ECOSYSTEM ENGINEERING BY A DOMINANT DETRITIVORE IN A DIVERSE TROPICALSTREAM

Prochilodus mariae (Characiformes: Prochilodontidae) is a detritivorou s fish distributed throughout the Orinoco river basin of South America . Spectacular migrations of these fishes occur at the end of the rainy season into the Andean foothills. Prochilodus ingest large quantities of sediments and may thereby modify habitats in neotropical streams. The major objectives of this study were (1) to explore experimentally the importance of Prochilodus in structuring a tropical stream in the Venezuelan Andean piedmont, and (2) to determine whether there was suf ficient ecological redundancy in a diverse and abundant assemblage of epibenthic fishes to compensate for the removal of Prochilodus. Commun ity structure was compared among three experimental treatments: (1) Pr ochilodus exclusion, (2) Prochilodus enclosure, and (3) the natural fi sh assemblage. Selective exclusion of Prochilodus resulted in striking changes in community structure as measured by patterns of sediment ac crual and the composition of algal and invertebrate assemblages. Highl y significant increases in total dry mass and in ash-free dry mass of sediments accruing on stream-bottom substrates were observed almost im mediately following the exclusion of Prochilodus. Moreover the composi tion of algal and invertebrate assemblages was significantly modified by Prochilodus. Taxa such as diatoms were reduced in number when Proch ilodus was present; in contrast, Prochilodus appeared to facilitate ni trogen-fixing cyanobacteria. Total invertebrate densities were greates t in the Prochilodus removal treatment; however, a variety of response s to the experimental treatments was observed among different taxa ana lyzed individually, including density reductions, increases, and no me asurable effects. This study suggests that the detritivore Prochilodus is a functionally dominant species in Andean foothill streams via sed iment-processing activities. Moreover, it provides little evidence to support the notion that strongly interacting species are limited to si mple systems with few food web components.