From environment to biomass: trait-mediated effects on submerged macrophyte biomass in a subtropical shallow restored lake

Abstract

Background

Recovery of submerged macrophyte communities is crucial for the restoration of shallow eutrophic lakes, as macrophyte biomass here plays a key role in maintaining a clear-water state. However, the influence of environmental conditions on macrophyte functional traits mediating biomass production is unclear. We examined trait-mediated responses of Vallisneria denseserrulata to abiotic (water and sediment) and biotic (fish, neighbor plants) variables in 20 isolated restored experimental units in the shallow urban Lake Yiai, China, to elucidate how these variables affected trait expressions and, consequently, the biomass of V. denseserrulata.

Results

Vallisneria denseserrulata biomass varied widely across the restored units and was positively related to the contents of organic matter and nitrogen in the sediment and negatively so with water total nitrogen, turbidity, and fish biomass. However, the biomass and diversity of neighboring plants showed no significant relationship with the biomass of V. denseserrulata. Path analysis further indicated that water, sediment, and fish variables indirectly influenced biomass by modulating the traits of V. denseserrulata. High sediment nutrients content was associated with conservative plant traits, such as a high carbon content and a high carbon to phosphorus ratio, and enhanced aboveground allocation, i.e., increased shoot biomass, leaf length, and leaf number, leading to higher total biomass. In contrast, elevated water nutrients and turbidity favored belowground allocation, reducing total V. denseserrulata biomass. Fish biomass indirectly suppressed V. denseserrulata biomass by degrading water quality and subsequently promoting belowground allocation, while no significant direct contribution by grazing was found.

Conclusions

Sediment, water, and fish control submerged macrophyte biomass through trait-mediated pathways. This study enhances our understanding of how environmental conditions affect the functional traits and biomass of submerged macrophytes, illustrating the role of trait-mediated processes in improving the success of submerged macrophyte recovery in subtropical shallow lakes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08445-6.

Introduction

Urban lakes have become the focus of ecological restoration efforts aimed at restoring their ecological functions and aesthetic value [1–3]. Common in-lake restoration strategies include chemical treatment (e.g., phosphorus precipitation) and biomanipulation, such as manipulating fish communities and reintroducing submerged vegetation, typically conducted after external nutrient loading reduction to speed up recovery [3–5]. Submerged macrophytes are particularly crucial in restoring shallow lakes from a turbid to a clear-water state, as they can stabilize sediments, reduce nutrient resuspension, suppress phytoplankton growth by shading and nutrient competition, and provide structural habitats for aquatic fauna [6, 7]. However, the recovery of submerged macrophytes remains inconsistent [4, 6]. In some lakes, macrophyte communities have successfully re-established and maintained a clear-water state, whereas in others, poor growth or limited expansion of submerged macrophytes has constrained their ecological role [8, 9]. As plant biomass and coverage are pivotal factors in determining the success of restoration, the inability to achieve adequate levels of biomass and coverage underscores the necessity of elucidating the mechanisms governing macrophyte biomass throughout the restoration process [10, 11].

Submerged macrophytes are strongly shaped by plant functional traits [12, 13]. These include morphological, physiological, and phenological characteristics affecting plant performance, mediating their resource acquisition and stress tolerance [14–16]. For example, higher biomass is often associated with higher canopy height and larger specific leaf area [13, 17]. However, the expression of these traits is not fixed but modulated by multiple interacting factors, including light climate, nutrient availability from both sediment and water (e.g., nitrogen and phosphorus), and biological pressures such as fish grazing and plant competition [10, 18, 19]. Trait-based approaches offer powerful insights into how plants adapt to dynamic environments [20, 21]. For example, under low-light conditions, Vallisneria natans increased plant height and exhibited leaf expansion with greater leaf length and leaf number to improve light capture [22, 23]. Similarly, submerged macrophytes growing in nutrient-poor sediments often develop longer and thinner roots to increase their surface area for nutrient uptake [24, 25]. These trait-mediated adjustments allow submerged plants to maintain growth and survival under fluctuating environmental constraints, ultimately influencing their dominance and role in stabilizing a clear-water state in shallow lakes.

Importantly, plant traits are interdependent, i.e., changes in one trait often involve trade-offs with others, reflecting fundamental allocation constraints and resource-use strategies [26, 27]. A key trade-off is between above- and below-ground biomass allocation, whereby plants adjust their biomass distribution and morphological investment to optimize resource capture under varying environmental conditions [26, 28]. According to the optimal partitioning theory, plants preferentially allocate biomass to the organ most effective in capturing the scarcest resource [29, 30]. For example, in nutrient-poor sediments, submerged macrophytes often allocate greater biomass to roots to enhance nutrient acquisition from sediments, whereas in nutrient-rich environments they shift allocation toward shoots and leaves to maximize light interception and photosynthetic capacity [31, 32]. Beyond these allocations, plants can also shift along a resource-conservative to resource-acquisitive continuum in response to environmental variation [27, 33]. Conservative traits, such as high carbon investment in structural tissues, promote long-term persistence, while acquisitive traits, such as nutrient-rich tissues, enable nutrient acquisition [17, 34]. Such trait plasticity enables a single species to dynamically adapt both biomass allocation and functional traits to optimize performance across diverse environments. Therefore, exploring these trait trade-offs not only provides valuable insight into how plants respond and adapt to environmental variability; it also helps elucidate how these adjustments influence plant biomass production and allocation.

