Warming Reduces Net Carbon Gain and Productivity in Medicago sativa L. and Festuca arundinacea

: High temperature stress imposes constraints on the productivity of agricultural systems, such as pastures, and predicted increases in global temperatures are set to exacerbate these limitations. Here, we sought to understand the impact of warmer growth temperature on gas exchange and net primary productivity for two widely cultivated pasture species. We grew a C 3 legume, Medicago sativa (lucerne), and a C 3 grass, Festuca arundinacea Schreb. (tall fescue), in a climate-controlled facility exposed to two temperature treatments (ambient: 26 ◦ C, aT; elevated: 30 ◦ C, eT). Soil water was maintained at non-limiting conditions in both temperature treatments to control for the confounding e ﬀ ects of warming on soil moisture. We found that warming reduced photosynthetic capacity and increased leaf dark respiration (R dark ) in lucerne, while tall fescue showed little physiological change at the leaf level, but increased ecosystem respiration (R eco ). Growth temperature had no signiﬁcant impact on the thermal optimum of photosynthesis (T opt ) or water use e ﬃ ciency in either species. Both species exhibited signiﬁcant reductions in productivity with warming; lucerne had greater reductions in shoot biomass, while tall fescue had greater reductions in root biomass. Our results highlight the potential for signiﬁcant declines in pasture productivity associated with even modest increases in average temperature and highlights the need for suitable management strategies and implementation of more heat-resistant cultivars. Improvements in photosynthetic performance for greater heat tolerance in lucerne, and traits associated with biomass allocation and root performance at higher temperatures in tall fescue, should be the focus for improving high temperature resistance in these plant species.


