Floral symmetry is associated with contrasting hydraulic strategies and trait coordination in angiosperm flowers

Abstract

Floral symmetry is widely recognized as a key driver of angiosperm diversification, yet how actinomorphic and zygomorphic flowers differ in their hydraulic architecture remains poorly understood. Clarifying symmetry-related differences in hydraulic strategies is essential for linking floral function with physiological and evolutionary divergence. Here, we measured 22 floral traits related to hydraulics, structural and reproductive investment across 39 woody species (21 actinomorphic and 18 zygomorphic) to evaluate how floral symmetry shapes hydraulic design. Our results revealed distinct hydraulic differentiation between the two symmetry types. Zygomorphic flowers exhibited significantly longer floral longevity (FL) and lower petal vein density (D_vein) than actinomorphic flowers. They also adopted a hydraulically safer strategy, characterized by more negative water potential at the turgor loss point (ψ_tlp), lower relative water content at turgor loss (RWC_tlp), and higher hydraulic capacitance (C_1,mass). Crucially, after accounting for native climatic variables, divergence in FL, ψ_tlp, and D_vein remained significant, indicating that these traits represent intrinsic features of floral symmetry, whereas C_1,mass and RWC_tlp function primarily as environmentally plastic responses. Functionally, actinomorphic flowers prioritized hydraulic supply capacity, whereas zygomorphic flowers combined enhanced hydraulic safety with extended longevity. Furthermore, zygomorphic flowers showed stronger trait coordination, forming integrated functional modules that likely contribute to greater hydraulic stability under fluctuating environmental conditions. Together, these findings demonstrate that floral symmetry is tightly linked to physiological function and that evolutionary shifts in symmetry may be accompanied by coordinated adjustments in hydraulic safety as well as structural and reproductive investment, thereby supporting divergent pollination strategies and life-history demands.

Highlights

• •Floral symmetry is associated with distinct hydraulic architectures and water-use strategies.
• •Zygomorphic flowers adopt conservative hydraulic strategies, while actinomorphic flowers rely on high vein density.
• •Zygomorphic flowers exhibit tighter hydraulic trait coordination and integration.

1. Introduction

Floral symmetry is a fundamental morphological feature of angiosperms and has played a pivotal role in their ecological adaptation and diversification (Busch and Zachgo, 2009; Citerne et al., 2010). Actinomorphy, characterized by a radial arrangement of floral organs with two or more planes of symmetry, is widely regarded as the ancestral state of angiosperms (Sauquet et al., 2017; Naghiloo, 2020). In contrast, zygomorphy, which possesses a single symmetry plane, has evolved repeatedly across lineages and is considered a key morphological innovation that accelerated angiosperm diversification (Endress, 1999; Sargent, 2004; Citerne et al., 2010; Reyes et al., 2016; Jiang and Moubayidin, 2022). Beyond morphology, floral symmetry is linked to contrasting ecological and physiological functions. The evolution of zygomorphy is thought to have been shaped by co-evolution with specialized pollinators, requiring precise floral orientation and structural reinforcement for accurate pollen placement (Culbert and Forrest, 2016; Yoder et al., 2020). In contrast to actinomorphic species that typically rely on generalized pollination (Fenster et al., 2004; Yoder et al., 2020), these architectural constraints are likely to impose distinct physiological demands related to heat balance, water use, and hydraulic safety, particularly under warmer or more variable climates (van der Kooi et al., 2019; Koski and Galloway, 2020; Wang et al., 2023). Consistent with this view, recent evidence indicates that environmental selection pressures contribute to geographic patterns in floral symmetry, with actinomorphic species declining in frequency toward warmer climates (Wang et al., 2023). If climate filtering acts on floral symmetry through these physiological pathways, native climatic conditions would also influence variation in floral hydraulic traits. Yet, despite the recognized ecological and evolutionary importance of floral symmetry, systematic comparative evidence on how symmetry types differ in hydraulic strategies—the coordinated suite of hydraulic traits related to floral water balance—remains limited.

Water status during anthesis is critical for maintaining floral function and reproductive success. Flowers require a continuous water supply not only to sustain turgor and structural integrity but also to regulate temperature through transpiration, which preserves floral rewards and pollinator attraction (Roddy et al., 2019; Patiño and Grace, 2002). However, compared with leaves, flowers generally possess weaker hydraulic architecture, including lower vein density, reduced hydraulic conductance (Roddy et al., 2013, 2016, 2023; Zhang and Brodribb, 2017), and limited capacity to restrict cuticular water loss (Bourbia et al., 2020; An et al., 2023). As such, floral organs are hydraulically vulnerable, with pedicels embolizing more readily than petioles or stems, allowing plants to shed short-lived flowers during drought to protect long-lived organs (Bourbia et al., 2020; Zhang et al., 2021c). These vulnerabilities are likely to become increasingly consequential under climate change as extreme drought and heat events intensify floral water stress (Teixido and Valladares, 2015; Carins-Murphy et al., 2023). Within this context, the Flower Economics Spectrum (FES) provides a unified framework for interpreting covariation among floral traits and the trade-offs shaping floral construction and function (Roddy et al., 2021). According to this framework, floral longevity and size are tightly coordinated with hydraulic traits. Longer-lived flowers incur higher construction costs and exhibit greater vein and vessel investment (Guo et al., 2023; Li et al., 2023; Paiva and Roddy, 2024; Zhou et al., 2025), and they show enhanced drought tolerance and embolism resistance (Zhang et al., 2017, 2021b). Larger corollas, although increasing pollinator attraction, impose higher carbon and water costs, thereby promoting adaptive strategies such as increased water-use efficiency and reduced minimum epidermal conductance to extend water residence time (Kelly et al., 2008; Lambrecht, 2013; Teixido and Valladares, 2014; Teixido et al., 2016; Paiva and Roddy, 2024).

