The C₄ Poaceae are a diverse group both in terms of evolutionary lineage and biochemistry. There is a distinct pattern in the distribution of C₄ grass groups with aridity, however, the mechanistic basis for this distribution is not well understood. Additionally, few studies have investigated the functional strategies of cooccurring C₄ grass species for dealing with aridity in their natural environments. We explored the coordination of leaf-level gas exchange, water use, and morphology among five co-occurring semiarid C₄ grasses belonging to divergent clades, biochemical subtypes, and size classes at three sites along a natural aridity gradient. More specifically, we measured pre-dawn and midday water potential, stomatal conductance, water use efficiency, and photosynthesis. Leaf tissue was also collected for analysis of stable isotopes of carbon and oxygen as well as for measurement of specific leaf area (SLA) and leaf width. Species differences in the responsiveness of stomata to changes in vapor pressure deficit were also assessed. It was expected that NAD-me species would maintain higher rates of photosynthesis, higher water use efficiency, and have more responsive stomata than other cooccurring species based on observed biogeographic patterns and past greenhouse studies. We found that Aristidoideae and Chloridoideae NAD-me-type grasses had greater stomatal sensitivity to VPD, consistent with a more isohydric strategy. However, midgrasses had both greater apparent water access and water use efficiency, regardless of subtype or lineage. PCK-type had less responsive stomata and maintained lower levels of photosynthesis with increasing aridity. There were strong interspecific differences in ¹³C, leaf width, and SLA, however these were not significantly correlated with water use efficiency. C₄ grasses in our study did not fit discretely into functional groups as defined by lineage, biochemistry, or size class. Interspecific differences, evolutionary legacy, and biochemical pathways are likely to interact to determine water use and photosynthetic strategies of these plants. Control of water loss via highly responsive stomata may form the basis for the dominance of certain C₄ grass groups in arid environments. These findings build on our understanding of contrasting strategies of C₄ grasses for dealing with aridity in their natural environments.

Species and Site Descriptions

To explore relationships between physiology, morphology, and grass subtype across a natural aridity gradient (across sites) and across species (within sites), we measured a suite of gas exchange parameters (photosynthesis, stomatal conductance, water use efficiency), water stress (leaf water potential), morphology (specific leaf area, leaf width), and bulk leaf stable isotopes of carbon and oxygen in a variety of C₄ grasses commonly found in North American semi-arid savannas. Five species representing different combinations of biochemical subtype, phylogenetic lineage, and physiognomy (Aristida wrightii Nash [NADP-ME, Aristidoideae, midgrass], Bouteloua curtipendula (Michx.) Torr. [NAD-ME/PCK, Chloridoideae, midgrass], Erioneuron pilosum (Buckley) Nash [NAD-ME, Chloridoideae, shortgrass], Eriochloa sericea (Scheele) Munro ex Vasey [PCK, Panicoideae, midgrass], and Hilaria belangeri (Steud.) Nash [PCK, Chloridoideae, shortgrass]) was studied across three sites representing a natural precipitation gradient on the Edwards Plateau, Texas (Appendix S1: Table S2). Measurements were made from May to September across three summer growing seasons (2019-2021) at three Texas A&M AgriLife Research Ranches (Figure 1c-f), Martin (30° 48' N, 99° 50' W; MAP= 630 mm), Sonora (30° 16' N, 100° 34' W; MAP= 570 mm) and Read (30° 32' N, 101°03' W; MAP= 480 mm). The sites are characterized as Low Stony Hill and Limestone Hill ecological sites which are dominated by midgrasses and shortgrasses, annual forbs, and woody-encroaching species (Prosopis glandulosa, Quercus virginiana, Juniperus ashei, and Juniperus pinchotii; Soil Survey Team, NRCS, 2022). At Martin, the woody vegetation is primarily dominated by P. glandulosa, while at Sonora and Read Juniperus spp. dominate. Many of the tallgrasses (Sorghastrum nutans, Schizachyrium scoparium) that were historically common at these sites have been extirpated due to a legacy of overgrazing until the 1960s. For the past few decades, the three sites have been either ungrazed or only lightly grazed with low stocking rates of goats. Deer are also present at all sites. Soils at Martin are primarily Tarrant-type soils with 1-8% slopes and very cobbly clay in the top horizon (Soil Survey Team, NRCS, 2022). At Sonora, soils in our plots are dominated by the Eckrant-Rock outcrop complex with 1-20% slopes and cobbly silty clay in the top horizon as well as the Tarrant-Valera complex with 0-3% slopes and very cobbly clay to 15 inches. Soils at Read are primarily Tarrant-rock outcrop complexes (1-15% slopes or dry, 8-30% slopes). Weather towers at each site were used to monitor temperature, humidity, and rainfall throughout the course of each growing season and measurement campaign using an EE181-L and TB4MM-L wired to a CR1000 datalogger (Campbell Scientific, Logan, UT). Plots for physiological measurements were located within approximately 500 m of each tower. Meteorological data were gap-filled using TexMesoNet sites with 25 km of each field site when necessary (Texas Water Development Board, 2023).