Introduction
Natural and anthropogenic factors have accelerated climate change, with global surface temperatures predicted to increase by 1.1-6.4 • C by 2100 [1]. Global warming can have significant impacts on plant carbon (C) assimilation (photosynthesis) and C loss (respiration) because these processes are highly temperature sensitive [2][3][4]. Photosynthesis and respiration regulate net primary productivity of ecosystems and the balance between these two processes determines whether an ecosystem is a net C sink or source [5]. High temperatures already impose major limitations on the productivity of temperate agricultural systems including pastures [6,7], and the response of these systems to predicted increases in global temperature will vary in both magnitude and direction across biomes, depending on species composition and prevailing climatic conditions [8][9][10][11][12][13]. A large to water stress [49], N fertilisation [50,51], shading [52], or only investigated daytime respiration [46]. Importantly, the responses of photosynthesis and respiration to temperature can be confounded by the effect of temperature on evapotranspiration and plant water status [53], which were rarely accounted for in previous studies.
Increases in air temperature will also affect the hydrological cycle, increasing aridity and drought frequency; hence, understanding plant-water relations in the context of warming is needed for predicting shifts in plant productivity. The ratio of carbon assimilated per unit of water lost, commonly referred to as plant water use efficiency (WUE), is a key ecosystem property linking global carbon and water cycles [54]; high WUE is a desirable trait in plant cultivar selection and breeding [55] as it may reduce the need for costly irrigation. Temperature can significantly influence plant WUE as warming results in higher evaporative demand and therefore increases transpiration rates, but may have little to no impact on photosynthetic carbon uptake [56,57], effectively reducing WUE. WUE can be estimated at a physiological level using leaf gas exchange measurements and is usually defined as the ratio of net light saturated photosynthesis (A sat ) to stomatal conductance for water vapour (g s )-intrinsic water-use efficiency (WUE i ; [58]) or the ratio of A sat to transpiration (E)-instantaneous water-use efficiency (WUE inst ; [59]). These metrics are however, significantly impacted by atmospheric vapour pressure deficit (VPD) [60], which is known to increase with temperature, potentially confounding the direct impacts of temperature. In order to address this issue, we used the optimal stomatal behavior parameter g1, which is inversely related to the marginal cost of water and VPD [61]. It has been shown that g1 is representative of plant water use strategy across plant functional types and ecosystems, and therefore allows WUEi to be compared across environmental conditions [61].
Here, we investigated the effects of warming on the photosynthetic capacity, leaf dark respiration (R dark ), daytime ecosystem respiration (R eco ), WUE and productivity of lucerne and tall fescue. Lucerne is a temperate legume and is the most important and most widely grown forage legume in the world with~32 Mha cultivated [62,63]. Lucerne provides high-quality feed for livestock and fixes atmospheric N 2 , thereby improving soil nutrient status, making it important for pasture productivity worldwide [64]. Tall fescue is highly productive and used extensively for forage in all classes of livestock and is adapted to a wide range of growing conditions [14,65]. We grew seedlings of lucerne and tall fescue in a climate-controlled facility in ambient temperature (26 • C, aT) or ambient temperature +4 • C (eT); aT represented the average daily maximum temperature for Richmond, NSW, Australia over the last 20 years, and eT represented the predicted maximum temperature increase of 4 • C for this region, within this century [7]. We hypothesized that warming would increase leaf level photosynthesis, but that concurrent increases in respiration would lead to a reduction in overall leaf-level C uptake, with resulting declines in productivity. We also hypothesised that tall fescue would be less negatively affected by warming because it has a higher T opt than lucerne [49,66]. Finally, we hypothesised that increased evaporative demand under eT would lead to higher levels of water loss via transpiration and an overall reduction in WUE for both species.
Soil was collected from the field at Western Sydney University (sandy-loam texture with a pH~5.6), sieved (5 mm), and mixed with quartz sand (soil to sand ratio of 7:3, v/v). Plastic pots (3.7 L, 15 cm diameter, 24 cm height) were filled with~3.9 kg soil mixture. Seeds were surface sterilized with a 1.25% bleach solution for 10 min, then rinsed 10 times with deionized water, and germinated in Petri dishes with sterilized water for 1 week (Zhang et al., 2020 in review). Five seedlings were transplanted into each pot and then thinned to four healthy individuals per pot after 2 weeks. All lucerne individuals were inoculated with Easy Rhiz soluble legume inoculant (Group AL, New Edge Microbials, NSW, Australia) and assessed for nodulation during the final harvest to ensure all lucerne individuals were well inoculated. Plants were watered using an automated irrigation system, with top-up hand watering conducted every other day to ensure soil water was maintained at similar levels between temperature treatments (80% field capacity), and was non-limiting for growth. Overall, the experiment contained 4 treatment combinations (2 temperatures × 2 species) and eight replicates for each treatment (divided between two growth chambers), generating 32 pots in total.
During the experiment, pots were randomly reallocated among positions within chambers every two weeks to minimize potential within-chamber effects. To mimic standard practices in pasture management associated with cut-and-carry operations, aboveground plant material was clipped~5 cm above the soil surface at 8 weeks after planting (WAP). At this time, fertiliser was added in the form of KNO 3 and KH 2 PO 4 , at a rate equivalent to 30 kg N ha −1 and 5 kg P ha −1 . Plants were then allowed to regrow for a further 8 weeks before they were harvested.

