Plant hydraulic traits primarily define the water regulation strategy, thus enabling better understanding of vegetation structure, function and dynamics under varying hydro-environments. Despite being intensively documented in woody species, variation and correlation of hydraulic traits across herbaceous species remain largely understudied.

Here, we report on the leaf hydraulics of nine herbs with contrasting growth forms (graminoid and forb). Traits quantifying drought resistance, including leaf water potential thresholds triggering xylem embolism (P_x), stomatal closure (P_gs) or leaf turgor loss point (P_tlp), and minimum conductance (g_min), together with leaf gas exchange, morphological traits and biomass allocation, were measured on pot grown plants. In addition, an in situ dry-down was imposed on four representative species, with leaf gas exchange, water potential and level of xylem embolism, being continuously monitored during dehydration to determine the dynamics of stomatal closure and leaf xylem embolism.

We found that the studied graminoids tended to be more drought tolerant than forbs, although the difference in hydraulic safety margin for stomatal closure (HSM_st) did not differ significantly between these growth forms. Across species, P_x was coordinated with P_gs and P_tlp, but was decoupled from gas exchange traits, including maximum photosynthetic rate and stomatal conductance. Furthermore, no correlations were found between hydraulic traits and specific leaf area or the ratio of aboveground to belowground biomass.

For plants that experienced in situ dehydration, stomatal closure always preceded the onset of xylem embolism in leaves. Moreover, species exhibited a distinct stomatal regulation strategy during the dehydration despite belonging to the same growth form.

Our findings contribute to the understanding of herb hydraulics, which will inform prediction on the dynamics of grassy ecosystems by providing traits data and guiding the classification of plant functional types in “grassy” ecosystems.

Plant materials and growth condition 

Nine herbaceous species representing two growth forms were chosen for the current study (Table 1). Species selected here are representative of understory herbaceous community within forest ecosystems across north China plain, and most species are cosmopolite, with the exception of Crepidiastrum denticulatum, Senna tora, and Elymus dahuricus, which occur sporadically worldwide, according to the occurrence record of Global Biodiversity Information Facility (www.gbif.org). All species are common weeds that have never been artificially selected. The distributional range of these species largely overlaps, yet there is considerable variation in the mean annual precipitation (MAP, mm) and mean annual temperature (MAT, ^oC) of the species climatic envelopes (i.e. bioclimate niche), with the MAP ranging from 434 to 1618 mm and the MAT ranging from 5.5 to 23.9 ^oC.

Seeds of the selected species were purchased from a commercial supplier that collects seeds from the field annually. Seeds were sown in hiko trays filled with water saturated vermiculite for germination, which typically took 2~3 weeks depending on species. When seedlings reached ca. 2 cm tall, plants were transplanted to square plastic pots (15 cm × 13 cm ×15 cm for length, width and height, respectively) filled with a mixture of multi-purpose compost and vermiculite (2:1), and then thinned to one plant per pot. Nutrients in the compost can sustain the growth of small plants for up to 12 months, thus ensuring the plant materials were not nutrient limited during growth. Each species was represented by 15~24 individuals.

The pots were then placed on an indoor vertical growth rack in a well-ventilated, controlled environment lab. The growth rack was modified by installing additional red and white full spectrum LED lights, which enabled approximately 800 μmol m^-2 s^-1 photosynthetic photon flux density (PPFD). The lights were programmed to illuminate 12 h per day from 6 am. Plants were irrigated manually every other day to avoid water limitation during development and were randomly positioned weekly to minimize the heterogeneity in growth environment. During the experimental period, lab temperature was maintained at 25 ^oC, and relative humidity (RH, %) fluctuated between 45%~50 %. Measurements commenced 4~6 weeks after transplanting.

Experimental design

Two complementary experiments were designed to test our hypotheses. The first experiment aimed to address the variation of key hydraulic traits across species, as well as their correlations, with carbon and morphological traits. For each species, we measured xylem vulnerability to embolism, leaf pressure volume (PV) traits, leaf minimum conductance and morphological traits, including specific leaf area as well as the ratio of aboveground to belowground biomass, from well-watered individuals. Then, a subset of individuals (6~8) of each species were exposed to progressive dry-down by withholding irrigation. Subsequently, leaf gas exchange characteristics and leaf water potential (Ψ_leaf, -MPa) were assessed periodically during dry-down until leaf stomatal conductance approached zero, thereby obtaining the response of leaf gas exchange to Ψ_leaf. For all species, measurements were conducted during the vegetative phase and no flowering was observed by the end of the experiment.   

The objective of our second experiment was to resolve the sequence of stomatal closure and initiation of xylem embolism. To this end, we monitored the dynamics of leaf gas exchange and development of xylem embolism in situ on slowly dehydrated plants. Four representative species were selected for the experiment (i.e., Eleusine indica, Cynodon dactylon, Crepidiastrum denticulatum and Medicago sativa; Table 1). Each species involved three replicas except for M. sativa, which was represented by two individuals.

