were significantly wn,P,rrirtlc~pristelll1 siniillata, Cyphochrira.r spp., M. lower in the lower reservoir (exact permutation digolepis. and Phetiacogastrr aff. niegnlostictio). test, p 5 0.05) and the downstream tributaries (p 5 Only nine species of juveniles were collected in at 0.05) than those upstream. Maximum values ob- least three downstream sites in 1993 and 1994, re- served upstream were four times those obtained spectively (Table 4).For some ofthem, their associ- downstream. The mean number of species per sam- ation with environmental variables differed be- ple reached minimum values in the lower reservoir tween years and when differences occurred. associ- and maximal values upstream. When compared by ations were often more significant after dam coni- permutation tests. mean residuals of the regression pletion than before. As an example, association for numbers of species verse sample area were high- between Eigenmaiiiiia IVresceiis and the type of , Ei/de.?. Comparisons of tht. residuals of regression for the numher of species versus sampling arca performed using an exact permutation test (Good 1093) with a modificJSASmacro (Foster 101)5 andFersonalcamrnunicatitin). Onlyprobahilitiesfnrsignificant Jifferencesare given. i Downstream 1093 Downstream 1904 Lower rcservciir llpper reservoir

Upstream l?W - 0.c1nn3 0.000 I - Reservoir (head) lW4 - 0.0179 il.(li1811 Reservoir (body) 1904 n. ooos n.n133 Downstream 1904 11.0197 245

3000 t MC Ps Figure 5. Ranked abundances of fish juveniles taxa in the pre- 1000 downstream 93 reservoir (1993) downstream and post-reservoir (1994) down- Ea 3001 ;m HO". stream tributaries, the lower and upper reservoir sites and the IO0 ...- upstream (1994) tributaries. Only the codes (Table 1) of the five most abundant species are presented. 30 i "s lo substrate changed from non significant in 1993 to I 3 I highly significant in 1994 (Table 4). As potential dif- 1 2 20 II I 1 3 5 10 30 50 100 ferences in habitat selection were detected before 3000 and after dam closure, analysis of fish-habitat rela- IOOO] Ea downstream 94 tionships was restricted to 1994 samples. - In the canonical correspondence analysis of fish- 300 cc. 100 - habitat associations (Figure 6), distance from the

30 - estuary, channel shape and water temperature were 10 - the most important environmental variables, i.e. the longest vectors (Ter Braak & Verdonschot 3- 1995). Three relatively distinct groups of sites were revealed by the CCA, corresponding to the down- 2 3000- (u stream, reservoir and upstream tributaries. Al- o3 1000- Ag lower reservoir 94 though a compromise of the numerous taxa-varia- > 300- bles interactions measured, the CCA summarises :c :c Ho 7.- IOO- well the main interacting factors that define habitat . M. of the various taxa. Upstream taxa, such as Cy- E 30- SI Hg.. & 10- . plzocharax spp., Characidiurn gr. fasciatum, Pseu- nE . dopristella sintulata, Astyanax cf. keithi, Moenkliau- 3- 5I 3 li I sia collettii, Moenkhausia, oligolepis, Phenacogaster z aff. nzegalostictus, Pinzelodella cristata, and Rivulus xiphidus, generally preferred asymmetrical channels (mainly two-stage) with moderate water veloc- 300 ities, elevated oxygen levels, and equal proportions of mud and sand (Figures 6,7a-b). 30 Downstream taxa, such as Naiznostonzus beckfor- 10 di, Bryconops sp3, Henzigrainnius uniliiieatus, Ei- geriniannia virescens and Eleotris ainblyopsis, also 31 I! I 20 preferred asymmetrical channels, though mainly I 2 3 5 10 30 50 100 narrower l-stage, as well as slow-to-elevated water 3000 - Mm cy velocities, lower oxygen concentrations and water IOOO-' . . Ph upstream 94 transparencies, and greater proportions of sand in 300- ps M~~u.. the bottom substrate (Figure 6,7c). Crenicichla sax- 100 - atilis (Cs), though ordinated amongst the reservoir 30 - taxa, is probably a remnant of the Sinnamary's orig- 10 - inal middle stretches, occurring in high numbers 3i % I both at the upper reservoir sites and in tributaries 14 1 downstream of the reservoir (Figure 5), but observ- 1 2 5 10 50 100 3 20 30 ed less often than expected in the reservoir (last Rank depth category, Figure 7d). Cs's ordination amongst reservoir taxa is probably due to its preferences for Ttrlde-/.Camparison olthe total frequency (IC)‘if environmental vari;ililcs mdjuvenile taxa (It 1present at downstream tributaries in 1W3 anti 19W. Only significant associations helwecn taxa ;ind variahlez (Fisher esaet text I :ire presented lor each taxa year and variahle. See Table 1 lor tiixa codes.