Photosynthetic Capacity
Gas exchange measurements were conducted on attached, recently fully expanded leaves (RFEL) using a portable open gas exchange system (LI-6400XT, LI-COR Inc., Lincoln, NE, USA). Leaf areas for all leaf samples of lucerne used in gas exchange analyses were calculated by capturing high resolution photographs on a pre-calibrated surface and then analysing leaf area using Image J (National Institutes of Health). Leaf area for tall fescue was kept constant by filling the cuvette during each measurement, taking care to prevent overlapping of leaves.
The response of net photosynthesis (A) to intercellular CO 2 concentration (Ci) was measured thirteen WAP (weeks after planting) or 5 weeks after clipping. Measurements were taken between 10:00 and 15:00 AEST, at a saturating PPFD of 1800 µmol m −2 s −1 , and at treatment specific mid-day air temperatures (26 or 30 • C). Leaf-to-air vapour pressure deficit (VPD L ) levels were on averagẽ 1.2 kPa and~1.6 kPa for plants under aT and eT, respectively. Steady state measurements of light-saturated photosynthesis (A sat ; µmol m −2 s −1 ), stomatal conductance (g s ; mol m −2 s −1 ) and transpiration (E; mmol m −2 s −1 ) were conducted at the beginning of each A/Ci curve at a cuvette reference [CO 2 ] (C r ) of 420 µmol CO 2 mol −1 [70]. A/Ci curves were subsequently conducted by measuring A sat at 15 additional C r points (5,50,100,150,200,250, 300, 420, 500, 650, 800, 1000, 1250, 1500 and 1800 µmol CO 2 mol −1 ) and each leaf was allowed~2 min to equilibrate to each C a before conducting measurements [71]. Time constraints resulted in measurements only being conducted on 6 of the 8 replicate plants per species, per temperature treatment, giving a total of 24 plants. A/Ci curves were parameterized using the Farquhar, von Caemmerer & Berry model of C 3 photosynthesis [2] for maximum rate of Rubisco carboxylation (V cmax ; µmol m −2 s −1 ) and the apparent maximum rate of electron transport for RuBP regeneration (J max ; µmol m −2 s −1 ). Analyses were conducted using the fitaci function within the plantecophys package [72] in R version 4.0 (R Development Core Team, 2020) without constraining mesophyll conductance [73]. To make V cmax and J max values comparable across species and growth temperatures, estimated V cmax and J max values were corrected for temperature and expressed at 25 • C using the default Tcorrect parameters included in the fitaci function. This allows for V cmax and J max values to be expressed at a common temperature using modified Arrhenius-type equations that involve previously determined values for activation energy (J mol −1 ), delta S (J mol −1 K −1 ) and peakedness of the curve (EdV) [12], for more details see https://github.com/RemkoDuursma/plantecophys/blob/master/vignettes/new_T_responses.Rmd. CO 2 -saturated assimilation rates (A max ) were determined at a C i of 1500 µmol CO 2 mol −1 by fitting each A/Ci data set to the Farquhar, von Caemmerer & Berry model [2].
The responses of leaf photosynthesis (A) to irradiance (photosynthetic photon flux density (PPFD)) were measured immediately following the completion of A/Ci curves on the same leaves. A/PPFD curves were measured by first exposing each leaf to a PPFD of 1800 mmol m −2 s −1 with sample chamber [CO 2 ] (C s ) at 400 µmol CO 2 mol −1 [74] until steady state of net CO 2 fixation was reached. Irradiance was then reduced in a stepwise manner and A was measured at 13 additional PPFD points (1400, 1200, 1000, 600, 300, 200, 150, 100, 80, 60, 40, 20 & 0 µmol m −2 s −1 ). Each leaf was allowed 2 min to equilibrate to each PPFD before measurements were taken. Measurements were conducted on 6 replicate plants per species per temperature treatment for a total of 24 plants. A/PPFD curve parameters were estimated using nonlinear least squares regression of a nonrectangular hyperbola (Equation (1); [74,75]) using the nlsLM function within the minpack.lm package in R version 4.0 (R Development Core Team, Vienna, Austria, 2020).
where A net and A max-light are the area-based net and maximum gross photosynthetic rates (µmol m −2 s −1 ) respectively, Φ is the apparent quantum yield (mol CO 2 mol −1 photons); R d is the daytime dark respiration rate (A net at PPFD = 0 µmol m −2 s −1 ); ∅ is the curve convexity (dimensionless). The light saturation point (LSP) was estimated as the PPFD when 80% of A max (model asymptote) was achieved and the light compensation point (LCP) was estimated from the x-axis intercept (A net = 0) [75]. The responses of A sat to leaf temperature (A/T curves) were measured 1 week after A/Ci curves were measured (14 WAP) on the same individuals used for A/Ci curves (6 replicates per species per treatment). Measurements were taken at a PPFD of 1800 µmol m −2 s −1 and 420 µmol CO 2 mol −1 . All plants to be measured were transferred to an empty glasshouse chamber, which was maintained at 10 • C, on the night before measurement. The A/T curves began at 09:00 and A sat was measured on all plants at 10 • C. A/T curves were subsequently analysed by measuring A sat at 6 additional temperatures (15,20,25,30,35 and 40 • C) by raising the temperature of the glasshouse chamber and the block temperature of the leaf gas exchange cuvette in a step-wise manner. Temperatures were raised every hour, and plants were allowed to acclimate to each temperature for a minimum of 30 min before measurements were taken.
The A/T curves were fit with an empirical temperature response function (Equation (2); [76,77]) which describes the temperature response in terms of the optimum (T opt ) and minimum temperatures, and the curvature coefficient q. All analyses were conducted using the nls function in R.
Here T mn and T mx are the minimum and maximum temperatures respectively, while T r is a reference temperature (20 • C) and q is the curvature coefficient [77].
Spot measurements of leaf gas exchange were conducted three times during the experiment: (1) 8 WAP, immediately prior to plants being clipped; (2) 12 WAP, one month following clipping; and (3) 16 WAP, at the end of the experiment. Measurements of A sat , stomatal conductance (g s ) and transpiration (E) were conducted on RFEL at respective mid-day growth temperatures (26 or 30 • C) using a photosynthetic photon flux density (PPFD) of 1800 µmol m −2 s −1 , a reference [CO 2 ] of 400 µmol CO 2 mol −1 and a flow rate of 500 µmol s −1 . VPD L was maintained between 1-2.0 kPa and each leaf Agronomy 2020, 10, 1601 6 of 19 was allowed 5-10 min to equilibrate before measurements were taken [71]. Measurements were taken between 10:00 and 15:00 AEST and conducted on the same four replicate individuals of each treatment at each time point.