Submerged macrophyte restoration in subtropical lakes often focus on a few species, in China, for instance, Vallisneria denseserrulata, a meadow-forming pioneer plant widely transplanted into restored lakes due to its rapid clonal growth, high tolerance to environmental variables, and strong nutrient uptake capacity [35]. However, the biomass of V. denseserrulata varies across restored lakes, and the reasons for this remain understudied, constraining the predictability and effectiveness of macrophyte-based lake restoration practices. Here, we quantified how abiotic (sediment and water nutrients, water turbidity, and light conditions and biotic (biomass and number of fish and biomass and diversity of neighbor plants) variables affect the traits of V. denseserrulata and evaluate trait-environment relationships mediating plant biomass in twenty restored units of Lake Yiai, China. We hypothesized that the biomass of V. denseserrulata would be influenced by both abiotic and biotic variables primarily through indirect modulation of plant functional traits. We anticipated that variations in abiotic and biotic variables would shape the expression of plant functional traits through trade-offs in the allocation of resources between above- and below- ground parts, as well as through shifts along the continuum from resource-conservative-to-resource-acquisitive strategies. By linking environment-trait-biomass relationships, we established a trait-based framework for understanding the growth mechanisms of V. denseserrulata and provide practical guidance to improve restoration effects in degraded shallow lakes.

Materials and methods

Study area and field sampling

Lake Yiai (30°25′−30°29′N, 114°50′−115°05′E) is a shallow subtropical urban lake located in Huangzhou District, Huanggang City, Hubei Province, China, with a total area of approximately 320 ha. In March 2021, the status of water quality (in 20 sampling sites), sediments (in 28 sampling sites), and submerged macrophytes (at 20 sampling sites) was investigated across Lake Yiai before restoration. Mean Secchi depth (SD) was 0.3 ± 0.1 m and mean depth (WD) 2.3 ± 0.4 m. Water total nitrogen concentration (WTN), total phosphorus concentration (WTP). and chlorophyll a (Chl a) were 9.6 ± 4.8 mg/L, 0.5 ± 0.4 mg/L, and 40 ± 27 µg/L, respectively. Sediment nitrogen content (STN) and phosphorus content (STP) were 3.6 ± 1.4 mg/g and 1.9 ± 0.7 mg/g, respectively, while sediment organic matter (SOM) was 10.5 ± 2.9%. Overall, water and sediment properties in Lake Yiai exhibited spatial variability (Table S1; coefficient of variation, CV > 30%). However, submerged macrophytes were not observed at the sampling sites.

Since March 2022, an ecological restoration project has been ongoing in Lake Yiai. Twenty units of the lake were isolated using water-proof enclosures, ranging in size from 0.9 to 13.3 ha (6.7 ha in average) and with a total restored area of 130 ha (~ 40% of the lake surface area) (Fig. 1). These restored units were mainly located along the lake’s littoral zone and had a mean depth of 2.1 m (range: 1.2–2.9 m), generally providing suitable conditions for submerged macrophyte restoration. In each restored unit, more than 95% of the fish, mainly omnivorous fish (e.g., Carassius gibelio and Cyprinus carpio), were removed by trap nets. Thereafter, submerged macrophytes, primarily V. denseserrulata, were transplanted across the entire restored area of each unit, at a density of 80 ind./m² and 50 ind./m² in total for other species (Hydrilla verticillata, Potamogeton malaianus, and Myriophyllum spicatum). The total number of plants of each species in each unit was pre-calculated, and transplantation was carried out using a broadcasting method, where plants (roots) were wrapped in sediment and then thrown into the water [36]. During transplantation, efforts were made to distribute the plants as evenly as possible across each unit according to the planned density. Consequently, V. denseserrulata and the other submerged macrophyte species were mixed and widely distributed throughout each unit, avoiding strict species zonation. All the plants were nursery-cultivated and supplied by Guangzhou Beishan Aquatic Eco-Science and Technology Ltd, with consistent growth prior to transplantation (V. denseserrulata shoot height: 25–35 cm; shoot height of other species: 40–50 cm).

Fig. 1.

Distribution of the twenty restoration units in Lake Yiai

Field sampling was conducted in each restored unit from August 25 to September 3, 2023, about three months after finishing plant transplantation. Temperature (T), dissolved oxygen (DO), specific conductivity (SPC), and pH in the water were 29.9 ± 1.8 °C, 6.7 ± 1.2 mg/L, 0.3 ± 0.03 mS/cm, and 8.5 ± 0.6, respectively. Submerged macrophytes were collected using a grab-type plant sampler with an open area of 0.19 m² (length: 50 cm, width: 38 cm) [37]. The sampler was lowered vertically to the sediment surface, closed to enclose the vegetation, and then lifted, collecting the entire plant including roots. In each restored unit, sampling was conducted along three transects spaced approximately 150 m apart. For each transect, three sampling sites were established: one in the middle and two approximately 5 m from the shoreline. Since the plants were planted as evenly as possible across each unit, this arrangement ensured that sampling captured within-unit spatial heterogeneity. Macrophytes collected from all sampling sites in the same units were pooled and subsequently identified based on Flora of China [38] and iPlant (https://www.iplant.cn/) to species level for further analysis.

Water depth (WD) was measured using a depth measuring rod, while SD was determined with a Secchi disc at the middle of each transect. After this, the ratio of SD to WD (SD/WD) was calculated. If the Secchi disc hit the bottom, the SD/WD ratio was set to 1.

Water and sediment samples were taken simultaneously in the middle of each transect. Water from the upper 30 cm was collected using a 5 L acrylic water sampler from the three sampling sites along each transect. These samples were subsequently thoroughly mixed, and 1-L water samples were taken to the laboratory for further analysis. Surface sediment (0–15 cm) was collected using a Peterson grab sampler from the same three sites, combined into a single composite sample per transect, and stored in a cooling box in the field.