Although the importance of floral hydraulic traits is widely recognized (Ashman and Schoen, 1994; Galen, 2005; Roddy et al., 2019), comparative analyses of hydraulic structure and function across floral taxa remain limited. Existing studies have focused on only a few major lineages: basal angiosperms maintain floral water balance through high water supply, whereas monocots and eudicots tend to adopt conservative strategies to reduce water loss (Roddy et al., 2016; Ke et al., 2024). Additionally, flowers of aquatic and terrestrial herbaceous species exhibit similar economic spectrum patterns but differ in hydraulically relevant structures such as petal thickness and epidermal cell size (Ke et al., 2023). Notably, floral symmetry is closely associated with environmental adaptability (Citerne et al., 2010; Wang et al., 2023), while hydraulic traits represent core functional traits mediating plant responses to environmental change (Tyree and Ewers, 1991). Recent studies further suggest that zygomorphic flowers may, on average, have larger corollas and longer lifespans than actinomorphic flowers (Li, 2024; Stephens et al., 2024). Within the FES framework, such increases in floral size and longevity would be expected to impose substantial carbon and water costs, theoretically necessitating coordinated adjustments in hydraulic safety and efficiency (Kelly et al., 2008; Lambrecht, 2013; Paiva and Roddy, 2024). Consequently, zygomorphic flowers may be expected to differ in hydraulic architecture from actinomorphic flowers. However, systematic comparative studies integrating hydraulic, economic, and reproductive traits to assess how actinomorphic and zygomorphic flowers differ in hydraulic strategies remain scarce. Addressing this gap is essential for understanding how floral morphology integrates with hydraulic strategies to shape ecological adaptation.

Here, we measured 22 floral traits related to floral water transport, drought tolerance, and reproductive characteristics across 21 actinomorphic and 18 zygomorphic woody species in Xishuangbanna tropical botanical garden, China. Our objective was to test how floral symmetry is associated with hydraulic strategies by addressing two questions: (i) Do actinomorphic and zygomorphic flowers differ in hydraulic and structural traits, and are these differences independent of native climatic conditions? Based on previous evidence indicating that zygomorphic flowers are larger and longer-lived than actinomorphic flowers, we hypothesized that zygomorphic flowers would adopt a more conservative hydraulic strategy. Specifically, we predicted that zygomorphic flowers would exhibit greater drought tolerance (e.g., more negative turgor loss point) and higher hydraulic capacitance. Because environmental selection pressures vary geographically and are known to shape both the distribution of floral symmetry and water availability, we further predicted that native climatic conditions would influence these floral traits. (ii) Do actinomorphic and zygomorphic flowers differ in the coordination among hydraulic, structural, and reproductive traits? Under the flower economics spectrum, floral traits are expected to align along coordinated axes reflecting shared construction and maintenance constraints. Because zygomorphic flowers typically incur greater structural and water-balance demands, we hypothesized that they would show stronger integration among hydraulic, structural, and reproductive traits (e.g., floral size, longevity). Such tighter coordination would reflect stronger functional constraints relative to actinomorphic flowers, which were expected to exhibit weaker or more variable patterns of trait covariation (Fig. 1).

Fig. 1.

Conceptual framework of this study. It was hypothesized that zygomorphic flowers would exhibit larger size, longer longevity, higher construction costs, and enhanced drought tolerance compared to actinomorphic flowers. Driven by greater structural and water-balance demands from their longer longevity, zygomorphic flowers were expected to show stronger trait integration and adopt a hydraulically safer strategy to maintain stability. However, both symmetry types were hypothesized to follow the fundamental hydraulic efficiency–safety trade-off. Trait abbreviations are shown in Table 1.

2. Materials and methods

2.1. Study site and materials

The study was conducted at the Xishuangbanna Tropical Botanical Garden (XTBG), Yunnan Province, China (21°55′N, 101°27′E, 570 m a.s.l.), located on the northern margin of the Southeast Asian tropics. The site experiences a typical tropical monsoon climate, with a mean annual temperature of 21.8 °C and annual precipitation of approximately 1493 mm. Seasonality is pronounced, with a wet season from May to October and a dry season from November to April (Cao et al., 2006).

We studied 21 actinomorphic and 18 zygomorphic species in the common garden at XTBG (Fig. 2 and Table S1), with all species maintained under relatively uniform environmental conditions. Floral symmetry was classified based on a functional-morphological assessment of the entire flower, including the perianth, androecium, and gynoecium, following the criteria outlined by Neal et al. (1998) and Yoder et al. (2020). These species were selected to represent phylogenetically diverse lineages with well-resolved phylogenetic positions, allowing incorporation of phylogenetic information into subsequent analyses. All selected species were abundant and widely distributed within XTBG, and readily accessible for sampling. In addition, species clearly exhibited one of the two floral symmetry types (actinomorphy vs. zygomorphy), which constitutes the core characteristic of interest in this study. From December 2024 to July 2025, we measured 22 floral traits across the 39 study species (Table 1), collecting samples for each species during its peak flowering period. For each species, floral trait measurements were conducted on 3–6 individuals.

Fig. 2.