Depth (ni) 1 !1-0.4 3‘ - (1371 13131371312 0 ~1.41-l1.S-I 4 h 3 3243 (1 3 4 2 3 3 7 1 2 2 73 :(I s 37 -1 23112 3 137 II (I I’ 3 o 3 (1.(133 Width (ni) 7 II-? .? 1 I 2.14 h5 S 4.1- 1o 14 (1 reservoir O (I o

Ti Temperature ( 73-23.‘) 2 II 7 CI 1 o I Ïl 1 (I 1 II (1 Il II o I II 24-34.4 43 3137304o 3 1133237 25-75.(2 37 1475132h 2 3 Il 4 2 2 o 5 n il 31-76.Q 7 o 1 3 (1 1 (1 2 Il 1 o 2 o 1 1 II 2 77 (1 (I o (I Il II (I Il II (I II (I (I II O (J I1 II (I o33

il il (I 3 (I 2 2 (1 II 1 o II o 0 Il 3 57 31s2.115 1317121 .i 7 3331733 II 122 o 7 I 1 33 (1 (1 2 1 Iï 1 2 II 3 13 13 !1 !1 O II Il O Il II (I II II II 0 O 0 II i1 0.(172 (1.(143 O.Il43 TI anspareney Low (I 3 (I 0 Moderate --17 I 0 High 43 55

C’hannel shape Symetrical 2 1 1~11il2(13nl(l1II1171’1 As. meander 4 8 4 5 3 I1 3 2 3 s 3 5 3 3 I4(1 3 3 5 AS. ‘-stages 4 I ‘~141114131l131311l

Mud (“,{li i (1-33 13 II Il 13 11 I2 I1 2 13 I3 I31 .? 34-hh 31 I I7 II I I7Il I I I o 2 11(I ’(1 (1 (-17-1 (l(1 7 (1 4 5 4 4 If1 4 h .3 3 1 I7 3 17 4 (m.1 (1.1133 247

elevated water temperatures, shallowwater depths, amongst the downstream species is probably due to higher oxygen concentrations and sandy bottoms its strongly preference for leaf cover and elevated (Figure 7d), electivities characteristic of reservoir water velocities, as well as its avoidance of turbid taxa. waters (Figure 7c), electivities characteristic of Indeed, reservoir taxa (Figure 6) appear to be downstream taxa. transitional in habitat electivities (Figure 7d), Overall, the proportion of juveniles demonstra- though generally avoiding water velocity and pre- ting strong habitat preferences (Fisher’s exact test) ferring various levels of leaf cover, ligneous debris, was higher for taxa more often than expected oxygen concentration, water depth and proportions caught in upstream sites (Figures 7a-b). Inversely, of mud and sand substrates. Copella cavsevennensis juvenile taxa more frequently observed in down- (Cc), though ordinated amongst the downstream stream and reservoir sites appeared less selective taxa (Figure 6), occurred more often than expected towards local environmental characteristics (Fig-

9 in the reservoir (Figure 7c), where it also occurs in ures 7c-d). higher numbers (Figure 5). Cc’s ordination

Q Table 4. Continued.

