Chamber‐based system for measuring whole‐plant transpiration dynamics

Abstract Most of our insights on whole‐plant transpiration (E) are based on leaf‐chamber measurements using water vapor porometers, IRGAs, or flux measurements. Gravimetric methods are integrative, accurate, and a clear differentiation between evaporation and E can be made. Water vapor pressure deficit (VPD) is the driving force for E but assessing its impact has been evasive, due to confounding effects of other climate drivers. We developed a chamber‐based gravimetric method, in which whole plant response of E to VPD could be assessed, while keeping other environmental parameters at predetermined values. Stable VPD values (0.5–3.7 kPa) were attained within 5 min after changing flow settings and maintained for at least 45 min. Species differing in life form and photosynthetic metabolism were used. Typical runs covering the range of VPDs lasted up to 4 h, preventing acclimation responses or soilborne water deficit. Species‐specific responses of E to VPD could be identified, as well as differences in leaf conductance. The combined gravimetric‐chamber‐based system presented overcomes several limitations of previous gravimetric set ups in terms of replicability, time, and elucidation of the impact of specific environmental drivers on E, filling a methodological gap and widening our phenotyping capabilities.

lysimeters are based on soil weight loss and are often unable to accurately separate soil evaporation from E. Satellite methods and Eddy covariance can capture water dynamics from field to entire regions, but increasing scale implicitly loses resolution and relies increasingly on assumptions to distinguish E from evaporation (Li et al., 2019).
Measuring E over time by differences in weight loss using laboratory balances is an integrative, reliable, and easily replicable approach, provided soil evaporation is minimized or accurately accounted for and the measuring time is short enough to avoid detectable gains in dry mass due to CO 2 assimilation (Cirelli et al., 2012;Cliffton-Brown & Jones, 1999). Chamber-based gravimetric measurements of whole plant E are rarely reported, probably due to the difficulty in maintaining consistent and stable conditions within the chamber (Fletcher et al., 2007), which can lead to the adjustment of plants to the unstable environmental conditions and subsequently to being misconstrued as "steady-state" E responses.
The difference in water vapor concentration between the leaf intercellular spaces and the atmosphere (represented by the atmospheric water vapor pressure deficit (VPD)) is the driving force for E. As a consequence of global warming, VPD has increased (Hatfield & Prueger, 2015). Indeed, values as high as 4.0 kPa associated with heat waves have been reported (Medina et al., 2017).
However, assessing the impact of VPD on E has been evasive, due to difficulties separating the actual effect of VPD from other climate drivers that vary concomitantly (Grossiord et al., 2020).
Assuming water availability in the root zone is not limiting, it follows that high VPD will result in high E, due to the increased gradient between the leaf and the atmosphere. Nevertheless, this trend is not always maintained as most species or even genotypes within a species may limit E at high VPD via stomatal control of water loss (Bunce, 1984). The apparent stomatal response to VPD is additionally complicated by the possible influence of internal and external factors such as sub-stomatic CO 2 concentration, air temperature, light intensity, or hormonal triggers on stomatal behavior (Asch et al., 1995;Bunce, 1997;Yong et al., 1997). The speed with and degree to which a plant responds to VPD has been shown to vary across species and even genotypes (Devi et al., 2010;Ocheltree et al., 2013;Tardieu & Simonneau, 1998). Plant responses of E to VPD can roughly be grouped by the hydrostability level against increased evaporative demand (Stocker, 1956). Isohydric, hydrostable plants, such as apple (Malus pumila) and rice (Oryza sativa), show a pronounced stomatal response to maintain leaf water potential as VPD increases (Jones, 1983), indicated by a nonlinear response to VPD. On the other hand, in anisohydric, hydrolabile plants, such as sunflower (Helianthus annuus) and barley (Hordeum vulgare), E responds mostly linearly to VPD, whereas stomata respond to changes in soil moisture and not to air humidity (Tardieu & Simonneau, 1998). The capacity of a plant to regulate E against increased VPD ultimately keeps water in the soil, which may be beneficial if drought conditions occur later in the season (Sinclair et al., 2005). Consequently, determining the independent effect of VPD on E for a species or variety will be advantageous when selecting or breeding for drought prone environments.
In this paper, we present a chamber-based gravimetric method to measure whole-plant E dynamics in a controlled environment in the laboratory, in which VPD can be accurately adjusted, while keeping other environmental drivers controlled. We tested and validated the system using contrasting plant species, differing in life form, habit, and photosynthetic metabolism. The result is an affordable, sensitive, reliable, and highly time-resolved method of analysis of E in response to VPD in combination with environments varying in light intensity and temperature.
2 | MATERIAL S AND ME THODS 2.1 | Description and set-up of the gravimetric monitoring system for whole-plant transpiration (MoSysT) MoSysT ( Figure 1) consists of two chambers: a main (80 cm wide x 80 cm deep x100 cm high) monitoring chamber (1) and an upstream (66 cm wide x 80 cm deep x100 cm high) air pre-mixing chamber (2).
Both air streams discharge into the pre-mixing chamber before supplying the main measuring chamber from a side entrance. Air streams are forced into the mixing chamber from the wet air depot and into the main measuring chamber by inside-access-PVC pipe ventilators (15 and 16) automatically regulated by the control of a custom-programmed software. Target chamber VPD levels were set by entering the corresponding temperature and humidity values into the software control interface. Based on these target values, the software constantly responds to incoming readings from sensors at 1-s interval, monitoring humidity and temperature in (13) and out (7) of the main chamber by auto-adjusting the wet air stream ventilator and heating devices in the main (10) and air pre-mixing chamber (11) via drive and power regulation, respectively. Air flow into the main chamber was always kept above 40 m 3 h −1 to ensure that the whole chamber volume is exchanged every minute to avoid feedback effects of transpiring plants.