Phylogenetic relationships of the 39 species studied. The colors of the tips indicate floral symmetry types: blue for actinomorphic species and red for zygomorphic species.

Table 1.

Floral traits measured in this study.

Trait	Abbreviation	Unit	Ecological significance
Floral longevity	FL	day	Floral display and pollinator attraction
Floral area	FA	cm2	Floral display and pollinator attraction
Flower mass per area	FMA	g m−2	Construction cost
Flower dry matter content	FDMC	g g−1	Construction cost
Saturated water content	SWC		Water storage
Petal thickness	PT	mm	Water storage
Hydraulic capacitance before turgor loss	C1,mass	mol H2O kg−1 MPa−1	Water storage
Hydraulic capacitance after turgor loss	C2,mass	mol H2O kg−1 MPa−1	Water storage
Vein density	Dvein	mm mm−2	Water transport
Vessel density	VD	no.mm−2	Water transport
Vessel lumen fraction	VLF		Water transport
Hydraulically weighted vessel diameter	Dh	μm	Water transport
Theoretical hydraulic conductivity	Kt	10−4 kg m−1 s−1MPa−1	Water transport
Vessel wall reinforcement	VWR		Water transport
Vulnerability index	VI		Water transport
Stomatal density	SD	no.mm−2	Water loss and gas exchange
Guard cell length	GCL	μm	Water loss and gas exchange
Minimum epidermal conductance	gmin	mmol m−2 s−1	Water loss
Relative water content at turgor loss point	RWCtlp		Drought tolerance
Saturated osmotic potential	πo	MPa	Drought tolerance
Water potential at the turgor loss point	ψtlp	MPa	Drought tolerance
Elastic modulus	ε	MPa	Drought tolerance

Note: Although FL and FA are classified as reproductive traits based on their primary role in reproduction, they are also influenced by underlying structural and hydraulic properties.

2.2. Floral longevity

Floral longevity (FL; days) was defined as the duration for which a flower remains open and reproductively functional. We randomly tagged 10–20 emerging buds from different individuals per species and monitored them twice daily (at dawn and dusk) until floral senescence. Floral longevity was calculated as the time interval from complete corolla opening to visible withering and discoloration of the corolla and stamens (Paiva and Roddy, 2024).

2.3. Morphological traits

Flowering branches were collected in the morning and transported to the laboratory, where the fresh weight (FW; g) of each flower was measured immediately. Flowers were then rehydrated to full saturation until a constant weight was achieved and recorded as saturated weight (SW; g). Subsequently, flowers were oven-dried at 70 °C for at least 48 h to determine dry weight (DW; g). Saturated water content (SWC) was calculated as (SW - DW)/DW.

For determination of floral area (FA; cm²), all floral parts were scanned using a flatbed scanner (Epson Perfection V850 Pro). Following scanning, samples were oven-dried at 70 °C for at least 48 h to determine dry weight (DW; g). Flower mass per area (FMA; g m⁻²) was calculated as the ratio of dry weight to floral area, and flower dry matter content (FDMC; g g⁻¹) was calculated as the ratio of dry weight to saturated weight. Additionally, petal thickness (PT; mm) was measured on fresh petals using a digital caliper (Nscing Es, Nanjing, China), with the mean of three measurements per petal recorded.

2.4. Petal anatomical traits

Petal stomatal traits were assessed using the nail polish imprint method. Clear nail polish was applied evenly to a 1 × 1 cm area adjacent to the primary veins on both the adaxial and abaxial epidermis of freshly opened flowers. After air-drying, the imprints were gently peeled off with forceps. Stomatal counts were performed under a light microscope (Leica DM2500, Leica Microsystems Ltd., Wetzlar, Germany), and fields containing stomata were photographed to measure guard cell length (GCL; μm). The imprinted areas were scanned to determine the area, and stomatal density (SD; no.mm⁻²) was calculated as the total number of stomata divided by the imprinted area.

For vein density analysis, freshly opened flowers were fixed in 70% FAA solution (formalin: acetic acid: 70% alcohol = 5:5:90), and 1 × 1 cm sections adjacent to the central petal vein were excised. For sympetalous flowers with basal fusion or narrow tubes, sampling was restricted to the corolla lobes. In contrast, for species with well-developed tubes, measurements were taken from both lobe and tube regions on separate replicates to derive an integrated species-level value. Vein lengths were measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA), and vein density (D_vein; mm mm⁻²) was defined as the total vein length per unit area (Ke et al., 2024).

2.5. Pedicel anatomical traits

Pedicel cross-sections were prepared using a standard paraffin embedding protocol. Flowers were collected at peak anthesis and fixed in 70% FAA, after which pedicels were dehydrated and embedded in paraffin. Transverse sections were cut using a rotary microtome (Leica RM2245, Leica Microsystems, China), deparaffinized with limonene, and stained with a 1:1 mixture of 1% safranin and 0.5% astra blue. Cross-sectional images were captured using the light microscope. The pedicel cross-sectional area, total vessel number (n), and vessel lumen area were measured using ImageJ software. Pedicel vessel density (VD; no.mm⁻²) was calculated as the number of vessels per unit pedicel cross-sectional area. The pedicel vessel diameter (D_i; μm) and the hydraulically weighted vessel diameter (D_h; μm) were calculated according to Meunier et al. (2020):