Experiment I: Variation of hydraulic, gas exchange and morphology traits across species

Leaf xylem vulnerability curves

For each species, 3~5 individuals of similar size were used for leaf xylem vulnerability curves (VCs) assessment. Prior to the day of measurement, soil of targeted plants was watered to field capacity and plants were covered with an opaque black plastic bag to ensure full rehydration and complete stomatal closure. Leaf embolism level was measured with the optical visualization (OV) technique https://www.opensourceov.org/) (Brodribb et al., 2016). This method was based on the variation of xylem optical attributes before and after the occurrence of embolism. The technique allows for the spatiotemporal pattern of embolism propagation within the vein network to be revealed, and is particularly suitable for plant material characterized by tender and fragile tissues such as herbs (Jacob et al. 2022); embolism thresholds generated by OV have been validated elsewhere (Brodribb et al., 2016; Chen et al., 2021; Gauthey et al., 2020).

For each individual plant, one recently mature and fully expanded leaf was placed into custom-built leaf clamps equipped with a digital camera (Cavicam, https://cavicams.com/), which was controlled by a single board computer (Raspberry Pi, Raspberry Pi Foundation), and photographed upon illumination. Glass microscope slides were used to flatten the leaf when necessary. A neighboring leaf was used for measuring Ψ_leaf using a psychrometer (ICT International, Armidale, NSW, Australia). Before measurement, the leaf was gently polished using an 800-grit sanding stick on the adaxial side to remove the epidermis. A custom-made holder was employed to fix the psychrometer, thus ensuring the chamber made full contact with the leaf. High vacuum grease was applied to seal the possible gaps between the chamber and the leaf. Following installation of these devices, the plant was cut above the soil level and allowed to desiccate until crispy. During dehydration, Ψ_leaf and images of the leaf were taken at 10 min intervals.

Images of leaves were processed using Image J software (National Institutes of Health, Bethesda, MD, USA). The protocol for image processing has been elaborated elsewhere (Li et al., 2019; Skelton et al., 2021). In short, images obtained from an independent leaf were stacked, and the difference between two consecutive images was revealed by the “Image difference” function in the OSOV toolbox. Pixels indicating the embolized area on the conduits were highlighted and then analyzed after noise was removed. Percentage of xylem embolism level (%) was calculated as the pixels on each image divided by the total embolized area during dehydration. Leaf VCs were then generated by plotting PLC against Ψ_leaf.

Gas exchange and water potential measurement

Gas exchange characteristics were measured between 9 am to 11 am on fully expanded, mature leaves of each plant. Leaf gas exchange was measured using a portable photosynthesis system (Model 6800, Li‐Cor, Lincoln, NE, USA) equipped with a red-blue LED light source and external CO₂ injector. The selected leaf was placed into a chamber supplied with 800 μmol m^-2 s^-1 PPFD and 420 μmol mol^-1 CO₂ in the air, with the temperature and vapor pressure deficit inside the cuvette being maintained at 25 ^oC and 1.2~1.5 kPa, respectively. Variables including maximum photosynthetic rate (A, μmol m^-2 s^-1), stomatal conductance (g_s, mol m^-2 s^-1), and transpiration rate (E, mmol m^-2 s^-1) were logged when these readings were visually stable, which typically took 10~15 mins depending on leaf water status. Leaves were removed from the plant after gas exchange measurements and Ψ_leaf was then immediately determined using a Scholander‐type pressure chamber (PMS Instrument, Corvalis, OR, USA). Maximum leaf photosynthetic rate (A_max), stomatal conductance (g_smax) and transpiration rate (E_max) were defined as the corresponding values recorded under well-watered conditions and averaged across datapoints measured at Ψ_leaf>-1 MPa. Measurements were taken once a day, starting from the well-watered state until the stomata were completely closed.

Pressure-volume traits

For each species, 4~6 leaves from 3~5 individuals were collected for pressure-volume (PV) curve measurements. Selected individuals were irrigated and placed in darkness overnight to allow for full rehydration. Leaf samples were detached from plants the next morning, covered with para-film if necessary, and then used for PV curve measurements. Leaf PV curves were generated following the standard method. Briefly, leaf samples were slowly dehydrated under laboratory conditions, and then leaf water potential (Ψ_leaf) and leaf fresh weight (FW, g) were regularly measured by pressure chamber and analytical balance (weighted to 0.001 g), respectively, until Ψ_leaf reached -2.0 MPa to -3.0 MPa and leaves were visually wilting, depending on the species. PV curves were analyzed following Lenz et al. (2006) to obtain leaf turgor loss point (P_tlp, -MPa) and osmotic potential at full turgor (P₀, -MPa).

