Sensitivity of phytoplankton metrics to sample-size: A case study on a large transitional water dataset (WISER)

Phytoplankton is a key element for the assessment of ecological and environmental status of aquatic ecosystems, also in the frame of the MSFD. The methods used to analyse phytoplankton communities are mainly based on time consuming cell counts, dictating a trade-off between the number of cells counted and the number of phytoplankton samples in a monitoring plan, whenever the financial budget, the available personnel or the response time are limited. We investigated the optimal sample size effort, intended as number of phytoplankton cells enumerated per sampling station, of many commonly used metrics. To this aim, precision and uncertainty of the metrics as a function of the sample size have been addressed using a case study carried out in the Lesina lagoon (Apulia, Italy), where 12000 cells per sample have been enumerated within five sampling stations. Overall, we show that some of the commonly used indices/metrics for the description of phytoplankton communities are strongly dependent upon the sampling effort (as number of enumerated individuals per sample), while other metrics are relatively independent. Metrics based on the number of taxa only (e.g. Taxonomic richness), on the ratio between number of taxa and number of individuals (e.g. Margalef's diversity index) or on the species evenness (e.g. Pielou's evenness index) strongly depend on the sample size and their uncertainty (in term of sampling variance) is usually large. On the other hand, metrics of taxonomic diversity accounting for the proportional abundance of each taxa (e.g. Shannon-Wiener's diversity index), metrics accounting for the dominance of the most abundant taxa (e.g. Berger-Parker's dominance index) and metrics derived from the individual size distribution (e.g. Index of Size spectra Sensitivity, ISS Phyto) are able to achieve high precision and low uncertainty already at small sample sizes (virtually less than 200 enumerated cells) and thus they minimize the allocated effort vs. the gained information. Accordingly to our observations, we provide recommendations about the metric selection and the optimal effort per sample to be allocated in phytoplankton monitoring plans.