Erhai Lake is a large subtropical lake located on the Yunnan–Guizhou Plateau in China (25^◦52′N and 100^◦06′E). It has a water surface area of 252 km² situated 1,966 m above sea level (a.s.l.), with an average water depth of 10 m and an average water retention time of 4.5 years. During the study period, the average water level of the lake was 1,965.03 m a.s.l. The region has a subtropical monsoon climate, with an annual mean temperature of 15°C, wet seasonal precipitation of 870 mm (May–October), and dry seasonal precipitation of 170 mm (November–April) (Wen et al., 2021). Erhai Lake is in the early stages of eutrophication and is changing from mesotrophic to eutrophic, with total nitrogen (TN) and phosphorus (TP) concentrations ranging from 0.60 to 0.80 mg L^-1 and 0.015 to 0.037 mg L^-1, respectively (Yin et al., 2021). The dominant submerged macrophyte species in the lake include Potamogeton maackianus, Vallisneria natans, Ceratophyllum demersum, and Hydrilla verticillata (Wen et al., 2022). There are 6 common, 10 rare, and 2 endangered species in Erhai Lake, including Utricularia aurea Lour. and Ottelia acuminata (Ye et al., 2018; unpublished data). The maximum colonization depth of the submerged macrophytes is less than 5.5 m (Wen et al., 2022).

We conducted the study in Erhai Lake in June and July of 2017. Three evenly distributed transects were set perpendicular to the shoreline in each of the eight bays (i.e., Shapin, Xizhou, Majiuyi, Erhaiyue, Xiangyang, Wase, Changyu, and Hongshan) within the lake. With the local socioeconomic development in recent decades, increasing anthropogenic discharge and agricultural non-point pollution around these bays lead to the increased sediment nutrient content, i.e., TP and TN concentrations increased by ca. 0.10% and 0.46% within 2010, respectively (Li et al., 2016). There are 23 rivers flowing into the lake, with one river and one water diversion tunnel flowing out of the lake (Yang et al., 2021a; Yang et al., 2021b; Hu et al., 2021). Macrophyte sampling occurred at 0.5 m water depth intervals across each transect using a rotatable reaping hook to the maximum colonization depth (i.e., 5 m depth) of macrophytes (Figure 2). The sampling area per sample was 0.2 m², and sampling was repeated three times at each depth. We sampled the submerged macrophyte communities from 642 quadrats (0.2 m²) (Figure 2). The macrophytes collected from each sample were washed, identified of their species level, drained, and weighed as fresh biomass. The total plant community biomass and the biomass of individual macrophytes at each sampling plot were expressed as the average value from three quadrats. The distribution depth of the submerged macrophytes was determined using a sonic depth finder. Water clarity at each quadrat across the water depth was measured in situ with the Secchi depth before sampling.

Based on our hypotheses, we applied the polynomial regression to explore the non-linear relationships of community biomass with the water depth and species richness. Since we assumed that species richness is negatively correlated with the water depth, unary linear regression analysis was used to explore their relationship. The macrophyte community biomass data were log(x + 1), transformed to improve the variance homogeneity and normality of data. In order to classify the distribution pattern of the submerged macrophyte community biomass along the water depth gradient, regression trees (RTs) were performed to divide the water depth into classes by applying the R package “Party.” RT models recursively partition data to find increasingly homogeneous subsets based on independent variable splitting criteria using variance-minimizing algorithms. The dependent data were partitioned into a series of descending left and right child nodes derived from parent nodes (Krzywinski and Altman, 2017). The RT divided the water depth into three classes, 0< water depth ≤1.5 m, 1.5< water depth ≤3.6 m, and 3.6< water depth ≤5 m (Supplementary Figures S1, S2), representing shallow, medium, and deep water areas, respectively. The species richness of the submerged macrophyte community was the average value at each sampling depth and was calculated using the R package “Vegan.” Simple linear regression was used to explore the relationship between the macrophyte community biomass and species richness of different water depth areas. Relative biomass is equal to the biomass of a species divided by the sum of the biomass of all species in each plot, and a species with a relative biomass of more than 5% was defined as a dominant species (Mouillot et al., 2013; Ma et al., 2017; Ratcliffe et al., 2017) (see Supplementary Figure S3 for biomass variation across the dominant species). The ratio of water clarity (i.e., Secchi depth) to water depth was calculated as a proxy for underwater light availability, and simple linear regression was applied to explore the relationship between the water clarity and the water depth ratio and water depth (Supplementary Figure S4). We used the ggplot2 package in R for graph plots. All statistical analyses and graph plots were performed using R 3.5.4 (R Core Team, 2019).

The macrophyte community biomass showed a unimodal relationship with the water depth gradient (R ² = 0.29; p < 0.001; Figure 3A). The macrophyte community biomass decreased as the water depth increased (R ² = 0.16; p < 0.001; Figure 3B). The water clarity to water depth ratio decreased with the increasing water depth (Supplementary Figure S4). Species richness also showed a unimodal relationship with the macrophyte community biomass (R ² = 0.06; p < 0.05; Figure 3C).