Taxa Hm Pf B2 Mc Mh Ev Ga Rx 4 Year 93 94 93 94 93 94 93 94 93 94 93 94 93 94 93 94 93 94 93 94 fc ft7 5 8 7 6 3 9 6 7 6 5 4 4 6 5 4 5 7 Sand (%) 0-33 76 645441646331123124 34-66 21 112011201110211020 67-100 13 001311120213131313 0.003 0.033 0.033 Leaf cover (%) 0-20 25 222312232323240314 21-40 21 112110201000101020 41-60 30 202030302020102020 61-80 24 122311232301022103 81-100 10 100000000010000000 Ligneous debris (%) 0-9 11 1 00100010111010 101 10-19 35 2 23422332323242 315 20-29 52 4 24041515120112 031 2 30 12 O 11200110100101 010 0.050 Distance from the estuary (km) 0-29 44 3 34320434221102 002 30-59 55 4 23433433423252 344 60-89 11 O 01010100010111 111 90-119 O0 O 000000000000 00 000 2 120 O0 O 000000000000 00 000 d 0.004 Distance from the river (m) O o1 010001000100010000 1-5 13 110201010111020202 6-100 I5 5 2 6451145342433254 101-1000 2 1 112110212101002001 0.04s cor. eoeff. 0.6 eigen values 0.4 0.2 O

I I I I I I O '7 -1 -0.5 0.5 1 1.5 d I

Fi,$i(reO.Canonical correspondence analysis (Chesscl et al. 1W7) biplot for the 42 fish taxa of (seeTahle 1 TOT codes) and 13 environmental variables (see 'Ihble 4 for variables and categories). The eigen values (a)and correlation coefficients are illustrated graphically 10 facil- itate evaluation. For clarity. the sites are indicated hy points (h),their ordination scores reflecting their geographical origins (upstream. reservoir. downstream l0?4 as labelled). 'I'hc fish taxa were grouped according to ordination to :lid interpretation. e

Discussion ton unpublished) but differ from other published lists of fish spccics represented in the River Sinnam- Jir\wile fish comtiiiiizit,~strirctirre ary. More than 60 tasa reportcd by Boujard 8i Ro- jas-Bcltan ( 19SX) for that river were not observed as The range of tasa captured during our investiga- juveniles in the tributaries we sampled. whereas 26 tions (Table 1) compare well with the juvenile tasa additional tasa were captured. Superimposed on caught in the Crique V6nus (Figure 1)by light traps potential erroneous identifications, which are com- (Ponton 1994)and other sampling methods (D. Pon- mon in south American freshwater fish studies (Va- 0.5 O Lf -0.5 cc

Nb Pf

Tg Ah

Ev Gc

Na Ea L

Figure 7. Habitat electivities and Fisher's exact test associations (:k = 0.01 < p 5 0.05; ** = 0.001 c p 5 0.01; *** = p i0.001) for fish taxa in tributaries and the Petit Saut reservoir of the River Sinnamary (see Table 1for codes). Each histogram represents the differencebetween the frequency of that taxon in the group of samples having that category of environmental variable and the frequency of that taxon in all , samples for fish species representative of upstream tributaries (a-b), downstream tributaries (c), and reservoir sites (d). See Table 4 for categories of each environmental variable. 05 0 Ee 5 -0 Hm

Sm Ri

ri SL Weitzman 1590)- two points may explain the Amongst the 14 most important estuarine tasa list- observed discrepancies. ed by Bou.iard (1953. Table ?, p. 737)-only individu- Firstly. the zone referred to by Boujard SL Rojas- als of the synbranchiforme S~W~~IIIZL~ZISniomzori!- Beltan (1988) as the upper (haut) Sinnamary is in rus were also observed as juveniles and adults in fact part of the current lower section of the reser- lower and upper reaches of the river (S. MGrigoux SL * voir. The accessibility of the upstream section ofthe D. Ponton unpublished). Thus, it seems unlikely Sinnamary was very difficult prior to the start of that juveniles of estuarine fish species use freshwa- dam construction, which opened up an area origi- ter tributaries and their associated areas as nurser- , nally accessible only by helicopter or boat + por- ies. tage. Secondly, most of the species listed by Boujard On the other hand. the taxa we observed com- SL Rojas-Beltan (19SS), and unreported in our Sam- pare well with those Tito de Morais et al. (1995) ples. are of estuarine origin. Boujard (19921 esti- found in the true upper reaches of the River Sin- matcd that about 60 fish species of the River Sin- namary. Taxa we did not collect correspond mainly namary can be found in the rivcr's estuary. to Serrasalmidao and large Siluriformes. whose 251