Respiration
Leaf dark respiration (R dark ) was measured using a portable open gas exchange system (LI-6400XT, Li-Cor, Lincoln, NE, USA) on the week following the measurements of A/T curves on the same individuals (15 WAP). All measurements were conducted at night in the dark between 22:00 and 24:00 on RFEL at respective night-time air temperatures (18 or 22 • C for aT and eT, respectively) using a photosynthetic photon flux density (PPFD) of 0 µmol m −2 s −1 , a reference [CO 2 ] of 420 µmol CO 2 mol −1 , humidity maintained at ambient chamber conditions and a flow rate of 300 µmol s −1 . Ecosystem respiration (R eco ) was analysed one day before the final harvest (16 WAP) on 4 replicates of each species under each temperature treatment. Each pot was fitted with a removable, airtight chamber, composed of a non-transparent PVC tube (15 cm in diameter × 35 cm in height, 7 L volume), allowing the headspace to be closed for gas measurements [78]. The measurement chambers were installed on each pot, water was added to the bottom tray during measurement to ensure no gas leakage occurred, and 20 mL gas was sampled after 0, 15 and 30 min using a syringe. Each gas sample was stored in a 12 mL vial and transported to the laboratory. All gas samples were analysed using a headspace autosampler (Teledyne Tekmar, Mason, OH, USA) connected to a 5890-gas chromatograph (Agilent/Hewlett-Packard, Palo Alto, CA, USA) equipped with a thermal conductivity detector (TCD) to quantify the concentration of CO 2 (Zhang et al., in review 2020). Rates of CO 2 flux were determined by measuring the change in gas concentration over time within the headspace (Collier, Ruark, Oates, Jokela, & Dell, 2014). All measurements were standardized by the total dry biomass (both aboveground and belowground biomass) within each pot.

Leaf Level Water Use Efficiency (WUE)
Instantaneous WUE (WUE inst ) was calculated from gas exchange measurements as the ratio of carbon assimilated (A sat ) to water lost through transpiration (E) [79]. Intrinsic WUE (WUEi) was calculated as the ratio of A sat to g s . Calculations of WUE are influenced by atmospheric vapour pressure deficit (VPD). In order to account for this, we used the g1 parameter from a model of stomatal conductance that accounts for VPD (Equation (3); [61]): where C s and D (kPa) are the CO 2 concentration and the VPD at the leaf surface, respectively, and A is the net assimilation rate [80].

Biomass and Shoot Nutrients
During the final harvest, plant shoots were cut at the soil surface, and placed in paper bags. The pots were completely emptied, then roots were washed and collected from the soil and placed in paper bags. All samples were dried at 65 • C for 48 h and weighed and a subset of the dried shoot biomass was separated, weighed and then ground to a fine powder using a mixer mill (Retsch ® MM200; Haan, Germany). The ground sample was analysed for P concentrations using an Epsilon-3× X-ray fluorescence spectrometer (PANalytical, EA Almelo, The Netherlands) as described in [81]. C and N concentrations were measured using an isotope-ratio mass spectrometer (Stable Isotope Facility, UC Davis, Davis, CA, USA; IRMS; Deltaplus XP and Delta C prototype Finnigan MAT, respectively, Finnigan MAT, Bremen, Germany; 0.1% precision). Due to the internal airflow system in our glasshouse, where expelled CO 2 from each chamber is recycled and CO 2 concentration is controlled by the addition of fresh air, the proportion of fresh air to expired air varied between chambers. Hence, Agronomy 2020, 10, 1601 7 of 19 the isotopic value from the respired CO 2 and fresh air varied, and therefore the plant isotopic value could not be interpreted and was excluded from the study.

