Seasonal succession of functional traits in phytoplankton communities and their interaction with trophic state

Understanding and explaining the structure of communities in response to environmental gradients is a central goal in ecology. Trait‐based approaches are promising but yet rarely applied to understand community dynamics in response to changing environmental conditions. Here, we investigate seasonal succession patterns of functional traits in phytoplankton communities and how nutrient reductions (oligotrophication) alter these patterns. We used phytoplankton data from 40 years of observation from the Rappbode Reservoir (Germany), which underwent a strong shift in trophic conditions, and translated taxonomic composition into functional traits by assigning trait values compiled from the literature. All studied traits (morphological, behavioural and physiological traits) responded to changing environmental conditions and showed consistent, reoccurring seasonal developments. The seasonal succession of phytoplankton communities was shaped by a trade‐off between small‐celled, fast‐growing species that are able to rapidly incorporate existing resources (r‐strategists) and large‐celled species with more complex and efficient mechanisms to exploit scarce mineral nutrients or acquire previously unexploited nutrient pools (k‐strategists). In summer, when nutrients were scarce, the k‐strategy was prevailing (important traits: phosphate affinity, nitrogen fixation, motility and mixotrophy). During the rest of the year, nutrients and turbulence were high and r‐strategists dominated (important traits: maximum growth rate and light affinity). A comparison between eutrophic and oligotrophic years revealed that the main features of functional trait succession were largely preserved, but intra‐annual fluctuations from spring to summer were stronger during eutrophic years. Nutrient reductions mainly affected functional traits and biomass in spring, while in summer the functional community composition changed little. Synthesis. This study provides for the first time a quantitatively supported functional template for trait‐based succession patterns in lakes under different nutrient conditions. By translating taxonomic composition into trait information, we demonstrate that the quantification of functional characteristics enables ecological interpretation of observed community dynamics and provides not only a testable template but also a powerful tool towards a more mechanistic understanding. The quantification of functional traits further improves the predictability of community shifts in response to changing environmental conditions and thus opens new perspectives for predictive limnology using lake ecosystem models.