0.5 O Hq -0.5 CZ

CY AV

Me Af

Pm Ps

B2 Ho

MC Mo

Figure 7. Continued. 0.5 ____ . , . . -. _I . .I . .. O PC ~0.5 Bb

Ra RX

F~,~III.L.7. Continued. adults occupy the main channel of the river: juve- law during a large part of the 1993-1994 rainy sea- niles of these species are rarely or never found in son and this strong disturbance (sensu Resh et al. tributaries even during surveys covering extended 1988) induced important modifications to the juve- periods of time (D. Ponton unpublished). Thus, the nile communities found in the tributaries at the end nurseries sf the young of at least some large Guia- of the rainy season. In July 1993, downstream trib- nese fish species remain to be found and our study utaries were dominated by Characiformes and 80% provides a partial view only of the juveniles that in- of the most abundant taxa were Characidae. Dom- habit the Sinnamary River. inance of young Characiformes at the end of thc rainy season appears to be a recurrent phenomenon from year to year before dam closure. When re-ana- lysing data obtained at that period of the year in 1991 and 1991 in the Venus Creek with a similar To our knowledge, our work is the first to document sampling protocol (Tito de Morais unpublished da- thc cffects of a hydroelectric dam on juvenile íish ta). we found that Characiforms accounted for communities in the tropics. Studies aimed at docu- 79.3% of the young fishes in July 1991 and 83.3% at menting the effects of dams on fish in tropical areas the end of June 1991. Contrastingly, after the first * usually focus only on the adults of large species year of dam operation, the young Characifornis caught by gill nets (see Petr 1978 for a review and werc no more dominant at the downstream sam- MCrona et al. 1987 for an example in the Ámazon pling sites: their relative abundance decreased from basin). Similarly. studies on the effects to fish young 83.4% to 35.9%. More precisely. of the five most are restricted mainly to specific taxa that art‘ of abundant tarta caught in 1993. the four abundant commercial interest as larvae, such as largc niigra- Characiformes taxa were replaced by one Lebiasi- tory catfishes in the Ámazon (Barthem et al. 1991) nidae and three Cichlidae species and only the per- or young Tilopia in Africa (Dudley 1974). ciform Eleotridae Eleotris ioiil~lyopsisremained Downstream of the dam. water levels remained abundant. 253

The habitat requirements of the progeny of the maintained high numbers of juveniles in down- four characids found to be abundant in downstream stream sites after dam operation. reaches before dam completion (Moenkhausia col- The juvenile fish communities we observed in the lettii, Pristella maxillaris, Pseudopristella siinulata reservoir differ strongly from those of the river’s and Hemigrammus ocellifer) seem unlikely to ex- tributaries. The progeny of Characiformes repre- plain by themselves the decrease in abundances of sented only 17.1 and 56.7% of the individuals in the Characiformes. None of the variables for which de- body and head parts of the reservoir, respectively viations from expected frequencies were found (Figure 4). Mean values of density and richness (Figure 7a) are likely to have been modified by dam were also lower especially in the lower part of the operation. The drastic modifications in the richness reservoir where lentic characteristics prevailed (Ta- and relative composition of the juvenile communi- bles 2,3). These lentic environments, characterised ties must most likely result from the differential rate by high temperatures and low oxygen concentra-

