Characterizing the drivers of seedling leaf gas exchange responses to warming and altered precipitation: indirect and direct effects

Climate change is expected to bring warmer temperatures and more variable precipitation patterns worldwide, patterns that will depend on the ability of the world's flora to take up carbon under these new conditions. We subjected deciduous tree seedlings growing in an old-field ecosystem in Massachusetts, USA to warming and altered precipitation. We found that leaf carbon uptake was greatest under the coolest, wettest conditions, an effect driven by increased soil water availability in these plots. Our findings suggest that warming may reduce leaf carbon uptake by decreasing soil moisture, an effect that will be exacerbated during drought periods.

Abstract

Anthropogenic forces are projected to lead to warmer temperatures and altered precipitation patterns globally. The impact of these climatic changes on the uptake of carbon by the land surface will, in part, determine the rate and magnitude of these changes. However, there is a great deal of uncertainty in how terrestrial ecosystems will respond to climate in the future. Here, we used a fully factorial warming (four levels) by precipitation (three levels) manipulation experiment in an old-field ecosystem in the northeastern USA to examine the impact of climatic changes on leaf carbon exchange in five species of deciduous tree seedlings. We found that photosynthesis generally increased in response to increasing precipitation and decreased in response to warming. Respiration was less sensitive to the treatments. The net result was greater leaf carbon uptake in wetter and cooler conditions across all species. Structural equation modelling revealed the primary pathway through which climate impacted leaf carbon exchange. Net photosynthesis increased with increasing stomatal conductance and photosynthetic enzyme capacity (V_cmax), and decreased with increasing respiration of leaves. Soil moisture and leaf temperature at the time of measurement most heavily influenced these primary drivers of net photosynthesis. Leaf respiration increased with increasing soil moisture, leaf temperature, and photosynthetic supply of substrates. Counter to the soil moisture response, respiration decreased with increasing precipitation amount, indicating that the response to short- (i.e. soil moisture) versus long-term (i.e. precipitation amount) water stress differed, possibly as a result of changes in the relative amounts of growth and maintenance demand for respiration over time. These data (>500 paired measurements of light and dark leaf gas exchange), now publicly available, detail the pathways by which climate can impact leaf gas exchange and could be useful for testing assumptions in land surface models.

Introduction

Globally, terrestrial carbon exchange represents the largest flux of carbon between the Earth’s surface and the atmosphere (Le Quéré et al. 2012; IPCC, 2013) and studies have shown that land-atmosphere carbon cycle feedbacks are a major source of uncertainty in the Earth System Models used to project climate change (Friedlingstein et al. 2013). The flux of carbon between the atmosphere and land surface is dominated by photosynthetic carbon uptake by vegetation, as well as carbon release from vegetation and soils through respiration. Photosynthesis and plant respiration are variable and strongly influenced by climatic conditions (Wu et al. 2011; Lu et al. 2012), but scientific understanding of how these fluxes will be altered by climate change remains limited (Arneth et al. 2010). This is, in part, because the responses of these fluxes to experimental conditions are complex, as they are influenced by the scales considered as well as interactions among driving variables (Smith et al. 2014).

Climate change is expected to result in warmer temperatures and changes in precipitation patterns for most of the world (IPCC, 2013), including the northeastern USA (Hayhoe et al. 2007). Warming directly influences plant gas exchange. Short-term (seconds to minutes) warming typically increases enzymatic rates and, subsequently, rates of photosynthesis and respiration up to a peak, beyond which rates decline. This optimum occurs at a lower temperature in photosynthesis than respiration. In response to longer-term warming, photosynthetic and respiratory enzyme activity may show an acclimation response, resulting in rates that differ from those expected from short-term responses alone. The acclimated rates may be higher, lower, or similar to those observed without warming (Atkin et al. 2005; Way and Yamori 2014; Yamori et al. 2014). Warming may also influence photosynthetic rates by increasing leaf vapour pressure deficit, which reduces stomatal conductance and, subsequently, photosynthesis (Ocheltree et al. 2014). Warming may also reduce net photosynthesis (i.e. photosynthesis minus leaf respiration) if it increases leaf respiration to a greater degree than photosynthesis.

A meta-analysis of 24 experimental warming studies found that warming tends to enhance ecosystem photosynthesis (Wu et al. 2011), an effect which was shown to occur in 68 % of leaf-level studies (Way and Yamori 2014). Meta-analyses of respiration are less conclusive, with ecosystem-level studies showing an increase in aboveground respiration (Wu et al. 2011), and leaf-level studies indicating a decrease in respiration with increasing growth temperature (Slot and Kitajima 2014). These differences likely arise because of time and spatial scale incongruences, which are difficult to account for within and when comparing across meta-analyses.

Both warming and altered precipitation can affect soil moisture and, consequently, soil water availability to plants; while precipitation affects soil moisture directly by adding water to the system, warming effects are indirect and occur as the result of changes in evapotranspiration of water from plants and soil (Harte et al. 1995; Seneviratne et al. 2010). Soil water availability has been shown to influence leaf gas exchange (e.g. Camberlin et al. 2007; Ignace et al. 2007; Li et al. 2007; Llorens et al. 2004). Increases in soil moisture can increase photosynthetic carbon uptake as stomata open further, allowing CO₂ to diffuse into leaves more quickly (Hsiao 1973; Chaves et al. 2009; Pinheiro and Chaves 2011). Conversely, stomata close in drier soils, slowing water loss and leading to a decrease in intercellular CO₂ and, thus, photosynthetic rates (Potter et al. 1993; Shaw et al. 2002; Wu et al. 2011; Chaves et al. 2009; Pinheiro and Chaves 2011).

Respiration responses to soil moisture, and drought in particular, have not been consistent across studies (Pinheiro and Chaves 2011); some studies show that drought may inhibit respiration, similar to its effects on photosynthesis (Ribas-Carbo et al. 2005; Galmes et al. 2007; Gimeno et al. 2010; Flexas et al. 2006), but drought may also increase respiration as result of increased respiratory demand for ATP (Atkin and Macherel 2009) or increased maintenance respiration (Gratani et al. 2007; Slot et al. 2008). Alternatively, other studies have found no effect of soil moisture changes on respiration (Galmes et al. 2007; Gimeno et al. 2010). The response to water availability likely depends in part on photosynthetic supply of substrate (Gifford 2003; Van Oijen et al. 2010; Pinheiro and Chaves 2011), but the coupling of photosynthesis and respiration under different soil water conditions has not been well studied, particularly in natural areas and in the context of climate change.

