Light and Water Conditions Co-Regulated Stomata and Leaf Relative Uptake Rate (LRU) during Photosynthesis and COS Assimilation: A Meta-Analysis

Abstract

As a trace gas involved in hydration during plant photosynthesis, carbonyl sulfide (COS) and its leaf relative uptake rate (LRU) is used to reduce the uncertainties in simulations of gross primary productivity (GPP). In this study, 101 independent observations were collected from 22 studies. We extracted the LRU, stomatal conductance (g_s), canopy COS and carbon dioxide (CO₂) fluxes, and relevant environmental conditions (i.e., light, temperature, and humidity), as well as the atmospheric COS and CO₂ concentrations (${\mathrm{C}}_{\mathrm{a},\mathrm{COS}}$ and ${\mathrm{C}}_{\mathrm{a},\mathrm{CO}2}$). Although no evidence was found showing that g_s regulates LRU, they responded in opposite ways to diurnal variations of environmental conditions in both mixed forests (LRU: Hedges’d = −0.901, LnRR = −0.189; g_s: Hedges’d = 0.785, LnRR = 0.739) and croplands dominated by C3 plants (Hedges’d = −0.491, LnRR = −0.371; g_s: Hedges’d = 1.066, LnRR = 0.322). In this process, the stomata play an important role in COS assimilation (R² = 0.340, p = 0.020) and further influence the interrelationship of COS and CO₂ fluxes (R² = 0.650, p = 0.000). Slight increases in light intensity (R² = 1, p = 0.002) and atmospheric drought (R² = 0.885, p = 0.005) also decreased the LRU. The LRU saturation points of C_a,COS and C_a,CO2 were observed when ΔC_a,COS ≈ 13 ppt (R² = 0.580, p = 0.050) or ΔC_a,CO2 ≈ −18 ppm (R² = 0.970, p = 0.003). This study concluded that during plant photosynthesis and COS assimilation, light and water conditions co-regulated the stomata and LRU.

1. Introduction

1.1. Global Change and the Terrestrial Carbon Fixation

Global industrialization has caused the enrichment of atmospheric greenhouse gases (GHGs) globally [1]. In the atmosphere, GHGs have changed the process of radiation transmission [2]; further, more radiation is being trapped in the atmosphere [3,4]. The long-lived GHGs have dominated 47% of global warming, in particular, CO₂ takes up 80% of this increase alone [5]. In this context, terrestrial and oceanic carbon sinks have been found to increase continuously over the last 60 years [6,7,8,9]. The terrestrial ecosystem responds to climate variability and climate change [10,11,12]. Global warming and land-cover changes have jointly accelerated carbon turnover [3,4]. The terrestrial carbon residence time decreased by one year in vegetation and by nine years in soil from the 1980s to the 2000s [13]. As the primary driver of terrestrial carbon accumulation or sequestration [14,15], GPP is not only the total carbon amount taken up by plants within photosynthesis [16] but also the largest CO₂ flux in global carbon cycling [17]. Variations in temperature and precipitation, as well as extreme events, have affected GPP and the ecosystem respiration (ER) [18,19,20]. Moreover, in the terrestrial carbon–nitrogen cycle, such as CO₂ fertilization [21], nitrogen deposition and other processes [22], a coupling interaction occurs with the systematic change of terrestrial carbon fixation [23]. Then, thickened tropospheric CO₂ subtly influences GPP [24,25]. In network observations and ecological modeling studies, to represent the carbon that is actually fixed by the undisturbed biosphere through plant photosynthesis, the net ecosystem productivity (NEP) is a central concept in terrestrial carbon cycling [26]. NEP was originally defined as the imbalance between GPP and the ER [27]. Observations from networks of eddy covariance (EC) and footprint analysis provide rapid and accurate estimations of the exchange of mass and energy [28]. Furthermore, fluxes of CO₂ and water (H₂O) are indicators of ecosystem carbon and hydrological balances [29]. The EC micrometeorological technique, as well as the ecology-based biometric methods, are used to efficiently monitor the net ecosystem exchange (NEE) [30]. NEE is the net atmosphere–biosphere exchange of CO₂ [31]. In the detailed mechanism, NEE is the sum of the ecosystem CO₂ fluxes that do not involve the organic carbon either in its sinks or sources, as well as NEP [27]. Normally, NEE approximately equals to the negative NEP [26].