Leaf minimum conductance

For each species, two leaves from each of the three randomly chosen plants were used to determine leaf minimum conductance (g_min, mmol m^-2 s^-1). Prior to the measurement, plants were rehydrated in darkness overnight, similar to material used for VC assessments. Then, leaf samples were cut from a plant and bench dried under 25^oC and 50% RH for up to 10 hours, until the relationship between leaf FW and time became linear. Leaf FW was measured periodically during dehydration. The g_min was calculated following Blackman et al. (2019).

Leaf morphology and biomass allocation

Specific leaf area (SLA, m² kg^-1) was determined from the samples used for g_min measurements. Prior to bench dehydration, leaves were flattened and photographed with a known scale, and leaf area was then estimated using the Image J software. Following the g_min measurement, leaf samples were oven-dried to gain the dry weight (DW, g). Leaf SLA was calculated as the ratio of leaf area to DM. The ratio of aboveground to belowground biomass (ABR) was measured on the plant material used for constructing VCs. Following the measurement of VCs, the aboveground portion of plants were collected, and the corresponding root was carefully rinsed free of soil. Samples were then oven-dried to obtain DW. For all plant material, oven drying was performed at 70 ^oC and lasted for at least 72 hours or until no more weight change was detected.

Experiment II: In situ dry-down of intact plants

For each individual subjected to in situ dry-down, three fully expanded leaves with similar size were labeled for leaf gas exchange, water potential and leaf xylem embolism level measurements, respectively. Plants were well-watered prior to the day of dehydration, and received no irrigation thereafter. For each plant, a Cavicam and psychrometer were installed on two labelled, neighboring leaves following the same protocol described above, with images and the water potential of leaves recorded every 10 mins. Leaf gas exchange was measured between 9 am to 11 am daily on the remaining labelled leaf, with the cuvette environment, including PPFD, vapor pressure deficit, temperature and CO₂ concentration, being set as detailed above. In situ dehydration ceased until the aboveground portion of plant became completely dry and fragile, with Ψ_leaf being no longer measurable by the psychrometer. Then, the leaf used for imaging was excised from the base, still allowing the Cavicam to continuously photograph until no embolism events were observed over the next 24 hours. Obtained images were processed according to the aforementioned protocol to estimate the dynamics of xylem embolism during dehydration.    

Data analysis

Leaf VCs were fitted by a three-parameter sigmoidal-exponential model to estimate leaf water potential inducing 12% and 50 % xylem embolism (P₁₂ and P₅₀, -MPa), respectively. Measured g_s during dehydration was plotted against Ψ_leaf, and the relationship was fitted with a nonlinear model with a local polynomial regression (loess) approach, to extract water potential thresholds at 50% and 90% loss of g_smax (P_gs50 and P_gs90; -MPa), respectively. All curve fittings were performed using the fitplc package. Hydraulic safety margin for stomatal closure (HSM_st, -MPa) was defined as P₁₂-P_gs90. To compare the P₅₀ of herbs with that of woody species, P₅₀ data measured from leaves of woody species, exclusively using the optical visualization technique (P50 OV), were sourced from Cardoso et al. (2022) (n=98). Additional P50 OV data for herbs were extracted from the data compiled in Huang et al. (2023) (n=6), as well as by Cardoso et al. (2022) (n=8). Note that such analysis was to compare the overall magnitude of variation in P50 between growth forms (e.g. Lens et al. 2016), instead of contrasting the difference in embolism resistance among coexisting woody species and herbs.

The difference in gas exchange traits, xylem embolism thresholds, pressure volume traits, g_min and morphological traits across species were compared using a linear model with the lm() function, followed by a Turkey’s HSD post hoc when needed. The difference in stomatal closure thresholds (i.e. P_gs50 and P_gs90) across species was assessed by comparing the overlap of confidence interval. Also, the linear model was used to test the difference in hydraulic traits (i.e. P₅₀, P_gs90, P_tlp, and HSM_st) between graminids and forbs with averaged trait values. Leaf P50 OV data of woody species and herbs was contrasted using the Welch’ t-test, which assumes unequal variance because of the uneven sample size within each group. Furthermore, to avoid the potential influence of phylogenetic relatedness on comparison, phylogenetic information of the compared species was obtained using the V.PhyloMaker2 package based on the “The Plant List” nomenclature system under scenario3 (Jin & Qian 2022; Figure S2), and were incorporated in the analysis. Thereafter, a phylogenetic independent contrast was conducted for P50 OV data using pic() function of the ape package. Data were tested with Kolmogorov-Smirnov and Levene’s test, for normality, equality, and consistency in distribution, and were log-transformed, if necessary. Correlative relationships among traits were fitted using linear regression with the lm() function, while multivariate analysis (i.e. principal component analysis, PCA) was performed using the procomp() function. All statistical analyses were conducted within the R computing environment (version 4.3.2, R Core Team, 2023).