Abstract
Here, we describe a novel device for programming and replenishing water transpired by potted plants. To test the robustness of the system, vines were subjected to progressive water stress (WS), the severity of which was maintained in relation to transpiration (Tc) of well-watered (WW) plants. Throughout the 40-day experiment, water supply in the WS treatment was progressively lowered to 70, 50, and 30% of WW Tc prior to rewatering. During the same stages, mean Tc of WS plants was 74, 48, 28, and 93% that of WW plants. Linear relationships between vine transpiration and water supply during the 40-day experiment (R2 = 0.95 for WW and 0.94 for WS) confirmed the reliability of the system in providing a water supply that closely tracked measured transpiration. The emptying volume of the cylinder tank was set at 265 mL and proved to be adequate for daily water losses, which ranged from ~300 to 2300 mL. In addition to relieving operators of laborious and time-consuming manual irrigation, the system provides the ability to adjust water supply to actual water use as measured concurrently in a grapevine-enclosure system and enables customization of the water supply according to the size and transpiration potential of each vine.
As traditionally rainfed viticultural regions experience more frequent, temporary summer drought, in-depth knowledge about the adaptation of grape genotypes to varying severity of water deficit is needed. In Italy, comprehensive surveys of the response of certain genotypes to water stress have been performed for a few varieties, especially Sangiovese and Montepulciano (Merli et al. 2014, Palliotti et al. 2014), but little is known about this subject for most varieties. This knowledge gap has adverse consequences for several reasons: (i) typically, when supplemental irrigation must be introduced in areas that have no previous experience with irrigation management, overuse is common, leading to obvious overirrigation and vine imbalances; (ii) gaps in knowledge about genotype-related vine water requirements and responses to water stress make it difficult to gauge the level of water deficit the plant can withstand without compromising grape yield and quality but while restricting excessive vegetative growth; and (iii) if direct measurements of soil or plant water status (i.e., predawn or midday leaf water potential) cannot be made, decisions about if, when, and how to irrigate are made according to empirical methods such as visual observation of vines.
A second issue is the choice of field versus pot studies, each of which has advantages and drawbacks. An advantage of field studies is that they are unbiased toward mature and fully productive experimental vines with no apparent root restrictions. However, in environments marked by variable summer rainfall, it is likely that severe stress will not occur in any given year unless a labor-intensive treatment, such as covering the berm early in the season to limit rainfall infiltration, is applied (Poni et al. 1994). Another major concern is that due to erratic root distribution in vineyards, targeting a given fraction of evapotranspiration replacement as the ratio of absorbed-to-delivered water is difficult. Site-specific assessment of vine water use is feasible using methods such as field lysimeters (Williams et al. 2003), sap-flow techniques (Ginestar et al. 1998), and trunk-diameter variation (Intrigliolo and Castel 2007), although these methods are time-consuming and allow only a limited number of vine replicates.
We chose a pot study because: (i) stress can be induced and relieved easily without interference from the surrounding climate; (ii) actual vine water use can be determined gravimetrically prior to water stress, thus allowing precise control of stress severity; and (iii) the use of pots assures that all water supplied is available for root uptake.
The proposed system is a “closed-loop” process in which the operator defines a general control strategy, after which the control system determines when and how much water to apply based on feedback from one or more sensors (Zazueta et al. 1993). In this type of system, feedback and control occur continuously. The majority of commercial microirrigation closed-loop control systems base irrigation decisions on sensors that measure soil moisture status (water potential or volumetric water content), use climatic data to estimate plant water use, or use a combination of these approaches. Systems that base decision making on plant water status are less common than those that use soil water status; a broadly marketed example is the Dynagage Flow32-1K Sap Flow system (Dynamax Inc., Houston, TX) in which the closed loop is established through a pump controller programmed to continuously deliver actual transpiration volumes as scanned through sap-flow readings (Van Bavel 1992).
Our aims in the present study were to (a) describe a new, custom-built system for running fully automated long-term water-deficit trials on potted plants by embedding a whole-canopy gas-exchange apparatus and a device for programmed water replenishment; (b) provide examples of the kinds of datasets the system can deliver; and (c) assess whether the automated water supply maintains soil and vine water status under the expected conditions.
