Combined data from 2000 to 2014 on the seasonal course of the spring bloom in Kiel Bay are shown in Figure 2. The biomass peak may occur in February or March, but the real peak is normally missed by low-frequency sampling. The resulting high variability of the data must be accounted for when calculating and interpreting the Dia/Dino index. In Figure 3, the seasonal maxima of biomasses of diatoms and dinoflagellates during each spring period are shown in one column, but they may occur on different dates. Biomass peak data below a distinct threshold may be due to inadequate sampling and have to be excluded from the analyses. Abnormal Dia/Dino indexes, identified mainly by comparison with the results of the above-described alternative Dia/Dino index, can be excluded most efficiently if a biomass threshold of 1000 μg/L for either diatoms or dinoflagellates was adopted (Figure 3). This threshold demarcated ~25% of the average peak biomass of diatoms plus dinoflagellates over the investigation period. It is valid for the Eastern Gotland Basin and Kiel Bay, but its validity has to be reassessed in other areas. Based on this threshold, the following years had to be excluded from the analysis: 1979, 1980, 1983, 1989, 1990, 1992, 2002, 2006, and 2014 for the Eastern Gotland Basin and 1989, 1992, 1996, 1997, 1998, and 2011 for Kiel Bay. These years are still provisionally shown in Figures 4A, 5A but marked with an “X” to demonstrate the effect of a threshold of 1000 μg/L. Outliers could be removed (Figure 5A) and sub-GES values caused by inadequate sampling are replaced by the more realistic data of the alternative Dia/Dino index for the years 1980, 1983, 1989, 1990, 2002, 2006 and 2014.

Figures 4A, 5A show examples of the Dia/Dino indexes calculated with the two different procedures: (1) based only on the springtime maxima of diatom and dinoflagellate biomass (wet weight), as shown in blue in the figures, and (2) based on the mean springtime values of diatom and dinoflagellate biomass (wet weight), as shown in red in the figures. Surprisingly, despite the different data basis, the results are similar and the correlation between the two methods is highly significant (Spearman rank correlation ρ = 0.99^**).

In principle, both the wet weight and the carbon content can be used to calculate the Dia/Dino index. A comparison of the two approaches for each sample for which data were available (n = 901) showed that the values of the Dia/Dino index based on wet weight were mostly higher than those based on carbon units (Figure 6).

Figure 3 shows that, in the Eastern Gotland Basin, diatom blooms occurred mainly in the 1980s and after the year 2000, whereas dinoflagellates dominated in the 1990s. In Kiel Bay, diatoms dominated the spring bloom, but dinoflagellates increased slightly over the course of the investigation.

The development of the Dia/Dino index over the investigation period is presented in Figures 4, 5. As mentioned in Section Basic Biomass Data, the Dia/Dino indexes of years marked with an “X” are not valid. The strong fluctuations may be smoothed by the use of 3-year moving averages (Figures 4B, 5B), in which each point takes the neighboring years into account and represents the mean of 3 years.

In the open Eastern Gotland Basin (Figure 4A), i.e., excluding the coastal areas, which are influenced by freshwater influxes, the ecosystem shift from the end of the 1980s is well-represented. It shows a decrease from 100% diatom dominance (in 1986) to an almost 0% diatom share (in 1991 and others) compared with dinoflagellates. The smoothed curves (Figure 4B) clearly reveal the low abundance of diatoms from 1990 to 2001, after which diatoms recovered albeit with some fluctuations within this trend.

In Kiel Bay, the Dia/Dino index was high, without any apparent trend (Figure 5A), except between 2012 and 2014, when a slight decrease occurred; this is best seen in the smoothed curve of Figure 5B.

Correlations between the Dia/Dino index and abiotic parameters were checked. Such analyses were already carried through by Wasmund et al. (2013) for the diatom decrease and a connection to the minimum winter temperature was detected. We found a highly significant negative Spearman rank correlation (ρ = −0.48^**) between minimum winter temperature and Dia/Dino index in the Eastern Gotland Basin, but not in Kiel Bay.

In Kiel Bay, the potential relationship to zooplankton could be tested because recent zooplankton (size: >100 μm) abundance data were available from three stations of Kiel Bay from 2013 to 2015 (n = 30). These data were compared with long-term phytoplankton biomass data from Kiel Bay for the years 2000–2014 (n = 214) together with data on dissolved inorganic nitrogen (DIN) and silicate. The monthly means of an extended spring period (January to May) are presented in Figure 2, which shows that phytoplankton blooms usually occurred from February to March and then collapsed due to nutrient (DIN) limitation at the end of March, when zooplankton abundance was still low and insufficient to control the diatom bloom.

