Shrimp pond zooplankton dynamics and the efficiency of sampling effort

The abundance and species composition of zooplankton assemblages were examined in a commercial Penaeus monodon (Fabricius) aquaculture pond in subtropical Australia. Numbers and biomass of zooplankton in the surface tows (142 µm mesh) varied from 2080 L^?1 (2466 µg L^?1) to < 1 L^?1 (15 µg L^?1). Peak zooplankton abundance and biomass occurred at the beginning of the shrimp farm season. The small copepod Oithona australis Nishida was dominant during these peaks in abundance although other species were common, including small, low‐salinity copepods in the first year and larger marine copepods in the second year. Irrespective of taxa or size, zooplankton abundance declined rapidly after the stocking of P. monodon postlarvae, indicating high levels of non‐selective predation. For the remainder of the grow‐out season, mean zooplankton abundance was below 1 L^?1 in the first season and below 3 L^?1 in the second season. The factors that maintain the abundance of these zooplankton assemblages at these relatively low levels are poorly understood but may include a deterioration pond water quality. At these lower levels of abundance, the copepods Acartia pacifica Steuer and Acartia sinjiensis Mori were relatively common during the middle phase of the production season, with barnacle nauplii increasing in abundance during the later stages of the grow‐out season. Analysis of fixed effects showed that there were significant differences in total zooplankton abundance between farm production seasons and occasion within season, and time (day or night). Analysis of random effects demonstrated that the dominant source of variation, in total zooplankton abundance, was day‐to‐day changes within sampling occasions. The complexity of temporal and spatial patterns in the abundance, distribution and composition of zooplankton assemblages in shrimp ponds presents significant challenges in designing sampling programmes that accurately quantify temporal or spatial trends. Our results have shown that sampling for more than four consecutive days, at more than one site, is necessary to accurately assess such trends.