Discussion
Despite long-standing interest in food web energetics (after Lindeman 1942) and trophic cascades (after Hairston et al. 1960), research on these two processes have been largely disconnected in the literature. Energy transfer was considered mainly by ecosystem ecologists as a diagram to understand ecosystem energetics and functioning (Chassotet al. 2010; Eddy et al. 2021), whereas trophic cascades were widely adopted by community ecologists to understand population dynamics (Layman et al. 2015; Jonsson 2017; Barbier & Loreau 2019). These two paradigms, however, address highly related processes of the same ecosystem, so a unified framework has increasingly been advocated by theoretical studies (DeBruyn et al. 2007; Barbier & Loreau 2019). In line with these theoretical advances, our study proposes to understand trophic cascades from an energetic perspective and combines mesocosm experiments and food chain models to disentangle two energetically relevant hypotheses. Our results showed strong associations between trophic cascade strength and predator efficiency but not with primary productivity (Fig. 5; Table S3, S4), supporting the energy transfer hypothesis rather than the productivity hypothesis. Thus, although primary productivity plays vital roles in food webs by providing ultimate energy for all trophic species (Oksanen & Oksanen 2000; Post 2002), an impeded energy transfer could limit the biomass of top predators and weaken their cascading effects (Mooney et al.2010; Kersch-Becker & Thaler 2015). These findings corroborate a recent study on grassland experiments, where a decreased energy flow to predatory beetles led to stronger suppressing effects of arthropod herbivores on plants, i.e., weaker cascading effects (Barnes et al. 2020).
In our experiment, the predator efficiency varies significantly among mesocosms with different zooplankton community compositions. Mesocosms dominated by D. magna (e.g., those initialized with D. magna and D. brachyurum ) showed the highest predator efficiency (Fig. 3c) and strongest trophic cascades. In comparison, mesocosms dominated by other zooplankton species (e.g., those initialized withS. dorrii and D. brachyurum , or with all three zooplankton species) showed lower predator efficiency and weaker trophic cascades in spite of higher total zooplankton biomass (Fig. 2b). This may be explained by the physical and chemical characteristics of D. magna, making them better food sources for predators (Vincent et al. 2020). Specifically, the lower motility of D. magna can increase the attack rate of predators (Rall et al. 2012), and its high body nutrient concentration can increase the assimilation efficiency (DeMott et al. 1998). Therefore, the lower predator efficiency may be attributed to the lack of high-quality resources in mesocosms without D. magna , or low amount of high-quality resources (i.e., lower biomass of D. magna ) in mesocosms with all three zooplankton species (Fig. S3). These results highlight the roles of species identity (O’Connor & Crowe 2005), rather than species richness per se (Duffy 2002), in mediating the strength of trophic cascades. Similar patterns have been reported for plant-herbivore interactions, where a higher primary productivity driven by plant diversity does not propagate to higher trophic levels due to the presence of inedible or anti-herbivory plant species (Brett & Goldman 1997; Davis et al. 2010).
While nutrient supply had no significant impacts on trophic cascades, it nevertheless imposed strong regulation on other processes of food webs. In line with previous studies (Teurlincx et al. 2017; Zhou & Declerck 2019), nutrient manipulations strongly shape the relative abundance of zooplankton species (Fig. S3) as well as their biomass and stoichiometry (Figs. 2a, S4). Moreover, the impaired performance of the predator under LP treatments, e.g., low biomass, production, and body P content, indicates a carryover effect of phytoplankton P limitation on top trophic levels (Boersma et al. 2008). Interestingly, we found that food chain efficiency (FCE) exhibited a unimodal pattern along the gradient of nutrient supply (Fig. 3d). The lower FCE in LP than MP treatments agrees with previous findings on the negative effects of nutrient limitation (Dickman et al. 2008; Atkinson et al.2021). But with further P enrichment, FCE started to decline, due to the relatively lower herbivore and predator efficiencies. This result may suggest a disadvantages of high nutrients for food web efficiency (Karpowicz et al. 2021), due to for example, differences in responding magnitudes between primary producers and apex consumers to nutrient enrichment (Fig. S2).
Our results may have useful implications for natural resource management. In particular, the classic biomanipulation approach for lake restoration is rooted in the theory of trophic cascades, which assumed that addition of piscivorous fish suppresses the biomass of planktivorous fish and releases the herbivorous zooplankton, thus controlling algal biomass via grazing (Carpenter et al. 1995). The importance of predator efficiency in determining trophic cascades revealed in the present study suggests that the biomanipulation approaches are most effective in systems with a higher efficiency of energy flow into apex predators, whereas eutrophic systems with low energy transfer efficiency may benefit weakly from the manipulating practices of piscivores (Langeland 1990; Brett & Goldman 1996). In addition, our results also highlight the importance of trophic interactions as drivers of ecosystem energy budgets, e.g., carbon related functions (Strickland et al. 2013; Schmitz et al.2018; Wyatt et al. 2021). For example, elevated efficiencies of energy transfer allowed increased biomass accumulation in apex consumers (McCauley et al. 2018) and therefore stronger trophic cascades, which may turn an ecosystem from carbon source into carbon sink (Schmitz & Leroux 2020).