The trends in the alternative Dia/Dino index, based on silicate consumption, are similar to those determined using the different biomass data (mean and maximum values; Figures 4, 5). The values of the alternative Dia/Dino index, shown in green in the figures, are, however, much higher than those of the standard Dia/Dino index. This is because the alternative Dia/Dino index involves the maximal possible diatom biomass that can grow on the basis of silicate. Accordingly, different GES thresholds have to be defined.

The definition of GES is an important precondition for the operationalization of an indicator. Frequently, the most pristine state is considered as defining a GES. Systematically acquired monitoring data have been available in the Baltic Sea only since the 1980s and only these could be used in this paper. However, whether the status of the 1980s is the GES is questionable, because “pristine” conditions in the sea may be those dating back at least to the early twentieth century. Fortunately, high-quality historical quantitative phytoplankton data from the beginning of the twentieth century are available from Kiel Bay and were compiled by Wasmund et al. (2008). Historical Dia/Dino indexes calculated from those data are shown in Table 1. Thus, between 1905 and1950, the historical Dia/Dino index was always >0.87, with a mean value of 0.94. If a deviation of 20% from the historical mean value is tolerated, then a reduction by 0.19 is allowed and the suggested GES threshold will be 0.75. This implies the continued clear dominance of diatoms over dinoflagellates and marks the mid-point between balanced diatom and dinoflagellate biomass (Dia/Dino index = 0.5) and total diatom dominance (Dia/Dino index = 1.0).

For the Eastern Gotland Basin, the reversal point between diatom and dinoflagellate dominance, which occurs at an index value of 0.5, is suggested, with a Dia/Dino index >0.5, i.e., diatom dominance, indicating GES. This is appropriate because the Dia/Dino index covers the whole range from 0 to 1. Based on our suggestions, these GES values have recently been approved by HELCOM (2016b).

As shown in Figures 4, 5, the values of the alternative Dia/Dino index are much higher than those of the standard Dia/Dino index. Consequently, different GES thresholds have to be defined. For this purpose, we compared the standard and the alternative Dia/Dino indexes for each year and tried a linear regression (Figure 7). It was applicable for Kiel Bay (r² = 0.593), but for the Eastern Gotland Basin, the best correlation was found with a polynomic equation (r² = 0.588). The correlations between the standard and alternative Dia/Dino indexes are highly significant (p < 0.001). Insertions of the standard GES values into the equations lead to GES values for the alternative Dia/Dino indexes of 0.84 and 0.94 for the Eastern Gotland Basin and Kiel Bay, respectively.

Assessments by the standard and alternative Dia/Dino indexes are in good agreement in most cases. Years with a poor GES according to the alternative Dia/Dino index in the Eastern Gotland Basin (1988, 1991, 1993–2001, 2008, 2012) were also identified as “poor” by the standard Dia/Dino index (Figure 4A). The standard Dia/Dino index indicated clearly a worse status than the alternative Dia/Dino index in some years (1984, 2004), probably due to inappropriate sampling, and should not be applied in these exceptional cases (see discussion in Section Restrictions due to Insufficient Biomass Data and Their Potential Rectification). In Kiel Bay, a GES was realized in all years after excluding those marked by “X” (Figure 5A).

The intensity of a bloom can be characterized by its maximum biomass and duration. These parameters are not accurately recorded by low-frequency routine sampling. However, for this indicator bloom intensity does not require precise quantification; rather, of greater relevance is whether the spring bloom is mainly formed by diatoms or dinoflagellates. For the Dia/Dino index, our results show that, for the specific areas investigated, the spring bloom is sufficiently represented if either diatoms or dinoflagellates reached a biomass threshold of 1000 μg/L. This value turned out to be the best fit to bring the standard Dia/Dino index and the alternative Dia/Dino index into best agreement (Figure 7). In other areas, this threshold may be modified depending on regional conditions.

