Experimental setup
The mesocosm are situated at Huazhong Agricultural University in Wuhan City, Central China (30°29′N, 114°22′ E). 10 cm of sediment was added which was collected from the top few cm of sediment in Lake Liangzi (N 30°11′3″, E 114°37′59″; sediment TN, 5.5 ± 0.4 mg g-1 and TP, 0.42 ± 0.08 mg g-1, dry weight, n = 5)), and was thoroughly mixed in a clean container before transferring it into the experimental containers. On top of this a layer 35 cm of tap water was added (TN, 2.438 ± 0.249 mg L-1; TP, 0.022 ± 0.006 mg L-1, n = 5). Each container received 12 V. spinulosa (shoot length:12.1 ± 2.0 cm, mean ± SD, n = 480) and 16 P. crispus turions (length:5.0 ± 0.5 cm, mean ± SD, n = 640) which were planted evenly distributed within each container.
The treatment with shading of 75% of the incoming sunlight (S1) was achieved by covering the containers with one layer of black sun-shading net. The nutrient solution used for the nutrient loading treatment (E1) was made by dissolving NaNO3 and KH2PO4 salts in demineralized water. Nutrients were added once a month to these E1 containers, simulating a high-level nutrient loading of 2 mg L-1 N and 0.2 mg L-1 P, a eutrophic state, in the range of concentration and ratio of earlier experiments by Coppens et al. (2016) and Jeppesen et al. (2007). For the herbivory treatment, adult snails (n = 5, size ranged from 1 to 3 cm) were added to each herbivory treatment container (H1), a density of snails similar to natural lakes around the Yangtze River (H. J. Wang et al., 2010), and no extra snails were added to the rest of the containers (H0). Due to snail mortality and reproduction, and some snails even recovered from the H0 treatment, snail biomass changed largely over time in the treatments. We thus used the final snail biomass achieved as a continuous variable to predict the response parameters in our experiment.