| INTRODUC TI ON
Understanding and explaining the structure and dynamics of biotic communities in response to environmental gradients is a central goal in ecology. As planktonic organisms in aquatic systems have short generation times (Collins, Rost, & Rynearson, 2014), are very dynamic and are highly influenced by abiotic factors as well as biotic interactions, they are well suited to study the reaction of communities to environmental changes. In temperate lake ecosystems, seasonal changes in environmental factors such as temperature, light intensity, nutrient concentration or grazers induce shifts in phytoplankton abundance and species composition (Bergquist, Carpenter, & Latino, 1985;Stomp et al., 2007;Tilman, Kilham, & Kilham, 1982;Vrede, Vrede, Isaksson, & Karlsson, 1999), referred to as seasonal succession. Seasonality is the presence of regular and periodic changes in a variable that recur on an annual time-scale. Explaining and predicting these distinct, reoccurring seasonal patterns has long been in the focus of freshwater ecologists (Margalef, 1978;Reynolds, 1984a;Sommer, Gliwicz, Lampert, & Duncan, 1986). Early theoretical models describe phytoplankton succession mainly as a consequence of turbulence and nutrient availability (Margalef, 1978;Reynolds, 1988). They predict the occurrence of r-strategists, which are characterized by small cell sizes and high maximum growth rates, under high nutrient and high turbulence conditions, as they prevail during spring. In summer, when nutrient availability and turbulence are low, k-strategists with larger cells, slow growth, but high nutrient affinities and diverse strategies for nutrient acquisition (e.g. mixotrophy, nitrogen fixation) are expected to dominate (Margalef, 1978;Reynolds, 1988). The most popular and widely cited conceptual model about plankton succession is the verbally formulated plankton ecology group (PEG) model, which provides a standard template to describe dynamics of total biomass and composition of plankton communities in response to specific driving environmental factors in the temperate zone (Sommer et al., 1986(Sommer et al., , 2012. For example, the PEG model predicts a shift from small, edible algae in spring towards larger, inedible algae in summer as a response to increased grazing pressure from zooplankton. Besides changes along the seasonal development, the species composition of phytoplankton communities has also been shown to vary along nutrient gradients, e.g. during oligotrophication (Anneville et al., 2002;Gaedke, 1998;Jeppesen et al., 2005). Interestingly, studies about oligotrophication focus mostly on inter-annual changes, while intra-annual changes in succession patterns with trophic status have rarely been addressed.
Community dynamics of phytoplankton along seasonal or along nutrient gradients are traditionally described taxonomically. As the basal level in taxonomy, species can be conceptualized by a characteristic information about morphological and physiological features, however, predictions at species level are notoriously difficult or maybe even impossible to make (Reynolds, 2000). Therefore, higher taxonomic units (e.g. diatoms, cyanobacteria) are widely used to evaluate phytoplankton distributions (Wetzel, 2001). However, phylogenetic classifications of organisms have the disadvantage that their ecological functions are heterogeneous within these higher taxonomic units and hence often do not reflect their ecological niche. For instance, species from the same taxonomic group might show very different ecological adaptations, while species from different taxonomic groups can share similar ecological strategies (e.g. mixotrophy or the ability to form colonies; Salmaso, Naselli-Flores, & Padisák, 2015).
Trait-based approaches are a promising tool to overcome these drawbacks and to better reflect the ecological properties of (and diversity within) a community. While much work has been done on classifying species into functional groups (e.g. Kruk et al., 2017;Kruk, Mazzeo, Lacerot, & Reynolds, 2002;Padisák, Crossetti, & Naselli-Flores, 2009;Reynolds, 1980;Reynolds, 1984a;Reynolds, Huszar, Kruk, Naselli-Flores, & Melo, 2002;Salmaso et al., 2015), the study of individual functional trait dynamics in natural communities and their links to abiotic drivers as well as to fitness and survival (e.g. maximum growth rate or phosphate affinity) is still in its early stages in aquatic ecology (Litchman, Edwards, Klausmeier, & Thomas, 2012;Litchman & Klausmeier, 2008;Litchman, Klausmeier, Schofield, & Falkowski, 2007;Weithoff, 2003). Functional traits can provide a mechanistic foundation for understanding and predicting community structure and dynamics across environmental gradients (Edwards, Litchman, & Klausmeier, 2013b;Thomas, Kremer, Klausmeier, & Litchman, 2012) and bridge from the level of organisms to that of ecosystems (Falkowski, Barber, & Smetacek, 1998;Litchman et al., 2015). However, studies about the seasonal demonstrate that the quantification of functional characteristics enables ecological interpretation of observed community dynamics and provides not only a testable template but also a powerful tool towards a more mechanistic understanding. The quantification of functional traits further improves the predictability of community shifts in response to changing environmental conditions and thus opens new perspectives for predictive limnology using lake ecosystem models.

K E Y W O R D S
freshwater ecology, functional groups, oligotrophication, plankton ecology group model, Rappbode Reservoir, seasonal dynamics, trait-based approaches dynamics of phytoplankton traits are rare, especially for physiological traits requiring detailed laboratory measurements. We are only aware of Klausmeier (2013a) andEdwards (2016), who studied the seasonality of maximum growth rate, light and nutrient utilization traits in a marine ecosystem.
To the best of our knowledge there are no studies investigating the seasonal dynamics of eco-physiological traits (i.e. derived from quantitative laboratory measurements, for simple binary traits refer to Weithoff, Rocha, & Gaedke, 2015) in a freshwater habitat. Our study aims at closing this knowledge gap and investigates to which extent eco-physiological traits conceptualize functional changes in phytoplankton communities along inter-and intra-annual environmental gradients in lakes. Additionally, we analyse how the impact of nutrient reductions alters the seasonal patterns of these functional traits. We take advantage of a 50-year-long, seasonally resolved dataset from the German Rappbode Reservoir, which underwent a strong and abrupt shift in trophic conditions in the nineties (Wentzky, Tittel, Jäger, & Rinke, 2018). This allows us to analyse functional trait succession under nutrient-enriched and nutrient-deficient conditions (average TP concentrations: 0.13 and 0.02 mg/L, respectively), without the confounding effects of geographical location and lake morphometry that are problematic when making cross-system comparisons (e.g. Edwards et al., 2013b). In contrast to previous studies (Edwards et al., 2013a(Edwards et al., , 2013bKlais et al., 2017;Kruk, Martínez, Nogueira, Alonso, & Calliari, 2015;Weithoff & Gaedke, 2016), we describe phytoplankton communities by a variety of relevant traits from independent categories, including morphological, behavioural and physiological traits (cell size, silica use, mixotrophy, motility, nitrogen fixation, buoyancy, ability to form chains and colonies, edibility for Daphnia, maximum growth rate, phosphate affinity and light affinity). With our trait-based approach we intend to achieve an understanding of the composition and dynamics of freshwater phytoplankton communities in response to seasonal and long-term environmental changes. Moreover, our goal is to generalize the existing patterns in order to provide a functional template for traitbased succession patterns in temperate lake ecosystems, which is quantitative and therefore largely extends the verbally formulated PEG model. Such a trait-based, quantitative approach will push forward research about seasonal phytoplankton developments, since it allows for a predictive community ecology that can be statistically tested and is capable of making comparisons across different environments.