$$D_{\mathrm{i}}=\sqrt{\frac{{4A}_{\mathrm{i}}}{\mathrm{\pi }}}$$

$$D_h={\left(\frac{\sum _{i=1}^n{D}_i^4}{n}\right)}^{\frac{1}{4}}$$

The vessel lumen fraction (VLF) was calculated as the ratio of total vessel lumen area to pedicel cross-sectional area. Vessel wall reinforcement (VWR) was determined as the square of the ratio of vessel wall thickness to vessel diameter. The pedicel cavitation vulnerability index (VI) was calculated as the ratio of the hydraulically weighted vessel diameter (D_h) to vessel density (VD). Theoretical pedicel hydraulic conductivity (K_t; 10⁻⁴ kg m⁻¹ s⁻¹MPa⁻¹) was calculated based on the Hagen–Poiseuille law (Tyree and Ewers, 1991), following Bourbia et al. (2020):

$$K_{\mathrm{t}}=\frac{\frac{\pi \rho }{128\eta }\sum _{i=1}^n\left(D_i^4\right)}{A}$$

where π is 3.14, ρ is the density of water (1000 kg m⁻³), η is the viscosity of water (1.002 × 10⁻⁹ MPa s⁻¹ at 20 °C), and A is the projected area of the flower connected to the pedicel.

2.6. Minimum epidermal conductance

Minimum epidermal conductance (g_min; mmol m⁻² s⁻¹) refers to the lowest conductance at which water vapor diffuses through the cuticle and incompletely closed stomata after stomatal closure (Roddy et al., 2016). It reflects the rate of unavoidable water loss through non-stomatal pathways under minimal transpiration conditions (Bourbia et al., 2020; Roddy et al., 2023). For each species, 3–6 fresh, intact flowers from different individuals were selected to measure g_min. Four species (Jasminum sambac, Gardenia sootepensis, Calotropis procera, and Thunbergia mysorensis) were excluded because the multiple incisions required to determine their initial floral area would compromise measurement accuracy. To determine g_min, intact petals were excised, and all cut surfaces were immediately sealed with cyanoacrylate glue to prevent water loss from the excision sites. The sealed petals were scanned to determine their initial floral area, and then placed in a dark, thermostatic climate chamber equipped with a fan to maximize boundary layer conductance. Following an initial dehydration period to ensure stomatal closure, petal weights were recorded every 15–20 min alongside chamber temperature and humidity for at least eight consecutive measurements. Minimum epidermal conductance was subsequently calculated from the mass-loss data following Roddy et al. (2023).

2.7. Pressure-volume curve

Pressure-volume (PV) curves were constructed for six freshly opened flowers from different individuals per species. Three species (Ochna thomasiana, Lagerstroemia tomentosa, and Rotheca myricoides) were excluded from measurements because their short pedicels prevented effective sealing and hindered observation of xylem exudation during pressurization. To construct PV curves, flowering branches were collected in the early morning, recut underwater to eliminate embolisms, and fully rehydrated. One flower was selected per branchlet, and its mass and water potential were measured repeatedly during dehydration using a pressure chamber (PMS1505D-EXP, PMS Instrument Company, Corvallis, OR, USA). After completion of measurements, samples were oven-dried at 70 °C for at least 48 h to determine dry mass (Roddy et al., 2019). PV curves describe the relationship between water potential and relative water content during dehydration, allowing derivation of key hydraulic parameters (Tyree and Hammel, 1972). These parameters include saturated osmotic potential (π_o; MPa), water potential at the turgor loss point (ψ_tlp; MPa), relative water content at the turgor loss point (RWC_tlp), and cellular elastic modulus (ε; MPa), all of which reflect drought tolerance and provide insight into the physiological mechanisms underlying plant responses to water stress (Bartlett et al., 2012; Roddy et al., 2019).

2.8. Statistical analysis

Data were log₁₀-transformed prior to analysis to improve normality and homoscedasticity. For saturated osmotic potential (π_o) and water potential at the turgor loss point (ψ_tlp), which are negative values, absolute values were taken before transformation. Because petals of five species lacked stomata, resulting in zero values for stomatal density (SD), a constant of 1 was added to all SD measurements prior to log₁₀ transformation. All analyses were performed using species-level means, except for linear mixed-effects models (LMMs), which were fitted using individual-level data. LMMs were fitted using the lmer function in the ‘lme4’ package to evaluate differences in floral traits between symmetry types (actinomorphic vs. zygomorphic) (Bates et al., 2015). In these models, floral traits were treated as response variables, floral symmetry as a fixed effect, and species identity as a random effect to account for non-independence among observations.

To characterize the native climatic background of each species, occurrence records were extracted from the Global Biodiversity Information Facility (GBIF, https://www.gbif.org/). Reliable native distribution information was unavailable in Plants of the World Online (POWO, https://powo.science.kew.org/) for three cultivated or hybrid taxa (Hibiscus rosa-sinensis, Michelia × alba, and Clerodendrum × speciosum); therefore, these species were excluded from climate-based analyses, resulting in a final dataset of 36 species. Occurrence records were cleaned using the ‘CoordinateCleaner’ package (Zizka et al., 2019) to remove potentially erroneous data points (e.g., coordinates in the sea, zero coordinates, and spatial outliers) and were further filtered to retain only records falling within the native distribution ranges delineated by POWO. To mitigate spatial sampling bias, records were spatially rarefied using a 10-km distance threshold with the ‘spThin’ package (Aiello-Lammens et al., 2015). Climatic variables—including mean annual temperature (MAT), temperature seasonality (TS), annual precipitation (MAP), and precipitation seasonality (PS)—were extracted for each occurrence from the WorldClim database (30-arcsec resolution) (Fick and Hijmans, 2017). The aridity index (AI) was calculated as MAP/(MAT + 10) (Liu et al., 2018; Song et al., 2022; Xu et al., 2025). Species-level mean values were then computed to represent each species' native climatic conditions. Because MAP and AI were highly correlated (r = 0.96, P < 0.001), and AI provides a more comprehensive measure of effective water availability by incorporating evaporative demand, only AI was retained for subsequent analyses. To quantify effects of floral symmetry and native climate on floral traits, we fitted multiple linear regression models with floral symmetry and selected climatic variables as predictors. Prior to analysis, all continuous predictors were standardized to facilitate direct comparison of effect sizes. Multicollinearity among predictors was assessed using variance inflation factors (VIF), with all values remaining below 5. Standardized regression coefficients (β) were reported to evaluate both the direction and relative magnitude of predictor effects. In addition, the relative contributions of floral symmetry and climatic variables to total trait variation were quantified.