If this biomass criterion is not met or if otherwise unusually low Dia/Dino indexes were obtained due to inadequate sampling, the alternative Dia/Dino index can be calculated and applied. An example of this latter case was provided by the year 2004 in the Eastern Gotland Basin. As seen in Figure 4A, this year's Dia/Dino index indicates sub-GES but it seems to have been an outlier. In fact, a diatom bloom developed during a large data gap in phytoplankton sampling between 23 March and 22 April 2004, as indicated by a drop in silicate consumption from 13.2 μM (20 March 2004) to 3.6 μM (3 May 2004). In this case, the alternative Dia/Dino index may be applied, which in 2004 yielded a value of 0.85 which still indicates GES.

If the true peak of the bloom is missed, the approaches based on both the mean and the maximum values are influenced. A calculation based on mean values also takes into account the duration of the bloom. Surprisingly, the two modes of calculation lead to fairly similar Dia/Dino index values.

Diatom blooms are short-lived and disappear quickly whereas dinoflagellate blooms typically last longer (Lips et al., 2014). Therefore, diatoms will be overrated if the calculation is based on maximum values, and dinoflagellates if the calculation is based on mean values. While the two modes of calculation seem to be more or less equivalent, if strict conditions are required then the mean values calculated over the 3-month spring season should be used.

Abundance and biovolume are determined in microscopic analyses. The conversion of biovolume into biomass (wet weight) is simple, as a constant factor for the density of the cytoplasm is assumed. However, the next step, the conversion from wet weight to carbon, is problematic because it depends on cell size (e.g., Menden-Deuer and Lessard, 2000). As some researchers choose to avoid this vague conversion, carbon biomass is often missing from their data. To utilize the most extensive dataset, we recommend working with the Dia/Dino index based on wet weight, which in most cases will yield higher values than those based on carbon units (Figure 6). This is because wet weight overrates diatoms, as their large vacuole contains only a small amount of carbon but is fully counted in the wet weight. However, especially in spring, when small diatoms dominate, the deviation is small.

The conditions in the different areas of the Baltic Sea are highly diverse, such that trends in the Dia/Dino indexes will differ depending on the sea area (Klais et al., 2011). It is therefore appropriate to adopt the HELCOM sea area structure (Figure 1) as the assessment units for the index and avoid further splitting into sub-areas to prevent reduction of the dataset for each unit.

The trends in the Dia/Dino index in the Eastern Gotland Basin impressively show the ecosystem shift that began at the end of the 1980s, as described by Wasmund et al. (1998) and Alheit et al. (2005). Möllmann et al. (2009) analyzed 52 biotic and abiotic variables in the central Baltic Sea and identified two stable states separated by a transition period lasting from 1988 to 1993. According to Klais et al. (2011), dinoflagellates dominate in the positive phase of the North Atlantic Oscillation (NAO). Our data confirm that the new state of the 1990s persisted for one decade and successively changed back to the earlier state.

In Kiel Bay, the Dia/Dino index was high over the whole period but decreased slightly during the last 3 years of the investigation (Figure 5B). Whether a trend will evolve with the continuation of the data series will be determined with the continued use of the Dia/Dino index. A similar ecosystem shift as in the Baltic Proper might also occur in the western Baltic Sea. In that case, the recent situation provides a baseline for identifying future changes in the ecosystem almost in real-time.

Identification of the reasons for the long-term changes in the Dia/Dino index may help to understand its ecological relevance. Wasmund et al. (2013) showed that diatom spring blooms were strong after cold winters in the Eastern Gotland Basin but not in Kiel Bay. Therefore, it was not surprising that we found a highly significant negative correlation between minimum winter temperature and Dia/Dino index for the Eastern Gotland Basin.

Wasmund et al. (2013) argued that the reduction in diatoms after mild winters cannot be caused by a direct physiological effect of a slight temperature change, but by temperature-sensitive hydrographical (bottom-up) or biological (top-down) impacts that respond strongly to moderate temperature changes. They discussed two hypotheses: (i) a lack of convective mixing after mild winters may inhibit diatom growth, and (ii) a higher overwintering zooplankton biomass could graze on the early diatom blooms.

We favor the latter, food-web hypothesis: Zooplankton growth is earlier and stronger after mild winters, when these organisms feed on the diatom bloom, whereas it occurs later and is unable to control the diatom bloom after cold winters (see references in Wasmund et al., 2013). In their study of sediments from the Bornholm Basin, Dutz et al. (2004) found high hatching rates of nauplii already in March and April, with strong increases in those rates as the incubation temperature increased. In mesocosm experiments, warming led to a reduced time-lag between the phytoplankton bloom and the microzooplankton biomass maximum (Horn et al., 2015). Therefore, early diatoms can be controlled by grazing, which implies lower and later algal peaks at higher temperatures (Gaedke et al., 2010).