| Study site and sampling
The Rappbode Reservoir is Germany's largest drinking water reservoir and is located in Harz Mountains, a mid-mountain reach in central-northern Germany. Three pre-dams discharge their water into the Rappbode Reservoir. The inflow volume is 120 × 10 6 m 3 and the residence time 344 days. It has an elongated shape with a length of 8 km, a maximum surface area of 3.95 km 2 , a mean depth of 28.6 m and a maximum depth of 86 m. The Rappbode Reservoir is a monoto dimictic water body, which underwent a re-oligotrophication process around 1990. Within a very short time period of 2-3 years, total phosphorus concentrations in the epilimnion declined from approximately 0.12 to 0.02 mg/L (Wentzky et al., 2018). The day when stratification set on decreased over the years and the stratification duration increased (for details on calculation of stratification see Wentzky et al., 2018): From 1980to 1990, the stratification period started on average at day 130 (±8 days), while the mean stratification onset was already at day 114 (±9 days) between 1996 and 2016 (oligotrophic period). The stratification offset increased from on average day 322 (±9 days) during the eutrophic period to day 336 (±20 days) during the oligotrophic period. As a result of earlier stratification onset and later stratification offset the mean stratification duration increased from 192 days (±10 days, eutrophic period) to 223 days (±25 days, oligotrophic period). For more details about the Rappbode system, we refer to Rinke et al. (2013), Friese et al. (2014) and Wentzky, Frassl, Rinke, and Boehrer (2019). onwards. More details about sampling methods and sample analysis are given in Wentzky et al. (2018). For further analysis, we calculated depth-weighted average values from the data collected at 0, 5 and 10 m depth in order to make them comparable with the mixed water samples collected at 0-10 m depth and both datasets were merged. These measurements cover most of the epilimnetic layer. In this study, we used data for phytoplankton community composition, soluble reactive phosphorus (SRP), water temperature, nitrate (NO 3 ), silica (Si), oxygen, pH and secchi depth (for details on measurement methods, see Wentzky et al., 2018).

| Trait selection
For this study we selected functional traits that are considered crucial for survival, growth or reproduction of phytoplankton (see  Thomas, and Yoshiyama (2010) and Klais et al. (2017). to be able to assign trait values to every member of the community we took advantage of a method developed by Bruggeman, Heringa, andBrandt (2009) andBruggeman (2011)

| Phytoplankton community data
To compare the seasonal development between nutrient-rich and nutrient-poor years, the dataset was split into two periods of equal length based on TP concentrations: The eutrophic period covered the nutrient-rich years between 1970 and 1990. During the eutrophic period the annual mean TP concentration was on average TA B L E 1 Overview about phytoplankton functional traits used in this study, including their trait type, range and categories, definition and ecological function. Trait type and ecological function are assigned according to Litchman and Klausmeier (2008)  Reservoir are presented in Table 1 in Supporting Information S2.