Pairwise trait correlations were assessed using Pearson correlation analysis via the rcorr function in the ‘Hmisc’ package (Harrell Jr, 2025). Floral trait networks were constructed using the ‘igraph’ package (Csárdi and Nepusz, 2006), where traits were represented as nodes and significant Pearson correlations (P < 0.05) as edges. Network topology was characterized by the number of edges, edge density, average path length, and modularity (Kohkaew et al., 2025; Liu et al., 2025; Wan et al., 2025). Standardized major axis (SMA) analysis was conducted using the sma function in the ‘smatr’ package to test whether trait scaling relationships differed between symmetry types (Warton et al., 2012). Furthermore, we performed Principal Component Analysis (PCA) using the PCA function in the ‘FactoMineR’ package and Permutational Multivariate Analysis of Variance (PERMANOVA) using the adonis2 function in the ‘vegan’ package to assess whether symmetry types differed in their multivariate trait distribution (Lê et al., 2008; Oksanen et al., 2025). A phylogenetic tree for the 39 species was constructed using the phylo.maker function in the ‘V.PhyloMaker2’ package (Jin and Qian, 2022). Given that only one floral trait (SD) had a Blomberg's K value greater than 1, and only four traits exhibited significant phylogenetic signals (Table S2), the overall phylogenetic conservatism of floral traits was considered weak. Consequently, phylogenetic effects were not included in subsequent data analyses. All statistical analyses were conducted in R v.4.4.1 (R Core Team, 2025).

3. Results

3.1. Comparison of floral traits between actinomorphic and zygomorphic flowers

Based on the results of linear mixed models (LMMs), most floral traits (17 out of 22) did not differ significantly between zygomorphic and actinomorphic species. In contrast, five traits showed significant divergence (Table S3). Specifically, compared with actinomorphic flowers, zygomorphic flowers had significantly longer floral longevity (FL; P = 0.018; Fig. 3a) and higher hydraulic capacitance (C_1,mass; P = 0.042; Fig. 3b). They also showed greater drought tolerance, reflected by a more negative water potential at the turgor loss point (ψ_tlp; P = 0.042; Fig. 3c), and a lower relative water content at the turgor loss point (RWC_tlp; P = 0.006; Fig. 3d). In addition, petal vein density (D_vein) was significantly lower in zygomorphic than actinomorphic flowers (P = 0.013; Fig. 3e and Fig. S1 and S2).

Fig. 3.

Differences in floral traits between actinomorphic and zygomorphic flowers. ∗, P < 0.05; ∗∗, P < 0.01. Trait abbreviations are shown in Table 1.

Multiple regression analyses (Table S4) revealed that the four native-habitat climatic variables (MAT, AI, TS, and PS) exhibited limited explanatory power for most floral traits. After accounting for these variables, floral symmetry remained a significant predictor of FL, D_vein, and ψ_tlp. In contrast, symmetry-related differences in C_1,mass and RWC_tlp were no longer significant once climatic conditions were included in the models. Variance decomposition (Fig. S3) further revealed that native-habitat climatic variables accounted for a notable share of variation in several traits, such as hydraulic capacitance and RWC_tlp. By contrast, floral symmetry explained a greater proportion of variation in a subset of core traits, including FL, D_vein, and ψ_tlp.

3.2. Trait network patterns and coordination in actinomorphic and zygomorphic flowers

Compared to actinomorphic species, zygomorphic species displayed a more densely connected trait network (Fig. 4), characterized by a higher number of edges and greater edge density, but lower average path length and modularity (Table S5). The identity of hub traits also differed between symmetry types. In actinomorphic flowers, flower mass per area (FMA) and hydraulic capacitance (C_1,mass) emerged as hub traits. In contrast, vessel wall reinforcement (VWR) was identified as the hub trait in zygomorphic flowers (Fig. S4).

Fig. 4.

Trait networks of (a) actinomorphic (n = 21) and (b) zygomorphic flowers (n = 18). Orange and blue lines represent positive and negative relationships, respectively. Purple nodes, economic traits; green nodes, hydraulic traits. Thicker lines indicate stronger relationships. Trait abbreviations are shown in Table 1.

Regarding associations among floral hydraulic traits, K_t was significantly positively correlated with D_h and VLF in both actinomorphic and zygomorphic flowers (Fig. 5a and b and Fig. S5 and S6). In zygomorphic flowers, significant positive correlations were found among K_t, D_vein, and g_min (Fig. 5c–e), and g_min was significantly positively correlated with GCL (Fig. 5f and Table S6); however, these relationships were not significant in actinomorphic flowers. Furthermore, significant positive correlations between D_h and VI, as well as between FMA and K_t, were observed in both symmetry types (Fig. 5g and h). Finally, a significant positive correlation between FMA and PT was observed exclusively in actinomorphic flowers (Fig. 5i and Table S7).

