Estimating macrozoobenthos abundance and biomass in river rapids: Accuracy analysis via randomization

Abstract

The accuracy of biomonitoring in river ecosystems depends critically on sampling methods. This is particularly relevant at the mesohabitat level—discrete morphological units such as rapids—which are characteristic of the near-pristine rivers of Eastern Fennoscandia and exhibit high spatial heterogeneity. This study examined how the total number of samples affects the accuracy of regional macrozoobenthos abundance and biomass estimates in such habitats. Based on 535 samples from Eastern Fennoscandia, randomization techniques and regression analysis were employed to relate sample size to estimate accuracy. The required number of samples was determined for confidence levels of 70%, 90%, and 95% and confidence interval widths ranging from 20% to 60% of the mean. At a 70% confidence level, achieving a 60% confidence interval width required at least 16 samples for abundance and 35 for biomass; for a 40% width, 35 samples were needed for abundance and 79 for biomass. The percentile-based method underestimated required sample sizes compared to the more reliable Cox method. Regression models (R² > 0.99) enabled the determination of optimal sample sizes for any desired confidence level (70–95%) and precision (20–60%). Biomass estimation consistently required 2–3 times more samples than abundance to achieve equivalent precision. These findings offer practical sample size guidelines for heterogeneous lotic habitats, where researchers must balance statistical reliability against logistical constraints. These findings offer practical sample size guidelines for heterogeneous lotic habitats, where researchers must balance statistical reliability against logistical constraints.