1.2. The Current State of Investigation on Photosynthesis and COS Assimilation

As a trace gas [32], the largest source of the COS atmospheric budget is the oceans of the Southern Hemisphere and the low latitude regions [33,34], as well as anoxic soils [35,36,37,38,39,40], i.e., directly from sulfur-containing organic matter or indirectly transferred from dimethyl sulfide (DMS) and carbon disulfide (CS₂) [41,42,43,44]. Meanwhile, anthropogenic activities are the largest influencing factor of regional COS emission [45,46,47,48,49]. Among the whereabouts of atmospheric COS, vegetation photosynthesis plays a major role in tropospheric COS consumption [50]. From the process point of view, when catalyzed by the enzyme carbonic anhydrase (CA), COS can be irreversibly fixed in plants and soil ($\mathrm{COS}+{\mathrm{H}}_2\mathrm{O}\to {\mathrm{HCOOS}}^{-}+{\mathrm{H}}^{+})$ [37,51,52,53,54,55,56]. In terms of the correlation between the photosynthetic hydrolysis rates of COS and CO₂ [37,51,52,53,54,55,56] or the temporal dynamics of the proportional volume mixing ratio [32,41,42,57], the canopy assimilation rate of COS is correlated with that of CO₂. Furthermore, unlike CO₂, COS cannot be emitted through respiration. The COS flux is employed as an indicator of GPP in ecosystems [58,59,60]. Stomatal conductance (g_s) can also be estimated by extrapolating the stomatal conductance of H₂O (g_s,H2O) from the stomatal conductance of COS (g_s,COS) [61,62,63]. Based on laser spectroscopy technology [64], high-frequency observations of trace gases make it possible to measure the chamber-based or the EC flux COS [65,66,67]. At the regional scale, when using the processes-based land surface models (LSMs), such as SiB 4 and ORCHIDEE [68,69,70], to estimate the spatiotemporal variations of GPP [68,71,72,73], transpiration and g_s [54,74,75], researchers take the atmospheric COS mixing ratio data of observation networks [58] and quantitative remote sensing as the input variables [76,77,78,79,80]. Global studies take the carbon assimilation systems, such as Carbon Tracker-Langrange, to recognize the sources and sinks of COS [71,81]. Recently, scientists are also making efforts to identify the missing source and sinks of global atmospheric COS [60,82,83,84]. It is urgent to precisely formulate the regulation mechanism of canopy COS assimilation, as well as the linkage between canopy COS flux (F_C,COS) and GPP. The following field observations have studied the relationship between environmental factors and COS assimilation. Site observation of bryophytes found that the biological fixation of COS is mostly regulated by meteorological factors [85]. In plant leaf photosynthesis, during hydrolysis, the catalysis of COS by the CA enzyme is light-independent [37,41] and more efficient than that of CO₂ [86]. During the light response of photosynthesis, CO₂ is preferentially catalyzed by RuBisCO over COS by a factor of 110 [51,87]. Moreover, the catalytic efficiency of the CA enzyme, which is highly abundant, exceeds that of RuBisCO overall [37,86,88,89]. At the leaf scale, stomata are not only the gateway for photosynthetic CO₂ assimilation but also the vapor flux in transpiration. At the same time, COS is assimilated through the diffusive pathway (e.g., the leaf boundary layer, stomata and mesophyll). From the process perspective, environmental conditions (e.g., light, temperature, leaf moisture conditions and the CO₂ concentration in the leaf boundary) have an important role in the conductance of both the stomata and mesophyll [90]. In response to leaf drought, to protect the mesophyll cells from excessive water loss, abscisic acid (ABA) is transferred to the stomata by the catheric sap flow [91], which stimulates the striction of guard cells and the stomata close. The gradient between the atmospheric CO₂ concentration (C_a,CO2) and the intercellular CO₂ concentration (C_i,CO2) is mainly regulated by the stomatal guard cells. Specifically, the mesophyll CO₂ concentration is controlled by the diffusive pathway. Then, the above processes regulate CO₂, COS and H₂O fluxes [92,93]. There are differences between the plant morphology (herbs, trees), tree foliage types (conifer, deciduous and evergreen) and the photosynthesis pathways (C3 and C4 plants). It has been estimated that 85% of photosynthetic CO₂ uptake is regulated by g_s and mesophyll conductance (g_m) [94]. Stomata play an important role in linking carbon and water processes [95]. On the other hand, because COS is a trace gas in the troposphere and the catalytic efficiency of CA enzyme is high, the intercellular COS concentration (C_i,COS) is kept at a very low level [41,51,96]. As they share the same assimilation pathway, COS and CO₂ are also maximally resisted by the stomata and the mesophyll [37,73,97]. Furthermore, based on parallel observations of the atmospheric concentration of COS (C_a,COS) and C_a,CO2 [56,98], recent studies found that COS can be used to obtain g_s across different biomes and scales [60,73,99]. From the perspective of both diel and seasonal variation, COS can not only be slightly continuously taken up by the partially closing stomata at night [65], but when some branches wither, COS can be released from these dead or drought-damaged components [65,100,101]. In the process of plant photosynthesis, by eliminating the interference of non-foliage components [90], the conductance of the leaf boundary (g_b) is caused by adhesion and COS. The COS assimilation follows ${\mathrm{F}}_{\mathrm{c},\mathrm{COS}}={(1/{\mathrm{g}}_{\mathrm{b}}+1/{\mathrm{g}}_{\mathrm{s}}+1/{\mathrm{g}}_{\mathrm{m}})}^{-1}·\left({\mathrm{C}}_{\mathrm{a},\mathrm{COS}}-{\mathrm{C}}_{\mathrm{i},\mathrm{COS}}\right)$ [51,56,60,73,97]. The pathway conductance controls the COS assimilation rate at the leaf scale [60]. Meanwhile, the COS flux is found to be regulated by radiation and the vapor pressure deficit (VPD) [102,103,104,105]. In the diffusional pathway, environmental conditions, such as radiation, the leaf VPD, relative humidity (rH), nitrogen deposition, salinity and aerosol concentrations, also affect stomata conductance [106,107,108]. In the process of COS, CO₂ and H₂O exchange, the molecule of COS is larger than CO₂ and the adhesional coefficient of COS is different from CO₂ [51,56]. Moreover, stomata strongly respond to COS, which is mediated by H₂S [51]. Further, in hydrolysis, the mesophyll COS is catalyzed by the CA enzyme [37,41], while in photoreaction, CO₂ is catalyzed by RuBisCO [51,87]. This causes a different diffusional coefficient between COS and CO₂ in each pathway. For other processes, because COS cannot be released by plant respiration, the difference is also shown in the photosynthetic feedback pathway between COS and CO₂ [109]. The assimilation processes of both COS and CO₂ also vary between species. It is important to explore the differential regulation mechanism of environmental conditions in the whole diffusional pathway of COS and CO₂ assimilation (Figure 1). In observational and modeling studies, water use efficiency (WUE), as the ratio of carbon assimilation (i.e., GPP) to water losses, has been employed to represent the coupling of the carbon and water cycles. Controlled by leaf stomata, WUE is broadly correlated with C_a,CO2 at the leaf boundary layer [110].