Fig. 5.

Relationships of pedicel theoretical hydraulic conductivity (K_t) with (a) hydraulically weighted vessel diameter (D_h), (b) vessel lumen fraction (VLF) and (c) vein density (D_vein), the minimum epidermal conductance (g_min) with (d) vein density (D_vein), (e) theoretical hydraulic conductivity (K_t) and (f) guard cell length (GCL), the hydraulically weighted vessel diameter (D_h) with (g) vulnerability index (VI), the flower mass per area (FMA) with (h) theoretical hydraulic conductivity (K_t) and (i) petal thickness (PT). Red and blue points and lines represent zygomorphic and actinomorphic flowers, respectively. ns, P > 0.05; ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001. Trait abbreviations are shown in Table 1.

For traits related to floral drought tolerance, ε was significantly negatively correlated with C_1,mass and positively correlated with RWC_tlp in both symmetry types (Fig. 6a and b). Notably, the slope of the relationship between ε and RWC_tlp was significantly steeper in actinomorphic flowers (Table S8). In addition, ε was significantly positively correlated with FMA in actinomorphic flowers (Fig. 6c), whereas this relationship was not significant in zygomorphic flowers. In contrast, zygomorphic flowers exhibited a significant negative correlation between ε and g_min (Fig. 6d). Moreover, ψ_tlp was negatively associated with both RWC_tlp and GCL in zygomorphic flowers (Fig. 6e and f), while these relationships were not significant in actinomorphic flowers.

Fig. 6.

Relationships of elastic modulus (ε) with (a) hydraulic capacitance before turgor loss (C_1,mass), (b) relative water content at turgor loss point (RWC_tlp), (c) flower mass per area (FMA) and (d) minimum epidermal conductance (g_min), the water potential at the turgor loss point (ψ_tlp) with (e) relative water content at turgor loss point (RWC_tlp) and (f) guard cell length (GCL). Red and blue points and lines represent zygomorphic and actinomorphic flowers, respectively. ns, P > 0.05; ∗, P < 0.05; ∗∗∗, P < 0.001. Trait abbreviations are shown in Table 1.

For associations involving floral reproductive traits, FA was significantly positively correlated with D_h and VI in both actinomorphic and zygomorphic flowers (Fig. 7a and b). FA was also positively associated with SWC and negatively associated with FDMC in zygomorphic flowers (Fig. 7c and d), whereas these relationships were not significant in actinomorphic flowers. FL was significantly positively correlated with both FMA and PT in actinomorphic flowers (Fig. 7e and f), but these correlations were not significant in zygomorphic flowers.

Fig. 7.

Relationships of floral area (FA) with (a) hydraulically weighted vessel diameter (D_h), (b) vulnerability index (VI), (c) saturated water content (SWC) and (d) flower dry matter content (FDMC), the floral longevity (FL) with (e) flower mass per area (FMA) and (f) petal thickness (PT). Red and blue points and lines represent zygomorphic and actinomorphic flowers, respectively. ns, P > 0.05; ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001. Trait abbreviations are shown in Table 1.

3.3. Multivariate analysis of actinomorphic and zygomorphic floral traits

Because several traits were missing for seven species, principal component analysis (PCA) was performed on 32 species with complete trait data. The PCA based on 22 floral traits revealed that the first and second principal components explained 24.5% and 17.2% of the total variance, respectively (Fig. 8). Specifically, PC1 was mainly associated with D_h, VI, FA and C_2,mass, while PC2 was primarily driven by FMA and K_t (Table S9). PERMANOVA indicated no significant difference between actinomorphic and zygomorphic flowers in their distribution within the multidimensional functional trait space (P = 0.454; Table S10).

Fig. 8.

Principal component analysis of 22 floral traits of 32 species. Blue circles represent actinomorphic species, and red triangles represent zygomorphic species. Trait abbreviations are shown in Table 1.

4. Discussion

By systematically comparing floral hydraulic, structural, and reproductive traits between actinomorphic and zygomorphic flowers under common garden conditions, this study provides new insights into how floral symmetry is associated with hydraulic strategies and trait coordination. Multivariate analyses revealed substantial overlap between the two symmetry types in overall trait space, suggesting that both are subject to shared constraints imposed by the floral economics spectrum. Despite this overlap, clear differences emerged in specific hydraulic traits and in the coordination among traits. Zygomorphic flowers were characterized by longer floral longevity, higher drought tolerance, and lower water supply capacity, and these differences remained significant after accounting for native climatic conditions. Moreover, trait network and correlation analyses indicated stronger integration among hydraulic, structural, and reproductive traits in zygomorphic flowers. Together, these results suggest that floral symmetry influences how key traits are coordinated to maintain water balance and functional stability during anthesis, rather than driving a complete reorganization of overall floral functional strategies.

