Discussion
In freshwater systems, the description of patterns and drivers of
plankton phenology - and the projection of climate change effects - has
been largely limited to verbal scenarios and qualitative graphical
models (Sommer et al. 2012; De Senerpont Domis et al. 2013; Berger et
al. 2014). Our study takes a step forward towards both a deeper
understanding of underlying drivers and a quantitative prediction of key
phenological events across climatic gradients, lake types and climate
change scenarios. With respect to the three questions raised in
(Kharouba & Wolkovich 2020), our model makes the following predictions.
1 - Phenological asynchrony, defined as the delay between the onset of
the phytoplankton bloom and the population maximum of Daphnia , is
highly variable across climatic gradients and lake types under current
climatic conditions. 2 - The degree of phenological asynchrony varies
systematically across Europe and is co-determined by physical lake
properties (in particular water transparency, lake depth, and their
product optical depth) that mediate how climate controls the onset of
the algal bloom. 3 - Under constant, uniform warming, phenological
asynchrony can predictably increase, remain unchanged or decrease, again
driven by on the factors that control the onset of the algal bloom.
In evaluating these predictions, one must keep in mind that the main
objective of our work is to identify general, continental-scale and lake
type-specific patterns of phenology and phenological synchrony, and not
to predict phenologies and their synchrony in any existing, real lakes.
Modeled lake types were simplified to one-dimensional water columns with
temporally constant light attenuation properties, and the driving
meteorology was obtained at a spatial grid resolution of 0.5°.
Observations from real lakes can therefore deviate from model
predictions, especially in lakes where local climatic conditions diverge
substantially from the grid-averaged meteorology. In Supplement S3
(Model validation), we illustrate this with an example from Lake
Windermere, where use of the local rather than grid-averaged meteorology
greatly improved the accuracy of TDM predictions. In contrast, our
approach successfully captures general trends in phenology related to
large-scale climatic gradients and their interaction with lake depth and
water transparency. Below, we illustrate this by comparing modelled with
observed phenologies from lakes covering a broad range of climate
regions and optical depths.
First, linear regressions of observed vs. predicted timings of both OAB
and TDM suggest that model predictions are unbiased, i.e. regression
slopes are not significantly different from 1 and intercepts not
different from zero (Supplement S3 Model validation). Unfortunately,
empirical data on the phenological delay between OAB and TDM are too
scarce for a similar regression analysis. Yet, the predicted wide range
in phenological asynchrony across lake types and geographic gradients –
as well as its inferred dependence on the dominant controlling processes
of OAB – are supported by observations from a broad range of lake
types.
For example, the longest delays are expected in lakes where our model
predicts that OAB is controlled by incident radiation. This is in line
with data from Loch Leven in northern Britain collected in 1979-2007,
where the median predicted, radiation-controlled, phenological delay of
~140 days compares well with the observed median delay
of ~120 days (Carvalho et al. 2015; Gunn et al. 2015).
Intermediate phenological delays are expected in lakes where our model
predicts that OAB is controlled by the timing of ice-off. In such lakes,
the phenological delay should thus increase in warmer years without ice
cover. Both of these expectations are in line with observations from
Lake Müggelsee in eastern Germany, where the phenological delay was
~74 days in the ice-covered year 1987 and
~98 days in the ice-free year 1988 (Shatwell et al.
2008), close to model predictions of 75 to 88 days, respectively.
Similar observations were made in Lower Lake Constance in southern
Germany, where phenological asynchrony was ~92 days in
the ice-covered year 2011 and ~116 days in ice-free year
2014 (IGKB 2012, 2016). Finally, the shortest phenological delays are
expected in lakes where our model predicts that OAB is triggered by the
onset of stratification. In such lakes, the phenological delay is also
predicted to be largely independent of optical depth. Both of these
expectations are in line with observations from the Sicilian Lake
Arancio (OD ~ 24) in 1991 and 1993 and Upper Lake
Constance (OD ~ 75) in 2011 and 2014, where both
observed (Naselli-Flores & Barone 1997; IGKB 2012, 2016) and predicted
phenological delays were ~60 days for these two lakes.
Our predictions are also in line with the general observation that
phenological responses to warming can vary greatly across space and
between different taxa at the same locations (Kharouba et al. 2018;
Roslin et al. 2021). More specifically, our analyses provide a
mechanistic understanding of why simple, ubiquitous phenological
responses to warming are not to be expected in pelagic producer-grazer
systems, and can thus explain why studies of the impacts of warming on
phytoplankton-Daphnia dynamics in different systems have come to
different conclusions (Winder & Schindler 2004; George 2012; Berger et
al. 2014; Straile et al. 2015).
The predicted extent of the variation in phenological asynchrony
suggests that Daphnia populations must be able to cope with large
temporal and spatial variability in the phenology of their resource and
that a single, optimal type of co-evolved phytoplankton-Daphniaphenology may not exist. It, therefore, seems unlikely that
warming-induced changes in phenological asynchrony must always have
negative effects on pelagic grazer populations. Consumer performance
does indeed not only depend on the degree of phenological synchrony with
its resources but also on the magnitude of the resource peak, which in
the case of phytoplankton strongly depends on the availability of
mineral nutrients and light (Jäger et al. 2008; Winder et al. 2012). A
recent review emphasized that, to date, almost no empirical study of
temperature-mediated phenological asynchrony has addressed the most
important consumer performance measure, i.e. population size (Samplonius
et al. 2021). Further steps in the projection of climate effects on
seasonal plankton dynamics, therefore, require a merging of the purely
physical approach presented here with models that quantitatively
describe trophic interactions in the plankton and their dependence on
temperature, light, and nutrient supply (Jäger et al. 2008; Schalau et
al. 2008; Kerimoglu et al. 2013; Uszko et al. 2017).
Changes in the phenological delay between the onset of the spring
phytoplankton bloom and the Daphnia population maximum have
consequences for lake ecosystem processes far beyond the
phytoplankton-Daphnia interaction. For example, a shorter spring
bloom implies a more rapid control of algal biomass by Daphnia ,
suggesting that sedimentation losses are less important under such
circumstances (Maier et al. 2019). Thus, changes in phenology and spring
bloom duration can affect algal export production to deeper waters and
the sediment, with consequences for food webs and biogeochemistry
(Kienel et al. 2017; Maier et al. 2019). Similarly, a shorter bloom
period and faster Daphnia growth can decrease grazing by
protozoans (Tirok & Gaedke 2006), and thus increase trophic transfer
from primary producers to fish (Caldwell et al. 2020) as well as impede
the development of toxic cyanobacteria in the bloom (Shatwell et al.
2008). The wide range of phenological asynchrony exposed in our study,
and its predicted responses to warming, are thus likely to affect lake
food web dynamics, energy, and nutrient fluxes in ways that remain yet
to be systematically explored. Our study provides predictions of the
phenological patterns that drive these processes as a function of
geographic location and lake type, and thus identifies space-for-time
(Pickett 1989) and lake type-for-time substitutions that can address the
ecological consequences of phenological delay.