Although soil moisture responses of leaf carbon exchange are well studied in potted plants, field studies are lacking, particularly in areas outside the Mediterranean region (Chaves et al. 2002, 2009; Pinheiro and Chaves 2011). As a consequence, model representation of soil moisture responses of leaf carbon exchange are still rudimentary, relying on simple scaling factors to adjust photosynthetic capacity (e.g. Oleson et al. 2010) or the relationship between photosynthesis and conductance (e.g. Zaehle et al. 2010) in response to moisture availability (Egea et al. 2011; Smith et al. 2014). Although simple, these formulations and the way in which they are implemented differ greatly between models (De Kauwe et al. 2013) and the different implementations have shown varying capacities to reproduce observed data (Keenan et al. 2010; Egea et al. 2011), indicating a need for more data describing these responses in the field (Smith et al. 2014).

In this study, we examined leaf-level gas exchange and growth in five species of tree seedlings in response to four levels of warming and three levels of precipitation across a single growing season. The species used were Betula lenta, Betula populifolia, Prunus serotina, Quercus rubra and Ulmus americana. These species vary in their current and projected ranges, with both Betula species being restricted longitudinally (suggesting precipitation sensitivity), and B. populifolia being restricted latitudinally (suggesting temperature sensitivity) relative to P. serotina, Q. rubra and U. americana, which have large current and projected ranges that span much of the eastern and midwestern USA (Prasad et al. 2007-ongoing; Iverson et al. 2008).

We hypothesized that warming would not directly influence observed rates of photosynthesis and respiration because of acclimation responses. We further hypothesized that both warming and precipitation change would influence soil moisture and that soil moisture would correlate positively with stomatal conductance and photosynthesis. We expected that the influence of soil moisture on respiration would mirror that of photosynthesis because of an increase in substrate supply for respiration. We expected that the soil moisture effect would be the result of a combination of stomatal and biochemical effects. As such, we expected warming to exacerbate the negative effect of reduced precipitation and to counteract the positive effect of added precipitation on leaf photosynthesis and respiration. In general, we expected the Betula species to be most sensitive to the climate treatments because of their smaller range sizes.

Methods

Research site Boston-Area Climate Experiment

All research was conducted at the Boston-Area Climate Experiment (BACE; Rodgers et al. 2012; Suseela et al. 2012), which is located in an old-field ecosystem at the University of Massachusetts’ Suburban Experiment Station in Waltham, Massachusetts, USA (42° 23′ 3″ N, 71° 12′ 52″ W). The site had a mean annual temperature of 9.3 °C and mean annual precipitation of 1180 mm, with similar amounts of precipitation falling in each month (NOAA National Climatic Data Center Cooperative Station ID 190535, January 1960–April 2009). The experimental area had a loam topsoil (0–0.3 m) over a gravelly sandy loam subsoil. The experiment had three blocks, each containing 12, 2 × 2 m plots (36 plots in total). Clear plastic ‘rainout’ shelters provided 50 % cover over the four ‘dry’ plots in each block, redirecting 50 % of the ambient rainfall to storage tanks. That captured rainwater was immediately distributed to an area encompassing the four ‘wet’ plots, creating three precipitation treatments: dry (50 % ambient), wet (150 % ambient), and ambient. Warming treatments were applied within the precipitation treatments. Ceramic heaters of different wattages were mounted 1 m above all corners of each plot. These supplied either no, low (200 W/heater), medium (600 W/heater) or high (1000 W/heater) levels of heating to each plot. Fake heater boxes were used to replicate non-warming effects of heaters (e.g. shading) in the unwarmed treatment. Infrared radiometres (Apogee Instruments, Logan, UT, USA) sensed the canopy temperature in the unwarmed and ‘high’ plots and a feedback control system (LabVIEW; National Instruments Corp, Austin, TX, USA) was set to maintain a 4 °C difference between the two. All heaters within a group of four plots were wired to the same circuit. The three precipitation treatments and four temperature treatments provided twelve climate treatments. All treatments were turned on by July 2008. The mean canopy temperature difference between the high warming and unwarmed plots across the 2011 growing season was 2.81 °C, with 0.27 °C greater warming achieved during the night than during the day, when greater convective heat losses often prevented the experimental infrastructure from achieving warming targets. The greater warming at night occurred on 61 % of the days across the measurement period. There was minimal variation in canopy temperature among precipitation treatments (±0.14 °C).

Plot setup and species composition

The BACE was constructed in an old-field ecosystem. Each plot contained a mixture of common, mostly non-native grass and forb species (Hoeppner and Dukes 2012). In addition, seedlings of eight native tree species (Acer rubrum, B. lenta, B. populifolia, Pinus strobus, Populus grandifolia, P. serotina, Q. rubra and U. americana) were planted in four subplots within the main plots in late April 2011. Four seedlings of each species were planted in each 2 × 2 m plot (one individual per species per 0.5 × 0.5 m subplot). Due to high mortality of the other species, responses were analyzed for only a subset of species (B. lenta, B. populifolia, P. serotina, Q. rubra and U. americana). All species’ ranges extended northward beyond 45°N. B. populifolia is typically not found further south than 40°N, whereas the other three species’ ranges extend close to or further south than 30°N. P. serotina, Q. rubra and U. americana have longitudinal ranges that extend from the East coast of the USA westward beyond 90°W. The two Betula species have a similar eastern edge, but typically do not extend westward beyond 80°W for B. populifolia or 85°W for B. lenta (Prasad et al. 2007-ongoing). The geographical differences correspond well with niche space differences for these species, which indicate that the Betula species are not able to tolerate the low levels of rainfall (below ∼500 mm/year) that P. serotina, Q. rubra and U. americana can tolerate. In addition, the upper end of mean annual temperature tolerance is lower for B. populifolia (∼10 °C) than the other species evaluated (Prasad et al. 2007-ongoing). Future projections suggest an increase in abundance of P. serotina, Q. rubra and U. americana in the northeastern USA due to northerly range shifts, but a decrease in abundance of both Betula species, particularly B. populifolia (Prasad et al. 2007-ongoing; Iverson et al. 2008).