1.3. Background and Main Goals of Investigations

As an improvement of the traditional partition method of NEE [111], the leaf relative uptake rate ($\mathrm{LRU}=\frac{{\mathrm{F}}_{\mathrm{COS}}\times {\mathrm{C}}_{\mathrm{a},\mathrm{CO}2}}{\mathrm{GPP}\times {\mathrm{C}}_{\mathrm{a},\mathrm{COS}}}$) has been proposed for use in estimating GPP from the COS flux [51,56,59,96,112]. LRU has been estimated to be in the range of 1.5~4.0 for many species [51,54,56,113]. On the other hand, the LRU is also an ecophysiological indicator used in estimating C_i,CO2/C_a,CO2 and g_s [56,90]. COS assimilation of different plant components [51], litters and soil responses vary distinctly during at diel and seasonal scales, leading to a fluctuation in the LRU value [90]. Further, the differential behavior of COS and CO₂ in physiological and meteorological processes causes small fluctuations in the plant assimilation ratio of the two, which further affects the LRU value [51,101,114,115]. Generally, LRU is controlled by g_s, as well as concentrations of either COS or CO₂ in the leaf boundary layer [54]. The environmental conditions, atmosphere and vegetation at certain stations may influence the LRU [51,65,88,116]. In the species-dependent LRU environment response process [65,101,113,114], light intensity, temperature and water limitation [115], as well as atmospheric conditions, such as the frictional wind speed, C_a,COS and C_a,CO2, are influential [51,68]. In this study, we hypothesized that regarding diurnal, daily and seasonal variation, the above environmental conditions (i.e., light, temperature, water conditions, C_a,COS and C_a,CO2) co-regulate g_s, photosynthesis and COS assimilation, and this species-dependent relationship will affect the LRU (Figure 1). By providing perspectives on the development of COS-constrained GPP simulations and the key physiological processes within the LRU regulation mechanism, this study aimed to determine the relationship between environmental conditions and LRU variations, to describe how stomata regulate the relationship between F_C,COS and canopy CO₂ flux (F_C,CO2) and to constrain the accuracy of LRU estimates.

Figure 1. Schematic diagram of COS and CO₂ assimilation during leaf photosynthesis. The diffusion processes of COS and CO₂ are generalized from Section 1.2. The current state of investigation on photosynthesis and COS assimilation by us. Blue font characters are the key environmental conditions we filtered from the literature. In the corresponding physiological processes, they act as regulators. The source literature is as follows: light intensity [37,41,51,86,87,90], rH [61,62,63,85,90,91,92,93,94,106,107,108,115], VPD [102,103,104,105] and C_a,COS and C_a,CO2 [37,41,50,51,52,53,54,55,56,60,73,87,97,98,109].

Figure 1. Schematic diagram of COS and CO₂ assimilation during leaf photosynthesis. The diffusion processes of COS and CO₂ are generalized from Section 1.2. The current state of investigation on photosynthesis and COS assimilation by us. Blue font characters are the key environmental conditions we filtered from the literature. In the corresponding physiological processes, they act as regulators. The source literature is as follows: light intensity [37,41,51,86,87,90], rH [61,62,63,85,90,91,92,93,94,106,107,108,115], VPD [102,103,104,105] and C_a,COS and C_a,CO2 [37,41,50,51,52,53,54,55,56,60,73,87,97,98,109].

2. Materials and Methods

2.1. Data Preparation

We used ${\mathrm{F}}_{\mathrm{C},\mathrm{COS}}$, ${\mathrm{F}}_{\mathrm{C},\mathrm{CO}2}$ and the relevant environmental conditions to explain the mechanism of COS assimilation during photosynthesis and the $\mathrm{LRU}$ regulation mechanism based on stomata. During the data preparation, we collected studies published during 1985~2020 from Google Scholar and Web of Science, which simultaneously observed g_s, F_C,COS, F_C,CO2, environmental conditions and the simulated LRU [14,51,65,90,96,101,113,114,115,117,118,119,120,121,122,123,124,125,126,127,128,129]. According to the PRISMA statement of systematic reviews and meta-analyses, when retrieving these studies, we used measurement, photosynthesis and g_s as the keywords, and then selected 228 studies. Then, we used measurement, photosynthesis and COS as the keywords, and then selected 58 studies. In the intersection of the two folders there were 25 studies that simultaneously observed photosynthesis, COS and g_s. Then, to constrain the uncertainty caused by the temporal variation of observations from instruments and the parameters, we paid no consideration to the belowground parts [90] and just chose measurements of leaves. Ultimately, 22 datasets from 15 sites were chosen for this meta-analysis. During measurements of the leaf gas exchange and ${\mathrm{g}}_{\mathrm{s}}$, these studies also measured relevant environmental conditions. We collected four categories of environmental conditions, and light conditions were represented by light intensity (μmol m⁻² s⁻¹) or photosynthetic active radiation (PAR, μmol m⁻² s⁻¹), temperature conditions were represented by the canopy or soil temperature (°C) and water conditions were represented by rH (%) and VPD (Pa). Meanwhile, the atmospheric COS and CO₂ concentrations were represented by C_a,COS (ppt) and C_a,CO2 (ppm).