Fish were sampled in each restoration unit using a multi-mesh gill net with eight mesh sizes (5, 10, 15, 20, 25, 30, 35, and 40 mm; each 10 m in length and 1.5 m in height), with the 5–15 mm sections targeting pelagic fish and the larger meshes benthic fish to ensure sampling fish of different sizes [39]. The nets were randomly placed in each unit for 2–3 h and then retrieved.

Sample analysis of submerged macrophytes

Biomass and morphological and stoichiometric traits of V. denseserrulata were measured to determine plant growth performance, resource allocation patterns, and nutrient status under the varying environmental conditions in the different units. Biomass was calculated in wet weight (WW), measured with an electronic balance, and expressed as WW gram per square meters (g WW/m²). For each transect, 15 V. denseserrulata individuals were randomly selected from the mixed samples (three sites per transect). Maximum leaf length (LLen), leaf width (LWid), and root length (RLen) were measured using a ruler, and root-to-leaf length ratio (RLratio) was subsequently calculated. The numbers of leaves and roots (LNum, and RNum) of each individual plant were counted manually. Finally, the plants were separated into shoots (leaves and stems) and roots and oven-dried at 60–70℃ to constant weight, followed by root-to-shoot ratio (RSratio) calculation. Dried samples of shoots and roots were mixed and ground to determine the contents of carbon (PC), nitrogen (PN), and phosphorus (PP) of the whole plant. Plant PC and PN were determined using an elemental analyzer (Flash EA 1112 series, CE Instruments, Italy), while PP was measured using sulfuric acid/hydrogen peroxide digestion and ammonium molybdate methods [40]. All plant elemental contents were expressed on a dry weight basis. Stoichiometric mass ratios between elements, including the carbon to nitrogen ratio (PCN), the carbon to phosphorus ratio (PCP), and the nitrogen to phosphorus ratio (PNP), were subsequently calculated.

The biomasses of neighbor macrophytes (NBiomass), including H. verticillata, P. malaianus, M. spicatum, and Ceratophyllum demersum, were measured using an electronic balance and expressed as grams per square meter (g WW/m²). The diversities of neighbor macrophytes (NShannon) were quantified using the Shannon-Wiener index based on species biomass rather than species individual counts [11]. The data are given in Supplementary materials Table S2 and Figure S1.

Water and sediment sample analysis

Water WTN, WTP, Chl a, and total suspended solids (TSS) concentrations were measured following standard methods [41]. For WTN, we used alkaline K₂S₂O₈ digestion UV spectrophotometry, and WTP was determined following the ammonium molybdate spectrophotometric method after digestion with K₂S₂O₈. Chlorophyll-a concentration was determined by acetone extraction and spectrophotometry. Total suspended solids were quantified gravimetrically.

Sediment STN and STP were determined after K₂S₂O₈ digestion, STN was measured by UV spectrophotometry, and STP using the ammonium molybdate spectrophotometric method. Sediment SOM was determined by loss on ignition. Samples were oven-dried at 105 °C to constant weight and then combusted at 550 °C for four hours. Percentage weight loss was calculated as SOM. Sediment nutrient contents were expressed on a dry weight basis. Details on water and sediment are provided in Supplementary materials Table S2 and Figure S1.

Fish catch analysis

After removing all fish from the net (see above), the fish were weighed using an electronic balance, and the number of individuals was counted manually. Fish catch was recorded as biomass per unit effort (BPUE, g/net/h) and number per unit effort (NPUE, ind./net/h). Fish BPUE or NPUE were calculated by dividing total fish biomass or total number of fish caught in each net by the effective net soaking time. Effective net soaking time was defined as the total time the nets were deployed in the water, excluding handling time during deployment and retrieval. Data on the composition and catch of each fish species can be found in Supplementary materials Table S2 and Table S3.

Statistical analysis

Pearson correlation analysis was performed to examine the relationships among plant traits and among environmental variables. Principal component analysis (PCA) was performed on all plant traits to reduce dimensionality and visualize trait covariation. The first two principal components (PC1 and PC2) were extracted and used for further analysis. The loadings of each trait on these components, as well as their relative contributions, are presented in Table S4. Relationships between plant biomass or plant trait PCs and environmental variables were assessed using partial least squares regression (PLSR), which is particularly suitable for addressing multicollinearity among environmental variables. Variable importance in projection (VIP) scores was calculated, with variables showing VIP > 1 considered influential environmental factors. For these variables, Pearson correlation analysis was used to examine relationships between plant biomass or trait PCs and individual environmental variables (Figure S2), and generalized linear models (GLMs) with Gaussian distribution were then fitted, including two-way interactions to evaluate combined effects. For plant traits, significant environmental variables were also projected onto the PCA biplot of plant traits using the envfit function to visualize their influence on trait structure. Partial least squares path modeling (PLS-PM) was applied to quantify direct and indirect effects among sediment (STN, STP, SOM), water (WTN, WTP, TSS, Chl a, SD, WD, SD/WD), fish (NPUE, BPUE), neighbor plants (NBiomass, NShannon), plant traits (PC1, PC2), and biomass. Variance inflation factor (VIF) was calculated within each group of environmental variables to check for potential multicollinearity (Table S5). Bootstrapping was employed to test path significance, and goodness-of-fit values were calculated to evaluate model performance. All statistical analyses were conducted using R (v4.3.1). Prior to analysis, the variables were log-transformed to improve normality and homoscedasticity.