| Ecological trait space of the phytoplankton community
After assigning trait values to each species, we transformed this trait matrix of species into a distance matrix using principal com- Of the 11 PC axes 4 were non-random. For easier visualization of the results only the first two axes were used. However, the trait scores for the first four axes are presented in Table 3 in Supporting Information S2.
To address co-variation between traits, a correlation matrix was calculated, using the Pearson correlation coefficient. The correlation matrix was visualized using the corrplot function from the r package corrplot (see Figure 1b).

| Seasonal development of environmental parameters, phytoplankton biomass and traits
The taxonomic composition of each sample in our dataset was trans-

| Synthesis of seasonal differences in trait composition
As a graphical method to synthesize the information obtained from the individual traits and to evaluate the importance of selected traits for the eutrophic and oligotrophic period, radar charts were created for each season, using the 'radarchart' function from the r-package fmsb (Nakazawa & Nakazawa, 2019

| Ecological trait space spanned by the species
Separating the phytoplankton species according to their functional traits in a PCA (Figure 1a)  nitrogen fixation were closely related because both only occurred in cyanobacteria. The silica use trait was located opposite of the traits nitrogen fixation and buoyancy, indicating a good separation between diatoms (mostly in the upper half of Figure 1a) and cyanobacteria (lower half of Figure 1a). Larger cell size was associated with motile and mixotrophic species. In contrast, species with smaller cell size occurred together with higher maximum growth rate, edibility for Daphnia, light affinity and chain-and colony-forming ability.
The traits mixotrophy and motility were ordinated in far distance to high maximum growth rate indicating a trade-off between mixotrophy and fast growth, or in other words, characterize mixotrophs as

| Seasonal development of environmental variables and phytoplankton biomass
The phytoplankton biomass and environmental parameters, including water temperature, soluble reactive phosphorus, nitrate, silica, oxygen, pH and secchi depth showed clear seasonal patterns ( Figure 2; Figure 1 in Supporting Information S2) and seasonality explained between 3.6% (Secchi depth) and 94.6% (water temperature) of variability in the data ( Table 2 in Supporting Information S2). As indicated by the GAMs, the seasonal development of biomass during the eutrophic period differed substantially from that of the oligotrophic period (Figure 2a). While eutrophic years showed a clear biomass maximum during spring between days 100 and 150, followed by a biomass minimum, representing the clearwater phase, seasonal fluctuations were less pronounced during oligotrophic years and biomass was more equally distributed over the season. Water temperature was very well explained by seasonality (>90% explained deviance, Table 2

| Seasonal development of phytoplankton functional traits
Many individual functional traits exhibited a recurrent seasonal pattern during the eutrophic as well as during the oligotrophic period, depicted by the GAMs of the annual time series (Figure 3; Figure 2 in Supporting Information S2). The variation in trait data explained by seasonality, varied between 10.1% and 63.4% (Table 2 in Supporting Information S2). Similar to the seasonal variations in abiotic variables, for most traits a more pronounced seasonality was found during eutrophic years. This was indicated by the larger differences in trait composition between spring and summer, shown by the radar plots (Figure 4), as well as by the higher explanatory power of the GAMs during nutrient-rich compared with nutrient-poor years ( Table 2 in (Figure 3f,g,k). Especially the increase in mixotrophy with oligotrophication was very prominent, which have gone up from less than 5% throughout the year in eutrophic years to almost 25% in nutrient-poor summers. In summary, the calculation of community-averaged traits (Figures 3 and 4) allowed for a quantitative assessment of changes in functional characteristics of the plankton community over seasonal and nutrient gradients.

| D ISCUSS I ON
The trait space spanned by the phytoplankton species (Figure 1a) showed that phosphate affinity, mixotrophy and motility increased with increasing cell size, while maximum growth rate and light affinity decreased (Banse, 1976;Finkel, 2001;Tang, 1995). This basically indicates a trade-off between r-strategists (small cell size, high maximum growth rate and light affinity, low efficiency of resource use) and larger celled k-strategists with slower growth rates, but more complex mechanisms for survival (high mixotrophy, motility and N-fixation) and high efficiency to use mineral nutrients (high P affinity; Grover, 1991;Huisman & Weissing, 1995;Leibold, 1997;Litchman & Klausmeier, 2001;Sommer, 1986b). These trade-offs among functional traits drive species replacements along environmental gradients and are therefore the basis for the seasonal succession patterns observed in Rappbode Reservoir.