Raw data were acquired from the tables, figures and appendixes of the publications using Engauge Digitizer (Free Software Foundation, Inc., Boston, MA, USA). Relationships between the environmental conditions and dependent variables (i.e., LRU, g_s, F_C,COS, and F_C,CO2) identified in the studies were also collected for verification of the results. We filtered the search results to find studies performed over the plant photosynthetic part of terrestrial ecosystems and classified the research in previous studies on the basis of their study sites. Then, these studies were categorized into subsets of biomes on the basis of their study sites. These biomes included needle-leaf forest (NLF), broadleaf forest (BLF), mixed forest (MXF), grassland (GRA), marsh (MSH), cropland dominated by C3 plants (CPL_C3) and cropland dominated by C4 plants (CPL_C4) (Figure 2).

In each biome group, the studies were separated according to their measurement frequencies. The data from studies that included data from diurnal variations were classified into groups as noon vs. morning and evening (N vs. ME), day vs. night (D vs. N) and forenoon and afternoon vs. dawn and dusk (FA vs. DD) data; FA vs. DD was included to avoid the influence of inverted light-impacted stomatal closures at midday [130]. Data from periods longer a day, such as daily and seasonal records, were employed only for testing. The summer solstice day and the autumnal equinox day (normally on 20 June and 23 September, DOY 171 and 266) were used as the beginning and end of summer, respectively. For comparison, PAR = 600 μmol m⁻² s⁻¹ was the criterion of sunny and cloudy conditions when separating the records of the control experiments (

Table 1. Control (Contr.) and treatment (Treatm.) groups at each timescale.

Group	Treatm.	Contr.	Treatm.	Contr.	Treatm.	Contr.
Diel and Seasonal	Daily		Seasonal
	PAR ≤ 600 μmol m−2 s−1	PAR > 600 μmol m−2 s−1	Before DOY 171, After DOY 266	DOY 171~266
Diurnal	N vs. ME		FA vs. DD		D vs. N
	6:00~9:00, 15:00~18:00	9:00~15:00	5:00~8:00, 17:00~20:00	8:00~11:00, 15:00~17:00	5:00~20:00	20:00~5:00
Control	PAR ≤ 600 μmol m−2 s−1	PAR > 600 μmol m−2 s−1
Total	All groups

). Notably, different biomes in the same study were classified as separate groups, and data from the same study site recorded at different temporal scales were separated into independent groups. A total of 101 separate studies were used to establish the dataset. The mean value of each variable and its sampling size and standard deviation were also collected. The sample size of each group and each indicator, as well as the characteristics of the variations in environmental conditions, are listed in Table S1.

2.2. Meta-Analysis

For different dependent variables, the changes in environmental conditions were quantified between the control and the treatment. To estimate the impact of environmental condition changes between the controls and treatments, statistical parameters (e.g., the count ($\overline{N}$), average value ($\overline{X}$) and standard deviation ($\overline{SD}$)) for both the controls and treatments were used to perform meta-analysis for each group with METAWIN 2.1 [131]. Then, the effect size (d) was calculated as follows [73]:

$$\mathrm{d}=\frac{{\overline{N}}_E-{\overline{N}}_c}{s}J$$

(1)

$$s=\sqrt{\frac{\left(n_E-1\right)\left(s_E^2\right)+\left(n_c-1\right)\left(s_C^2\right)}{n_E+n_c-2}}$$

(2)

where the subscripts E and C denote experimental and control groups, respectively; ${\overline{N}}_i$ is the mean; n_i is the sample size; and $s_i^2$ is the variance for the experimental (i = E) or control (i = C) groups; J is used to correct for bias due to small sample size [73].

A response ratio (RR) was also calculated as follows [132]:

$$\mathrm{RR}=\ln \left(\frac{\overline{X_t}}{\overline{X_c}}\right)=\ln \left(\overline{X_t}\right)-\ln \left(\overline{X_c}\right)$$

(3)

where $\overline{X_t}$ and $\overline{X_c}$ are the means of a variable in the treatment (Tables S2 and S3) group and the control group, respectively [132]. Its variance ($v$) was estimated as follows:

$$v=\frac{S_c^2}{n_c\overline{X_c^2}}+\frac{S_t^2}{n_t\overline{X_t^2}}$$

(4)

where $n_c$ and $n_t$ are the sample sizes in the control and treatment groups, respectively; S_c and S_t are the standard deviations in the control and treatment groups, respectively.

