INTRODUCTION
Since Tilman & Downing (1994)’s pioneering work, many studies have so far investigated biodiversity–ecosystem functioning (BEF) relationships and demonstrated that ecosystem productivity often increases with increasing biodiversity (Tilman et al., 2014). These results attract ecological and societal interests as biodiversity loss leads to a decline in ecosystem goods and services for humanity by decreasing ecosystem productivity (Balvanera et al., 2006; Harrison et al., 2014; Millennium Ecosystem Assessment, 2005), stimulating biodiversity conservation and restoration. In the assessment of BEF relationships, biomass is often used as a proxy for productivity due to difficulties inin situ measurements (e.g., Duffy et al., 2015; Gamfeldt et al., 2015; Scurlock et al., 2002). Despite unclear causal mechanisms, positive relationships between biodiversity and biomass are found in many observational studies in nature (Duffy et al., 2017). From a meta-analysis on non-experimental and experimental data, Waide et al. (1999) concluded that such a positive BEF is commonly found for terrestrial consumers, whereas the dominant pattern is unimodal or non-significant for aquatic consumers.
There are two possible explanations for the scarcity of positive BEF in aquatic consumer communities. One is the dominance of top-down effects on consumer biomass in aquatic food webs, i.e., the negative effect of predators on herbivore biomass is stronger in aquatic than in terrestrial systems (Shurin et al., 2002; Strong, 1992). Another is the difference in vertical diversity between these two systems, i.e., a greater number of trophic levels (TLs) in aquatic than in terrestrial food webs due to wider body size distribution and higher metabolic efficiency in aquatic systems that permits more extended trophic links (Rossiter et al., 2017). Since the energy entering consumer communities becomes progressively attenuated with increasing TLs (Lindeman, 1942), we may underestimate ecosystem productivity if substituted with community biomass measurements in systems with multitrophic interactions.
Recently, multitrophic perspectives have provided new insights into empirical studies on BEF (Eisenhauer, 2017; Eisenhauer et al., 2019; Soliveres et al., 2016). In multitrophic systems, a comprehensive understanding of BEF relationships requires the measurement of energy flows but not community biomass in food webs because the former can be a common currency for ecosystem productivity across a variety of ecosystem types (Barnes et al., 2018; Hines et al., 2015). Supported by size-based food web models, the vertical diversity hypothesis (VDH) states that functional diversity across TLs can enhance energy flows within a community (Schneider et al., 2016; Wang & Brose, 2018). However, this hypothesis has been poorly tested empirically (Hines et al., 2015; Thompson et al., 2012a). Although experimental approaches are effective in testing the VDH, simplified experimental systems with small species pools and few TLs may lack the power needed to demonstrate that more energy travels to higher TLs in real food webs with high species richness and trophic network complexity. Barnes et al. (2016) employed the energy-based approach to explore BEF relationships in non-manipulative terrestrial food webs with multitrophic interactions, classifying consumers into given trophic groups based on ecological traits. However, realized consumer trophic positions (TPs) vary broadly depending on the ecological context, as is often the case for aquatic consumers (Nakazawa et al., 2010). Such flexibility in TPs can generate analytical uncertainty in energy-based BEF.
Stable isotope analysis is a powerful tool for accurately estimating consumer TPs (Chang et al., 2014; Jennings et al., 2002b; Post, 2002b). Using this technique, many ecologists measure food chain length (FCL) defined as top predator TPs (Post, 2002a). FCL is a convenient indicator for empirical studies but may not accurately capture the overall energy flow within a food web as not all energy flow distributions among trophic links are fully considered (Figure 1). To overcome this limitation, Ishikawa et al. (2017) recently proposed an alternative indicator to characterize food web topology, integrated trophic position (hereafter, i TP), which is defined as the summed TPs of all consumer species in a focal food web weighed by the relative biomass of each species (see Eq. 5 for details). The i TP indicates how many times consumer biomass undergoes metabolic turnover, on average, as a whole community. Moreover, it reflects dynamical changes in the biomass pyramid via trophic interactions, e.g., a decrease in prey biomass due to predation can be offset by increasing predator TP and predator biomass. Therefore, the i TP can provide more appropriate and quantitative estimates of tropic energy flows within a food web (Figure 1).
Here, we test the VDH, i.e., whether and how functional diversity can enhance overall trophic energy flow, by applying the i TP to complex real food webs. We focus on spatial variation in a meta-community of coastal benthic macroinvertebrates in the ancient Lake Biwa, which exhibits high biodiversity with endemism (Kawanabe et al., 2012; Timoshkin et al., 2011). Its huge lake basin has hundreds of tributary rivers, which greatly vary in land use and local human population sizes (Kohzu et al., 2009; Ohte et al., 2010). Thus, its coastal habitats have high spatial heterogeneity in nutrients and basal resource availability (Karube et al., 2010; Sakai et al., 2013). Because benthic macroinvertebrates are sedentary or less mobile, we comparei TP across a wide variety of local communities that share the species pool within a meta-community but greatly vary in species composition in response to local environments (Shibata et al., 2014). Since land use can be a strong driver that alters species composition, a between-community comparison provides an opportunity to explore ecosystem consequences of biodiversity change across anthropogenically induced environmental gradients at the landscape level (Barnes et al., 2014; Gossner et al., 2016; Thompson et al., 2012b). Comparing our results with previous findings that the i TP decreases in more diverse communities of stream benthic macroinvertebrates (Ishikawa et al., 2017), we discuss a possible mechanism to explain such a negative BEF in contrast to the VDH. To deepen our mechanistic understanding of observed patterns in biodiversity and i TP for the coastal meta-community, we finally employed a size-based food web approach, which enables us to link size-specific community properties, such as abundance, biomass, and TP, to ecosystem functioning based on the scaling theory (Cohen et al., 2003; Jennings & Mackinson, 2003; Jennings et al., 2002b).