Materials and Methods
Water-supply apparatus
The new water-supply device is embedded in the whole-canopy gas-exchange system described in detail by Poni et al. (2014). The system can concurrently monitor up to 12 chambers, and the switching interval is set at 90 sec. Figure 1 shows schematic and actual views of the automated water-supply apparatus, and Figure 2 shows the electrical wiring of the solenoid valve control system to the AM64 multiplexer. Items and relative costs needed to configure the system to measure vine transpiration gravimetrically or by whole-canopy gas exchange are shown in Table 1.
Actual view (A) and diagram (B) of the mechanical and hydraulic components of the water-supply system. 1: connection to the water pipeline; 2: valve for manual regulation of tank-filling speed; 3: solenoid valve for automated tank filling; 4: cylinder tank; 5: cylinder tank shaft; 6: adjustable volume of water supply; 7: stirrup connected to the cylinder shaft; 8: adjustable shaft dead end (adjustment is made through the stirrup); 9: spring for cylinder discharge; 10: valve for manual regulation of cylinder discharge speed; 11: solenoid valve for irrigation supply; 12: vine pots.
Layout for electrical wiring of the solenoid valves. CR10X: Campbell data logger; AM416: Campbell multiplexer; C: channel; GND: ground; RES: reset; CLK: clock; Com: common; H: high excitation voltage; BD649: Darlington transistor equipped with a protection diode; A: solenoid valve for filling cylinder tank; B: solenoid valve for vine water supply.
List of components and relative current costs (USD) for assessing vine transpiration by gravimetric or whole-canopy gas-exchange methods. Material and equipment quantities are calculated for a system that accommodates 12 potted vines and an ambient reference.
The core of the system is the cylinder tank (Figure 1, component 4), which is programmed to deliver water to the vine based on real-time measurements of vine transpiration. In brief, the system set-up, functioning, and operations are as follows: (i) the volume of water contained in the tank is determined by the length of shaft run (5) × the cylinder cross-section. Acting on the adjustable shaft dead end (8), the tank volume (6) to be loaded and delivered at any watering event can be set from 10 to 460 mL; (ii) when solenoid valve 3 is commanded to open, valve 2 is manually regulated so that water pulled under pressure from the urban pipeline fills the tank in ~15 sec; (iii) when valve 3 is shut off and one of the solenoid valves is concurrently opened (11), valve 10 is manually regulated so that the cylinder tank is pushed by the spring and shaft (9) to empty in ~40 sec; (iv) valve 3, the function of which is to fill the tank, is controlled by a Darlington transistor (item A in Figure 2) connected to the data logger through control line C4; (v) each of the 12 solenoid valves on the bench (11) is controlled by another Darlington transistor (item B in Figure 2) that functions to deliver water to each pot; cycling activation of the solenoid valves (11) is accomplished by the multiplexer AM416 via the control line Com H2 connected to the C3 clamp on the data logger (Figure 2). In summary, each watering event is activated by the following three steps: (1) the data logger activates the opening of valve 3 for 20 sec to allow filling of the cylinder from the urban pipeline; (2) the data logger deactivates valve 3 and waits for 0.5 sec; (3) the data logger activates the opening of one of the solenoid valves on the bench (11) and waits for up to 50 sec until the cylinder tank driven by the springs (9) returns to its initial, normally closed condition.
The first decision required is the working volume of the cylinder tank, which must initially be set by the operator at a value between 10 and 470 mL and adjusted against the expected total diurnal vine transpiration. Volumes that are too small must be avoided because a significant proportion of the supplied water would evaporate from the soil surface and thus be unavailable for root absorption. Moreover, depending on the soil texture and infiltration rate, a low-volume supply may wet only the upper part of the root system. Conversely, when used under low vine-water demand, volumes that are too high may result in only one application per day, thus increasing diurnal soil-moisture fluctuations. Here, daily vine transpiration varied between ~300 and 2300 mL depending on the severity of the water deficit, and the volume of the cylinder tank was set at 265 mL.