The impact of the change in environmental conditions on Hedges’ d or $\ln \left(\mathrm{RR}\right)$ was considered significant if the 95% confidence interval (CI) did not include zero. The total heterogeneity (${\mathrm{Q}}_{\mathrm{total}}$) of each group was classified into within-group heterogeneity (${\mathrm{Q}}_{\mathrm{within}}$) and between-group heterogeneity (${\mathrm{Q}}_{\mathrm{between}}$). If the probability value ${\mathrm{Q}}_{\mathrm{between}}\le 0.05$, the difference between the groups or biomes was considered significant. In detail, ${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d}}$ is the ${\mathrm{Q}}_{\mathrm{between}}$ value of ${\mathrm{Hedges}}^{’}\mathrm{d}$, and when the groups are fixed to biomes, ${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d},\mathrm{biomes}}$ was used. The response function that fitted the data from the most studies was chosen to represent the dependent variables’ responses to the changing environmental conditions. The mean effect size, 95% CI and sample number were calculated for each biome and group with MATLAB R2016b (The MathWorks, Inc., Natick, MA, USA).

2.3. Statistical Analysis

To investigate the changes in environmental conditions regulation of stomata and the responses of F_C,COS and F_C,CO2, the relationships between F_C,COS and variation in C_a,COS or C_a,CO2 and the relationships between F_C,CO2 and variation in C_a,COS or C_a,CO2 were compared. These comparisons further clarified whether the LRU would be adjusted along with the variation in g_s. First, we chose groups set by diurnal variations and compared the linear, quadratic or cubic regression between the Hedges’ d of each dependent variable and the change in each environmental condition, as well as its reciprocal transformation. Significant intercorrelations were obtained to explain the impact of the change in environmental conditions on each dependent variable. Second, we performed the aforementioned regression between the Hedges’ d value of the LRU, g_S, F_C,COS and F_C,CO2 in pairs to clarify the roles of F_C,COS and F_C,CO2 regarding how changes in environmental conditions regulate g_S and the synchronous variation in the LRU; these results corroborated the correlation between F_C,COS and F_C,CO2, as well as how the LRU responded to the stomata. Scatter plots with regression lines were created with SigmaPlot 12.5 (Systat Software, Inc., San Jose, CA, USA).

3. Results

3.1. Changes in Environmental Conditions

Among the diurnal variation groups, in each LRU, g_s, F_C,COS and F_C,CO2 group, the light intensity was increased in 100%, 73%, 89% and 90% of the studies, respectively, in each group (Figure 3a). Moreover, 89%, 63%, 71% and 69% of the studies included a warming process, respectively (Figure 3b). Although characterized by VPD, the water conditions at the canopy boundary were increased in 100% of the overall studies, and the rH values showed that the atmospheric moisture was greatly diminished in 75%, 50%, 71% and 60% of the studies, respectively (Figure 3c,d). For C_a,COS and C_a,CO2, 90%, 50%, 74% and 71% of the studies showed enhanced C_a,COS, respectively (Figure 3e). In contrast, 86%, 50%, 36% and 42% of the studies showed diminished C_a,CO2, respectively (Figure 3f).

3.2. Changes in the Leaf LRU, g_s, F_C,COS and F_C,CO2

The meta-analyses showed that there were significant differences in the LRU between the groups categorized by time scale, which reflected the tendency of environmental conditions (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d}} < 0.01$). Furthermore, no significant discrepancy was observed between the groups at the same time scales (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d}}=0.56$) (Figure 4a). Under the aforementioned environmental condition changes, the LRU decreased in the “Total”, “Control” and the N vs. ME groups among the groups categorized by time scale, which was significantly stronger in the NLF, MXF and CPL_C3 biomes. Solely on the aspect of light intensity, the decrease of LRU was noticed in the lower light “control” group. Moreover, no significantly different LRU responses among biomes were obtained (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d},\mathrm{biomes}}$= 0.12) in response to environmental condition changes. A change in the LRU was associated with enhanced g_s due to stomatal opening in all the groups, except the daily time scale group. Although no significant differences were found between biomes (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d},\mathrm{biomes}}=0.73$), the g_s in MXF, CPL_C3 and CPL_C4 showed the strongest response to environmental condition changes (Figure 4b). For the photosynthetic gas fluxes, no significantly different F_C,COS responses to environmental condition changes were found between the groups categorized by time scale (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d}}=0.44$). After experiencing the strongest variation of VPD, C_a,COS and C_a,CO2 (Figure 3d–f), F_C,COS was diminished in most samples of N vs. ME, FA vs. DD, D vs. N and “Total” groups; meanwhile, only in N vs. ME was this decrease significant. The diminished F_C,COS was also found in the NLF, BLF, GRA and CPL_C3 biomes. In contrast, in the biomes of MXF, MSH and CPL_C4, F_C,COS was found to have increased. At the same time, significantly different F_C,COS responses were found between biomes (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d},\mathrm{biomes}}=0.00$) (Figure 4c). Moreover, no significant differences in the F_C,CO2 response were found between the different temporal groups (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d}}=0.74$). F_C,CO2 decreased significantly in all the N vs. ME, FA vs. DD, D vs. N and “Total” groups, which reflected F_C,CO2 diurnal variations. Significantly different F_C,CO2 responses were found between biomes (${\mathrm{Q}}_{\mathrm{b},{\mathrm{Hedges}}^{’}\mathrm{d},\mathrm{biomes}}=0.00$). In NLF, BLF, GRA and CPL_C3, the environmental conditions had an evident negative impact on F_C,CO2. However, F_C,CO2 increased in MXF and MSH under the same environmental conditions variation (Figure 4d).