Y of reproduction success and/or of survival during tions, appear to be especially detrimental to the ju- early life among species in relation to the low water veniles that prefer lotic conditions (Figure 7a-b). levels induced by dam operation. Miniature or The progeny of taxa more frequently observed in v small-sized Characiformes taxa found in the Sin- the reservoir’s littoral zone appear to be less selec- namary River are mostly phytophils (Ponton & Tito tive of local habitat characteristics (Figure 7c) con- de Morais 1996) and most of them reproduce during firming that eurytopic species cope with environ- the period of highest water levels (Breder & Rosen mental change better than do stenotopic species 1966, Munro 1990, D. Ponton unpublished). Low (see review in Poff & Ward 1990). For example, downstream flow during impoundment reduced the young stages of Krobia guianensis dominated large- intensity and duration of flooding events during ly the upper and lower reservoir samples but were most of the 1993-1994 rainy season. As a conse- also present in tributaries from other sections of the quence, adults of small-sized characids may have Sinnamary. The generalist tendencies of K. guia- encountered difficulties in accessing suitable nensis (Figure 7c) provides them a great ubiquity in spawning grounds andfor their young may have the River Sinnamary and may explain their wide been unable to benefit from food and shelter usu- distribution in the black water rivers of Surinam ally found in flooded areas. (Ouboter & Mol 1993). Despite alteration of flow regime, the progeny of some species remained very abundant in all the downstream sites. For example, the high reproduc- Perspectives tive success of Eleotris anzbliopsis (Figure 5) may be explained by the fact its youngs first occur at the end Impoundment imposed a ‘press disturbance’ (sensu of the dry season (D. Ponton unpublished observa- Yount & Niemi 1990, Niemi et al. 1990) on a large tions), thus this species had already reproduced be- section of the River Sinnamary. In the next years, fore dam closure. Moreover, the reproductive guild this disturbance is expected to drive the fish com- (sensu Balon 1981) to which the species belongs munity towards a new biologically accommodated r may also bring some advantages towards flow alter- state where tolerant species will dominate and sen- ations. Eleotris oxycephala, another species of this sitive species will be lacking. This succession will bottom dwelling genus, is a nest spawning speleo- take years, as many biological processes are in-