Soil moisture measurements

The average relative extractable water (θ_R) was monitored weekly at the site. θ_R is an estimate of the ratio of total extractable water (θ_T) to maximum extractable water (θ_T,max) available for uptake by plants across multiple soil layers throughout the rooting zone. θ_T was calculated using a function described by Vicca et al. (2012):

$${\theta }_{\mathrm{T}}=({\theta }_{S1}-{\theta }_{S1,\text{wp}})H_1+({\theta }_{S2}-{\theta }_{S2,\text{wp}})H_2+\mathrm{\dots }({\theta }_{\text{Sn}}-{\theta }_{\text{Sn},\text{wp}})H_{\mathrm{n}}$$ (1)

where θ_Sn is the soil water content of a given layer n, θ_Sn,wp is the soil water content at the wilting point for layer n, and H_n is the thickness of layer n. Here we measured θ_S at three layers (0–0.3, 0.45 and 0.60 m) using time domain reflectometry (100; Campbell Scientific, Logan, UT, USA) and permanently installed waveguides. θ_S across the 0-0.3 m range was estimated using vertical waveguides, while horizontal waveguides at 0.45 m depth were used to estimate θ_S across the 0.3–0.525 and 0.525–0.75 m range, respectively. An estimate of 0.14 m³/m³ volumetric water content (VWC) was used for θ_S,wp at 0–0.3 m depth, and 0.08 m³/m³ VWC was used for depths below 0.3 m (Saxton and Rawls, 2006). For θ_T,max, estimated field capacity values of 0.28 m³/m³ VWC for 0–0.3 m depth and 0.18 m³/m³ VWC for depths below 0.3 m were used in place of θ_S. In some cases, θ_S at a given depth could not be estimated due to equipment error. In those cases (9.5 % of data), we gap-filled the data using the linear relationship between values at that depth and the depth directly above it, which were strongly correlated in all cases. In cases where θ_R was estimated below zero or above one, estimates were set to zero and one, respectively.

Gas exchange measurements. Gas exchange analyses were performed during three separate measurement periods within the middle of the growing season of 2011: late spring (5–8 June, day of year (DOY) 156–159), early summer (29 June–2 July, DOY 180–183), and midsummer (28–31 July, DOY 209–212). Two individuals per species were randomly chosen in each main plot; however, for some species during some measurement periods, one or no individuals of a species were measured in a given plot due to seedling death or lack of suitable leaves. Snapshot measurements of leaf carbon and water exchange, including net photosynthesis (A_n), transpiration (E) and stomatal conductance (g_s), were taken during midday hours (between 1000 and 1600 h) on the youngest fully expanded leaf of each seedling using a LI-6400 portable photosynthesis system (LI-COR Inc., Lincoln, NE, USA). Light within the chamber was set to a saturating level (1500 µmol m⁻² s⁻¹ PAR) provided by red/blue LED lights within the chamber of the LI-6400. Cuvette CO₂ concentrations were set to 360 µmol mol⁻¹. Leaf temperatures inside the cuvette were set to the leaf temperature read inside the cuvette by the internal thermocouple immediately following clamping on to the leaf and allowed to stabilize before readings were taken. Leaf vapour pressure deficit (D_leaf) was allowed to stabilize, but not held constant.

Dark respiration (R_d) measurements were taken the night following the photosynthesis measurements on the same leaf as the previous days’ photosynthesis measurements. Measurements were taken at least two hours after sunset with similar cuvette CO₂ and airflow settings as the photosynthesis measurements. Temperatures inside the cuvette were set to the temperature read by the internal thermocouple following closure of the cuvette.

Data analysis and statistics. We estimated daytime respiration (R_l) by standardizing R_d values to rates at the leaf temperature observed for A_n using a variation of a temperature-dependent temperature sensitivity formula described by Tjoelker et al. (2001) such that:

$$R_{\mathrm{l}}=R_{\mathrm{d}}{(3.22-0.046T_{\mathrm{d}})}^{\frac{T_{\mathrm{l}}-T_{\mathrm{d}}}{10}}$$ (2)

where R is the standardized rate, T_l is the leaf temperature at the time of the photosynthesis measurement (i.e. temperature in light), and T_d is the leaf temperature at the time of the respiration measurement (i.e. temperature in dark). From this, gross photosynthesis (A_g) was calculated by adding R_l to A_n (e.g. A_g = A_n + R_l). Then, the ratio of carbon lost to respiration to carbon gain through photosynthesis (R/A) was calculated as R_l divided by A_g. We did not consider light inhibition of dark respiration, given its variability under differing environmental conditions (e.g. Kroner and Way, 2016) and the relative insensitivity of A_g to this effect.

To explore the role of biochemical and stomatal effects on our photosynthesis results, we calculated maximum rate of Rubicsco carboxylation (V_cmax) using the one-point method (De Kauwe et al. 2015). This method operates under the assumption that our measured photosynthetic rates were carboxylation, rather than electron transport or phosphate utilization limited, at 1500 µmol m⁻² s⁻¹ PAR. We used the Farquhar et al. (1980) model to calculate V_cmax such that:

$$V_{\text{cmax}}=\frac{A_{\mathrm{g}}\left(C_{\mathrm{i}}+K_{\mathrm{m}}\right)}{C_{\mathrm{i}}-{\Gamma }^{\ast }}$$ (3)

where C_i is in the intracellular CO₂ concentration as measured by the LI-6400, K_m is the Michaelis-Menten constant given by:

$$K_{\mathrm{m}}=K_{\mathrm{c}}\left(1+\frac{o_{\mathrm{i}}}{K_{\mathrm{o}}}\right)$$ (4)

Γ*, K_c and K_o were estimated using leaf temperature using the equation

$$f\left(T_{\mathrm{k}}\right)=a^{\ast }\hspace{0.17em}\text{exp}\hspace{0.17em}\left(\frac{b\left(T_{\mathrm{k}}-298.15\right)}{298.15T_{\mathrm{k}}}\right)$$ (5)

where T_k is the leaf temperature in Kelvin, R is the gas constant (8.314 J mol⁻¹ K⁻¹), and parameters a (the rate at 25 °C) and b (J mol⁻¹) describe the shape of the curve and are taken from Bernacchi et al. (2001). We then estimated the degree of stomatal limitation by calculating a modified C_i (C_i,mod) that assumes no stomatal limitation as:

$$C_{\mathrm{i},\text{mod}}=C_{\mathrm{a}}$$ (6)

where C_a is the external CO₂ level (i.e., 360 µmol mol⁻¹) (Long and Bernacchi 2003). Finally, the degree of stomatal limitation (l) was estimated as:

$$l=1-\frac{A_{\mathrm{n}}}{A_{\mathrm{n},\text{mod}}}$$ (7)

where A_n,mod is the rate of net photosynthesis (A_n,mod) calculated using C_i,mod and equation (3) (Farquhar and Sharkey 1982).