| Functional traits quantitatively show a change from r-to k-strategists from spring to summer
The development of phytoplankton traits showed distinct reoccurring patterns over the season, which are conceptualized in Figure 5. These successional trait patterns were largely retained with trophic status, which is considerable given the large differences in nutrient concentrations between the two trophic periods (average TP concentrations: 0.13 mg/L for eutrophic and 0.02 mg/L for oligotrophic years). All traits, except the edibility for Daphnia trait (which is discussed separately below), clearly mirrored the environmental pressures over the year, e.g. high P affinity during P limitation in summer and high light affinity during light limitation in spring. Major differences in functional trait composition exist between the summer period, when the reservoir was strongly stratified and times when a large mixing layer was present. In spring, when turbulence and nutrient input was high, species with small cell sizes and high growth rates (r-strategists) dominated (Gaedke, 1992;Reynolds, 1984b;Sommer et al., 1986). Silica users were also most abundant under wellmixed conditions such as in spring. This was probably because silica users have high sedimentation velocities due to their siliceous cell wall and were therefore favoured by turbulence preventing them from sinking out of the photic zone (Sommer, 1984;Trimbee & Harris, 1984). The mixing of the water column and the poor light conditions in spring gave a competitive advantage to species with high light affinities (Edwards et al., 2013a;Yoshiyama, Mellard, Litchman, & Klausmeier, 2009), i.e. the ability for more efficient utilization of low light, since they are better adapted to fluctuating light conditions. Phosphate affinity and alternative strategies for mineral nutrient acquisition, such as the traits nitrogen fixation and mixotrophy were less relevant in spring, since nutrient availability was high. Also the proportion of motile and buoyant species was lower in spring since cells were moved upwards towards the light by turbulence and hence investing in motility was not necessary (Jäger, Diehl, & Schmidt, 2008;Visser, Massaut, Huisman, & Mur, 1996).

F I G U R E 5
Seasonal patterns of phytoplankton biomass and the importance of different phytoplankton traits during eutrophic (left) and oligotrophic (right) years. The thickness of the horizontal bars indicates the seasonal change in relative importance of the phytoplankton traits cell size, maximum growth rate, light affinity, silica use, phosphate affinity, nitrogen fixation, motility and mixotrophy In contrast, turbulence and nutrients were low in summer and light penetrated deeper into the water column. In response to the changed environmental conditions phytoplankton developed different functional strategies to survive. In agreement with predictions from ecological theory (Litchman & Klausmeier, 2001;Margalef, 1978;Wirtz & Eckhardt, 1996), the summer community shifted towards slower growing species with larger cell sizes and higher tolerances towards periods of nutrient stress (k-strategists). The nutrient limitation in summer provided opportunities for phosphate affine phytoplankton and the development of more complex nutrient acquisition strategies such as mixotrophy and nitrogen fixation.
Organisms also invested in motility, which was either realized by the possession of flagella or by the regulation of buoyancy to overcome sedimentation losses and nutrient deficiency by migrating to deeper waters, which are important stressors during stratification in summer. This agrees with experiments, which observed a replacement of sinking taxa with buoyant and flagellated taxa with decreasing mixing depth (Jäger et al., 2008;Reynolds, Wiseman, Godfrey, & Butterwick, 1983).
In summary, our results quantitatively show a shift from r-strategists (small cell size, high maximum growth rate and low efficiency of nutrient use) in spring to k-strategists (large cell size, slow growth rate and complex mechanisms of resource acquisition) in summer, which is in line with verbal descriptions of the typical successional sequence observed in temperate lakes (Margalef, 1978;Reynolds, 1984a;Sommer et al., 1986). The major advancement of our analysis is to put these findings into a quantitative framework using functional traits. This allows not only to provide a quantitatively characterized functional template for trait-based succession patterns ( Figure 5) but moreover also provides a testable framework that is prone to advanced statistical and experimental analysis.