Statistics
All statistical analyses were carried out using the software R version 4.0.0 (R Core Development Team 2020). In order to calculate species differences and any interaction between temperature and species, we applied a two-way ANOVA with temperature and species as fixed effects and chamber as a random effect, using the "lmer" function from the R-package 'lme4' [82]. We tested for statistical differences between temperature treatments by conducting pair-wise t-tests for each species using the "emmeans" function from the R-package "emmeans" [83]. All data were analysed for normality and homoscedasticity and, where necessary log-transformed to fit these assumptions.

Carbon Balance
At the leaf level, we observed a significant interaction between species and growth temperature for A sat (p = 0.03; Table 1); warming resulted in a small (non-significant) increase in A sat for tall fescue (+11%; Figure 1a), but significantly reduced A sat for lucerne (−31%, p = 0.03). At the leaf level, warming had no significant impact on R dark in tall fescue, but significantly increased R dark in lucerne by 173% (p = 0.02; Figure 1b). Overall, we found that warming significantly increased the ratio of R dark to A sat in lucerne by 218% (p = 0.01; Figure 1c), but had no significant impact in tall fescue. At the whole pot level, warming did not affect the R eco for lucerne, but led to a 56% increase in tall fescue (p = 0.03; Figure 1d). Table 1. F values of linear mixed effects models and their interactions on parameters for lucerne (Medicago sativa) and tall fescue (Festuca arundinacea), with Species (S) and growth temperature (T) as fixed effects and growth chamber as the random effect. Statistical significance denoted by * (p < 0.05), ** (p < 0.01) & *** (p < 0.001). Variables were log transformed to meet assumptions of normality. # G1 is a modelled output parameter, hence could not be tested for statistical significance. (+11%; Figure 1a), but significantly reduced Asat for lucerne (−31%, p = 0.03). At the leaf level, warming had no significant impact on Rdark in tall fescue, but significantly increased Rdark in lucerne by 173% (p = 0.02; Figure 1b). Overall, we found that warming significantly increased the ratio of Rdark to Asat in lucerne by 218% (p = 0.01; Figure 1c), but had no significant impact in tall fescue. At the whole pot level, warming did not affect the Reco for lucerne, but led to a 56% increase in tall fescue (p = 0.03; Figure 1d). Error bars indicate ± standard error. Statistically significant differences between temperature treatments are portrayed as * (p < 0.05), ** (p < 0.01).

Photosynthetic Capacity
Warming had no significant impact on V cmax (Figure 3a) or J max (Figure 3b) for both species, but led to a significant decline in V cmax25 (−44%: Figure 3b) and J max25 (−33%: Figure 3d) for lucerne. Warming had no significant impact on A max (Figure 3g; Table 1), apparent quantum efficiency (Φ) (Supplementary Materials Table S1), A max-light or light compensation point (LCP) (µmol m −2 s −1 ) for either species. We did, however, observe a significant interaction between species and temperature for A max (Table 1; p = 0.046), J max (p = 0.03) and J max25 (p = 0.033): Tall fescue exhibited an increase in A max (+22%: Figure 3g), J max (+21%: Figure 3c) and J max25 (+6%: Figure 3d), but lucerne exhibited a decrease in A max (−19%), J max (−31%) and J max25 (−33%) under eT. We also observed a significant interaction between species and temperature for V cmax25 (p = 0.012): lucerne exhibited a much larger decrease in V cmax25 (−46%: Figure 3b) than tall fescue (−18%). Warming had no significant impact on J max : V cmax at the growth temperature (Figure 3e) or when corrected to 25 • C (Figure 3f) for both species. Spot measurement of gas exchange conducted throughout the experiment showed no significant change in rates of A sat between temperature treatments for both species ( Figure S1).