We calculated rates of V_cmax (V_cmax,25) and R_d (R_d,25) standardized to 25 °C. V_cmax,25 was calculated using equation (5), with the b parameter set to 65 330 J mol⁻¹, from Bernacchi et al. (2001). R_d,25 was calculated using equation (2).

Relative extractable water (θ_R) from the beginning of May (DOY 121) to the end of the experiment was analyzed using a mixed model analysis of variance with precipitation treatment, warming treatment, and their interaction as fixed factors. The experimental block and day of measurement (continuous) were considered random variables in the model. As precipitation treatments were nested within blocks and warming treatments were nested within precipitation treatments, these relationships were also included as random effects in the models.

Response variables A_n, R_d, R_d,25, E, g_s, R/A, A_n/g_s, V_cmax, V_cmax,25 and l were analyzed using mixed model analyses of variance with precipitation treatments, warming treatments and species as fixed effects and included all possible interactions. The experimental block, the individual, and the measurement week were included as random effects. Again, the nested relationships of precipitation treatments within blocks and warming treatments within precipitation treatments were included as random effects in the models. All model fitting was done using the ‘lmer’ procedure in the ‘lme4’ package (Bates et al. 2015) in R (R-Development-Core-Team 2009). Following model fitting, we calculated Wald χ² statistics and performed type-II Wald tests for each fixed effect using the ‘Anova’ function in the ‘car’ package (Fox and Sanford 2011) in R. Least squared means were calculated using the ‘lsmeans’ function in the ‘lsmeans’ package (Lenth 2016) in R. Post-hoc comparison of means were done using Tukey’s Least Squared Difference tests using the ‘lsmeans’ package (Lenth, 2016) in R.

We also used structural equation modelling to examine the primary components directly and indirectly influencing A_n and R_d. The analysis determined the components influencing T_leaf, E, D_leaf, θ_R, A_n, g_s, R_d and V_cmax. In addition to the predicted variables, the treatment types were used as explanatory variables. The path was determined using hypothesized relationships with a goal of capturing the primary determinants of A_n and R_d. The precipitation treatment was converted to a continuous variable using rainfall amount (i.e. 0.5, 1, 1.5). Warming treatment was converted to a continuous variable using the heater wattage (i.e. 0, 200, 600, 1000). All variables were scaled before fitting the model. All species were included in the model. The path analysis was conducted using the ‘sem’ function in the ‘lavaan’ package (Rosseel 2012) in R. All analyses were performed using R version 3.2.1 (R-Development-Core-Team 2009).

Results

Treatment effects on soil moisture

Both the warming (P < 0.01; Table 1) and precipitation (P < 0.01; Table 1) treatments had strong effects on θ_R in the plots. On average across the measurement dates, θ_R decreased by 27 % in the drought plots compared with ambient and increased by 61 % in the wet plots compared with the ambient plots. Warming increased θ_R by 9.5 % in the low warming plots, but decreased θ_R by 14 and 37 % in the medium and high warming plots, respectively. There was no interaction effect of warming and precipitation on θ_R (Fig. 1 and Table 1).

Table 1.

Relative extractable water (θ_R) mixed model results.

	Df	χ2	P-value
Precipitation (P)	2	17.01	<0.001
Warming (W)	3	13.03	0.005
P × W	6	0.99	0.986

P-values < 0.05 and 0.10 are bolded are bolded and italicized, respectively. Key: Df, degrees of freedom; χ², Wald’s chi squared statistic.

Figure 1.

Mean relative extractable water (θR; unitless) in the added precipitation (AdP; blue, solid lines), ambient precipitation (AmP; grey and black, dashed lines), and reduced precipitation (brown, dotted lines) over the course of the experiment. Darker colours within each precipitation treatment indicate higher levels of warming (NW, no warming; LW, low warming; MW,  medium warming; HW, high warming). Background bars indicate leaf gas exchange measurement periods. Means are for each plot type during each measurement date (n = 3). Mixed model results related to this figure can be found in Table 1.

Leaf gas exchange responses to climate treatments

Reduced precipitation significantly decreased net photosynthesis (A_n) by 27 %, while added precipitation increased A_n by 14 % compared with ambient (P < 0.01; Fig. 2, Table 2, and see Supporting Information), effects consistent across all species (P > 0.05; Table 2). Warming had a marginally significant effect on A_n (P =  0.062; Table 2), decreasing rates by 4.9, 8.9 and 22 % in the low, medium and high warming plots, respectively, compared with ambient (Fig. 2; Table 2). Gross photosynthesis (A_g) responses to precipitation were similar to those of A_n, decreasing by 21 % in response to reduced precipitation, and increasing by 11 % in response to added precipitation (P < 0.01; Table 2). Warming did not influence A_g (P > 0.05; Table 2). The maximum rate of Rubisco carboxylation, at ambient (V_cmax) or standardized temperature (V_cmax,25), did not respond significantly to the treatments or their interactions in any species (P > 0.05 in all cases; Table 2). The responses of A_n, A_g, V_cmax and V_cmax,25 did not differ among species (P > 0.05 in all cases; Table 2).

Figure 2.

Least squared mean (±SE) net photosynthesis (An; top) and intrinsic water use efficiency (An/gs; bottom) in the added (AdP; blue, solid lines), ambient (AmP; grey, dashed lines), and reduced precipitation (brown, dotted lines) across each of the four warming treatments and all measurement dates and species. Mixed model results related to this figure can be found in Table 2.

Table 2.

Mixed model results for parameters related to leaf CO₂ uptake.