E. phi1 (Breder & Rosen 1966). E. amblyopsis maybe- volved. For example,it is likely that some species or have in a same manner and this reproductive habits taxa of the juvenile community observed in the lit- on the bottom of streams may confer a certain inde- toral zone of the reservoir have been favoured by pendence from flow variations. The cichlids Nan- abundant new food resources. Indeed, impound- nacara anomala, Crenicichla saxatilis and Krobia ment has induced the rapid death of flooded trees, guianensis are also nest spawners (Sterba 1963, and large amounts of leaves have accumulated. This Breder & Rosen 1966,Werner 1982),and all of them leaf litter is known to provide cover (Lowe-McCon- nel1 1964) and a source of food for fish (Henderson tions from year to year, a characteristic shared by 6: Walker 1990, Walker 1995). On the long term, this most of the rivers of French Guiana (Ph. Vauchel, value is expected to decrease in the future as the personal communication). This high natural envi- leaves will decompose and will not be renewed as ronmental variability may have given fish commu- long as riparian vegetation does not develop. Thus, nities inhabiting these rivers a great persistence, re- taxa for which early survival had been increased fol- sistance, and rate of recovery (e.g. Poff & Ward lowing impoundment may encounter harsher con- 1990). Future investigations will demonstrate ditions as the reservoir evolve (Balon 1974). whether Guianese fish communities show a greater The other main question remains whether and resiliency than those of more predictable systems. how the downstream fish community will recover from the perturbation induced by impoundment. Little published information is available on the ef- Acknowledgements fects of river damming on the early life of neotropical fishes, and no data exist on the resiliency of fish This investigation was funded by Electricit6 de communities in these geographical areas. The ratc France (Contract no. GP 7530) and ORSTOM. In- of recovery is expected to depend strongly on the stitut Français de Recherche Scientifique pour le duration of downstream flow modifications (e.g. Développement en Coopération. which also gra- Kingsolving & Bain 1993). A long period of flow al- ciously provided GHC with travel to and subsist- teration, i.e. a press disturbance (sensu Kingsolving ence in French Guiana. Watcr levels in the River 6: Bain 1993).will slow down recovery and even ini- Sinnamary and in the reservoir were provided re- pede it. As a consequence, the first step towards res- spectively by Ph. Vauchel, Hydrological Section. toring the capacity of downstream tributaries to ORSTOM Cayenne and Electricite de France, Cen- sustain a rich community of juveniles. and thus fish tre National d'Equipement Hydraulique. Amén- species. is to require dam operators to simulate nat- agement de Petit-Saut. We thank L. Lauzanne and ural discharge fluctuations as soon as impoundment J.C. Bron for their invaluable contributions to this will be completed. Water level oscillations in the study. N.Brehm, G. Elfort. A. Mallet, S. Papet and River Sinnamary will re-create the flooded areas J. Raffray helped at one moment or another in the where phytophil species can reproduce and other field. juveniles can feed and find shelter from predation. A large part of the fish community resiliency would also depend on whether short flow increases. References: cited such as those during the 1993-1994 rainy season (Figure 21, are sufficient to act as a refugium from Balon. E.K. lQ74. Fish production of a tropical ecosystem. pp. 14S-74X. 1~: the long lasting perturbation. Many South Amer- E.K. Balon 6r A.G. Coche (ed.)Lake Iíariha: A Man-Made Tropical Ecosystem in Central Africa. Manogra- ican fish species arc of small or very small size and phiac Riologicae 73, Dr W. Junk Puhlishers. The Hague. most miniature species present a life span of 1 year Balon, E.K. 1W. Additions and amcndmcnts to the classifica- or less (Weitzman 6i Vari 1988). Thus. poor repro- tion of reproductive styles in fishes. Env. Bid. Fish. h: .?77-38Q. ductive success of these species or ISW survival of Rartheni. K.B.. M.C.L. de Brito Rihein1 & M. Petrere Jr. lrJQl. their early life stages due to successive periods of Life strategies ol scime long-distancc migratory catfish in relation to hydroelectric dams in the Amazon hasin. Bid. Con- drastic flow regulations may impede their ability to wrv. 55: 339 345. persist downstream of the dam. On the othcr hand. ßayley. P.B. IQSS.Fxtcirs aflcctinggrnwth rates of young tropical small Characiformes and Cyprinodontiformes prc- flacrdplain lishes: seasonality and density-dependence. Fnv. Rial. Fish. 21: 127-142. sent traits such as early maturation. continuous re- B production and small hrnod size. which reflects rap- Bettoli. P.W. M.J. M:iccina. IQ% Sampling with toxicants. pp, 117: H.K. Murphy & D.W. Willis (ed.) Fisherics'li.cl1- id colonisation abilities (Winemiller 1989). Morc- niqucs. American Fisheries Uctheada. IW. Society, over, fish species of the River Sinnamary have ßoujard, T. Space-time organi~ationof rivcrinc fish ~0171- evolved in a system that possrsscs high flow varia- mimitics in Frcnch iiuiana. Env. Uiol. Fish. 34: 735-14b. 255

