Abstract

Fish assemblages of temperate lakes are structured primarily by an inter- action between piscivory and a small number of environmental variables, but tropical floodplain assemblages have often been viewed as unpredictably structured. We tested the predictability of fish assemblage structure in floodplain lakes of the Orinoco River, Ven- ezuela, in relation to 22 variables describing environmental variation at the microhabitat, habitat, and supra-lake levels. Fish species abundances were estimated through electro- fishing surveys of 20 lakes in three regions for the early and late dry seasons of each of two consecutive years. Canonical correspondence analyses indicated that assemblage struc- ture was predictably related to only four descriptors of lakes: transparency, conductance, depth, and area. Discriminant function analyses revealed that transparency (''clear'': Secchi transparency .20 cm; or ''turbid'': Secchi transparency #20 cm) was tightly associated with the numerical density of six major taxa (82% classification accuracy) and the numerical density of piscivorous species (89% accuracy). Depth and area probably derived their significance from causal relationships to transparency and availability of cover, whereas the influence of conductance arose incidentally through an association with biogeographical zonation. Mantel tests indicated that similarity in structure of assemblages was not strongly related to the distance between lakes. Transparency was a remarkably reliable predictor of species composition. Fish with sensory adaptations to low light were dominant in turbid lakes, whereas visually oriented fishes predominated in clear lakes; seasonal change in- volved decline in the proportion of visually oriented fishes concomitant with a decline in transparency. The effect of transparency on assemblage structure was probably mediated by the relationship of transparency to visibility of prey. The structuring of Orinoco fish assemblages by piscivory, under the influence of transparency as controlled by depth and area, contrasts with previous views emphasizing random assemblage variation in neotropical floodplain lakes and extends the applicability of a conceptual model originally developed for temperate lakes.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call