Figure legends

Figure 1 - A) Schematic depiction of how the absence (Control=blue line) and the presence of foundation species (Dreissena polymorpha = orange line, Myriophyllum spicatum= green line) is hypothesized to affect a generic ecosystem variable (e.g. chlorophyll-a concentration). From these three time series, we can generate an expectation for what would be a perfectly additive effect on the ecosystem variable (black line = (Myriophyllum +Dreissena ) - Control). If the recorded time series from ecosystems with both foundation species present is lower than the expected additive effect, both species would interact synergistically; if it is higher, both species would interact in an antagonistic fashion. B) Dreissena polymorpha , “Zebra mussel” (Photo credit: N. Sloth). C) Myriophyllum spicatum , “Eurasian water millfoil” (Photo credit: P. DynowskiI). D) Schematic of the experimental ponds: the ponds are approximately 4 m in diameter and have a deep (1.5 m) and a shallow end (0.5m), where we planted macrophytes and mussels. In the middle of each pond we placed a multiparameter sonde at 1 m depth to monitor ecosystems dynamics.

Figure 2 - Effect of foundation species on mean of ecosystem parameters in Phase 1, Phase 2 and Phase 3 (left, middle, and right panel, respectively). The lines indicate the respective average of all four ponds per treatment per hour (mean ± SE), the gray shading indicates the disturbance phases, and the coloured bars underneath the time series indicate whether a treatment was significantly different from the control (one linear model per hour: (orange=Dreissena main effect [D treatment], green=Myriophyllum main effect [M treatment], purple=interactive effect [MD treatment]). The data stem from a seven day long sliding window (168 data points). The black line indicates the predicted additive response based on the sum of the separate macrophyte or mussels treatment with the control subtracted (e.g. (Macrophyte Chl + Dreissena Chl) - Control Chl = predicted additive response).

Figure 3 - Effect of foundation species on variance (Coefficient of variation: CV) of ecosystem parameters in Phase 1, Phase 2 and Phase 3 (left, middle, and right panel, respectively). The lines indicate the respective average of all four ponds per treatment per hour (mean ± SE), the gray shading indicates the disturbance phases, and the coloured bars underneath the time series indicate whether a treatment was significantly different from the control (one linear model per hour: (orange=Dreissena main effect [D treatment], green=Myriophyllum main effect [M treatment], purple=interactive effect [MD treatment]). The data stem from a seven day long sliding window (168 data points).

Figure 4 - Effect of foundation species on mean metabolic rates of the ecosystems in Phase 1, Phase 2 and Phase 3 (left, middle, and right panel, respectively). The lines indicate the respective average of all four ponds per treatment per hour (mean ± SE), the gray shading indicates the disturbance phases, and the coloured bars underneath the time series indicate whether a treatment was significantly different from the control (one linear model per hour: (orange=Dreissena main effect [D treatment], green=Myriophyllum main effect [M treatment], purple=interactive effect [MD treatment]). All rates were calculated using Odum’s diel oxygen technique (Staehr et al.2010). The black line indicates the predicted additive response based on the sum of the separate macrophyte or mussels treatment with the control subtracted.

Figure 5 - Effect of foundation species on variance (Coefficient of variation: CV) of ecosystem metabolism in Phase 1, Phase 2 and Phase 3 (left, middle, and right panel, respectively). The lines indicate the respective average of all four ponds per treatment per hour (mean ± SE), the gray shading indicates the disturbance phases, and the coloured bars underneath the time series indicate whether a treatment was significantly different from the control (one linear model per hour: (orange=Dreissena main effect [D treatment], green=Myriophyllum main effect [M treatment], purple=interactive effect [MD treatment]). The data stem from a seven day long sliding window (168 data points).