Measurements at each of the three sites were made once in 2019 and 2020 and twice in 2021 within the same weeklong period to avoid temporal differences between sites, and were only made on full-sun days approximately 2-weeks post rainfall events to avoid intervals when grasses were senescing or curling. Repeated growing season measurements were not made in 2019 and 2020 due to repeated drought events at some sites. We did not attempt to characterize growing season “dry-down” as these are pulse-driven systems with variability in timing and intensity of summer rainfall events but instead sought to characterize the behavior of species across sites when most physiologically active. All measurements were made on the top-most fully expanded leaves of perennial grass tillers currently in flower in order to ensure consistency of phenological stage across species and sites for each measurement campaign.

Water Relations

A Scholander-type pressure chamber (Model 600, PMS Instrument Co., Albany, OR) fitted with a grass compression gland and base was used to make predawn (Ѱ_pre) and midday (Ѱ_mid) measurements of leaf water potential. Leaf blades were cut slightly above the ligule using a razorblade, wrapped in plastic, and pressurized in the chamber until sap was visibly exuded from veins, and equilibrium pressure was recorded. Four leaf blades were used for each of the measurement intervals at each site. Predawn measurements were made between 4:30 and 6:30 AM. Midday measurements were made between 1:30 and 3:30 PM to assess the water status of grasses during the most stressful point in the day.

Gas Exchange

            At each of the three sites, two approximately 20 x 30 m plots were established with two grass patches per species in each plot used for gas exchange measurements (net photosynthesis, A_net; stomatal conductance to water vapor, g_s; transpiration, E; instantaneous water use efficiency, WUE_i (A_net/E) and intrinsic water use efficiency, WUE_g (A_net/g_sw). A LI-6800 Portable Photosynthesis System fitted with a 2x3 cm side-to-side clear-top chamber and 3x3 cm light source was used for gas exchange measurements. Chamber settings were adjusted to mimic ambient conditions: light source was set to 1,500 µmol m⁻² s⁻¹, source carbon dioxide to 420 ppm, fan speed to 10,000 rpm, flow rate of 700 µmol s⁻¹, and temperature and relative humidity adjusted to track ambient conditions measured using a Kestrel 3000 Weather Meter throughout the day. Measurements were made beginning at approximately 9:30 AM each day after the dew had dissipated and warm-up and system tests had been performed. The infrared gas analyzers were matched in between each measurement and instantaneous measurements were logged three times per leaf blade once steady-state conditions were reached in the chamber (stable ΔCO₂ and ΔH₂O, positive intercellular CO₂ [C_i]). The middle section of one or two leaf blade(s) was placed in the chamber for each species per patch per plot (8 individuals per species per day). Leaf tissue within the chamber was collected and placed in plastic bags with wetted paper towels on ice until corrections could be made for the leaf area within the chamber at the end of each field day using an Epson Perfection V39 flatbed scanner. Species measurements were randomly rotated so that the same species was not measured in the same order or during the same time of day. Gas exchange measurements were discarded if accidentally logged before reaching a steady state or if very low negative values of C_i were observed.

