The indicator uses mean zooplankton size and total stock (MSTS) for evaluating whether good environmental status is achieved or not. The indicator uses the parameter mean zooplankter size (mean size) which is presented as a ratio between the total zooplankton abundance (TZA) and total biomass (TZB). This metrics is complemented with an absolute measure of total zooplankton stock, TZA or TZB, to provide MSTS. Thus, MSTS is a two-dimensional, or a multimetric, indicator representing a synthetic descriptor of zooplankton community structure.
Data period: The MSTS evaluations are currently restricted to the analysis of zooplankton communities observed during June-September. This seasonal time period was chosen because it is covered most extensively by the monitoring sampling programmes supplying the data; moreover, this is also the period of the highest plankton productivity as well as predation pressure on zooplankton (Johansson et al. 1993; Adrian et al. 1999). The structure of the marine food web is naturally variable; therefore, the indicator is designed to detect changes in the community structure that significantly deviate from the natural variability during the growth season.
Control charts: The time series of the MSTS components (mean size and total stock) for each zooplankton community are analyzed with cumulative sum (CuSum) control charts. The CuSum methods are designed to detect persistent small changes when the long-term mean changes in observed processes or periods. A control chart uses information about the natural variation of the process that is evaluated to examine if the process, i.e. the structure of the zooplankton community, is moving beyond the expected stochastic variability which is defined as desirable tolerance. If the process is in control, i.e. the zooplankton community structure is not affected by pressures, then subsequent observations are expected to lie within the tolerance boundaries. The hypothesis that the process is in control is rejected if the observations fall outside the desired tolerance boundaries. As a test statistic, control charts employ the controlling mean (μ) and specify control limits of n × standard deviations (σ) above and below the mean or the confidence intervals (CI). The upper and lower control limits are defined using a conservative approach of ±5σ for μ estimated for either RefConFish (reference conditions for fish) or RefConChl (reference conditions for chlorophyll a concentrations).
All datasets used for setting the thresholds values for evaluating status are >30 years of observations. The normality of each data series is first tested for normality (D'Agostino & Pearson omnibus normality test, Shapiro-Wilk and Kolmogorov-Smirnov normality tests). As both mean size and total zooplankton biomass often deviate significantly from the normal distribution, the values can be transformed using Box-Cox procedure and all calculations are then carried out on the transformed data. Once a controlling mean (μi) and standard deviation (σi) have been specified based on the chosen period used to determine the baseline against which status evaluation is made, indicator values (xi,t) within the time series are standardized to z-scores (zi,t) as:
The approach for setting the reference period used a window of the available data corresponding to the selected reference period, i.e. years representing sub-basin specific reference conditions for (i) food webs not measurably affected by eutrophication; these are based on environmental quality ratio (EQR) and historical data on chlorophyll a (HELCOM 2009) when defining RefConChl, and (ii) high feeding conditions for zooplanktivorous fish when defining RefConFish (Assessment protocol figure 1).
The μi and σi are defined based on the conditions during the reference period.
Assessment protocol figure 1. Examples for setting RefConChl and RefConFish using long-term variability in chlorophyll a expressed as ecological quality ratio (EQR) in the northern Baltic Proper (modified from HELCOM 2009) (left) and body condition index (Fulton's K) of sprat in the ICES subdivision 27 (right) used to identify time period (green area) when zooplankton community was sufficient to efficiently transfer primary production to secondary consumers.
To investigate trends in accumulated small changes for the zooplankton mean size and total stock over long time periods, the CuSum charts (Assessment protocol figure 2) are constructed by first determining a decision-interval CuSum (DI-CuSum) that is calculated by recursively accumulating negative deviations (one-sided lower CuSum) as:
with Si=0 = 0. The k value is the allowance value in the process, expressed in z units, reflecting natural variability of the mean shift one wishes to detect. Thus, deviations smaller than k are ignored in the recursions. The default choice of k = 0.5 is considered appropriate for detecting a 1-σ shift in the process mean (Lucas 1982).
Assessment protocol figure 2. CuSum analysis of mean size (A) and total zooplankton biomass, TZB (B) using data series for station B1 (Askö station, Western Gotland Basin). The data are normalized to z-scores (right Y axis, open symbols). The threshold values are shown as dashed blue lines (-5σ from the mean for the reference period; σ is standard deviation) and the reference period (years) is indicated as a black bar on the top. The lower CuSum (solid blue line) indicates accumulated changes in the mean size and TZB; the CuSum lines are crossing the respective good status threshold values in 1995 (mean size) and 1999 (TZB). According to this chart, from 1995 onwards, MSTS indicates food web structure being in not good status.
A strategy that was used for obtaining an overall status assessment when several datasets are available for an assessment unit is based on the integrated datasets. Since all zooplankton data are generated by national laboratories following HELCOM-Monitoring Manual guidelines and standardized gears and analysis methods, the data used for MSTS calculations are likely to be comparable. In order to arrive at a meaningful decision scheme, the main properties of the datasets should be considered. This includes issues such as length of the time series, their variability within defined reference periods, length of the time series overlapping with the reference periods, statistical properties of yearly mean values (i.e. number of samples contributing), quality control practices in the analyzing laboratories, etc. These issues were carefully considered and discussed before this two-stage assessment algorithm (first, comparing the datasets, and, second, generating integrated data for the assessment unit) was applied.
The indicator is evaluated using HELCOM assessment scale 2, which is consists of 17 Baltic Sea sub-basins. In the future it should be further discussed whether a higher spatial resolution (i.e. separating coastal and offshore areas) is needed.
The assessment units are defined in the HELCOM Monitoring and Assessment Strategy Annex 4.