If a diatom bloom fails, the persisting nutrients enhance dinoflagellate growth, resulting in a prolongation of the spring bloom, a delay and reduction in sedimentation, and the prolonged persistence of suspended material in the pelagial (Heiskanen, 1998; Tamelander and Heiskanen, 2004). This sequence of events will be manifested as a better temporal match between the spring bloom and developing copepods and as an intensified microbial food web, as more nutrients remain in the mixed surface layer after stratification of the water column (Heiskanen, 1998).

This mechanism does not occur in small and shallow Kiel Bay, where the spring bloom develops early, irrespective of the temperature, while zooplankton develops in April and is thus unable to control the diatom spring bloom because of a mismatch in their occurrences (Smetacek, 1985). This scenario was confirmed by results presented in Section Trends of the Dia/Dino Index in Selected Sea Areas (Figure 2).

Not only grazing by zooplankton but also infections by parasites may control the diatom blooms in the Baltic Proper. Diatoms are a preferred host for fungal parasites in spring (Gutiérrez et al., 2016). Peacock et al. (2014) found abundant infesting stages of parasites only at water temperatures >4°C. Therefore, diatoms are not only subject to zooplankton grazing but also to infections especially after mild winters. They survive better in colder waters. This observation provides strong support for our food-web hypothesis.

The typical diatom spring bloom grows rapidly in response to new nutrients, but when the supply of the limiting nutrient is exhausted, the diatoms starve and quickly sink without significant decomposition (von Bodungen, et al., 1981; Tallberg and Heiskanen, 1998). At this stage they provide food to the benthos, which leads to enhanced metabolic activities in the sediment (Graf et al., 1982) and to the growth of suspension-feeding bivalves (own data, not shown). The resulting stimulation of the natural purification processes of the Baltic Sea is an important ecosystem service provided by these organisms (Karlson et al., 2007).

The effect of a stronger sedimentation will depend on the region. Particles that sink to oxic bottom areas feed the benthos, whereas those that reach deep anoxic basins devoid of zoobenthos will be subject to anaerobic bacterial degradation processes, leading to a stronger oxygen deficit (Spilling and Lindström, 2008). The effects of oxygen deficit will be discussed in Section Indicator of Eutrophication.

Whether diatoms or dinoflagellates are the most desirable food depends on different factors. Representatives of both groups may be appropriate as food, but those that are toxic must obviously be avoided and those that are of reduced nutritional quality, mucus-producing, or covered by thick cell walls are undesirable (Mitra and Flynn, 2005).

Traditionally, diatoms were considered the most important primary food source in the marine food web, and dinoflagellates of only minor importance (Kleppel, 1993; Turner, 1997). For example, in the southern Benguela upwelling system, egg production by adult female copepods was higher in waters dominated by diatoms than by microflagellates (Walker and Peterson, 1991). Recently, low food quality of diatoms (Ask et al., 2006; Dutz et al., 2008; Koski et al., 2008) and the inhibitory effects of this group of organisms (Turner et al., 2001; Ianora et al., 2003; Barreiro et al., 2011) have also been described, leading Uusitalo et al. (2013) to conclude that diatoms can generally be considered as poor-quality food. According to Vehmaa et al. (2012), however, a food-quality approach based on class level is too simplistic; rather, phytoplankton taxa must be considered at the species level.

Within our study we could not resolve the question of food quality. Therefore, we regarded for the Dia/Dino index primarily the aspect of vertical mass transport and considered it as an indicator of the food pathway. A high Dia/Dino index indicates stronger sedimentation and therefore a better nutrition of the benthos, and a low Dia/Dino index a better nutrition of the pelagic components of the food web.

In case of doubt whether diatoms or dinoflagellates are more beneficial for the ecosystem, the most pristine state is considered as GES. We were able to use historical data from Kiel Bay to define the GES there (Table 1). The suggestions of GES thresholds were carefully checked by HELCOM expert teams and confirmed to be reasonable by an intensive study of historical data by Wasmund (2017).