Photosynthetic Capacity
Warming had no significant impact on Vcmax (Figure 3a) or Jmax (Figure 3b) for both species, but led to a significant decline in Vcmax25 (−44%: Figure 3b) and Jmax25 (−33%: Figure 3d) for lucerne. Warming had no significant impact on Amax (Figure 3g; Table 1), apparent quantum efficiency (Φ) (Supplementary Materials Table S1), Amax-light or light compensation point (LCP) (µmol m −2 s −1 ) for either species. We did, however, observe a significant interaction between species and temperature for Amax (Table 1; p = 0.046), Jmax (p = 0.03) and Jmax25 (p = 0.033): Tall fescue exhibited an increase in Amax (+22%: Figure 3g), Jmax (+21%: Figure 3c) and Jmax25 (+6%: Figure 3d), but lucerne exhibited a decrease in Amax (−19%), Jmax (−31%) and Jmax25 (−33%) under eT. We also observed a significant interaction between species and temperature for Vcmax25 (p = 0.012): lucerne exhibited a much larger decrease in Vcmax25 (−46%: Figure 3b) than tall fescue (−18%). Warming had no significant impact on Jmax: Vcmax at the growth temperature (Figure 3e) or when corrected to 25 °C (Figure 3f) for both species. Spot measurement of gas exchange conducted throughout the experiment showed no significant change in rates of Asat between temperature treatments for both species ( Figure S1).  Growth temperature had no significant impact on the short-term temperature response of photosynthesis for either species (Figure 4; Table 1). T opt in tall fescue grown under aT and eT was 29 • C and 28 • C, respectively, while T opt in lucerne was~30 • C for individuals grown under both aT and eT. We observed no significant difference in the T opt between species and no interaction between species and growth temperature.
Agronomy 2020, 10, x FOR PEER REVIEW 10 of 19 Growth temperature had no significant impact on the short-term temperature response of photosynthesis for either species (Figure 4; Table 1). Topt in tall fescue grown under aT and eT was ~29 °C and 28 °C, respectively, while Topt in lucerne was ~30 °C for individuals grown under both aT and eT. We observed no significant difference in the Topt between species and no interaction between species and growth temperature.

WUE
Despite a small trend toward a reduction in WUE for both species under warming, no significant change in WUEi (Figure 5a) or WUEinst (Figure 5b) was observed in either species (Table 1). We also observed no significant change in WUEi normalised for VPD and expressed as g1 (Figure 5c).

WUE
Despite a small trend toward a reduction in WUE for both species under warming, no significant change in WUE i (Figure 5a) or WUE inst (Figure 5b) was observed in either species (Table 1). We also observed no significant change in WUE i normalised for VPD and expressed as g 1 (Figure 5c). Growth temperature had no significant impact on the short-term temperature response of photosynthesis for either species (Figure 4; Table 1). Topt in tall fescue grown under aT and eT was ~29 °C and 28 °C, respectively, while Topt in lucerne was ~30 °C for individuals grown under both aT and eT. We observed no significant difference in the Topt between species and no interaction between species and growth temperature.

WUE
Despite a small trend toward a reduction in WUE for both species under warming, no significant change in WUEi (Figure 5a) or WUEinst (Figure 5b) was observed in either species (Table 1). We also observed no significant change in WUEi normalised for VPD and expressed as g1 (Figure 5c).

Shoot Stoichiometry
Warming had no significant impact on shoot N for tall fescue, but led to a 38% increase in lucerne (Figure 6a, p = 0.049). Warming had no impact on shoot P for either species, but caused a significant increase in N:P ratio (+69%, p = 0.043) for tall fescue. Both tall fescue and lucerne incurred a significant reduction in C:N ratio by 26% (Figure 6d, p = 0.009) and 21% (p = 0.026), respectively.

Shoot Stoichiometry
Warming had no significant impact on shoot N for tall fescue, but led to a 38% increase in lucerne (Figure 6a, p = 0.049). Warming had no impact on shoot P for either species, but caused a significant increase in N:P ratio (+69%, p = 0.043) for tall fescue. Both tall fescue and lucerne incurred a significant reduction in C:N ratio by 26% (Figure 6d, p = 0.009) and 21% (p = 0.026), respectively.