| Plant cultivation and post-chamber measurement and harvest
A group of 10 plant species differing in habit, and photosynthetic metabolism was used (Table 1). All plant species were raised from seeds directly sown into nursery pots filled with standard compost and commercially available fertilized garden soil. After establishment, four seedlings of each species were transplanted into 2-L F I G U R E 1 Diagram of the controlled environment gravimetric whole-plant transpiration monitoring system. The system comprised the following components: (1) Main chamber; (2) Air pre-mixing chamber; (3) Water tank with floating ultrasonic nebulizer; (4) Dehumidifier; (5) Computer laptop unit/data logging and control software; (6) temperature and humidity sensor; (7) Temperature and humidity logger with visual display; (8) Balances with interface to laptop unit; (9) Light emitting diode; (10) Main chamber heating unit; (11) Air pre-mixing chamber heating unit; (12)   The transpiration rate was calculated as: where Δ weight(g) is the change in weight between two consecutive measurements; MW H20 is the molecular weight of water; t(s) is the time between two consecutive weight recordings in seconds; and LA(m 2 ) is the leaf area of the plant in m 2 .
The leaf conductance was calculated according to Buckley (2005): where E is the transpiration rate as calculated in (Equation 1).
Initially, four species, two common C4 pasture grasses in the tropics and subtropics, Brachiaria brizantha (Hochst. ex A. Rich.) Stapf. and Setaria sphacelata var. Narok, and two wide-spread C3 pasture legumes, Trifolium repens (L.) and Lotus corniculatus L., were used. To test the stability of the system, we exposed the plants to

| Statistical analysis and selection criteria of steady-state E
We used a statistical selection criterion to identify the period of constant (steady-state) whole-plant transpiration rate. The procedure assumes that leaves pass through a transition period while adapting to any new chamber environment (here consecutively increasing/ decreasing levels of VPD) until a steady-state E was reached. We

| RE SULTS
MoSysT was able to reach VPD values ranging from 0.5 KPa to 3.7 KPa (Figure 3, Appendix S1), which covers a wide range of conditions including very dry environments. Five different and stable VPD values could be reached and maintained for at least 45 min.
The internal adjustment of the system to dry and wet air streams was fast, within 5 min. a new stable VPD was reached in the chamber and could be maintained within +/− 10% of the targeted value (1) E = Δ weight(g) ∕ MW H20 * t(s) * LA m 2 * 1000 (2) G L = E ∕ VPD * 100 we were able to detect a sudden increase in E shortly after each VPD increment in the chamber. This sudden increase lasted between 3 and 5 min after which, a steady-state E corresponding to the specified VPD was reached (Figure 4). This "overshooting" transient response was neither observed in S. sphacelata nor in B. brizantha.
Overall, E increased with VPD in all species (Figure 4) Transpiration rates at specific VPD values were similar regardless of whether they were obtained by increasing or decreasing VPD (see Appendix S2) and this response was observed in all species. This result can be graphically seen in Figure 5 (Appendices S1 and S2) in which there was no distinction in the response of E to VPD regardless of the sequence of VPDs used.
The E response to VPD followed a saturation curve and the data were best fit by applying quadratic hyperbolic models. We compared the single hyperbolic rectangular model with the linear model fit, and r 2 was consistently higher following the nonlinear model (not shown).