3.3. The Dependent Variables as Regulated by Environmental Conditions

On the groups categorized by diurnal cycling (N vs. ME, FA vs. DD, D vs. N), the dependent variables (such as LRU, g_s, F_C,COS and F_C,CO2) were regulated by the environmental conditions (light intensity, temperature, rH, VPD, C_a,COS and C_a,CO2). Of the environmental conditions, the LRU was significantly influenced by variations in light intensity (R² = 1, p = 0.002), VPD (R² = 0.885, p = 0.005), C_a,COS (R² = 0.58, p = 0.05) and C_a,CO2 (R² = 0.97, p = 0.003) (Figure 5a–f). There was an exponential relation between light intensity and the LRU (Figure 5a). On the other hand, increased VPD reduced the LRU (Figure 5d). As C_a,COS increased, the LRU increased until a certain turning point of C_a,COS and then decreased. The regression suggested the LRU saturation point could be ΔC_a,COS ≈ 13 ppt (Figure 5e). In contrast, between LRU and ΔC_a,CO2, the LRU saturation point could also be ΔC_a,CO2 ≈ −18 ppm (Figure 5f).

As shown by the linear regression of the change in environmental conditions to g_s, the stomata were opened significantly only by light intensity (R² = 0.58, p = 0.01) (Figure 6a).

The fitted F_C,COS increased significantly with a decrease in light intensity and was significantly decreased with a slight increase in light intensity (R² = 0.23, p = 0.01) (Figure 7a,b).

In addition, F_C,CO2 decreased linearly with increasing temperature (R² = 0.80, p = 0.00) (Figure 8b). In terms of moisture conditions, F_C,CO2 increased linearly with increasing rH (R² = 0.83, p = 0.00) (Figure 8c). However, at the canopy boundary layer, the change in VPD did not show any significant correlation with either flux (Figure 7d and Figure 8d). From the perspective of gas exchange, F_C,COS was not significantly limited by C_a,COS (Figure 7e). Nevertheless, there was a compensation point for C_a,CO2 assimilation (ΔC_a,CO2 ≈ 10 to 40 μmol m⁻² s⁻¹), under which, F_C,COS was the weakest (R² = 0.30, p = 0.04) (Figure 7f). F_C,CO2 decreased linearly with increased C_a,COS (R² = 0.19, p = 0.04) but increased linearly with increased C_a,CO2 (R² = 0.19, p = 0.04) (Figure 8e,f).

3.4. The Role of F_C,COS and F_C,CO2 in the Mechanism of LRU and g_s Regulation

In the diurnal cycling groups, there was no significant correlation between either the $\mathrm{Hedges}’\mathrm{d}$ of F_C,COS or the $\mathrm{Hedges}’\mathrm{d}$ of F_C,CO2 with the $\mathrm{Hedges}’\mathrm{d}$ of the LRU (Figure 9a,b).

However, F_C,COS was significantly correlated with g_s (R² = 0.34, p = 0.02) (Figure 10a). The $\mathrm{Hedges}’\mathrm{d}$ of g_s exhibited a compensation point at a certain level of the $\mathrm{Hedges}’\mathrm{d}$ of F_C,COS (Figure 10a).

The linear correlation between the $\mathrm{Hedges}’\mathrm{d}$ of F_C,COS and the $\mathrm{Hedges}’\mathrm{d}$ of F_C,CO2 was determined (R² = 0.65, p = 0.00) (Figure 11a). Ultimately, when doing correlation analysis, there were not enough samples to separate the control group or the groups of CPL_C3, CPL_C4 and MXF from the total. Surprisingly, there was no significant correlation between the $\mathrm{Hedges}’\mathrm{d}$ of g_s and the $\mathrm{Hedges}’\mathrm{d}$ of the LRU (Figure 11b).

4. Discussion

4.1. Co-Response of LRU and Stomata to the Variations in Environmental Conditions

The changes in environmental conditions under diurnal cycling could be characterized as increases in light and temperature. Drought was characterized by a decrease in rH or an increase in VPD. For most studies investigating either LRU or F_C,COS, C_a,COS and C_a,CO2 increased, while for most studies that investigated F_C,CO2, C_a,CO2 decreased.

The g_s models (i.e., the Ball–Berry model) already simplified the regulating effects of environmental conditions on g_s at the leaf scale, and the “big leaf” model cannot be directly applied at the canopy scale [133]. However, this study indicated that under increasing light, the LRU response may be the opposite of the increase in g_s, even though there is no evidence to illustrate the correlation between g_s and the LRU. Under the change in environmental conditions, significantly opposite responses of the LRU and g_s were found, mostly in the MXF and CPL_C3 biomes; the significance was confirmed by comparison with the control groups. Only in the MXF, CPL_C3 and CPL_C4 biomes was g_s increased. Moreover, an increase in g_s with a decrease in the LRU was found in the N vs. ME group and the controls. Due to the lack of data, the LRU regulation mechanism in the GRA and CPL_C4 biomes was unclear. Due to the lack of separation between evergreen and deciduous BLF and the strong influence of moist-oxic soil [43], the correlations between the LRU and g_s in the NLF, BLF and MSH biomes were nonsignificant.

From the response of LRU to the change in environmental conditions, the decrease in the LRU with the increase in light and VPD found in this study was consistent with the findings of [115]. During an increase in light intensity, the decrease rate of the LRU slowed, as the LRU became saturated when exposed to strong light [115]. Moreover, the saturation point was greater than 380 μmol m⁻² s⁻¹. At the canopy scale, the main factors that are synergetic with g_s include the diffuse light fraction and VPD [73,121,134], which showed a trade-off relationship with the LRU. LAI influences g_s at the canopy scale as well [73,121,134]. Stomata are typically open under elevated C_a,COS conditions [51,124], and in this study, the LRU reached peak values when ΔC_a,COS ≈ 13 ppt, as well as when ΔC_a,CO2 ≈ −18 ppm.

