Less is more – Basic quantitative indices for fish can be achieved with reduced gillnet sampling

Abstract Obtaining a representative sample is fundamental to the assessment of fish communities. In this study we used data from 99 different surveys of 29 artificial lakes in the Czech Republic to evaluate the precision of fish community indicator estimates based on reduced gillnet scenarios derived from triplet gillnets installed in each particular locality, depth strata and habitat. Forty-four indicators (based particularly on numbers and biomass per unit of effort – NPUE and BPUE, respectively) were compared between the full sampling design (three gillnets at every sampling location) and six reduced scenarios which were combinations removing one or two gillnets from the triplets. The whole community comparison found the accuracy (R2) between the full sampling design and reduced scenarios usually to be above 0.9. Strong accuracy was also found for species richness and mean fish weight. In individual species comparisons, accuracy above 0.75 was found for common species in a given habitat, whereas accuracy for less common species sometimes dropped below 0.5. Setting one gillnet in an appropriate depth strata and lake locality appears to be sufficient for unbiased overall NPUE, BPUE, and species composition estimates, and at the same time, reduces sampling effort and minimizes unwanted fish mortality. Capturing less common species in a given habitat is more or less accidental and additional sampling effort increased the probability of catching these species. On the other hand, methods targeting rare species of interest may be considered instead of increasing gillnet effort above the CEN standard.