Samples from stone surfaces were collected in pools within four unpolluted hillstreams (two shaded and two unshaded) in monsoonal Hong Kong (lat. 23°N) to elucidate the extent of spatial (within and among streams) and temporal (seasonal) variations in algal biomass and assemblage composition. Sampling continued for over 12 months, incorporating the dry season when streams were at baseflow, and the wet season when spates were frequent. We anticipated that algal biomass would be lower in shaded streams and during the wet season, with associated seasonal differences in assemblage composition or relative abundance of different growth forms (e.g. erect versus prostrate). Benthic chlorophyll a (a proxy for algal biomass) varied among streams from an annual mean of 11.0-22.3 mg m -2. Dry-season standing stocks were 18% higher than during the wet season when spate-induced disturbance reduced algal standing stocks. Algal biomass varied significantly at the stream scale, but not at the pool scale, and was lower in unshaded streams, where standing stocks may have been limited by high densities of algivorous balitorid loaches (mainly Pseudogastromyzon myersi). An overriding effect of grazers on algal biomass could also have reduced variations resulting from spateinduced disturbance. Significant differences in assemblage composition among streams, which were dominated by diatoms and cyanobacteria (totally 82 taxa) were not systematically related to shading conditions. Seasonal variations in algal assemblages were statistically significant but rather minor, and did not involve major shifts in composition or growth form caused by spate-induced disturbance. The abundance of filamentous cyanobacteria in all the streams may have been due to 'gardening' by balitorid loaches that removed erect or stalked diatoms and favoured cyanobacteria that persist through basal regeneration of filaments. This explanation requires validation through manipulative experiments.

1. Balitorid loaches are widespread and highly diverse in Asian streams, yet their life history and ecology have received little attention. We investigated seasonal (wet versus dry season) and spatial variation in populations of algivorous Pseudogastromyzon myersi in Hong Kong, and estimated the magnitude of secondary production by this fish in pools in four streams (two shaded and two unshaded) over a 15-month period. 2. Mean population densities of P. myersi ranged from 6.0 to 23.2 individuals m )2 , constituting more than half (and typically >70%) of benthic fishes censused. Abundance was c. 25% greater in the wet season, when recruitment occurred. Significant density differences among streams were not related to shading conditions and were evident despite small-scale variations in P. myersi abundance among pools. Mean biomass varied among streams from 0.85 to 3.87 g ash-free dry weight (AFDW) m )2 . Spatial and seasonal patterns in biomass and density were similar, apart from some minor disparities attributable to differences in mean body size among populations. 3. All four P. myersi populations bred once a year in June and July, and life spans varied from 24 to 26 months. Populations consisted of three cohorts immediately after recruitment but, for most of the study period, only two cohorts were evident. Cohort-specific growth rates did not differ significantly among streams but, in all streams, younger cohorts had higher cohort-specific growth rates. 4. Secondary production of P. myersi estimated by the size-frequency (SF) method was 2.7-11.5 g AFDW m )2 year )1 and almost twice that calculated by the increment-summation (IS) method (1.2-6.6 g AFDW m )2 year )1 ). Annual P ⁄ B ratios were 1.17-2.16 year )1 (IS) and 2.73-3.22 year )1 (SF). Highest production was recorded in an unshaded stream and the lowest in a shaded stream, but site rankings by production did not otherwise match shading conditions. Wet-season production was six times greater than dry-season production, and daily production fell to almost zero during January and February. Cool temperatures (<17°C) may have limited fish activity and influenced detectability during some dry-season censuses. Estimates of abundance and annual production by P. myersi are therefore conservative. 5. Comparisons with the literature indicate that the abundance and production of P. myersi in Hong Kong was high relative to other benthic fishes in tropical Asia, or their temperate counterparts in small streams. Manipulative experiments are needed to determine the

1. Manipulative experiments were carried out in four Hong Kong streams (two shaded, two unshaded) to investigate the impact of grazing by an algivorous fish, Pseudogastromyzon myersi, on benthic algal biomass and assemblage composition. Experiments were conducted and repeated during both the dry and wet seasons to determine whether spate-induced disturbance modified any grazing effect. Treatments comprised fish exclusion and inclusion via closed and open cages, with a no-cage treatment used as a control for the cage effect. Treatments were maintained for 4 weeks in each experimental run. 2. Grazing by P. myersi reduced benthic algal biomass and the organic matter content of periphyton in open cages and the no-cage treatment relative to closed cages. The similarity between open-cage and no-cage treatments was evidence that the overall difference among treatments was caused by limiting fish access to closed cages and not merely an artifact of caging. Grazing effects were broadly similar in all streams, but there was a significant statistical interaction between treatments and seasons. 3. Analysis of dry-season data matched the overall trend in inter-treatment differences, confirming the effects of grazing by P. myersi on algal biomass and periphyton organic matter. Significant differences in algal assemblage composition between closed-cage and no-cage treatments during the dry season reflected reductions in the abundance of erect, stalked diatoms (Gomphonema) and filamentous cyanobacteria (Homeothrix). Removal of these vulnerable overstorey algae by P. myersi resulted in greater abundance of understorey diatoms (Achnanthes and Cocconeis) in the no-cage treatment in all streams during the dry season. The composition of algal assemblages in open cages was intermediate between the other two treatments. 4. Although fish densities were greater in all streams during the wet season, spateinduced disturbance obscured grazing effects and there were no significant differences among treatments attributable to fish grazing. Seasonal variation in impacts of P. myersi grazing provides support for the harsh-benign hypothesis, and confirms that biotic factors are less important controls of stream algal biomass and assemblage structure during periods (i.e. the wet season in Hong Kong) when abiotic disturbances are frequent or intense.

The algivorous balitorid loach Pseudogastromyzon myersi is abundant in Hong Kong streams where it may exert top-down control on benthic algal assemblages as reported for grazing fish in temperate and neotropical streams. Dietary selectivity by P. myersi was investigated in two shaded and two unshaded streams during the wet and dry seasons, thereby allowing for variation in the potential bottom-up influences of light and flow on algae. Fish stomach contents were compared with the benthic algal assemblages to assess selectivity and to test whether algal growth form influenced susceptibility to grazing. Diatoms and filamentous cyanobacteria dominated the diets of the fish, regardless of stream or season; most of the remainder of the stomach contents comprised fine particulate organic matter (FPOM). Stalked diatoms (Gomphonema) and filamentous cyanobacteria (Homeothrix) were the most important dietary items, with the former selected during the dry season and the latter in the wet season. Adherent diatoms (Achnanthes) were underrepresented in the diet, and filamentous chlorophytes were rarely eaten. Seasonal changes in diet were minor. Interstream variations reflected differences in the proportions of Gomphonema, Homeothrix and FPOM ingested, and were unrelated to shading. Grazing by P. myersi may influence algal composition and productivity by removing loosely attached diatoms and facilitating growth of filamentous cyanobacteria.