Scales of spatiotemporal variability in macroinvertebrate abundance and diversity in monsoonal streams: detecting environmental change

1. We quantified spatial and temporal variability in benthic macroinvertebrate species richness, diversity and abundance in six unpolluted streams in monsoonal Hong Kong at different scales using a nested sampling design. The spatial scales were regions, stream sites and stream sections within sites; temporal scales were years (1997–99), seasons (dry versus wet seasons) and days within seasons. 2. Spatiotemporal variability in total abundance and species richness was greater during the wet season, especially at small scales, and tended to obscure site‐ and region‐scale differences, which were more conspicuous during the dry season. Total abundance and richness were greater in the dry season, reflecting the effects of spate‐induced disturbance during the wet season. Species diversity showed little variation at the seasonal scale, but variability at the site scale was apparent during both seasons. 3. Despite marked variations in monsoonal rainfall, inter‐year differences in macroinvertebrate richness and abundance at the site scale during the wet season were minor. Inter‐year differences were only evident during the dry season when streams were at base flow and biotic interactions may structure assemblages. 4. Small‐scale patchiness within riffles was the dominant spatial scale of variation in macroinvertebrate richness, total abundance and densities of common species, although site or region was important for some species. The proportion of total variance contributed by small‐scale spatial variability increased during the dry season, whereas temporal variability associated with days was greater during the wet season. 5. The observed patterns of spatiotemporal variation have implications for detection of environmental change or biomonitoring using macroinvertebrate indicators in streams in monsoonal regions. Sampling should be confined to the dry season or, in cases where more resources are available, make use of data from both dry and wet seasons. Sampling in more than one dry season is required to avoid the potentially confounding effects of inter‐year variation, although variability at that scale was relatively small.