Results

Vallisneria denseserrulata biomass and environmental variables

The biomass of V. denseserrulata varied substantially across the restored units – from 25 to 5667 g WW/m², with a mean biomass of 3130 g WW/m² (Fig. 2a). Biomass was primarily influenced by SOM, STN, NBiomass, BPUE, WTN, TSS, and Chl a (VIP > 1 for all variables, Fig. 2b). Specifically, biomass increased significantly with SOM (Fig. 2c) but decreased markedly with both WTN and BPUE (Fig. 2d, e). Two-way interactions showed that high SOM mitigated the negative effects of WTN (SOM × WTN: Estimate = 8.0) and BPUE (BPUE × SOM: Estimate = 3.9, Table S6) on biomass, whereas simultaneously high WTN and BPUE further reduced the biomass (BPUE × WTN: Estimate = −3.7, Table S6).

Fig. 2.

Vallisneria denseserrulata biomass and its relationship with environmental variables. a Spatial variation in plant biomass across sampling sites. b Environmental variables with variable importance in projection (VIP) > 1 identified by partial least squares regression (PLSR), with their effects evaluated using Pearson correlation analysis (red, positive; blue, negative). Asterisks denote significance levels (*P < 0.05; **P < 0.01). c–e Linear relationships between plant biomass and sediment organic matter (SOM), water total nitrogen (WTN), and fish biomass per unit effort (BPUE). Abbreviations: WTN, water total nitrogen; TSS, total suspended solids; Chl a, chlorophyll a; STN, sediment nitrogen content; SOM, sediment organic matter content; BPUE, fish biomass per unit effort; NBiomass, neighbor plant biomass

Variation and correlations of plant traits

Leaf morphological traits were relatively stable across the restored units, CV in LLen (15%), LWid (12%), and LNum (13%) being relatively low. However, root traits varied largely relative to RLen (19%) and RNum (26%) (Table 1). Plant PC and PN varied insignificantly (PC: mean 307 mg/g, CV 13%; PN: mean 25 mg/g, CV 14%), whereas PP exhibited marked variations (4.2 mg/g, CV 31%), resulting in greater variation in PCP and PNP (36–38%) (Table 1). Biomass allocation also differed among sites, with the RSratio ranging from 0.17 to 0.46 (mean 0.25 g/g, CV: 27%) and the RLratio from 0.13 to 0.21 (mean 0.16 cm/cm, CV: 16%).

Table 1.

Summary statistics of plant biomass, morphological traits, stoichiometric traits, and allocation ratios of V. denseserrulata. Values represent minimum (Min), maximum (Max), mean, standard deviation (Sd), and coefficient of variation (CV, %). Plant elemental contents and RSratio are expressed on a dry weight basis

Trait	Abbreviation	Unit	Min (Site)	Max (Site)	Mean	Sd	CV (%)
Maximum leaf length	LLen	cm	47.6 (U14)	91.5 (U16)	76.7	11.8	15.3
Maximum leaf width	LWid	cm	0.6 (U20)	0.8 (U11)	0.7	0.1	12.1
Leaf number per shoot	LNum	-	8 (U14)	15 (U12)	11	1	13.2
Maximum root length	RLen	cm	8.8 (U1)	17.2 (U17)	12.2	2.3	18.6
Root number per shoot	RNum	-	34 (U14)	107 (U12)	67	17	25.9
Plant carbon content	PC	mg/g	204.1 (U15)	355.3 (U2)	306.6	38.3	12.5
Plant nitrogen content	PN	mg/g	19.1 (U17)	31.8 (U7)	25.1	3.5	14.1
Plant phosphorus content	PP	mg/g	2.4 (U4)	7.3 (U14)	4.2	1.3	31.1
Plant carbon to nitrogen ratio	PCN	mass ratio	8.4 (U15)	16.5 (U18)	12.4	2.2	17.4
Plant carbon to phosphorus ratio	PCP	mass ratio	31.7 (U15)	137.0 (U2)	80.4	28.8	35.8
Plant nitrogen to phosphorus ratio	PNP	mass ratio	3.6 (U16)	11.3 (U4)	6.6	2.5	38.4
Root-to-shoot biomass ratio	RSratio	g/g	0.17 (U11)	0.46 (U19)	0.25	0.07	27.4
Root-to-leaf length ratio	RLratio	cm/cm	0.13 (U11)	0.21 (U20)	0.16	0.03	16.4

Leaf traits were positively correlated, while RLen was negatively related to PN and PNP but positively to PCN (P < 0.05, Fig. 3b). The RSratio decreased with leaf traits, whereas the RLratio increased with RLen and declined with PNP (P < 0.05, Fig. 3b). Plant PC was negatively correlated with PP and positively with all elemental ratios (P < 0.05, Fig. 3b).

Fig. 3.

(a) Heatmap showing standardized values of plant functional traits across restored units (U1-U20) and (b) Pearson’ s relationship among plant traits. In (a), each cell represents the relative value of a trait at a given site, red reflecting higher values and blue lower values. In (b), positive and negative correlations are indicated by red and blue, respectively, with circle size and color intensity representing correlation strength. Asterisks denote significance levels (*P < 0.05; **P < 0.01; ***P < 0.001). Abbreviations: LLen, maximum leaf length; LWid, maximum leaf width; LNum, leaf number per shoot; RLen, maximum root length; RNum, root number per shoot; PC, PN, PP, plant carbon, nitrogen, and phosphorus contents, respectively; PCN, PCP, PNP, plant carbon to nitrogen, carbon to phosphorus, and nitrogen to phosphorus stoichiometric ratios, respectively; RSratio, root-to-shoot biomass ratio; RLratio, root-to-leaf length ratio

Trade-offs among plant traits and variables

Principal component analysis (PCA) results for plant traits showed that the first two principal components (PC1 and PC2) explained 36% and 27% of the variance, respectively (in total 63%) (Fig. 4a).