Boujard, T. & R. Rojas-Beltan. 1988. Zonation longitudinale du Lowe-McConnell, R.H. 1987. Ecological studies in tropical fish peuplement ichtyque du fleuve Sinnamary (Guyane Fran- communities. Cambridge Tropical Biology Series, Cambridge çaise). Rev. Hydrobiol. Trop. 21: 47-61. University Press, Cambridge. 382 pp. Boujard, T., M. Pascal & EJ. Meunier. 1990. Micro-répartition Mérona, B. de, J.L. de Carvalho & M.M. Bittencourt. 1987. Les spatio-temporelle du peuplement ichtyologique d’un haut effets immédiats de la fermeture du barrage de Tucurui (Bré- bassin fluvial de Guyane, L‘Arataye. Rev. Ecol. (Terre Vie) 45: sil) sur l’ichthyofauneen aval. Rev. Hydrobiol. Trop. 20: 73-84. 357-373. Munro, A.D. 1990. Tropical freshwater fish. pp. 145-239. Ziz: A.D. Breder, C.M. Jr. & D.E. Rosen. 1966. Modes of reproduction in Munro, A.P. Scott & T.J. Lam (ed.) Reproductive Seasonality fishes. Natural History Press, Garden City. 680 pp. in Teleosts: Environmental Influences, CRC Press, Boca Ra- Chessel, D., J.D. Lebreton & N. Yoccoz. 1987. Propriétés de l’a- ton. nalyse canonique des correspondences. Une utilisation en hy- Niemi, G.J., P. De Vore, N. Detenbeck, D. Taylor, A. Lima, J. drobiologie. Rev. Stat Appliqu. 35: 55-72. Pastor, J.D. Young & R.J. Naiman. 1990. Overview of case Copp, G.H. 1989. The habitat diversity and fish reproductive studies on recovery of aquatic systems from disturbance. Env. function of floodplain ecosystems. Env. Biol. Fish. 26: 1-26. Manag. 14: 571-587. Copp, G.H., G. Guti, B. Rovny & J. cerny. 1994. Hierarchical Ouboter, P.E. & J.H. Mol. 1993. The fish fauna of Suriname. pp. analysis of habitat use by O -I- juvenile fish in the Hungarian/ 133-154.62: P.E. Ouboter (ed.) The Freshwater Ecosystems of Slovak flood plain of the River Danube. Env. Biol. Fish. 40 Suriname, Kluwer Academic Publishers, Dordrecht.

~ 329-348. Petr. T. 1978. Tropical man-made lakes - their ecologicalimpacts. Dudley, R.G. 1974. Growth of Tilapia of the Kafue floodplain, Arch. Hydrobiol. 81: 368-385. Zambia: predicted effects of the Kafue Gorge dam. Trans. Poff, N.L. & J.V. Ward. 1990. Physical habitat template of lotic Amer. Fish. Soc. 2: 281-291. systems: recovery in the context of historical pattern of spatio- Foster, G.A. 1995. The randomization test applied to the means temporal heterogeneity. Env. Manag. 14629-645. of two independent samples. SAS Observations, Second Ponton, D. 1994. Sampling neotropical young and small fishes in Quarter 1995: 51-62. their microhabitats: an improvement of the quatrefoil light- Gauch, H.G., Jr. 1982. Multivariate analysis in community ecol- trap. Arch. Hydrobiol. 131: 495-502. ogy. Cambridge University Press, London. 298 pp. Ponton, D. & L. Tito de Morais. 1996. Recueil de données bibli- Géry, J. 1969. The freshwater fishes of South America. pp. 828- ographiques sur les stratégies de reproduction et les premiers 848. Zii: E.J. Fittkau, D. Illies, K. Klinge, G.H. Schwabe & H. stades de vie des poissons du fleuve Sinnamary (Guyane Fran-