F I G U R E 2
Schematic representation of the algorithm followed to select steady-state transpiration rates (E). (1) Identification and removal of outliers, (2) calculation of the slope of the regression line; (3) Hypothesis system for the selection of slopes not significantly different from zero; (4) Further selection criteria in case more than one 10 consecutive measurements complied with the pre-requisite in (3).
As expected C4 species tended to show lower E values for each VPD than C3 species ( Figure 6, Table 1). The response of E to VPD was highly variable and dependent on the species, not only in terms of absolute E values attained, but also in the shape of the response to VPD.
Whereas in the C4 species the response showed no signs of saturation, in the C3 species on the contrary, a saturating response was observed. Contrarily, leaf conductance was either constant or decreased with VPD ( Figure 7, Table 1) in all species but in C. virgata, in which leaf conductance increased in a curvilinear fashion in response to VPD.

| DISCUSS ION
This study introduced a gravimetric method coupled to an air-mixing system in which E can be accurately determined and environmental conditions were effectively controlled. To our knowledge, this is the first time a fully controlled system is presented that is able to cover a wide range of VPD values from <1 kPa, to almost 4 kPa. It can reproducibly and accurately measure transpiration rate at the whole-plant level, and thus characterize the response of E to short term changes in VPD independently of other environmental drivers such as temperature and light, which have been confounding factors of the impact of VPD on E.
The frequency (1-min interval) of data acquisition allowed the identification of steady-state rates, as well as the detection in some F I G U R E 3 Representative time course of temperature, relative humidity and water vapor pressure deficit (VPD) during transpiration measurements with five successively VPD levels. was not observed in the grass species used in this study. One might speculate that species-specific anatomical differences in the stomatal structure (i.e., guard and subsidiary cells) could be responsible as has been pointed out by Franks and Farquhar (2007) and Gray et al. (2020).

Time (min)
One of the main advantages of MoSysT is the control of environmental drivers that can affect E concomitantly with VPD; these include not only light and temperature, but also fertilizer supply or soil water content. As typical runs in the MoSysT were shorter than 4 h, possibilities of any significant soil drying eliciting acclimation responses including early stomatal closure could be minimized. This can be noted in the reversibility of the response of E to VPD regardless of whether VPD was set from low to high values or in the opposite direction (Figures 4 and 5). Short duration of experiments using the MoSysT also prevent detectable weight gains by photosynthesis in fast-growing species.
The nonlinearity of the response of E to VPD and its reversibility ( Figure 4) are two main conditions that support the hypothesis of a direct response of stomata and E to VPD regardless of water status of the leaf, the feedforward response of stomatal aperture postulated by Schultze et al. (1972). Nevertheless, the feedforward response can be modulated by the water status of the plant (Schultze & Küppers, 1979) caused, for example, by mid/long-term soil water deficit. Despite that E increases with VPD, restricted or limited E at high VPD has been reported in several plant species and genetic variation has been identified in several crop plants (Broughton & Conaty, 2022;Fletcher et al., 2007;Gholipoor et al., 2010;Shekoofa et al., 2014). In most species reported here, particularly C3 ones, limited E at high VPD could be also deduced as the best fit for the response to VPD was a saturation quadratic hyperbolic model.
Limitation of E at high VPD has been postulated as an effective mechanism to increase water use efficiency and to conserve water early in the growing season for use during drought events later in the season and thus it is a desirable trait for crop improvement in drought prone environments.

| Concluding remarks
MoSysT is an affordable (total cost is ca. 10,000€) and reliable system that can be used to study the interaction between atmos-  instance, how diverse is the structure of the guard-subsidiary cell complex and how determinant it is for the response of E to VPD? (Gray et al., 2020). How is the E to VPD response modulated by other underlying stressors such as light, high or low temperature, soil water deficit, nutrient deficiencies, salinity? Is the ABA involved in stomatal closure after the "overshooting" response locally synthesized or is it translocated from roots? (Bauer et al., 2013). These and other knowledge gaps can be addressed using the MoSysCT.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data supporting the findings of this study are available in the supplementary data of this article.