		An		Ag		An/gs		Vcmax		Vcmax,25
	Df	χ2	P-value	χ2	P-value	χ2	P-value	χ2	P-value	χ2	P-value
Precipitation (P)	2	9.62	0.008	17.48	<0.001	1.20	0.545	0.80	0.670	0.81	0.667
Warming (W)	3	7.33	0.062	3.80	0.284	4.08	0.253	3.35	0.341	3.67	0.299
Species (S)	4	59.07	<0.001	66.67	<0.001	150.12	<0.001	73.53	<0.001	57.10	<0.001
P × W	6	5.06	0.536	7.34	0.290	14.02	0.029	9.84	0.131	8.99	0.174
P × S	8	8.68	0.370	13.89	0.085	15.51	0.0499	9.08	0.336	6.48	0.594
W × S	12	13.52	0.332	14.66	0.260	9.14	0.691	9.43	0.666	6.59	0.883
P × W × S	24	22.19	0.568	15.83	0.894	14.77	0.927	21.79	0.592	18.55	0.775

P-values < 0.05 and 0.10 are bolded are bolded and italicized, respectively. Key: Df, degrees of freedom; χ², Wald’s chi squared statistic; A_n, net photosynthesis; A_g, gross photosynthesis; g_s, stomatal conductance; V_cmax, maximum rate of Rubisco carboxylation at the ambient (i.e. measurement) leaf temperature, V_cmax,25, maximum rate of Rubisco carboxylation standardized to a leaf temperature of 25 °C.

Transpiration (E) increased with increasing precipitation (P < 0.01) and decreased marginally with warming (P < 0.10; Table 3). There was no interaction between precipitation and warming (P > 0.05; Fig. 3 and Table 3) and treatment effects did not vary by species (P > 0.05 in all cases; Table 3).

Table 3.

Mixed model results for parameters related to leaf transpiration and stomatal conductance

		E		gs		l
	Df	χ2	P-value	χ2	P-value	χ2	P-value
Precipitation (P)	2	18.84	<0.001	23.03	<0.001	4.30	0.116
Warming (W)	3	6.38	0.095	11.91	0.008	24.21	<0.001
Species (S)	4	155.83	<0.001	109.30	<0.001	281.94	<0.001
P × W	6	7.41	0.284	16.21	0.013	36.71	<0.001
P × S	8	3.99	0.858	4.25	0.834	2.64	0.955
W × S	12	15.26	0.223	12.10	0.438	19.31	0.081
P × W × S	24	18.14	0.796	20.28	0.681	21.35	0.618

P-values < 0.05 and 0.10 are bolded are bolded and italicized, respectively. Key: Df, degrees of freedom, χ², Wald’s chi squared statistic; E, transpiration; g_s, stomatal conductance; l, stomatal limitation to photosynthesis.

Figure 3.

Least squared mean (±SE) transpiration (E; top), stomatal conductance (gs; middle) and stomatal limitation to photosynthesis (l; bottom) in the added (AdP; blue, solid lines), ambient (AmP; grey, dashed lines) and reduced precipitation (brown, dotted lines) across each of the four warming treatments and all measurement dates and species. Mixed model results related to this figure can be found in Table 3.

Stomatal conductance (g_s) increased with increasing precipitation (P < 0.01; Table 3) and decreased with increasing warming (P < 0.01; Table 3), particularly in the reduced precipitation plots (precipitation x warming interaction: P < 0.05; Fig 3 and Table 3). Stomatal limitation of photosynthesis (l) increased with increased warming in the reduced precipitation plots, resulting in l values that were higher under reduced, compared with ambient and added precipitation in the medium and high warming plots (precipitation x warming interaction: P < 0.05; Figure 3 and Table 3). Treatment effects on l did not differ significantly by species (P > 0.05 in all cases; Table 3).

Intrinsic water use efficiency (A_n/g_s) increased with warming in the reduced precipitation plots, and decreased with warming in the added precipitation plots (precipitation × warming interaction: P < 0.05; Fig. 2 and Table 2). There was a weak precipitation by species effect (P = 0.0499; Fig. 4 and Table 2). Post-hoc analyses revealed that reduced precipitation alone marginally increased A_n/g_s in B. lenta compared with ambient precipitation (P = 0.058; Table 5). The within-species precipitation effect was not significant in any other case (P > 0.05 in all cases; Table 5).

Figure 4.

Least squared mean (±SE) ratio of net photosynthesis to stomatal conductance (An/gs), leaf respiration in dark (Rd), and the ratio of Rd to gross photosynthesis (Rd/Ag) in each species in the reduced precipitation (brown), ambient (AmP; grey) and added (AdP; blue) precipitation plots. Mixed model results related to this figure can be found in Tables 2 (An/gs) and 4 (Rd and Rd/Ag). Results from related Tukey’s tests can be found in Table 5.

Table 5.

Results from precipitation by species interaction tukey’s tests.

