Minimizing human error in macroinvertebrate samples analyses for ensuring quality precision in freshwater monitoring...

Benthic macroinvertebrates are often used in ecological quality monitoring. However, due to the large number of samples and specimens, sample processing (sorting/identification) is a labor-intensive task that is susceptible to errors. These errors can consequently lead to biased assessment results. We conducted the first audit of the Greek National Water Monitoring program. Totally, 444 samples were sorted at the laboratory by primary sorters and macroinvertebrate identification was conducted mainly at family level by primary taxonomists, having different taxonomic expertise. The Percentage Sorting Efficiency (PSE), Percentage of Taxonomic Disagreement (PTD), and the Relative Percentage Difference (RPD) were calculated to determine differences between auditing stages. Control charts were used to determine the process changes of the personnel (sorting: PSE index and identification: PTD index) as a calibration check. Additionally, national ecological indices/metrics were calculated to identify how they are affected by errors. All samples except from one had PSE values higher than 90%. The most common overlooked families were Chironomidae, followed by Baetidae and Gammaridae due to their high abundances. Average values of the PTD index for the total number of samples was 5.75% and 1.86% in each phase, respectively. The PTD values decreased between the two phases due to the gained experience of primary taxonomists during the 1st phase. The average action control limit was 95% for the PSE values and 14% for the PTD values. Overall, our ecological quality results indicated that the sorting error was less important than the identification one as the latter may lead to different ecological quality classifications. Our results show that our auditing procedure is effective and increases the quality and accuracy of the sample analysis procedure. It also highlights that human error should not be neglected since it may affect the ecological quality results and especially the good/moderate boundary which leads to rehabilitation measures.