Discussion
Here we sought to investigate the effect of climate model-predicted increases in temperature on the physiology and productivity of two widely cultivated, and economically important [62,67,68], pasture species. We found that a 4 °C increase in growth temperature had a more significant impact on the temperate legume (lucerne) than on the C3 grass (tall fescue), despite similar climate niches in pastures. Lucerne experienced reductions in photosynthetic capacity and increased Rdark under eT, while tall fescue showed an increase in Reco despite little change in physiological function at the leaf level. The results for lucerne support our first hypothesis, that photosynthetic upregulation under warming would be counterbalanced by a corresponding increase in respiration rates; however, this was not observed in tall fescue. Despite some physiological adjustment in tall fescue, both species exhibited significant declines in biomass productivity, contrary to our second hypothesis that tall fescue would be less negatively impacted than lucerne, due to its higher normal growing temperature of 15-30 °C. We also found little evidence to support our third hypothesis, that warming would lead to a reduction in plant water use efficiency.
Despite numerous global predictions suggesting widespread gains in photosynthesis with increasing temperatures, research examining these relationships in temperate pastures and related species have shown mixed results [84][85][86]. Here, we observed a significant interaction between species and growth temperature; lucerne exhibited a reduction in photosynthetic capacity under eT, while tall fescue exhibited a small increase. Our findings are in line with previous studies on tall fescue that observed little change in photosynthetic capacity between 25 to 30 °C [46] or found an increase in photosynthesis between 15-35 °C over the short term [14,87]. Studies have however, found