Response	Species	Contrast	Estimate	SE	Df	t-value	P-value
An/gs	B. lenta	RP-AmP	22.07	8.93	18.4	2.47	0.058
		RP-AdP	13.35	8.90	17.9	1.50	0.315
		AmP-AdP	−8.72	8.58	15.9	−1.02	0.577
	B. populifolia	RP-AmP	−11.70	9.57	24.0	−1.22	0.452
		RP-AdP	4.47	9.37	21.8	0.48	0.883
		AmP-AdP	16.17	9.10	19.6	1.78	0.203
	P. serotina	RP-AmP	12.52	11.31	45.6	1.11	0.515
		RP-AdP	11.58	9.96	27.7	1.16	0.485
		AmP-AdP	−0.94	11.01	40.7	−0.09	0.996
	Q. rubra	RP-AmP	7.60	15.45	126.5	0.49	0.875
		RP-AdP	0.92	15.57	128.7	0.06	0.998
		AmP-AdP	−6.68	11.30	45.0	−0.59	0.825
	U. americana	RP-AmP	6.50	8.15	13.1	0.80	0.711
		RP-AdP	7.94	8.14	13.0	0.98	0.605
		AmP-AdP	1.43	7.93	11.8	0.18	0.982
Rd	B. lenta	RP-AmP	−0.68	0.16	50.7	−4.19	<0.001
		RP-AdP	−0.49	0.16	49.0	−3.01	0.011
		AmP-AdP	0.19	0.16	43.6	1.23	0.443
	B. populifolia	RP-AmP	−0.25	0.16	47.2	−1.54	0.282
		RP-AdP	−0.25	0.16	48.3	−1.52	0.292
		AmP-AdP	0.00	0.16	46.8	0.00	1.000
	P. serotina	RP-AmP	−0.20	0.17	57.8	−1.18	0.471
		RP-AdP	−0.16	0.17	50.6	−0.93	0.621
		AmP-AdP	0.04	0.17	61.0	0.26	0.965
	Q. rubra	RP-AmP	−0.14	0.24	160.0	−0.59	0.827
		RP-AdP	−0.06	0.24	160.4	−0.23	0.970
		AmP-AdP	0.08	0.20	104.9	0.42	0.907
	U. americana	RP-AmP	0.33	0.15	36.4	2.23	0.079
		RP-AdP	0.21	0.15	34.4	1.44	0.334
		AmP-AdP	−0.12	0.14	32.7	−0.84	0.682
Rd,25	B. lenta	RP-AmP	−0.90	0.21	75.8	−4.37	<0.001
		RP-AdP	−0.58	0.21	71.3	−2.79	0.018
		AmP-AdP	0.32	0.20	65.3	1.66	0.229
	B. populifolia	RP-AmP	−0.13	0.20	69.9	−0.66	0.790
		RP-AdP	−0.20	0.21	70.2	−0.96	0.605
		AmP-AdP	−0.06	0.20	69.9	−0.32	0.945
	P. serotina	RP-AmP	−0.24	0.22	85.1	−1.11	0.513
		RP-AdP	−0.21	0.21	71.6	−1.02	0.569
		AmP-AdP	0.02	0.22	87.0	0.11	0.994
	Q. rubra	RP-AmP	−0.33	0.31	212.4	−1.07	0.536
		RP-AdP	−0.42	0.31	211.3	−1.34	0.374
		AmP-AdP	−0.09	0.26	156.3	−0.34	0.937
	U. americana	RP-AmP	0.55	0.19	54.4	2.97	0.012
		RP-AdP	0.39	0.18	51.1	2.14	0.093
		AmP-AdP	−0.16	0.18	48.6	−0.89	0.650
Rd/Ag	B. lenta	RP-AmP	−0.05	0.05	19.1	−0.97	0.602
		RP-AdP	0.06	0.05	18.6	1.24	0.444
		AmP-AdP	0.11	0.05	16.3	2.31	0.083
	B. populifolia	RP-AmP	0.10	0.06	25.6	1.79	0.194
		RP-AdP	0.16	0.05	23.5	2.92	0.020
		AmP-AdP	0.06	0.05	21.6	1.12	0.512
	P. serotina	RP-AmP	0.04	0.07	47.2	0.54	0.852
		RP-AdP	0.05	0.06	28.6	0.88	0.657
		AmP-AdP	0.02	0.06	42.1	0.24	0.969
	Q. rubra	RP-AmP	0.08	0.09	131.5	0.90	0.642
		RP-AdP	0.11	0.09	133.2	1.17	0.471
		AmP-AdP	0.03	0.07	48.0	0.39	0.921
	U. americana	RP-AmP	0.15	0.05	14.0	3.13	0.019
		RP-AdP	0.15	0.05	13.5	3.13	0.020
		AmP-AdP	<0.01	0.05	12.4	−0.02	1.000

Contrast indicates the two treatments being compared within each species. Estimates are least squared mean differences in the response variable between the two treatments. SE are standard errors of the least squared mean. P-values < 0.05 and 0.10 are bolded are bolded and italicized, respectively. Key: Df, degrees of freedom; RP, RP; AmP, ambient precipitation; AdP, added precipitation; A_n, net photosynthesis; g_s, stomatal conductance; R_d, leaf dark respiration at the ambient (i.e. measurement) leaf temperature; R_d,25, leaf dark respiration standardized to a leaf temperature of 25 °C; A_g, gross photosynthesis.

Leaf dark respiration, at ambient (R_d) and standardized (R_d,25) temperatures, responded to altered precipitation in some, but not all, species (precipitation × species interaction: P < 0.01; Fig. 4, Table 4, and see Supporting Information). Post-hoc analyses revealed that R_d of B. lenta was lower under reduced (P < 0.05) and ambient (P < 0.05) compared with added precipitation, but was unaffected by precipitation in the other species (P > 0.05; Table 5). Post-hoc analyses indicated that R_d,25 showed a similar response for B. lenta (Table 5). Interestingly, R_d,25 of U. americana was significantly and marginally higher in the reduced compared with ambient (P < 0.05) and added (P =  0.093) precipitation plots, respectively (Table 5). R_d,25 was unaffected by altered precipitation in the other species (P > 0.05 in all cases; Table 5).

Table 4.

Mixed model results for parameters related to leaf respiration.

	Df	Rd		Rd,25		Rd/Ag
		χ2	P-value	χ2	P-value	χ2	P-value
Precipitation (P)	2	6.10	0.047	4.77	0.092	8.33	0.016
Warming (W)	3	2.11	0.549	1.43	0.700	12.17	0.007
Species (S)	4	45.03	<0.001	45.60	<0.001	52.52	<0.001
P × W	6	9.15	0.165	8.92	0.178	6.04	0.419
P × S	8	25.06	0.002	31.03	<0.001	15.61	0.048
W × S	12	12.60	0.399	7.29	0.838	9.33	0.674
P × W × S	24	16.78	0.858	18.60	0.773	22.81	0.531

P-values < 0.05 and 0.10 are bolded are bolded and italicized, respectively. Key: Df, degrees of freedom; χ², Wald’s chi squared statistic; R_d, leaf dark respiration at the ambient (i.e. measurement) leaf temperature, R_d,25, leaf dark respiration standardized to a leaf temperature of 25 °C, A_g, gross photosynthesis.

The ratio of R_d to A_g (R_d/A_g) increased by 5.6, 5.5 and 16 % in low, medium and high warming plots, respectively, compared with the ambient warming plots (P < 0.01; Fig. 5 and Table 4). Reduced and added precipitation increased R_d/A_g by 14 % and decreased R_d/A_g by 9.3 %, respectively, compared with the ambient precipitation plots (P < 0.05; Fig. 5 and Table 4). There was a weak precipitation by species interaction (P =  0.048; Fig. 4 and Table 4). Post-hoc analyses revealed that the strongest increase in R_d/A_g with decreased precipitation was observed in B. populifolia and U. americana (Fig. 4 and Table 5).

Figure 5.

Least squared mean (±SE) dark leaf respiration (Rd; top) and ratio of dark leaf respiration to gross photosynthesis (Rd/Ag; bottom) in the added (AdP; blue, solid lines), ambient (AmP; grey, dashed lines) and reduced precipitation (brown, dotted lines) across each of the four warming treatments and all measurement dates and species. Mixed model results related to this figure can be found in Table 4.