Fig. 4.

(a) Principal component analysis (PCA) biplots of plant traits. Axes indicate the first two principal components (PC1 and PC2) with variance explained. (b) PCA with significant environmental variables fitted using the envfit function. (c-d) Loadings of each trait on PC1 and PC2, sorted by contribution from largest to smallest, indicating the relative influence of individual traits on the respective axes. (e-f) Relationships between plant trait PC1, PC2 and environmental variables based on partial least squares regression (PLSR). Environmental variables with VIP > 1 were identified as key drivers and their effects assessed using Pearson correlation analysis (red, positive; blue, negative). Asterisks denote significance levels (*P < 0.05; **P < 0.01; ***P < 0.001). Abbreviations: LLen, maximum leaf length; LWid, maximum leaf width; LNum, leaf number per shoot; RLen, maximum root length; RNum, root number; PC, PN, PP, plant carbon, nitrogen, and phosphorus contents, respectively; PCN, PCP, PNP, plant carbon to nitrogen, carbon to phosphorus, and nitrogen to phosphorus stoichiometric ratios, respectively; RSratio, root-to-shoot biomass ratio; RLratio, root-to-leaf length ratio. WD: water depth; WTN: water total nitrogen; TSS: total suspended solids; Chl a, chlorophyll a; STN: sediment nitrogen content; STP: sediment phosphorus content; SOM: sediment organic matter; BPUE: fish biomass per unit effort

PC1 represented the acquisitive-conservative traits trade-off. It loaded positively with PP, RLen, and LWid (acquisitive) and negatively with PC and PN, PCP, and PNP (conservative) (Fig. 4c). Sediment nitrogen content, STP, SOM, WD, and BPUE were negatively correlated with PC1, favoring conservative traits, while Chl a was positively correlated with PC1, favoring acquisitive investment (Fig. 4d). Only STN and SOM were significantly correlated with PC1 (Fig. 4e), and their interactions had no significant effect on PC1 (Table S6).

PC2 represented the aboveground-belowground trade-off axis, which exhibited negative correlations with leaf morphological traits, such as LNum, LLen, and LWid (aboveground), and positive correlations with PN, PP, RSratio, and RLratio (belowground) (Fig. 4d). Water nitrogen concentration, TSS, Chl a, and BPUE were positively correlated with PC2, promoting belowground allocation, while SOM and STN were negatively correlated with PC2, favoring aboveground investment (P < 0.05, Fig. 4f). The SOM × TSS interaction was negative (Estimate = −7.9), indicating a shift toward more aboveground traits, while TSS × WTN (Estimate = 7.5), BPUE × TSS (Estimate = 2.5), and BPUE × WTN (Estimate = 3.5) were positive, suggesting increased investment in belowground traits under these combined conditions (Table S6).

Influencing pathways of various driving factors on plant biomass

The results of partial least squares path modeling (PLS-PM) showed insignificant direct effects of fish (positively associated with NPUE and BPUE), sediment (positively with STN, STP, and SOM), water (positively with WTN, WTP, TSS, Chl a, and WD; negatively with SD and SD/WD) on macrophyte biomass (Fig. 5a and b). However, these factors significantly affected the traits of V. denseserrulata, which in turn indirectly determined macrophyte biomass (Fig. 5a; Table 2). In contrast, neighboring plants had no significant direct or indirect effects on V. denseserrulata biomass (Fig. 5a).

Fig. 5.

(a) Partial least squares path modeling (PLS-PM) illustrating the hypothesized causal relationships across environmental variables (fish, sediment, water, neighbor plants), plant traits, and plant biomass. Edge colors indicate the sign of standardized path coefficients (red for positive, blue for negative), line styles denote significance (solid lines: significant, dashed lines: non-significant), and values are standardized direct effects. R² values within nodes indicate the proportion of variance explained for each latent variable. (b) PLS-PM outer model displays the relationships between latent variables and their measured indicators (manifest variables). Ellipses represent observed variables with loadings (red for positive, blue for negative)

Table 2.

Standardized direct, indirect, and total effects in the PLS-PM analysis showing the influence of environmental drivers (fish, sediment, water, neighbor plants) and plant traits on biomass and interrelations among latent variables. Values are standardized path coefficients

Relationship	Direct	Indirect	Total
Water
Sediment → Water	0.09	0	0.09
Fish → Water	0.58	−0.01	0.57
Sediment
Fish → Sediment	−0.15	0	−0.15
Neighbor
Water → Neighbor	0.19	0	0.19
Sediment → Neighbor	−0.28	0.02	−0.27
Fish → Neighbor	0.14	0.15	0.29
Trait
Water → Trait	0.44	0.03	0.47
Sediment → Trait	−0.61	0	−0.61
Fish → Trait	−0.01	0.39	0.39
Neighbor → Trait	0.17	0	0.17
Biomass
Water → Biomass	0	−0.32	−0.32
Sediment → Biomass	0.04	0.42	0.46
Fish → Biomass	−0.16	−0.3	−0.46
Neighbor → Biomass	−0.19	−0.1	−0.29
Trait → Biomass	−0.61	0	−0.61

Sediment had a pronounced negative impact on traits (path coefficient = −0.61), fostering more conservative characteristics (PC1) and traits oriented toward aboveground growth (PC2), ultimately enhancing biomass (indirect effect = 0.42, Fig. 5a; Table 2). In contrast, fish and water exerted a negative impact on biomass through water-mediated effects on traits. Notably, water had a significant positive effect on traits (path coefficient = 0.44), resulting in more acquisitive characteristics (PC1) and greater belowground-oriented traits (PC2), which in turn indirectly diminished biomass (Fig. 5a; Table 2).