Stable Isotopes (δ¹³C, δ¹⁸O)

            Approximately 15-20 leaf blades were collected at each site per plot per species for stable isotope analysis. Leaf tissue was dried at 65ºC for three days and ground using a Retesch Oscillating Mixer Ball Mill MM400 to homogenize whole tissue samples (approx. 10-20 leaf blades). Bulk leaf δ¹³C was determined using an Elemental Analyzer (Costech Analytical Technologies, Inc., Valencia, CA, USA) coupled to an Isotope Ratio Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Oxygen stable isotopes (δ¹⁸O) were also determined using an Elemental Analyzer coupled to an Isotope Ratio Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Isotopic ratios of bulk leaf tissue were expressed as δ¹³C or δ¹⁸O and calculated as follows:

                                                δ = [(R_sample/R_standard) – 1]/1000

where R_sample and R_standard are the ratios of heavy to light isotope of sample leaf tissue (¹³C/¹²C or ¹⁸O/¹⁶O) and international standards respectively (here Vienna Pee Dee Belemnite, VPDB and Vienna Standard Mean Oceanic Water, VSMOW). All analyses were corrected to international standards with in-house reference materials and conducted at the Stable Isotopes for Biosphere Science Laboratory (http://sibs.tamu.edu).

Morphology

            We also measured specific leaf area (SLA) and leaf width (W_leaf) in order to understand intraspecific and interspecific patterns in C₄ leaf characteristics along the aridity gradient. Eight leaf blades per species (4 per plot) were collected at each site, placed in a plastic bag with a wet paper towel, and scanned using an Epson Perfection V39 flatbed scanner at the end of each field day. ImageJ was used to calculate the surface area of each leaf. Leaves were then dried for 3 days in a 65ºC oven and weighed. Specific leaf area was expressed as the leaf dry mass (g) per one-sided leaf area (cm²) of each individual leaf blade. Leaf widths were obtained from the leaf area corrections made for gas exchange measurements, with width calculated from the average of the widths at two ends and midpoint of each 3 cm leaf clipping.

Statistical Analyses

            In 2020 and 2021, our westernmost site (Read) experienced abnormally dry conditions throughout the growing season and in the months preceding (U.S. Drought Monitor, 2023), which severely limited leaf-out of C₄ grasses and prevented measurements at this site. For this reason, we ran a series of two-way analyses of variance (ANOVA, type III sums of squares to account for unbalanced data) with site and species as the main effects and A_net, g_sw, E, WUE_i, WUE_g, Ψ_pre, Ψ_mid, SLA, width, δ ¹³C, δ¹⁸O as response variables for the 2019 measurement campaign only in order to evaluate water relations and physiology with increasing aridity across the gradient, species-specific responses, and potential differences in species responses across sites. Gas exchange data were transformed using Box-Cox transformations to meet the assumptions of ANOVA (A_net, g_sw, WUE_i, WUE_g square root transformed, and E log-transformed). Post-hoc Tukey tests were performed to separate statistically different means in cases in which main effects were found to be significant, with α = 0.05 for all analyses. Since the full gradient could not be assessed in subsequent years, data for all measurement campaigns (July 2019 [Read, Sonora, Martin], June 2020 [Sonora, Martin], May 2021 [Sonora, Martin], September 2021 [Sonora, Martin]) were pooled and relationships between gas exchange, water relations, isotopes, and morphological variables assessed with Pearson correlations to understand interactions between physiological variables and water status of different grass species. To assess the sensitivity of stomata to changes in aridity, we also examined the logarithmic relationship between stomatal conductance and mean daytime VPD, as well as the degree of isohydry based on relationships between midday and predawn water potential for each species. We also examined the relationship between leaf width or SLA with WUE to assess whether features of morphology were associated with higher water use efficiency among our study species. Variables for regression relationships were averaged for each species by location, measurement date, plot, and patch. All statistical analyses were performed in R version 3.1.1 (R Development Core Team 2014).