Discussion
Here we sought to investigate the effect of climate model-predicted increases in temperature on the physiology and productivity of two widely cultivated, and economically important [62,67,68], pasture species. We found that a 4 • C increase in growth temperature had a more significant impact on the temperate legume (lucerne) than on the C 3 grass (tall fescue), despite similar climate niches in pastures. Lucerne experienced reductions in photosynthetic capacity and increased R dark under eT, while tall fescue showed an increase in R eco despite little change in physiological function at the leaf level. The results for lucerne support our first hypothesis, that photosynthetic upregulation under warming would be counterbalanced by a corresponding increase in respiration rates; however, this was not observed in tall fescue. Despite some physiological adjustment in tall fescue, both species exhibited significant declines in biomass productivity, contrary to our second hypothesis that tall fescue would be less negatively impacted than lucerne, due to its higher normal growing temperature of 15-30 • C. We also found little evidence to support our third hypothesis, that warming would lead to a reduction in plant water use efficiency.
Despite numerous global predictions suggesting widespread gains in photosynthesis with increasing temperatures, research examining these relationships in temperate pastures and related species have shown mixed results [84][85][86]. Here, we observed a significant interaction between species and growth temperature; lucerne exhibited a reduction in photosynthetic capacity under eT, while tall fescue exhibited a small increase. Our findings are in line with previous studies on tall fescue that observed little change in photosynthetic capacity between 25 to 30 • C [46] or found an increase in photosynthesis between 15-35 • C over the short term [14,87]. Studies have however, found marked reductions in photosynthetic capacity for tall fescue after long-term exposure to 35 • C [88], even in a putative heat-tolerant cultivar [6]. Our results for lucerne disagree with previous literature, which suggests that lucerne has a broad temperature optimum and experiences little to no change in carbon assimilation between 5-30 • C [20,47,84,89,90] and others who observed the highest rate of photosynthesis at 34 • C [91]. Our results do, however, agree with those of Pearson & Hunt (1972) [92] who found a steady decline in photosynthesis between 10 & 40 • C. Differences in these studies may be due to differences in light intensity, other environmental conditions and timing of measurements. Furthermore, many of the earlier experiments were conducted at the whole pot level and may not account for temperature-derived changes in soil C flux. Our results for lucerne are similar to field-based experiments, in which experimental warming increased photosynthesis in the grass Festuca ovina, but reduced photosynthesis in lucerne [23].
We found little evidence of photosynthetic acclimation with both species having a similar T opt of 30 • C at both growth temperatures. Our results agree with Brown and Radcliffe (1986), who observed a T opt for photosynthesis of 30 • C. Therefore, air temperatures that generate leaf temperatures above 30 • C will most likely lead to reductions in photosynthesis. Our results are in contrast to previous studies on similar species, which observed that T opt of photosynthesis was influenced to a greater degree by growth temperature than either species or climate-of-origin, with plants grown under lower temperatures showing markedly different T opt when grown at temperatures ranging from 15 to 35 • C [48,85,87,89].
At the leaf level, warming significantly increased R dark in lucerne, but had no impact on tall fescue. This is in line with previous long-term studies on lucerne, which observed a 54% increase in R dark when grown at 25 & 30 • C [93] and a 60% increase in respiration when lucerne was grown at 21 • C and 34 • C [91]. A previous study on tall fescue by Yu et al., (2012) [46] observed an increase in day-time respiration when plants were exposed to an increase in temperature from 25 • C to 30 • C; however, this effect disappeared after 28 days of high temperature exposure. Our results are similar to a meta-analysis [4] which found that warming, ranging from 0.5 to 29.6 • C, generally led to a greater increase in respiration for C 3 forbs (+57%) than grasses (+20%).
At the level of the plant-root-soil system, we found warming significantly increased R eco for tall fescue but had no impact on lucerne. Given little change in leaf-level respiration (R dark ), the increase in R eco for fescue under eT may be driven by increased respiration belowground (i.e., root or microbial respiration). Previous studies have shown contradictory results, with a recent meta-analysis suggesting a general increase in belowground respiration with temperature [94] while other studies have found no consistent effect [95,96] acclimation over time [97] or a reduction in respiration under warming [98]. Many of these studies, however, did not control for changes in soil water content or plant biomass in response to warming. Furthermore, previous studies have found that warming significantly increases belowground C allocation via root exudates under N limiting conditions [99][100][101]. In our study, tall fescue was severely limited by N while this was not the case in lucerne, due to high levels of nodulation and N fixing activity in lucerne (personal communication H. Zhang). The paradigm that short-term stimulation of respiration under warming declines with exposure time, and that respiration acclimates to growth temperature over the long-term [3,8,22,[43][44][45], was not observed in our species after four months of growth at warmer temperatures.
Our results indicate that warming amplified net C loss for both species, but in different ways: via belowground respiration in tall fescue and via leaf respiration in lucerne and, the overall trend towards a reduction in net C was highlighted by the significant reductions in biomass for both species. Our results for lucerne are similar to studies that suggested an optimum temperature for growth between 15-25 • C in the daytime and 10-20 • C in the night-time [14,89,91], but contrasts with studies showing no reduction in growth between 10-30 • C, with marked reductions being observed only at temperatures around 40 • C [92,102,103]. Previous studies on tall fescue have observed similar trends with festucoid species observed to have an optimum growth temperature between 20-27 • C [85,87,104,105]. The significant declines in biomass for tall fescue, despite little change in photosynthetic capacity, suggests that whole-plant productivity may be more sink-limited than source-limited [106], with productivity regulated by tiller production & leaf canopy orientation [107]. The significant decline in productivity for both species under warming may also be linked to the significant reduction we observed in C:N ratio, which suggests a decline in long term nitrogen use efficiency [108].
Growth temperature had no significant impact on leaf level water use efficiency in either species. This may be due to our careful control of soil moisture, thereby removing potential indirect effects of temperature on WUE via changes in soil moisture and the transient effect of temperature on g s which disappeared over the long-term [6]. Our results are consistent with other long-term studies in cotton (Gossypium hirsutum L.) grown at 28 and 32 • C [109] and in soybean [Glycine max (L.) Merr.] grown at 28 and 32 • C [110], while field studies on rice observed a reduction in WUE between 26 & 30 • C [111]. Short-term studies on tall fescue have observed a decrease in WUE inst in plants exposed to high temperature stress (35 • C) of 20 days when compared to those grown at 20 • C [6], while Yu et al., 2012 found no significant change in WUE for tall fescue grown at 30 • C after 28 days of exposure, when compared to those grown at 25 • C.

Conclusions
For plants to successfully acclimate to higher growth temperature, they must increase photosynthetic capacity to counteract temperature-driven increases in respiration. Numerous studies have suggested that warming may have a stimulatory impact on photosynthetic capacity globally, leading to enhanced net C balance of leaves and increasing primary productivity and terrestrial biomass (see Liang et al. 2013 and references therein).
However, our results obtained under controlled environment conditions in a greenhouse suggest that in the warm climate of Australia, an increase in temperature may lead to an overall reduction in net C balance and large reductions in the productivity of pasture systems. Breeding and development of more heat-tolerant cultivars of economically important pasture species may be required to maintain current levels of pasture productivity.