Discussion

This study elucidated the shaping of V. denseserrulata biomass by environmental variables through determination of plant functional traits in twenty restored units in urban Lake Yiai. As hypothesized, water, sediment, and fish variables indirectly influenced biomass by modulating plant functional traits rather than through direct effects. Nutrient-rich sediments promoted resource-conservative traits and aboveground allocation, increasing biomass, whereas elevated water WTN and TSS and higher fish BPUE shifted traits toward belowground allocation and acquisitive strategies, reducing biomass (Fig. 6). However, neighboring macrophytes had no significant direct or indirect effects on the biomass of V. denseserrulata, which is inconsistent with the proposed hypothesis.

Fig. 6.

Schematic diagram illustrating the trait-mediated effects of environmental variables on V. denseserrulata biomass during the restoration of Lake Yiai. Enhanced sediment nutrient availability promotes conservative functional traits and aboveground biomass allocation, thereby increasing the overall plant biomass. Conversely, elevated water nutrient concentrations, turbidity, and fish biomass prompt acquisitive traits and belowground allocation, leading to biomass reduction

Plant trait variations and trade-offs

Our study revealed distinct patterns in V. denseserrulata trait variation and trade-offs across the restored units, reflecting coordinated adjustments in resource acquisition and allocation strategies under heterogeneous environmental conditions. Plant PC and PN were relatively stable (CV: 12‑14%), whereas phosphorus-related stoichiometric traits, such as PP, PCP, and PCP (CV > 30%), were unstable. Carbon, primarily constituting structural compounds such as cellulose and lignin, remained relatively stable to maintain tissue integrity and basic plant architecture, whereas phosphorus, ubiquitous in cellular metabolism, played a crucial role in supporting plant growth and metabolic processes [42, 43]. In our study, the plant P content correlated significantly and negatively with PC and PCP (Fig. 3b), and these traits exhibited clear trade-offs along the first principal component axis (PC1, Fig. 4a), indicating a resource trade-off between nutrient acquisition and carbon conservation. According to the Growth Rate Hypothesis (GRH), high phosphorus content generally supports high growth rates. However, the plants with high phosphorus content in our study did not exhibit high biomass. This discrepancy can be explained by relatively higher limitation of nitrogen than of phosphorus [29, 42, 43], as indicated by the low sediment nitrogen to phosphorus molar ratio (mean = 7.1). Moreover, GRH was primarily developed for algae and zooplankton and did not account for the effects of size or structural carbon allocation on higher plants [44]. With increasing plant size, nutrient contents decreased due to dilution, while carbon accumulation continued to rise to support the larger-sized plants [45, 46].

Our results revealed that the root traits of V. denseserrulata exhibited greater variability than leaf morphological traits (Table 1), highlighting the pivotal role of belowground traits in mediating plant performance. As a clonal submerged macrophyte species, V. denseserrulata depends on a robust and extensive root system to access sediment nutrients, not least under competitive or nutrient-depleted conditions [25, 35]. In our study, the relative scarcity of nitrogen in sediment compared to phosphorus likely led to adjustments in root traits to optimize the sediment nitrogen uptake, thereby enabling the plant to effectively cope with spatial heterogeneity in resource distribution [25, 47]. PC2 represents a trade-off between aboveground and belowground allocation. Higher PC2 scores were associated with increased RSratio and RLratio. When plants allocate more biomass to roots, they also adjust their leaf morphology, likely to balance nutrient acquisition with photosynthetic capacity under heterogeneous environmental conditions [26, 28]. These trade-offs are consistent with optimal partitioning theory, which predicts that plants allocate a greater amount of biomass to the organ responsible for acquiring the most limiting resource [29, 30]. Collectively, these results suggest multidimensional trait plasticity of V. denseserrulata, manifested via conservative-acquisitive strategies (PC1) and aboveground-belowground allocation trade-offs (PC2), allowing adaptation to diverse environments.

Trait-mediated links between environment and biomass

The biomass of V. denseserrulata was related to sediment nutrients, water quality, and fish, and these environmental variables affected macrophyte growth indirectly through functional traits rather than through direct effects. These findings are consistent with meta-analyses showing that plant functional traits mediate biomass responses to global environmental changes [12, 13]. In our study, sediment STN and SOM emerged as the most significant positive drivers of biomass. Enhanced sediment nutrients provide a stable and readily available nutrient source that supports V. denseserrulata growth and biomass accumulation [18, 48, 49]. Nutrient-rich sediments promoted resource-conservative traits, which were characterized by increased tissue PC and PCP, as well as greater aboveground allocation (longer and more numerous leaves; Fig. 4). These trait adjustments are consistent with previous observations that nutrient-rich sediments encourage investment in structural tissues and aboveground organs, enhancing light interception and photosynthetic capacity [16, 28]. This alternation in biomass allocation probably directly contributed to the greater whole-plant biomass, as larger leaf size and enhanced structural robustness improve the carbon gain and competitive ability in submerged environments [26, 50].

