Introduction
Trophic cascades characterize the top-down effects of predators propagating down trophic levels, which represent one of the most classic paradigms in community ecology and have stimulated tremendous amount of experimental and theoretical studies (Carpenter & Kitchell 1988; Poliset al. 2000; Carpenter 2010; Ripple et al. 2016). Mounting evidence have demonstrated the prevalence of trophic cascades in nature (Pace et al. 1999; Shurin & Seabloom 2005) and their far-reaching impacts on ecosystem functioning and services (Estes et al. 2011; Stricklandet al. 2013; Walsh et al. 2016; Stock et al. 2017). Despite many previous studies to reveal various biotic and abiotic factors that potentially influence the strength of trophic cascades (Schmitz et al. 2004; Borer et al. 2005; Bruno & O’Connor 2005; Otto et al. 2008), the inconsistent outcomes, for example the idiosyncratic effects of predator diversity (Bruno & O’Connor 2005; Otto et al. 2008), suggest a continued uncertainty of its ultimate mechanisms.
One long recognized potential explanation is the productivity hypothesis, which predicts that the strength of trophic cascades increases with primary productivity (Fig. 1a). Primary productivity provides the essential energy and elements for upper trophic levels within a food web. Such bottom-up processes determine not only the trophic position of the apex predator, i.e., food chain length (Post 2002), but also the accumulation of its biomass (Moore & De Ruiter 2012; Barbier & Loreau 2019). By supporting higher biomass of apex predators, a higher primary productivity can potentially trigger a stronger trophic cascade. This hypothesis, historically illustrated as the ecosystem exploitation hypothesis (EEH) (Oksanenet al. 1981), has been supported by positive relationships between primary productivity and trophic cascade strength found across various studies (Wootton & Power 1993; Su et al. 2021). However, evidence that does not support the relationship between primary productivity and the strength of trophic cascades was also reported in other studies (Borer et al. 2005). These contrasting findings are striking as they have been found even within the same realm, e.g., freshwaters (Borer et al. 2005; Su et al. 2021).
An implicit assumption underlying the productivity hypothesis is unimpeded vertical energy fluxes, where energy fixed by primary producers can reach higher trophic levels efficiently. This assumption, nevertheless, rarely holds in nature (Lindeman 1942; DeBruyn et al. 2007). In natural ecosystems, only a small proportion of energy from low trophic levels can be converted into biomass of high trophic levels, due to inefficiencies in food ingestion, assimilation, and biomass production (Barneche & Allen 2018; Eddy et al. 2021). For instance, food webs often involve many less edible plants (Oksanen & Oksanen 2000) or anti-predative herbivores (Degerman et al. 2018), which can significantly hamper vertical energy flow to apex predators (Stiboret al. 2004). Inefficiency in energy transfer can significantly influence the biomass distribution of food webs (de Ruiter et al.1995; McCauley et al. 2018; Barbier & Loreau 2019) and thereby alter the strength of trophic cascades (Heath et al. 2014; Galiana et al. 2021). In particular, too low energy transfer may decouple biomass production at higher trophic levels from primary production, violating assumptions underlying the productivity hypothesis (Brett & Goldman 1997; Davis et al.2010). In such cases, the strength of trophic cascade depends mainly on the efficiency of energy transfer (energy transfer hypothesis; Fig. 1b), rather than the amount of energy supply characterized by the primary productivity. While the transfer efficiency hypothesis has been formulated in mathematical models (DeBruyn et al. 2007), it has not been tested explicitly.
In this study, we conduct a field mesocosm experiment, combined with mathematical models, to test the aforementioned two hypotheses underlying trophic cascades (i.e., the productivity hypothesis and energy transfer hypothesis; Figure 1). We constructed mesocosms hosting a three-level food chain and quantified the strength of trophic cascade by simulating loss of top predator. We simultaneously manipulated nutrient supply and zooplankton community composition to generate variation in primary productivity and energy transfer efficiencies, respectively. Because both primary productivity and energy transfer processes are necessary for maintaining higher biomass of apex predators, we expected a weakened relationship between primary productivity (resp. energy transfer efficiency) and trophic cascade strength when energy transfer efficiency (resp. primary productivity) is impeded (Fig. 1). We end our discussion with scientific and practical implications of our results.