| Edibility trait shows unexpected seasonal pattern
The seasonal development of the edibility trait, i.e. the susceptibility towards grazing by Daphnia, as well as the ability of algae to form chains and colonies was surprising as it was contrary to expectations and widespread belief. Theories about plankton succession, observations from lakes as well as modelling studies predict that the edibility of phytoplankton decreases after the clearwater phase towards summer and the algae composition responds to the increased grazing pressure by changing to less-edible, grazing-resistant species (Gaedke, 1998;Lampert, Fleckner, Rai, & Taylor, 1986;Sommer et al., 1986;Vanni & Temte, 1990;Wirtz & Eckhardt, 1996), which is, e.g. attained by the ability to form chains, colonies or filaments (Gliwicz, 1977). We observed the opposite pattern with low edibility and high coloniality during spring and an increase in algae edible for Daphnia and low coloniality later in the year, when grazing pressure is expected to be high (Sommer et al., 1986), both in nutrient-rich and -deficient years. In line with our observation, also studies from other lakes reported an increase in inedible algae in the absence of severe grazing and higher shares of edible algae when grazing pressure was high (Agrawal, 1998;Carpenter, Morrice, Elser, Amand, & MacKay, 1993), which contradicts predictions of defence theory (Coley, Bryant, & Chapin, 1985;Fagerstrom, Larsson, & Tenow, 1987;Porter, 1973). Agrawal (1998) hypothesizes that this paradox outcome might be explained by selective and size-specific grazing by zooplankton. As herbivores vary in their ability to consume the same phytoplankton species (Lundstedt & Brett, 1991), taxa that are edible to one grazer may be inedible to another. Hence edibility and resistance are specific to the particular grazer species, which can have opposing impacts on the phytoplankton composition (Knisely & Geller, 1986;Sommer et al., 2001). In this study edibility by Daphnia herbivores was considered. Possibly grazing pressure by other grazers, such as protozoans and calanoid or cyclopoid copepods, had a higher impact, resulting in algae being more edible towards Daphnia in summer. For example, Rhodomonas spp. (130 µm 3 cell volume) and Cryptomonas spp.
(1,500 µm 3 cell volume) were characterized as rather edible to Daphnia, but have been shown to be spared by copepod grazing (Sommer et al., 2001). Hence, high grazing pressure by copepods in summer might have triggered an increase in those algae species, which were inedible to copepods, but edible to Daphnia.
This shows that the edibility of algae is predator specific and thus difficult to define, making generalizations about the edibility of algae as proposed by the PEG model (Sommer et al., 1986) difficult.
Moreover, it is possible that the unexpected trends in the colony formation and edibility for Daphnia trait were due to reasons other than grazing pressure. For example, the low abundance of colonial and filamentous organisms during summer stratification might be related to higher sinking velocities of colonies (Reynolds, 2006) rather than to grazing pressure. Since different traits are not completely independent from each other and therefore not freely combinable, the unexpected trends in coloniality and edibility might have been shaped by trends in other more important traits. This would indicate that losses by grazing were not as important in shaping the phytoplankton communities (top-down) and that the seasonal phytoplankton dynamics in the Rappbode system were primarily regulated by resource availability (bottom-up).