Elevated WTN and TSS lead to lower V. denseserrulata biomass, likely by altering trait trade-offs to greater belowground allocation (RSratio and RLratio) as well as plant PN and PP, as observed. Elevated nutrients, particularly nitrogen, in the water column was associated with increases in PN and PP, indicating enhanced nutrient uptake from the water column [18]. However, when water nutrient concentrations surpass optimal levels, plant growth may be inhibited, as observed in mesocosm experiments [49]. This inhibition might result from stimulation of the proliferation of phytoplankton and filamentous green algae, which compete for light and nutrients [19, 51]. In our restored units, positive correlations between WTN and WTP and Chl a (Figure S1) provided indirect evidence supporting this mechanism. Furthermore, high TSS exacerbates light limitation by attenuating underwater irradiance, restricting photosynthetic carbon assimilation [22], as indicated by the significant negative correlation between TSS and SD in our restored units (Figure S1). Under these conditions, V. denseserrulata demonstrates a trait-mediated allocation strategy biased toward belowground biomass, contrary to classical optimal partitioning theory, which would predict increased shoot allocation under light limitation [29, 30]. V. denseserrulata has low light requirement and can sustain basic photosynthetic activity under low irradiance [23, 52]. Instead of investing in aboveground growth, which may be inefficient under persistent shading, the plant reallocates biomass to belowground structures and nutrient-rich tissues to optimize the acquisition of the sediment nutrients that remain accessible [23, 32]. This adaptive allocation may help maintain metabolic activity, support clonal propagation, and store resources for survival and future regrowth. However, it may reduce aboveground growth, which in turn will limit leaf photosynthesis and the overall carbon gain, ultimately reducing whole-plant biomass.

Fish BPUE showed a negative relationship with V. denseserrulata biomass through trait-mediated pathways. In Lake Yiai, the fish assemblage was composed of planktivorous species (e.g., Hypophthalmichthys molitrix, Hemiculter leuciclus), omnibenthivorous species (e.g., C. gibelio, C. carpio), herbivorous species (e.g., Parabramis pekinensis), and piscivores (e.g., Culter alburnus) (Table S3). Herbivorous fish may feed directly on aboveground plant tissues, leading to changes in functional traits such as reduced shoot growth and increased root allocation [53]. This shift in biomass distribution reduces the plant’s aboveground photosynthetic capacity. Benthivorous fish disturb sediments and resuspend nutrients, thereby indirectly altering water quality and light availability. These changes ultimately trigger trait adjustments and constrain plant growth [54, 55]. In our study, fish, especially BPUE, influenced the biomass of V. denseserrulata indirectly by deteriorating the water quality, resulting in changes in plant traits (Fig. 5), rather than via direct herbivory. These trait adjustments involved greater belowground allocation and increased tissue nutrient contents, as evidenced by a higher RSratio and elevated PN and PP. Collectively, these alterations restricted the aboveground photosynthetic capacity and ultimately led to a decrease in whole-plant biomass [18, 55].

Neighboring plants, primarily canopy-forming species, showed no significant direct or indirect effects on V. denseserrulata biomass, likely due to their relatively low biomass, averaging only ~ 20% of the V. denseserrulata biomass (Table S2). The low biomass of neighboring plant species (canopy-forming species) and a different vertical spatial distribution relative to V. denseserrulata (a rosette-type species) may limit their ability to exert strong suppressive effects [56]. Moreover, environmental factors, such as sediment nutrient content, water quality, and fish, had a significant influence on V. denseserrulata growth, which could also mask any potential effects of these neighboring macrophytes [57]. Nevertheless, the observed negative effects of neighboring plants (high biomass and low density) on the biomass of V. denseserrulata may favor V. denseserrulata’s belowground allocation and suppress its growth (Fig. 5). Therefore, control of the biomass of neighboring plant species, especially canopy-forming macrophytes, may be considered, while maintaining moderate plant diversity in the early phase of lake management after lake restoration until vegetation is well-established [6, 10].

Implication for lake restoration

Our findings emphasize the importance of sediment quality, water conditions, and fish-mediated interactions for influencing functional trait responses and promoting V. denseserrulata growth during summer restoration in subtropical lakes. Moreover, an enhanced sediment nutrient content may increase plant biomass and buffer the negative effects of water nutrient enrichment and fish activity, whereas the combined impact of high nutrient levels and fish biomass can exacerbate biomass decline. Therefore, successful lake restoration in subtropical lakes requires integrated management strategies [3, 4, 58]. Reducing external nutrient inputs through watershed-scale controls remains essential for limiting the overall nutrient loading and preventing further eutrophication [8, 58], which helps V. denseserrulata develop key functional traits such as longer and more leaves to effectively capture light. Additionally, managing fish communities to reduce the populations and biomass of benthivorous species is critical for minimizing sediment resuspension and enhancing water clarity, which in turn promotes aboveground growth and maintains balanced nutrient allocation in plants [3, 54]. Overall, these measures promote a favorable expression of plant functional traits, supporting a robust V. denseserrulata biomass and enhancing ecosystem resilience in subtropical lake.