Path analysis

The structural equation modelling was designed to indicate the pathways, indirect and direct, by which the precipitation and warming treatments impacted leaf A_n and R_d. A_n did not respond directly to the treatments (P > 0.05 in both cases; Table 6), but rather responded indirectly via responses to g_s (positive; P < 0.01; Table 6), V_cmax (positive; P < 0.01; Table 6), and R_l (negative; P < 0.01; Table 6). Interestingly, T_leaf (or D_leaf in the case of g_s) and θ_R determined each of these three factors (P < 0.05 in all cases; Table 6). θ_R was directly related to the climate manipulations (P < 0.05 in both cases; Fig. 6 and Table 6). However, T_leaf was poorly predicted by the climatic treatments (r² = 0.077), increasing with greater precipitation (P < 0.05) and decreasing with greater E, but showing no response to warming (P > 0.05; Fig. 6 and Table 6).

Table 6.

Results from structural equation modelling

Dependent variable (r2)	Independent variable	Standardized coefficient	Z-value	P-value
θR (0.189)	Precipitation	0.43	11.358	<0.001
	Warming	−0.137	−3.608	<0.001
Tleaf (0.077)	Precipitation	0.209	5.105	<0.001
	E	−0.103	−2.218	0.027
	Warming	0.017	0.423	0.672
Dleaf (0.360)	Tleaf	0.583	17.077	<0.001
gs (0.327)	θR	0.336	8.328	<0.001
	Dleaf	−0.309	−8.001	<0.001
	An	0.091	1.67	0.095
	Warming	−0.018	−0.518	0.604
	Precipitation	−0.007	−0.184	0.854
E (0.728)	gs	0.789	33.728	<0.001
	θR	0.14	6.021	<0.001
Vcmax (0.188)	Tleaf	0.383	9.953	<0.001
	θR	0.216	5.173	<0.001
	Precipitation	−0.036	−0.837	0.403
	Warming	−0.025	−0.658	0.511
Rl (0.175)	An	0.414	7.982	<0.001
	Tleaf	0.369	9.523	<0.001
	θR	0.237	5.385	<0.001
	Precipitation	−0.176	−4.071	<0.001
	Warming	0.046	1.189	0.235
An (0.611)	Vcmax	0.623	20.787	<0.001
	gs	0.62	18.211	<0.001
	Rl	−0.306	−8.242	<0.001
	Precipitation	−0.006	−0.213	0.831
	Warming	−0.001	−0.026	0.979
Rd (0.158)	θR	0.361	8.256	<0.001
	An	0.147	3.69	<0.001
	Precipitation	−0.152	−3.549	<0.001
	Tleaf	0.094	2.429	0.015
	Warming	0.073	1.877	0.06

Variable key: θ_R, relative extractable water; g_s, stomatal conductance; V_cmax, maximum rate of Rubisco carboxylation; R_l, leaf respiration in light; A_n, net photosynthesis; R_d, leaf respiration in dark; D_leaf, leaf vapour pressure deficit; T_leaf, leaf temperature. Independent variables are ordered by the absolute value of the standardized coefficient for each dependent variable.

Figure 6.

Path resulting from the structural equation modeling described in the text. Solid blue and dashed red arrows denote significant (α = 0.05) positive and negative relationships, respectively. The size of the arrow is indicative of the strength of the relationship. Non-significant relationships are not shown. The widths of the box outlines are related to the r2 value for each parameter assessed. Boxes outlined in dots indicate variables not predicted by the model. Key: Precip., precipitation treatment; Warming, warming treatment; θR, relative extractable water; E, leaf transpiration, Dleaf, leaf vapour pressure deficit; gs, stomatal conductance; Vcmax, maximum rate of Rubisco carboxylation; A_n, net photosynthesis; R_l, leaf respiration in light; R_d, leaf respiration in dark and T_leaf, leaf temperature. Linear coefficient values, Z values, P-values, and r² values for all parameters and relationships tested can be found in Table 6.

In contrast, R_d was influenced both directly and indirectly by the climatic changes. The strongest determinant of R_d was θ_R (positive; P < 0.01; Table 6), but T_leaf (P < 0.01; Table 6) and A_n (P < 0.01; Table 6) also increased R_d. Interestingly, the precipitation treatment had a direct negative influence on R_d (P < 0.01; Table 6), indicating that precipitation has differing influences on R_d depending on time scale (i.e. positive in short term, but negative in long term; Fig. 6 and Table 6). The warming treatment had a marginally significant positive influence on R_d (P  = 0.060; Table 6), indicating that warming can have both short- and, to a lesser degree, long-term effects on R_d.

Discussion

We used a climate manipulation experiment in an old-field ecosystem to examine the responses of net photosynthesis (A_n) and dark respiration (R_d) to warming and altered precipitation. Our goal was to not only to examine the responses, but to also to probe the mechanisms underlying them. Confirming theoretical understanding (Lin et al. 2012), A_n was controlled by a combination of stomatal conductance (g_s), leaf biochemistry (i.e. V_cmax), and leaf respiration (i.e. R_l). Leaf dark respiration was less sensitive to climate than A_n, due to offsetting long- and short-term effects of soil moisture.

In accordance with our original hypothesis, the three primary determinants of A_n were heavily influenced by direct (via altered precipitation) and indirect (via warming) soil moisture effects. Soil moisture significantly increased under increasing precipitation and decreased under warming over the course of our experiment. The precipitation response is not surprising. However, the warming response, while not as strong, indicates that warming-induced reductions in soil moisture may exacerbate the plant gas exchange responses to drought observed in precipitation manipulation-only studies (e.g. Wu et al. 2011; Yan et al. 2016). Reductions in soil moisture reduced g_s and V_cmax, the limiting enzymatic process at the light levels assessed (Long and Bernacchi 2003), which would contribute to the reduced A_n seen under warming and lower precipitation. However, soil moisture also reduced R_l, which would should have increased A_n, but was a weaker driver than g_s and V_cmax.

The leaf temperature at the time of measurement also influenced each of the three drivers of A_n. Stomatal conductance decreased with increased vapour pressure deficit (D_leaf), which is positively correlated with leaf temperature. The enzymatic processes V_cmax and R_l accelerated with increasing leaf temperature, as has been seen in many studies (e.g. Ryan 1991; Bernacchi et al. 2001; Medlyn et al. 2002a,b; Atkin et al. 2005). These results confirm theoretical understanding (Lin et al. 2012). Nonetheless, in contrast with our expectations, warming did not directly influence leaf temperatures at the time of measurement due to the measurement cuvette blocking incoming thermal radiation from the heaters. Instead, warming-induced reductions in soil moisture reduced transpiration, which led to higher leaf temperatures.

