RESULTS
A total of 27 benthic macroinvertebrate taxa were found in the coastal waters of Lake Biwa. Taxonomic richness varied from 0 to 13 (5.55 ± 2.92, N = 33). In coastal communities, oligochaetes and chironomids were the most dominant taxa in numerical abundance (pi= 0.45 ± 0.28 and 0.24 ± 0.25, respectively; Table S1), whereas grazing snails (Semisulcospira spp.) and a suspension-feeding bivalve (Unio douglasiae biwae ) were the most dominant in biomass abundance (Bi /BT = 0.17 ± 0.22 and 0.11 ± 0.25, respectively; Table S2).
Based on isotope data (Tables S3 and S4), TP was estimated for each taxon at each site (Table S5). Most grazer and deposit-feeder taxa were located around TP = 2 (Figure 2a). Among community members, carnivorous leeches and a freshwater prawn exhibited the highest TP (3.39 ± 0.50 and 3.28 ± 0.44, respectively), which are regarded as top predators in macroinvertebrate communities. In contrast, carnivorous insects, i.e., stoneflies and dragonflies, showed lower TPs than expected from their carnivorous habits, whereas grazing and suspension-feeding gastropods showed higher TPs than expected from their herbivorous habits, which are regarded as trophic omnivores. When all data were pooled for the whole coastal community, individual TPs significantly increased in body mass, but the slope of TP against the body mass was gentle (slope = 0.17, P < 0.001, regression analysis; Figure 2a). Such a gentle slope could be ascribed to the omnipresence of grazing and suspension-feeding gastropods with large body masses and intermediate TPs. When these data were excluded from the regression analysis, the body mass-TP slope became steeper (slope = 0.30, P = 0.009). As observed in the general patterns of allometric scaling, body mass and numerical abundance demonstrated a clear negative correlation (P < 0.001; Figure 2b). The numerical abundance-TP slope was also weakly negative (P = 0.028; Figure 2c), but the biomass abundance-TP slope was flat (P = 0.14; Figure 2d), as observed in another lake food web (Cohen et al., 2003).
Community biomass varied from 0.00 to 6515.28 mg/m2(2309.70 ± 2207.64 mg/m2, N = 33; Table S6). The basal resource availability defined by the phytoplankton and benthic algal biomasses impacted the biomass of each functional group. The benthic algal biomass showed bottom-up effects only on the chironomid biomass (r = 0.39, P = 0.039). Contrarily, the phytoplankton biomass had no bottom-up effects on any taxa but rather a negative effect on grazing insects and crustaceans (r = –0.46, P = 0.01). The phytoplankton biomass showed a strong negative correlation with DO in coastal waters (r = –0.54, P < 0.001), which was critical to the abundance of grazing insects and crustaceans (r = –0.59, P < 0.001) and prey taxonomic richness (r = –0.44, P = 0.015).
For BEF, the community biomass was significantly correlated with taxonomic richness (P < 0.001, N = 30; Figure 3a). This correlation, however, is partly due to the added species biomass to communities with low numerical abundance and thus low taxonomic richness, as shown by the community biomass almost reaching an asymptote at a value of 4 for taxonomic richness. Even when using the diversity index H ’ to consider numerical abundance and equality, the BEF relationship remained weak and varied greatly among communities with lower H ’ (P = 0.013, N = 29; Figure 3b). Using i TP, however, the positive BEF became stronger with a higher significance level (P = 0.002, N = 26; Figure 3c). In contrast, the variation in H ’ could not account for any change in FCL, which is defined as the maximum TP for each local community (P = 0.20, N = 26; Figure 3d).
The two top predator taxa more frequently appeared in local communities with higher prey richness (Wald = 4.17, odds ratio = 2.20, P = 0.041, logistic regression analysis) but not with higher biomass (P = 0.41) or greater abundance (P = 0.57) of prey taxa, except for predator-resistant gastropods. Although the i TP was positively correlated with the maximum TP (r = 0.42, P = 0.03, N = 26), multiple regression analysis revealed that the within-lake variation in the i TP was best explained by the relative biomass abundance (Bi /BT ) of deposit feeders, excluding deposit-feeding bivalves, with lower TPs and higher numerical abundance (t = –3.18, adjustedR 2 = 0.27, P = 0.004), rather than by top predators with the highest TP. Neither the i TP nor the maximum TP was affected by the basal resource availability.