While numerous studies have explored the relationships between plant functional traits and environmental variables [20, 23], trait-based approaches have rarely been integrated into the evaluation and guidance of submerged macrophyte restoration. Existing restoration efforts in both subtropical and temperate lakes have largely focused on either macrophyte coverage or standing biomass [3, 6, 8]. However, when pronounced declines in biomass or coverage become evident, restoration processes have often already failed. In this study, we found that environmental factors primarily influenced plant biomass indirectly through their effects on functional traits (Fig. 5), highlighting traits as key mechanistic links between environmental variability and biomass responses. As such, functional traits can serve as early-warning indicators of restoration performance, offering greater foresight and sensitivity than plant biomass alone [59, 60]. From a practical restoration perspective, synchronous monitoring of key functional plant traits (e.g., PC and PP) can help detect environmental stress early and adjust restoration practices in time. Moreover, our results provide new insights into plant material selection for restoration from a functional trait perspective. Reverse screening and pre-cultivation of V. denseserrulata populations, whose trait match local environmental conditions (e.g., high RSratio of V. denseserrulata for high WTN and TSS conditions), may help maintain a more stable trait expression after transplantation. Such trait-informed selection has the potential to buffer biomass fluctuations, reduce risks associated with excessive trait plasticity, and ultimately enhance the resilience and sustainability of submerged macrophyte restoration.

However, some limitations should be acknowledged. Our study was conducted in summer at one location (latitude). Previous studies have demonstrated that temperature strongly regulates plant functional traits and growth [61, 62]. For example, Stuckenia pectinata exhibited higher canopy height, LNum, and biomass in the warmer Mersin than in the colder Ankara region of Turkey, and warming experiments showed that trait responses were more pronounced in Ankara [63]. Consequently, the environment-trait-biomass relationships observed here may differ from those in lakes situated in other climatic zones. In temperate lakes, a shorter growing season and a lower summer temperature may constrain trait plasticity and lead to different trait trade-offs. In addition, differences in dominant macrophyte species and fish assemblages across climate zones may further modulate these relationships [64]. Future studies comparing subtropical and temperate lakes are therefore needed to test the generality of trait-based restoration frameworks and to refine trait-guided management strategies in different lake types.

Conclusion

This study highlights the critical role of sediment quality, water conditions, and fish-mediated interactions in influencing the functional traits and biomass of V. denseserrulata during the restoration of Lake Yiai. Enhanced sediment nutrient availability drove a shift toward resource-conservative traits and increased allocation to aboveground biomass, which facilitated an overall higher plant biomass. In contrast, elevated water nutrient, turbidity, and fish biomass had a negative impact on plant biomass, primarily through water quality degradation and trait-mediated pathways, including increased belowground allocation as a stress response. Our findings underscore the pivotal role of plant functional traits as mediators of environmental effects on macrophyte performance. Therefore, a trait-based framework provides valuable insights for optimizing submerged macrophyte restoration and promoting sustainable subtropical lake rehabilitation, for instance by monitoring key functional traits as early-warning indicators and selecting plant populations with suitable trait combinations for restoration.

Abbreviations

Min

Minimum

Max

Maximum

Sd

Standard deviation

CV

Coefficient of variation

SD

Secchi depth

WD

Water depth

SD/WD

Secchi depth to water depth ratio

WTN

Water total nitrogen concentration

WTP

Water total phosphorus concentration

Chl a

Chlorophyll a

TSS

Total suspended solids

STN

Sediment nitrogen content

STP

Sediment phosphorus content

SOM

Sediment organic matter

NPUE

Fish number per unit effort

BPUE

Fish biomass per unit effort

NBiomass

Biomass of neighbor plants

NShannon

Diversity of neighbor plants

Lleaf

Maximum leaf length

LWid

Maximum leaf width

LNum

Leaf number per shoot

RLen

Maximum root length

RNum

Root number per shoot

PC

Plant carbon content

PN

Plant nitrogen content

PP

Plant phosphorus content

PCN

Plant carbon to nitrogen ratio

PCP

Plant carbon to phosphorus ratio

PNP

Plant nitrogen to phosphorus ratio

RSratio

Root-to-shoot biomass ratio

RLratio

Root-to-leaf length ratio

WW

Wet weight

T

Temperature

DO

Dissolved oxygen

SPC

Specific conductivity

PCA

Principal component analysis

PC1/PC2

The first two principal components

PLSR

Partial least squares regression

VIP

Variable importance in projection

VIF

Variance inflation factor

GLMs

Generalized linear models

PLS-PM

Partial least squares path modeling

GRH

Growth Rate Hypothesis

Authors’ contributions

ZML conducted the investigation, performed formal analyses, curated the data, and drafted the original manuscript. SF, JWL, and XQY contributed to the investigation. BHG, HH, and ZWL participated in the investigation and contributed to manuscript review and editing. XLH, KYL, QLW, and EJ contributed to manuscript review and editing. JLY conceived the study, acquired funding, conducted the investigation, and contributed to both the original draft and the manuscript review and editing. All authors read and approved the final manuscript.

Funding

This study was funded by the National Key Basic Research and Development Program (2023YFF1304501), National Natural Science Foundation of China (42277067), Jiangxi Provincial Natural Science Foundation (20242BAB23063), Jiangsu Provincial Science and Technology Planning Project (No. BK20231516), and State Key Laboratory of Lake and Watershed Science for Water Security (NKL2023-KP02). Erik Jeppesen was supported by the Yunnan Provincial Council of Academicians and Experts Workstations (202405AF140006).

Data availability

Data will be made available on request.

Declarations

Ethics approval and consent to participate

This study was part of ecological monitoring under the Huanggang Lake Yiai Water Environment Comprehensive Restoration Project (Project ID: 20204211027701023095), authorized by the Huanggang Development and Reform Commission of Hubei Province. All sampling was carried out with the appropriate permissions from the local government, following national ethical standards and relevant institutional guidelines. Submerged macrophytes were nursery-cultivated, and no wild plants were collected. Fish sampling involved routine ecological monitoring and only common, non-endangered species were sampled. No laboratory experimentation on animals was conducted.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.