4.2. Coresponse of Canopy COS and CO₂ Fluxes to Variations in Environmental Conditions

At timescales under diurnal cycling, F_C,COS and F_C,CO2 both decreased in FA vs. DD and D vs. N under the changes in environmental conditions. However, F_C,COS increased in the MXF, CPL_C3, and CPL_C4 biomes, which was also found in most studies of NLF. In addition, in the N vs. ME group, F_C,CO2 in MXF and MSH increased in the FA vs. DD and D vs. N groups, and F_C,CO2 decreased in NLF, BLF, GRA, and CPL_C3.

In this study, when the light intensity was reduced by 300~700 μmol m⁻² s⁻¹, F_C,COS was composed of continuing assimilation by the leaves. For instance, stomata open in the dark and cause COS assimilation [135], which makes up 17% of daily F_C,COS [118]. Under a minor increase in light intensity, F_C,COS significantly decreased. However, F_C,CO2 increased under cooler conditions. In croplands and mid-latitude forests, F_C,COS also decreased under water shortage [65,101,136]. Water shortage, an increase in temperature, an increase in C_a,COS and a decrease in C_a,CO2 were all main factors resulting in a decrease in F_C,CO2, as the tissue water content is important in canopy CO₂ assimilation [137]. However, an slightly increase of C_a,CO2 value also indicated the lowest F_C,COS, as there was a hysteresis between F_C,COS and F_C,CO2 fluctuations within diurnal cycles [96]. Simultaneous increases in F_C,COS and F_C,CO2 were found only in MXF. In contrast, in NLF and CPL_C3, only F_C,CO2 decreased. In the nonstomatal structures, the COS assimilation rate increased with water shortage [138,139,140]. For example, in some bryophytes, mosses, liverworts and lichens, as the hydrolysis process is catalyzed by the light-independent CA enzyme [37] and is affected by reverse hydrolysis, the net absorption rate of COS is independent of light conditions [85]. Bidirectional hydrolysis has also been observed in some fungi and bacteria [141,142,143]. This retrohydrolysis does not cause COS emissions and influence the F_C,COS of the higher plants overall [51,56,144]. For higher plants, although there is a reverse hydrolysis process in the stomata opening process, resistance and osmotic pressure stopped COS from being emitted; the light-independent CA enzyme is still the key to the absorption rate of COS during photosynthesis [145,146]. Enhanced frictional wind at sunrise caused COS emission from the canopy gap; furthermore, soil microorganism breakdown boosted by light seems to be the reason for the significantly increased F_C,COS. Recent studies also discussed whether F_C,COS would be affected by the diversity of the enzyme CA between different species [147,148]. When the COS concentration in the subcavities of the mesophyll (C_m,COS) is close to zero, F_C,COS is strongly affected by C_m,COS [144,149]; however, no correlation was found between the changes in C_a,COS and F_C,COS.

4.3. The LRU Variation Mediated by the Canopy COS and CO₂ Fluxes under the Regulation of Stomata

Most of the gas exchange in leaflets is controlled by the stomata [73,150]. In this study, at the canopy scale, no evidence was found regarding the correlation between the LRU and F_C,COS or F_C,CO2. In the diurnal cycling groups, increased g_s and decreased F_C,COS were found to occur concurrently, except in the case of N vs. ME. This difference in N vs. ME could have been due to the stomatal closure that occurs at midday [130]. However, in non-wetting vegetation, such as NLF, MXF, CPL_C3 and CPL_C4, moderate stomatal closure and increased enzyme CA activity caused an increase in the F_C,COS of N vs. ME. The LRU and g_s were found to respond in opposite ways to the changes in environmental conditions in MXF and CPL_C3. Moreover, the LRU was found to differ between species. In the NLF and CPL_C3 groups, the LRU decreased with increasing F_C,COS and decreasing F_C,CO2. In contrast, increases in both F_C,COS and F_C,CO2 were found in GRA in which the LRU was found to decrease.

In the process of changing environmental conditions, under a minor increase in light intensity, F_C,COS decreased. Furthermore, synchronization was observed between the decrease in F_C,CO2 caused by warming and the decrease in the LRU caused by an increase in light. In contrast, g_s increased when the light intensity increased. Recent studies noted that pathway conductance and CA enzyme activity colimited F_C,COS [73]. Furthermore, the response of stomata to temperature was not very sensitive [151], as the CO₂ concentration in the subcavities of the mesophyll and the activity of enzyme CA colimited conductance throughout the whole pathway [138]. This study found a linear trade-off relationship only between F_C,CO2 and temperature. However, there was no correlation between F_C,CO2 and g_s because GPP and ER were not partitioned from F_C,CO2. Meanwhile, water shortage resulted in decreased F_C,CO2 (consistent with the decreases in rH and VPD).