| Nutrient reductions affect biomass and functional traits mainly during spring
While the general succession patterns of functional traits were independent of nutrient regime, the extent of the seasonal changes in functional traits from spring to summer clearly differed with trophic status. Phytoplankton biomass and functional traits exhibited lower fluctuations along the season during oligotrophic years, as the differences between the traits in spring and summer were relatively small.
In contrast, in eutrophic years seasonality of biomass and traits was more pronounced and the differences between spring and summer conditions were large. The increase in seasonal changes in traits with nutrient concentration was expected, as eutrophic systems usually show larger seasonal fluctuations in biomass and phytoplankton cell size spectra and more successional stages (Gaedke, Seifried, & Adrian, 2004;Kalff, 2002;Sommer, 1986a;Sommer et al., 1986).
Comparing the seasonal biomass development between the two trophic states, it became also evident that the strong phytoplankton spring bloom found in eutrophic years vanished with oligotrophication, while summer biomass changed little (or even became higher).
This contradicts the PEG model which expects the disappearance of summer blooms with oligotrophication, while the magnitude of the spring bloom is less affected (Sommer et al., 1986). Water residence time and internal lake processes might be a reason for the differences in biomass patterns between Rappbode Reservoir and, e.g. Lake Constance, which was a major study site for the development of the PEG model. The Rappbode system has a shorter residence time (approximately 1 year) and external nutrient loads are more important than in Lake Constance given its much longer residence time (4.3 years). However, the relative importance of external nutrient inputs versus internal nutrient processing changed during the oligotrophication process in Rappbode Reservoir. While in the eutrophic phase high external inputs restored high nutrient conditions during the cold season and induced a massive spring bloom followed by high downward nutrient export by sedimentation (Wentzky et al., 2018), this pulsed regime got largely replaced by internal processing in the oligotrophic phase. High shares of motile and mixotrophic species during the oligotrophic period reduced sedimentative losses and speeded up internal nutrient recycling and finally lead to a more dampened succession with less pronounced spring blooms and a higher persistence of algal communities throughout the growing season. These observations comply with findings from the re-oligotrophication in Lake Constance, where significant internal processing and nutrient regeneration have been documented (Gaedke & Straile, 1994;Tilzer, Gaedke, Schweizer, Beese, & Wieser, 1991).
A study from Lake Constance also showed that differences in the functional composition after nutrient reduction were most apparent during nutrient limitation in summer (Weithoff & Gaedke, 2016

| CON CLUS IONS
The study provides a quantitatively supported functional template for phytoplankton succession in temperate lakes under different nutrient regimes ( Figure 5). In line with conceptual models (Margalef, 1978;Sommer et al., 1986), we quantitatively showed that succession patterns of plankton communities were mainly driven by a trade-off between small-celled, fast-growing species that are able to incorporate existing resources at a reasonable short time (r-strategists) and large-celled species with more complex and efficient mechanisms to exploit scarce mineral nutrients or acquire previously unexploited nutrient pools (k-strategists). Moreover, the seasonal development of functional traits mirrored environmental pressures over the year. For example, phosphate affinity and mixotrophy peaked during phosphorous limitation in summer, while maximum growth rate and light affinity were high during the mixing season when light was limiting but nutrients were highly available. Noteworthy, the main features of functional trait succession were independent of nutrient regime and the seasonal development of functional properties of the community was similar during oligotrophic and eutrophic conditions.
Distinct changes in functional composition occurred, however, and seasonal differences during oligotrophic years were generally less pronounced over the year. Spring communities in the oligotrophic state moreover showed clear sign of nutrient limitation and therefore showed more functional resemblance with summer communities than under eutrophic conditions.
The study shows that translating species into functional traits by assigning trait values compiled from the literature provides a powerful method towards a more predictive community ecology.
Functional traits can be applied to translate information about taxonomic composition into ecologically interpretable functions and eco-physiological processes that can be linked to resource competition, succession and ecosystem dynamics. It enables ecological interpretation of observed phytoplankton community dynamics by quantification of functional characteristics and improves the predictability of community shifts in response to changing environmental conditions. This should open new perspectives for predictive limnology using lake ecosystem models.
Our method of assigning static trait values to each algal species does not take intraspecific trait variability into account, which can sometimes be significant (Bolius, Wiedner, & Weithoff, 2017;Malerba, Heimann, Connolly, & Leroux, 2016;Morabito, Oggioni, Caravati, & Panzani, 2007). However, in contrast to measuring traits directly on the natural community and thus including intraspecific trait plasticity, our method has the advantage that it can be applied to historic taxonomic data. This allows to follow long-term trends of the community from a functional perspective, e.g. to study the response to eutrophication or climate change. Moreover, our method can also explore patterns in traits, which are not directly measurable on the natural community (e.g. physiological traits such as maximum growth rate).
In summary, the reduction in taxonomic complexity to the common currency of functional traits allows assessing community structure in historic datasets, but the method can also be used to make comparisons across different environments and habitats. As trait-based approaches can serve as a unifying concept in ecology, we strongly encourage researchers to take advantage of them.

ACK N OWLED G EM ENTS
We thank the water supply works 'Wasserwerk Wienrode' and ' Talsperren