Uncertainty in phytoplankton-based lake ecological status classification: Implications of sampling frequency and metric simplification

Abstract

We examined the influence of sampling frequency and the elimination of selected metrics from the Phytoplankton Multimetric for Polish Lakes (PMPL index) on lake ecological status classification. The effect of sampling frequency was analysed by calculating the official PMPL index using data from four sampling periods (848 lake-years; 478 Polish lakes), and compared the Pmpl-based classification to that assessed from PMPL variants calculated using one, two and three sampling periods. The effect of metric elimination was analysed by comparing the official three-metric PMPL classification with two variants, a one-metric PMPL including only the concentration of chlorophyll-a and a two-metric variant including total phytoplankton biomass and summer cyanobacterial biomass. Correlation strength between the official PMPL and simplified variants weakened by reducing sampling frequency but varied by period combinations. Uncertainty of ecological status classification increased in both simplification types but it was greatest when using the one-metric (chlorophyll-a based) variant. Sampling reduction from four to three periods resulted in 0.5–6.5% of lakes being over- or under-classified (total lake-year misclassification: 3.7–6.7%). Highest uncertainty of classification to at least good status occurred when the spring sampling period was excluded (6.5%). The risk of misclassification increased to 4.2–14.9% when two sampling periods were used. When classification was based only on a one-period (late summer) PMPL, 12.1% of lake-years were misclassified to at least good and 9.3% to below-good ecological status. The risk of misclassification to at least good ecological status was 2% when the two-metric (biomass-based) PMPL index was used and 28.9% when the one-metric (chlorophyll-a based) was used. Lowest misclassification error (0.5%–5.2%) occurred when a three-period (spring-August-October), three-metric PMPL index was used, but as this 5.2% would result in substantial status under-classification, we recommend further research to possibly optimize a three-period PMPL variant to reduce the costs of lake ecological monitoring in Poland.