In the diurnal cycling groups, there was a synergetic correlation between F_C,CO2 and the variation in the rH or C_a,CO2. However, F_C,CO2 exhibited a trade-off with temperature or C_a,COS. Then, stomata closure was controlled by the C_a,COS decrease [51,124]. Moreover, enhanced friction wind caused COS emission from the canopy and induced a quadratic correlation between C_a,CO2 and F_C,COS. Therefore, the variation in F_C,COS was smaller than that in F_C,CO2 [96]. However, no significant correlation was found between the LRU and either F_C,COS or F_C,CO2. Environmental conditions such as light intensity, VPD, C_a,COS and C_a,CO2 were the key factors influencing LRU variation. As evidenced by the results from NLF, MXF and CPL_C3, C_a,COS and C_a,CO2 increased in the diurnal cycling groups in non-wetting biomes. During the diurnal variation, the canopy COS absorption flux mainly occurred in the daytime. First of all, a minor increase in light intensity enhanced g_s, and enhanced friction wind caused COS emission from the canopy gap; furthermore, F_C,COS decreased. Furthermore, as the intercomparison of global studies revealed, as C_a,CO2 increased by 10 to 40 μmol m⁻² s⁻¹, F_C,COS decreased to its lowest and rebounded. The LRU was influenced by both C_a,COS and C_a,CO2. In contrast, as the stomata closed, the COS diffusion pathway was inhibited due to water shortage at midday; even if C_a,COS and C_a,CO2 increased, F_C,COS was still decreased. Finally, F_C,COS was the lowest when the stomatal openings were small. Notably, the increase and decrease in F_C,COS could both be related to an increase in g_s. The regulation mechanism is important in formulating the LRU time series and in upscaling the LRU into the ecosystem relative uptake (ERU) and the atmospheric relative uptake (ARU) [122]. However, the nonsignificant relationship between the Hedges’ d of g_s and the Hedges’ d of the LRU limited our ability to illustrate how the variations in environmental conditions regulate stomata and fluxes and thus further influence the LRU.

4.4. Employing the LRU to Constrain the Uncertainties in COS-Based GPP Simulations

A potential improvement to the partitioning of NEE [152] is based on the function relating F_C,COS, the GPP and the LRU [43,122]. The quantum cascade is an advanced technique employed in the continuous measurement of F_C,COS; in this system, ${\mathrm{R}}_d$ is much smaller than in other systems [54,56,153]. However, during the COS-based simulation of the GPP, COS emissions also occur in dehydrating leaves [85], which would lead to systematic noise. Moreover, the partial nocturnal closure of stomata, which is the reason for the depletion of the nighttime C_a,COS [65,73,101,113,114,129,154,155], was also observed in certain species [101,113]. Overall, at night, even though COS hydrational assimilation on the water surface and in oxic soils, as well as oxidative release in anoxic soils [141], participate in COS circulation, the terrestrial nighttime F_C,COS is mostly influenced by the plant canopy [65,73,113,129,156]. It is also worth quantifying the influence of COS nighttime flux on the LRU at the diurnal scale. Recent studies have drawn attention to the fluctuation of F_C,COS [73,114]. These studies noted that even though F_C,COS was considered to be regulated by the activity of the enzyme CA and the LAI of the plant canopy, both diurnal and seasonal fluctuations in C_a,COS and C_a,CO2 were correlated with F_C,COS [73,114]. Additionally, C_a,COS and C_a,CO2 peaked at night when F_C,COS was low [157,158]. During the heavy turbulence at dawn, as COS and CO₂ were replenished in the canopy and subcanopy airspace, the canopy storage effect also influenced the simulated F_C,COS and F_C,CO2 [73]. It is also worth exploring whether this process influences the LRU.

Further studies should be performed to identify the additional key factors that regulate LRU variation [159] and the influence of NEE partitioning. Layered observations led to a proposal to partition F_C,COS into leaf assimilation and soil exchange components [42]. Moreover, soil COS consumption, which is linearly correlated with soil temperature and makes up 34~40% of the nighttime F_C,COS in the boreal forest, reached its saturation point at 20% water content [118,156]. It is worth further exploring the considerable differences between the environmental conditions regulation mechanism for COS leaf assimilation and that for COS soil exchange.

5. Conclusions

During plant photosynthesis and COS assimilation, light and water conditions co-regulated the stomata and LRU. Overall, at diurnal scales, F_C,COS was controlled by the stomata and was decreased by water shortage in non-wetting biomes; at the same time, the leaves were exposed to increased C_a,COS. However, under variations in the environmental conditions, a three-step response occurred in F_C,COS. First, slightly increased light increased the g_s, as well as the COS emission from the canopy gap; then, F_C,COS was decreased. Second, when the stomata were opening, since the subcanopy emission of COS and the enzyme CA catalysis rate of COS exceed CO₂, a C_a,CO2 increase of 10 to 40 μmol m⁻² s⁻¹ was a compensation point for COS assimilation, at which point, F_C,COS was the lowest. Third, the stomata closed under both high light intensity and high temperatures, and F_C,COS was decreased as well. Although this study provided no evidence of the correlation between the LRU and g_s, they were shown to be regulated by the variations in environmental conditions in opposite ways, at least in MXF and CPL_C3. Moreover, the LRU decreased under slightly increased light and atmospheric drought. The LRU saturation points of C_a,COS and C_a,CO2 were observed when ΔC_a,COS ≈ 13 ppt or ΔC_a,CO2 ≈ −18 ppm. In further studies, we will formulate the quantitative model of the LRU regulation mechanism based on the above-mentioned conclusion using g_s, C_a,COS, C_a,CO2, light and water conditions.