Figure legends
Fig. 1. Predicted timings of phenological events and the phenological delay between them in four representative lake types across Europe under reference climate conditions (median values for 1979-2009). Lake types are indicated by their maximum depth (zmax ) and light extinction coefficient (Kd ). Shown are (a-d) the onset of the algal bloom (OAB) in day of year (doy), (e-h) the timing of theDaphnia maximum (TDM, doy), (i-l) the phenological delay between TDM and OAB (PLD, days), and (m-p) the dominant process controlling OAB. Black lines delimit regions in which lakes develop ice cover in at least 16 out of the 31 simulated years. Regions, where the light threshold for OAB is exceeded already on the first day of the year or where the temperature threshold for TDM is not reached in more than 50% of the simulated years, are marked in grey. Supplement S1 provides figures on the geographic distributions of OAB, TDM, PLD, and the controlling processes for all 16 lake types investigated.
Fig. 2. Impact of geographical factors (latitude, longitude, elevation) and optical depth (OD) on the timing and asynchrony of phenological events across 1907 European locations and 16 lake types.(a-l) The relationships (± 1 standard error) of the onset of the algal bloom (OAB), the timing of the Daphnia maximum (TDM), and the phenological delay between them (PLD), with latitude, longitude, elevation and OD, as determined by general additive models (GAM) including these four factors as smooth functions. Y-axes scale in days, where the dashed line at 0 indicates the isoline of no effect of the respective independent variable. Standard errors are too small too be visible in all panels. Standard deviations (days) of OAB (m) , TDM (n) and PLD (o) across the 16 lake types at each location. Proportion of the variance in OAB (p) , TDM(q) and PLD (r) that is explained by GAMs including as independent variables only geographical factors (‘geo’), only ’OD’, or both (i.e. the full model, ‘geo+OD’).
Fig. 3. Predicted effects of a uniform temperature increase by +4°C on the timings of phenological events and the phenological delay between them in four representative lake types (median values of 31 simulation years). Lake types are indicated by their maximum depth (zmax ) and light extinction coefficient (Kd ). Shown is the difference in days between the warming and reference scenarios for (a-d) the onset of the algal bloom (OABdiff), (e-h) the timing of theDaphnia maximum (TDMdiff), and (i-l) the phenological delay between them (PLDdiff).(m-p) Dominant processes controlling OAB in the reference and warming scenarios. The controlling process changes between the two scenarios in green and purple regions but remains the same in blue, yellow and red regions. Black lines delimit regions in which lakes develop ice cover in the reference (dotted lines) and warming (solid lines) scenarios. Regions, where the light threshold for OAB is exceeded already on the first day of the year or where the temperature threshold for TDM is not reached in more than 50% of the simulated years in at least one of the scenarios, are marked in grey. Supplement S1 provides corresponding figures on OABDiff, TDMDiff, PLDDiff and the controlling processes for all 16 lake types investigated.
Figure 4: Factors and processes mediating predicted shifts in phenology between the reference and warming scenarios. (a)Median temporal shifts (days, with error bars indicating the 20 % and 80 % quantiles) between the warming and reference scenarios for the onset of the algal bloom (OABdiff), the timing of theDaphnia maximum (TDMdiff), and the phenological delay between them (PLDdiff) in lakes that differ in the dominant processes controlling OAB in the two scenarios. (b)Frequency distribution of PLDdiff in lakes that differ in the dominant processes controlling OAB in the two scenarios.(c-f) The relationship of PLDdiff to latitude, longitude, elevation and optical depth (OD), as determined by a general additive model (GAM) including these four factors as smooth functions. Grey shading (only visible in panel e) indicates ± 1 standard error. Y-axes scale in days, where the dashed line at 0 indicates the isoline of no effect of the respective independent variable. Proportion of the variance in OABdiff (g) , TDMdiff (h), and PLDdiff(i) that is explained by GAMs including as independent variables only latitude, longitude and elevation (‘geo’), only ’OD’, both (‘geo+OD’), or, alternatively, the five categories of dominant processes controlling OAB from panels (a) and (b).