Community biomass positively correlated with species richness both in shallow and deep water areas. However, the relationship was significant only in deep water (p < 0.05). The slope of biomass against species richness was small in shallow water but became steeper in deep water (Figure 4). The relationship between community biomass and species richness was insignificant at the medium water depth (p > 0.05; Figure 4).

In the shallow water area, dominant macrophytes species included P. maackianus (35.4%), followed by C. demersum (18.3%), Myriophyllum spicatum (13.7%), P. lucens (9.7%), H. verticillata (8.2%), Potamogeton wrightii (5.5%), and V. natans (5.2%). In the medium water area, the most dominant species were P. maackianus (59.5%), followed by C. demersum (11.7%), V. natans (8.4%), P. lucens (7.6%), and H. verticillata (5.9%). In the deep water area, the most dominant species comprised P. maackianus (35.7%), followed by V. natans (34.5%) and C. demersum (26.5%). The relative biomass of V. natans increased with the increasing water depth. P. maackianus had the highest relative biomass in the medium water depth, whereas C. demersum had the lowest relative biomass (Figure 5; Supplementary Figure S3).

We found that species richness decreased with the increasing water depth. In addition, there was a significant decrease in the water clarity to water depth ratio (i.e., light availability) with the increasing water depth (Supplementary Figure S4). These findings imply that the increasing light limitation with the increasing water depth negatively affects macrophyte species richness (Chisholm et al., 2013). At the medium water depth, the macrophyte community composition changed dramatically compared to that in shallow water. As light availability became more limited in deeper water, the submerged macrophytes likely competed for light and space, and dense canopy-forming species, such as P. maackianus, displaced the neighboring species and eventually became dominant at the medium water depth area. Previous studies also showed that canopy-forming species could elongate their stems and form a dense canopy to gain a competitive advantage over non-canopy-forming species (Fu et al., 2014b; He et al., 2019). However, at the deep water area, the macrophyte community structure was dominated primarily by a few species tolerant to low-light conditions, including P. maackianus, C. demersum, and V. natans, thus reducing species richness. Consistent with our findings, previous studies also reported reduced species richness in deep water areas as only a few species can withstand low-light conditions and grow (He et al., 2019; Chou et al., 2022).

Our results showed an overall unimodal relationship between macrophyte biomass and species richness, likely due to the fact that the species richness effect on community biomass was enhanced at harsher low-light conditions. These findings support our prediction that environmental gradients can modulate species richness–biomass relationships at a finer spatial scale. For instance, in shallow water, macrophyte species richness did not significantly affect the community biomass, likely due to the weak effects of complementarity (e.g., species-complementary resource use) (Jiang et al., 2008; Jiang et al., 2009). At the medium water depth, the community biomass was, on average, the highest compared to that at other water depths, and the insignificant species richness–biomass relationship likely resulted from the competitive exclusion (i.e., two species competing for the same limited resource may not coexist) between dominant and subordinate species (Huston, 1997; Tilman et al., 1997). In deep water, the numbers of species reduced sharply due to a further decrease in the underwater light availability, and only a few low-light tolerant species, i.e., P. maackianus, C. demersum, and V. natans, were able to colonize (Wen et al., 2022). These species can adapt to low-light conditions through different adaptive strategies (e.g., P. maackianus allocates more biomass to stems and forms dense canopies, while V. natans increases both the leaf chlorophyll content and leaf biomass) and occupy different water-column spaces (Chen et al., 2016; He et al., 2019; Wen et al., 2022). In addition, P. maackianus and M. spicatum can provide feedback to improve the environmental conditions, such as water clarity by inhibiting the growth of phytoplankton and periphyton through allelopathy or nutrient competition (Liu et al., 2020) and facilitate the growth of other species that are less tolerant to low-light conditions. The stronger species richness–biomass relationship in the deep water area in our study indicates that interspecific interactions among species may be conducive to the community biomass at low-light environments. Our findings agree with previous studies, which demonstrated that positive species interactions among macrophytes could promote growth and reproduction under adverse environments (Le Bagousse-Pinguet et al., 2012; Hao et al., 2013; Liu et al., 2020). For instance, Hao et al. (2013) found that the positive interaction between P. maackianus and M. spicatum due to their morphological differences or niche complementarity has enhanced their survival in harsh environments (e.g., severe eutrophic conditions). Hence, our results support the stress-gradient hypothesis, which suggests that species that are dominant in tough conditions will reinforce each other. However, in favorable environmental conditions, they exhibit competitive interactions (Bertness and Callaway, 1994; Steudel et al., 2013).

Our research provides evidence that the effects of macrophyte species richness on community biomass vary with the water depth in a large subtropical lake, suggesting that water depth can modulate the species richness–biomass relationship. In addition, our results imply that maintaining multiple macrophyte species with adaptive strategies to low-light conditions is important for enhancing the overall macrophyte community biomass in light-limited environments. Consistent with our findings, previous studies also reported a hump-shaped relationship between community biomass and water depth (Lewerentz et al., 2021; Ma et al., 2021), suggesting that the submerged macrophyte biomass can be regulated by managing water levels in freshwater lakes. Our results have significant management implications for the lake macrophyte biodiversity conservation and productivity.