Besides the food web hypothesis as a possible explanation for the temporal decrease in diatoms, their limitation by silicate might be contemplated based on the long-term decrease in silicate concentrations in the Baltic Sea over the last 40 or even 100 years (Papush and Danielsson, 2006; Conley et al., 2008). At present, this hypothesis can be rejected, as during a 30-year period there has been no trend in silicate concentrations in the Baltic Proper (Wasmund et al., 2013). The silicate concentrations after the winter fluctuated between 10 and 25 μmol/L and were high enough to support the formation of a diatom bloom.

Silicate is currently not of concern because nitrogen and phosphorus are the drivers of eutrophication. However, with ongoing input of nitrogen and phosphorus, the relative concentration of silicate will decrease such that silicate may become the limiting nutrient for diatom growth (Danielsson et al., 2008; Ferreira et al., 2011). Indeed, in the 1980s, silicate was nearly depleted by the spring blooms in the Baltic Proper (Wasmund et al., 2013). Therefore, the Dia/Dino index will decrease with eutrophication and a high Dia/Dino index will indicate a less eutrophied status.

Strong diatom blooms, i.e., high Dia/Dino indexes, may help mitigating eutrophication as they remove nutrients by intensive sedimentation and deposit them in the sediment. Moreover, mediated by higher sedimentation rates, water transparency increases which is beneficial for the phytobenthos in the shallow areas. However, high rates of sedimentation have negative effects on the oxygen regime and connected processes in the deep basins of the Baltic. Under reducing chemical conditions nutrients are released from the sediment and may enhance eutrophication (Conley et al., 2002; Karlson et al., 2007). Anoxic areas cover, however, only a small part of the total Baltic area recently (Feistel et al., 2016). Borders between oxic and anoxic regions are not visible in the sea's intermediate and upper layers because of intensive transport and mixing processes. Therefore, we do not differentiate between oxic and anoxic regions for assessments based on the Dia/Dino index, which examines only the upper water layer. We assume that the comparably large oxic areas have the greater impact.

A summary of the main effects of a high Dia/Dino index on an ecosystem is given in Table 2.

Limitations caused by undersampling are general phenomena connected with field data. Inadequate data caused by missing the diatom blooms may be excluded by a biomass threshold that has to be reached; they may be replaced by silicate consumption data which allow the calculation of an alternative Dia/Dino index. These limitations and potential rectification are extensively discussed above.

The Dia/Dino index is not applicable in areas where diatoms or dinoflagellates do not occur for natural reasons. Diatoms can generally grow throughout the Baltic Sea, but dinoflagellates are restricted to marine and brackish waters. Therefore, the Dia/Dino index is not applicable in freshwater-influenced areas such as lagoons (Olenina, 1998; Gasiūnaitė et al., 2005), large river plumes (Wasmund et al., 1999), and parts of the Gulf of Bothnia (Klais et al., 2011).

The validity of the Dia/Dino index may be reduced if other phytoplankton groups gain dominance in spring. In the Baltic Proper, the mixotrophic ciliate Mesodinium rubrum may become the dominant species in spring (Wasmund and Siegel, 2008; van Beusekom et al., 2009), and in Kiel Bay dictyochophyceae may compete with dinoflagellates. To keep the indicator simple and common for the whole Baltic area, other, sporadically dominating algal groups have been excluded from the assessment protocol.

This indicator is robust, as it is based on only two groups that can be identified easily, even without extensive taxonomic expertise, on the basis of approved routine methods. It reflects a wide range of naturally occurring changes, by covering the whole theoretically possible range in values from 0 to 1. The calculation is simple and traceable, and the necessary data can be acquired from existing monitoring programs. Phytoplankton data have been collected for many years in the COMBINE monitoring program of HELCOM and diverse national monitoring programs and are stored in accessible data banks, e.g., the COMBINE database hosted at the International Council for the Exploration of the Sea (ICES: http://dome.ices.dk/views/Phytoplankton.aspx).

As anthropogenic activities have only an indirect effect on the ratio of diatoms and dinoflagellates, the Dia/Dino index cannot be readily manipulated, except for cases of severe eutrophication that may lead to silicate limitation. It serves primarily as a descriptive trend indicator.

Parts of the work (Kiel Bay data processing) were parts of a research project supported by Bundesamt für Naturschutz (BfN), with funds from the German Bundesministerium für Umwelt, Naturschutz, Bau, und Reaktorsicherheit. The evaluation of data from the southern part of the Eastern Gotland was supported by statutory activity of the National Marine Fisheries Research Institutes (NMFRI) and by Polish National Environmental Monitoring (PMŚ). The funding agencies have never exert influence on our work and the conclusions. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.