Our results provide insight for model development of moisture-stomata-photosynthesis relationships. Currently, many models simulate photosynthesis and stomatal conductance as interdependent, instantaneous responses (e.g. through coupled schemes; Ball et al. 1987; Collatz et al. 1991, 1992; Leuning 1995; Medlyn et al. 2011) in which g_s is a function of A_n and D_leaf (or relative humidity), while A_n is typically simulated as function of light, T_leaf, soil moisture, and other leaf traits such as leaf N and leaf age. Our path analysis of field data (Fig. 5) found that moisture and D_leaf were the primary drivers of g_s, which, in turn, drove A_n (along with leaf temperature). A_n was not found to be a significant driver of g_s. This result indicates that A_n might be best simulated as a function of g_s, rather than separately as a function of a coupled scheme.

In addition, we found significant changes in the A_n/g_s ratio with climate. This result implies that models should consider including climatic responses as part of the coupled A_n–g_s scheme. However, more model-data comparisons at the leaf (e.g. Egea et al. 2011) and larger scales (e.g. Keenan et al. 2010; De Kauwe et al. 2013) are necessary to fully evaluate these responses.

Our climate treatments had a less pronounced effect on R_d than A_n. Precipitation increased R_d and R_d,25 in the B. lenta and decreased R_d,25 in U. americana, but had little effect on R_d of other species. Warming had little effect in general, which contrasts with results of a recent warming-only study examining leaf respiration in similar species (Reich et al. 2016). Our path analysis allowed us to gain some insight into the drivers of R_d responses to climate over varying time scales. In the short term, increased photosynthesis, temperature, and soil moisture acted to increase R_d. This temperature response is widely observed (Atkin et al. 2005) and, given that photosynthesis provides the substrate for respiration (Gifford, 2003), it is not surprising that higher A_n would lead to higher R_d. The positive response to soil moisture may indicate an increase in growth demand for respiratory products under more favourable (i.e. wetter) conditions. Of interest, and counter to the soil moisture response, was the observed increase in R_d with decreasing amounts of precipitation. This result may reflect a longer-term response to prolonged drought stress that increased the demand for maintenance respiration, as has been seen in other studies (Gratani et al. 2007; Slot et al. 2008; Atkin and Macherel 2009). A combination of offsetting long- and short-term responses likely contributed to the weak response observed when considering the climate treatments alone. These results suggest that the effect of soil moisture on R_d varies with the duration of exposure to a given soil moisture level. This time dependency of the soil moisture response may complicate the interpretation of the effects of other treatments such as warming (e.g. Reich et al. 2016), particularly when those other treatments have consequences for soil moisture. Targeted measurements within multi-factor experiments could help to pinpoint the mechanisms underlying respiration responses to climate.

Combined with the photosynthesis results, the weak response of respiration resulted in a decrease in the ratio of dark respiration to gross photosynthesis (R_d/A_g) under cooler and wetter conditions. Similar precipitation results have been seen in response to seasonal change in soil moisture in Mediterranean species (Gratani et al. 2008); however, the response of the respiration-photosynthesis relationship to water availability is understudied in the field outside of the Mediterranean region (Chaves et al. 2002, 2009; Pinheiro and Chaves 2011). Our results suggest that, for seedlings in the northeastern USA, future net leaf carbon uptake will likely decrease unless precipitation can raise soil moisture levels enough to counteract the negative effects of warming on soil moisture. However, as we were not able to measure the respiration of other tissues, we cannot address the whole-plant biosphere-atmosphere interaction.

We expected that the Betula species would be most sensitive to our climate manipulations due to their relatively restricted current and projected ranges (Prasad et al. 2007-ongoing; Iverson et al. 2008). Although the magnitude of gas exchange rates did differ by species, there were very few species-specific treatment responses. Nonetheless, Betula was generally more sensitive to soil moisture than other species. For instance, while all species tended to increase A_n/g_s in response to a combination of hotter and drier conditions, B. lenta increased this ratio in response to reduced precipitation alone, an effect seen in previous studies on other Betula species (Wang et al. 1998). This indicates that B. lenta may have a more risk-averse drought strategy than the other species examined here. Similarly, B. populifolia, U. americana, and, to a lesser degree, B. lenta showed the greatest sensitivity of R_d/A_g to precipitation. In these species, R_d/A_g rates increased with drought, indicating a risk-averse switch in resources from carbon acquisition to maintenance and growth of tissues. Although there were notable similarities among species examined, the few species-specific differences that did exist suggest that further examination of a more diverse range of plant species, particularly those that exist in narrow climatic ranges or at range edges (Reich et al. 2015), may reveal more differential responses.

Conclusions

Our findings indicate that warming and altered precipitation influence leaf carbon exchange primarily through indirect effects on environmental conditions and underlying processes. Leaf respiration was less sensitive than photosynthesis to our treatments, indicating that net leaf carbon uptake was increasing in wetter and, to a lesser degree, cooler conditions. As such, the net effect of climate change on carbon uptake in these species is likely to depend on whether the benefits associated with any precipitation increases can counteract the negative effects of temperature increases. More analyses, particularly of respiration responses in stems, roots and soil, are needed to explore the influence of these findings on ecosystem-level carbon uptake. These data could prove useful for evaluating the larger-scale models used to make future projections of climate-carbon feedbacks and, to facilitate such use, are included in the TRY (www.try-db.org) and open-access Purdue University Research Repository (https://purr.purdue.edu/publications/2213/2).

Sources of Funding

Our work was funded by the National Science Foundation (DEB-0546670 and DEB-1146279), US Department of Energy’s Office of Science (BER), through the Northeastern Regional Center of the National Institute for Climatic Change Research, National Aeronautics and Space Administration (NNX13AN35H), US Department of Agriculture (2015-67003-23485), and the Purdue Climate Change Research Center.

Contributions by the Authors

N.G.S. and J.S.D. conceived of the project. N.G.S., G.P. and C.E.G. took the measurements. N.G.S. analysed the data. All authors contributed to the writing of the article.

Conflicts of Interest Statement

None declared.

Supporting Information

The following additional information is available in the online version of this article — Figure S1. Raw net photosynthesis (An) and leaf dark respiration (Rd) data for all species, treatments, and days of year.