4.1. Symmetry-related differences in floral hydraulic traits

Our results indicate that zygomorphic flowers adopt a more conservative and hydraulically safer strategy than actinomorphic flowers. This pattern reflects coordinated adjustments between hydraulic safety and floral longevity rather than a simple causal sequence, consistent with the Flower Economics Spectrum, which emphasizes integrated trade-offs among floral construction, longevity, and hydraulic function (Roddy et al., 2021; Paiva and Roddy, 2024). Consistent with a recent global synthesis of 1452 angiosperm species (Stephens et al., 2024), zygomorphic flowers in our study exhibited significantly longer anthesis, confirming this as a general feature of floral zygomorphy. Crucially, multiple regression analyses demonstrated that this divergence in floral longevity persisted even after accounting for native climatic conditions, suggesting that it is more closely associated with floral architecture than with short-term environmental plasticity. Extended longevity is likely advantageous for specialized pollination, particularly in warm, high vapor pressure deficit (VPD) environments, and was coupled with enhanced hydraulic safety. Specifically, zygomorphic flowers maintained a significantly more negative turgor loss point (ψ_tlp) irrespective of native climatic contexts. Given that ψ_tlp represents a critical threshold for dehydration tolerance (Bartlett et al., 2012), the robustness of this pattern across climates indicates a strong link to zygomorphic floral architecture.

In contrast to the climatic independence of ψ_tlp, water storage traits exhibited a pattern consistent with environmental plasticity. Although zygomorphic flowers exhibited higher mass-based hydraulic capacitance (C_1,mass) and lower relative water content at the turgor loss point (RWC_tlp) in analyses without climatic controls, these differences disappeared after accounting for climatic variables. This finding implies that, unlike the critical physiological threshold defined by ψ_tlp, hydraulic capacitance functions as a phenotypically plastic trait tracking environmental water availability. Distinct from such plasticity, zygomorphic flowers were characterized by lower petal vein density (D_vein), a pattern that persisted irrespective of climatic variables. Given the dual hydraulic and mechanical roles of floral veins (Zhang et al., 2018, 2021a), this reduced investment in vascular tissues, coupled with environmentally modulated water storage, implies that zygomorphic flowers rely less on sustained high-flux water supply and rigid venation-based support, and more on internal water buffering to maintain turgor and structural integrity during anthesis.

Taken together, these symmetry-related differences align with global biogeographic patterns of floral traits. Although mean floral longevity generally declines toward lower latitudes across angiosperms (Song et al., 2022), zygomorphic flowers are disproportionately concentrated in low-latitude regions (Wang et al., 2023) and often exhibit comparatively longer anthesis. Our results suggest that this apparent divergence can be reconciled by considering floral hydraulics: in warm, high-VPD environments, longer-lived zygomorphic flowers likely require intrinsically safer hydraulic thresholds (more negative ψ_tlp) and greater reliance on internal water buffering to sustain turgor during anthesis.

4.2. Divergent hydraulic trait relationships between actinomorphic and zygomorphic flowers

Across both actinomorphic and zygomorphic flowers, consistent hydraulic relationships were observed, indicating shared functional constraints. In both symmetry types, theoretical hydraulic conductivity (K_t) increased with hydraulically weighted vessel diameter (D_h) and vessel lumen fraction (Fig. 5a and b), while larger vessels were associated with higher VI (Fig. 5g). This pattern reflects a fundamental efficiency - safety trade-off, whereby flowers rely on wider vessels to enhance water transport efficiency at the cost of increased embolism susceptibility under water stress, regardless of floral symmetry. Similarly, ε showed a negative association with C_1,mass and positive association with RWC_tlp in both symmetry types (Fig. 6a and b). These relationships highlight the central role of cell wall properties in floral water regulation. Compared to leaves, flowers generally possess lower vein density and hydraulic conductance and function largely as hydrostatic systems, maintaining turgor through adjustments in cell wall elasticity and tissue water storage capacity (Roddy et al., 2019; Aun et al., 2024). While a higher ε limits the capacity for elastic water storage and release, it prevents excessive cell deformation under declining water potentials, allowing floral tissues to retain a relatively high water content even after turgor loss (Cheung et al., 1975; Bartlett et al., 2012).

Despite the shared hydraulic strategies described above, actinomorphic flowers exhibited significantly weaker coupling among hydraulic traits, as reflected by their simpler trait network structure (Fig. 4a). In actinomorphic flowers, significant correlations were largely limited to positive relationships between FMA and both K_t and petal thickness (Fig. 5h and i), while correlations among other hydraulic traits were generally weak or absent. These findings imply that actinomorphic flowers rely more strongly on structural investment to support water transport and tissue function. Consistent with this interpretation, the significant positive correlation between FMA and ε (Fig. 6c) indicates that actinomorphic flowers maintain morphological stability by developing relatively more rigid tissues, likely at the expense of hydraulic flexibility. This decoupled hydraulic strategy may be related to the broader pollination context of actinomorphic flowers. Actinomorphic flowers are usually visited by a wider variety of pollinators (Fenster et al., 2004; Yoder et al., 2020) and show lower pollination precision than zygomorphic flowers (Citerne et al., 2010; Nikkeshi et al., 2015). Under such conditions, moderate petal deformation caused by transient water deficits may have limited consequences for pollination success, as multiple access routes for pollinators remain available. Consequently, actinomorphic flowers may tolerate greater variability in water status, exhibiting more flexible and less tightly coordinated hydraulic regulation than zygomorphic flowers.

