Top-down control of benthic communities by resident fishes (balitoroid loaches and Gobiidae) within pools in Hong Kong hillstreams

Abstract

Balitoroid loaches and stream-resident gobies are diverse and abundant in tropical Asian streams, yet the ecology of these benthonic fishes is little known. I undertook a mark-recapture study to determine the extent and pattern of fish movement among stream pools, to assess the likelihood that resident fishes had persistent impacts on benthic communities. It took place in a single Hong Kong hillstream during the non-breeding period (including the entire dry season), and involved three balitoroids and a goby. The mark-recapture study was coupled with pool-scale manipulative removals of macroconsumer guilds (herbivorous fishes, predatory fishes and shrimp), separately and together, in three hillstreams during the dry season in order to test for top-down effects on periphyton, benthic macroinvertebrates, and leaf-litter breakdown rates.

Tagging did not significantly affect the survival and growth of fishes, and recapture rates were high (58% overall), especially for the abundant Rhinogobius duospilus (Gobiidae: 70% of 412 tagged) and Pseudogastromyzon myersi (Balitoridae: 57% of 762 tagged). At the end of the study, 78% of recaptured gobies and 62% of recaptured P. myersi remained within their initial release pools, although few moved up to 46 m and (R. duospilus) and 101 m (P. myersi) distant, giving rise to strongly leptokurtic distribution of displacements. The sedentary behaviour of all study species was more marked than has been reported for most other small benthic fishes, suggesting they have persistent ecological effects within stream pools. Among fishes that did leave their home pool, most P. myersi moved upstream whereas gobies tended to move downstream. Large adult P. myersi were highly sedentary, and individuals that travelled further showed lower growth increments.

Standing stocks of herbivorous fishes (totally two species: 1.5-58.2 indiv m-2; 1.6-34.2 g wet weight m-2) and predatory fishes (totally four species: 7.3-11.1 indiv m-2; 24.1-38.0 g wet weight m-2) within pools were high, generally exceeding those reported from most other streams. Predatory palaemonid shrimp were also abundant (8.9-19.0 indiv m-2). Despite their high densities and generally sedentary behaviour, top-down control of benthic communities by macroconsumers in all three study streams was minor to non-existent. Removal of herbivorous fishes revealed that they had a limited influence on algal biomass. Removal of predatory fishes had no effects, presumably because the remaining shrimp depleted surplus prey made available by the absence of fishes. Even removal of the entire predator guild gave rise only to a small increase in insect scrapers (mainly baetid mayflies). The abundance of most macroinvertebrate taxa, overall macroinvertebrate assemblage composition, periphyton condition (including silt accrual), and leaf-litter breakdown were unaffected by removal of either or both herbivorous and predatory macroconsumer guilds. The inclusion of three different study sites and the pool-scale approach, while slightly compromising within-stream replication and hence statistical power, should have provided realistic results on the importance of top-down control of stream benthic communities by macroconsumers. Given the high standing stocks of fishes and shrimp, as well as their sedentary behaviour, the small effect sizes and very weak top-down control cannot be readily explained.

i