Sioli (ed.) Biogeography and Ecology in South America, Dr. çaise). Rev. Hydrobiol. Trop. (in press). , W. Junk Publishers, The Hague. Resh,V.H.,A.V.Brown,A.P. Covich,M.E. Gurt2,H.W. Li, G.W. Géry, J. 1977. Characoids of the world. T.F.H. Publications, Nep- Minshall, S.R. Reice, A.L. Sheldon, J.B. Wallace & R.C. Wiss- tune City. 672 pp. mar. 1988. The role of disturbance in stream ecology. J.N. Am- Gilderhus, P.A. 1972. Exposure times necessary for Antimycin er. Benthol. Soc. 7: 433-455. and rotenone to eliminate certain freshwater fish. J. Fish.,Res. Schwassmann, H.O. 1992. Seasonality of reproduction in Ama- Board Can. 29: 199-202. zonian fishes. pp. 71-81. In: W.C. Hamlett (ed.) Reproductive Gilderhus, P.A. 1982. Effects of an aquatic plant and suspended Biology of South American Vertebrates, Springer Verlag, clay on the activity of fish toxicants. N. Amer. J. Fish. Manag. Berlin. 2: 301-306. Sterba, G. 1963. Freshwater fishes of the world. Viking Press, Good, P. 1993. Permutation tests. A practical guide to resampling New York. 878 pp. methods for testing hypotheses. Springer-Verlag, New York. Ter Braak, C.J.F. 1986. Canonical correspondence analysis: a 225 pp. new eigenvector technique for multivariate direct gradient Granville, J.J. de. 1979. Végétation. pp. 12-13. Zii: ORSTOM (ed.) analysis. Ecology 67: 1167-1179. Atlas des DOM-TOM, IV La Guyane, ORSTOM, Paris. Ter Braak, C.J.F. & P.F.M. Verdonschot. 1995. Canonical corre- Henderson, P.A. & I. Walker. 1990. Spatial organization and spondence analysis and related multivariate methods in aq- population density of the fish community of the litter banks uatic ecology. Aquat. Sci. 57: 255-289. 1 within a central Amazonian blackwater stream. J. Fish Biol. Thioulouse, J. 1989. Statistical analysis and graphical display of 37: 401-411. multivariate data on the Macintosh. Cabios 5: 287-292. ? IGngsolving, A.D. & M.B. Bain. 1993. Fish assemblage recovery Tito de Morais, L., M. Lointier & M. Hoff. 1995. Extent and role I along a riverine disturbance gradient. Ecol. Applic. 3: 531-544. for fish populations of riverine ecotones along the Sinnamary Lillefors, H.W. 1967. On the Kolmogorov-Smirnov test for nor- River (French Guiana). Hydrobiologia 303: 163-179. mality with mean andvariance unknown. J. Amer. Stat. Assoc. Val, A.L. & V.M.F. de Almeida-Val. 1995. Fishes of the Amazon 62 399-402. and their environment: physiological and biochemical aspects. Lowe-McConnell, R.H. 1964. The fishes of the Rupununi savan- Springer-Verlag, Berlin. 224 pp. na district of British Guiana, South America. Part 1. Ecolog- Vari, R.P. & S.H. Weitzman. 1990. A review of the phylogenetic ical groupings of fish species and effects of the seasonal cycle biogeography of the freshwater fishes of South America. pp. on the fish. J. Linn. Soc. Zool. 45: 103-144. 356

381-393. [ri: G. Peters 6( H.Hutterer (ed.) Vertehrates in thc Part 3: Crcriicichlrr sn\-ntilis camplcx. Trop. Fish. Habb. 31: 75- Tropics, Museum Alexander Kocnig, Bonn. 76. Walker. 1. 1?9S. Amazonian streams and small rivers. pp. 167- Westby. G.W.M. IHSS. The ecology. discharge diversity and pred- 193. Irz: J.G.Tundisi. C.E.M. Bicudo &T.M.Tundisi (ed.) Lim- atory hehaviour of gymnotiforine electric fish in the coastal nology in Brazil. Acad. Sci.. Rio de Janeiro. streams of French Guiana. Behav. Ecol. Sociobiol. 23: 341- Weitzman. S.H. Sr R.P. Vari. 1988. I'vliniaturization in South 354. American freshwater fishes: an overview and discussion. Proc. Winemiller. K.O. 1989. Patterns of variatiun in life history among Biol. Soc. Wash. 1111: 444465. South American fishes in seasonal environments. Oecologia 115-141. Welcamme. R.L. 1979. Fisheries ecology of floodplain rivers. s1: Longman. London. 317 pp. Yount, J.D. 6: G.I. Nicini. 199Cl. Recovery of lotic communities Welcomnie. K.L. 19S.5. Rivcr fisheries. FAO Tech. paper 263. and ecosystems from disturhance - il narrative review of case Rome. 303 pp. studies. Env. Manag. 14: 547-569. Werner, LI. 1983. A look at the pike cichlids. gcnus Crrrzicichlrr.