In contrast, zygomorphic flowers exhibited tightly coordinated relationships among structural investment, water transport, and transpirational water loss (Fig. 4b), indicating stronger functional integration among floral traits and supporting our second hypothesis. In zygomorphic flowers, FMA was positively correlated with K_t (Fig. 5h), suggesting that enhanced water transport capacity is associated with higher construction costs, consistent with previous findings (Roddy et al., 2016). Because flowers typically have few or no stomata, floral water loss is largely controlled by epidermal and cuticular traits, with minimum epidermal conductance (g_min) being a key control on floral water balance (Roddy et al., 2016, 2023; Aun et al., 2024). Accordingly, significant positive correlations among K_t, D_vein, and g_min (Fig. 5c–e) formed a tightly linked hydraulic module, a pattern also reported previously (Roddy et al., 2016). Moreover, guard cell length was positively correlated with g_min and negatively correlated with ψ_tlp (Fig. 5, Fig. 6f), indicating that larger stomata are associated with greater residual water loss and increased susceptibility to turgor loss at higher water potentials. Zygomorphic flowers also showed negative correlations between ε and g_min (Fig. 6d), as well as between ψ_tlp and RWC_tlp (Fig. 6e), suggesting coordinated regulation of tissue elasticity and epidermal water loss (Bartlett et al., 2012). Together, these patterns indicate that zygomorphic flowers maintain water balance through precise integration of water transport, storage, and loss. Because their reproductive success depends strongly on maintaining floral shape and positional accuracy, even minor wilting can disrupt pollination efficiency (Nikkeshi et al., 2015; Stewart et al., 2022), likely favoring tight hydraulic - structural coordination to preserve morphological stability throughout anthesis.

4.3. Association between floral reproductive and hydraulic traits

Our study revealed clear associations between hydraulic traits and key reproductive traits, including floral size and longevity, providing further insight into how floral function is supported by hydraulic strategies. Floral size and longevity are central components of reproductive success: larger flowers generally enhance pollinator attraction, while extended floral longevity increases the likelihood of pollen transfer and outcrossing (Primack, 1985; Ashman and Schoen, 1994). While flower–pollinator interactions have been extensively studied, the physiological mechanisms underpinning variation in floral size and longevity remain unclear. Across both symmetry types, floral size was positively correlated with D_h (Fig. 7a), indicating that larger flowers require greater hydraulic efficiency to sustain water supply, consistent with previous findings (Ke et al., 2023, 2024). However, floral size was also positively associated with VI (Fig. 7b), highlighting a clear efficiency-safety trade-off whereby increased transport capacity entails higher susceptibility to hydraulic failure.

Despite these shared patterns, the coordination between reproductive traits and hydraulic or economic traits differed markedly between symmetry types. In actinomorphic flowers, floral longevity was positively correlated with both FMA and petal thickness (Fig. 7e and f), consistent with a structural investment strategy along the floral economics spectrum (Roddy et al., 2021). Higher FMA reflects greater dry matter investment per unit area and is generally associated with enhanced water retention and longer internal water residence time, which can prolong floral function under water stress (Zhang et al., 2017; Zhou et al., 2025). Increased petal thickness further contributes to water storage and may reduce water loss through thicker epidermal and cuticular layers (Guo et al., 2023; Paiva and Roddy, 2024). These results suggest that actinomorphic flowers primarily rely on structural investment to sustain floral longevity. In contrast, in zygomorphic flowers, floral size was positively correlated with saturated water content and negatively correlated with flower dry matter content (Fig. 7c and d). This suggests that size enlargement in zygomorphic flowers may primarily be achieved through increased water content rather than greater carbon investment, forming a “hydraulic skeleton” in which floral shape is maintained by turgor pressure rather than rigid carbon-based tissues. Such a strategy reduces carbon costs per unit area and is consistent with the ecological role of flowers as short-lived, display-oriented organs supported by dynamic water storage (Roddy et al., 2019, 2023).

Finally, floral hydraulic strategies are likely to interact with a broader spectrum of reproductive traits beyond the floral display traits measured here, such as seed mass and pollen production (E-Vojtkó et al., 2022; Xu et al., 2025). For instance, the high carbon investment associated with dense venation in actinomorphic flowers may impose trade-offs with resource allocation to other reproductive sinks. Conversely, the conservative hydraulic strategy of zygomorphic flowers implies a distinct pattern of resource partitioning, potentially prioritizing physiological buffering to support reproductive development under water deficit. Integrating these broader reproductive dimensions into floral hydraulic frameworks represents a promising avenue for future research.

5. Conclusions

In conclusion, our comparative analysis of 22 functional traits across 39 species reveals symmetry-related differences in hydraulic strategies and floral trait coordination between actinomorphic and zygomorphic flowers. Zygomorphic flowers adopt a conservative strategy, underpinned by intrinsically longer floral longevity and greater drought tolerance, while showing environmentally responsive modulation of hydraulic capacitance. In contrast, actinomorphic flowers rely on higher vein density to maintain water balance. Beyond the differences in individual traits, the two symmetry types also differ markedly in patterns of trait integration. Zygomorphic flowers display tightly coupled networks linking hydraulic, structural, and reproductive traits, whereas actinomorphic flowers exhibit weaker and more flexible trait coordination. Together, these results highlight a close association between floral symmetry and physiological organization. Future studies extending these comparative approaches across broader environmental gradients and water availability regimes will be essential to test the generality of the patterns reported in the present study and to further elucidate how water stress contributes to the diversification of floral functional strategies and symmetry types.

Author contributions

FJ-L, YK and JL-Z conceived and designed the study. FJ-L performed the experiments. FJ-L, YK, XY-H, YB-Z and JL-Z analyzed the data. FJ-L wrote the draft of the manuscript. All authors read, commented and approved the final version of the manuscript.

Data availability statement

The data that support the results of this study are available from the corresponding author upon reasonable request.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.