The system operates on the principle that the daily water supply to the vines can be replaced based on concurrent measurements of canopy transpiration. This makes water replenishment sensitive to large fluctuations in water use, which can occur depending on evaporative demand (e.g., cloudy days with low air vapor pressure deficit [VPD] versus clear days with high VPD) or as a result of new leaf development. For a water stress (WS) experiment, the system can be programmed to supply a group of plants with a specified fraction of the water delivered to a well-watered (WW) treatment. The replenishment coefficient that sets the fraction of water to be metered to WS can be adjusted according to prestress transpiration rates of each vine so that, for example, a more vigorous WS plant will receive more water than one with lower vigor.
Plant material and treatment layout
Data for system testing were derived from a water-stress experiment conducted in 2014 on twelve 2-year-old nonfruiting Vitis vinifera L. cv. Sangiovese grafted on SO4 rootstock and grown outdoors in 40-L pots. The pots were filled with a loam soil with 41% sand, 39% loam, 20% clay, pH 8.02, and organic matter content of 1.22%. Water content by volume, calculated after Saxton and Willey (2005) based on soil texture and organic matter, was 26.3% for maximum pot capacity and 12.9% for wilting point. Pots were painted white before the trial to limit radiation-induced overheating, and each vine was fertilized twice (one week before and two weeks after budbreak) with 5 g of Greenplant (15% N, 2.2% P, 20.7% K + 1.2% Mg + micronutrients) (Green Has Italia, Cuneo, Italy).
Four shoots per vine were allowed to grow from the two 2-node spurs retained at winter pruning. Shoot growth was directed upward along the catch wires to fill all available space while providing optimal light exposure and minimizing mutual shading. Twelve vines of vertically positioned shoots were arranged along a single, 35° NE–SW-oriented row and randomly assigned to a WW or WS treatment.
All vines were kept well watered until day of year (DOY) 174 (23 June) by supplying 1950 mL/vine daily, which corresponded to the actual mean canopy transpiration (Tc) measured by the whole-canopy system over the four days before beginning restricted irrigation. Starting on DOY 175, a progressive water deficit was imposed on half of the vines by programming the water-supply system to deliver 70% of WW Tc to each vine until DOY 183. From DOY 184 to DOY 196, the supply was reduced to 50% of WW Tc. Maximum stress (30% of WW Tc) was applied from DOY 197 to DOY 199; rewatering with 100% of Tc was performed on DOY 200 (19 July). During water stress, the pot surface in both treatments was covered with a plastic sheet to prevent infiltration of rainfall and to minimize losses due to soil evaporation.
Gas-exchange and vine measurements
Whole-canopy transpiration measurements were taken using the multichamber system described above. The chambers were set up on each vine and operated continuously (24 hr) from DOY 171 (20 June, 4 days before beginning the stress treatment) until DOY 209 (28 July, 9 days after rewatering). The flow rate fed to the polyethylene chambers (0.602 m3 ± 0.060) was set at 10 L/s and was kept constant throughout the measuring season; a complete volume air change occurred at intervals of ~60 sec. Tc (mmol H2O/sec) was calculated from flow rates and water vapor differentials after Long and Hallgren (1985). Upon dismantling the system on DOY 208, the vines were entirely defoliated and the surface of each blade was measured in the laboratory with a leaf-area meter (LI-3000A; LI-COR, Lincoln, NE).
Progression of water stress was monitored by measuring predawn leaf water potential (Ψpd) on DOY 169, 184, 188, 197, 200, and 209. Measurements were taken before sunrise on three leaves per vine using a Scholander pressure chamber (Model 3500; Soilmoisture Equipment Corp., Santa Barbara, CA). On DOY 184, 188, 197, and 200, the chambers were snipped for quick access to the foliage and then were immediately resealed with transparent tape.
Statistical analysis
The degree of variation around means was given as standard error (SE). Linear regression analysis was used when appropriate. The SyStat Software (San Jose, CA) was used.
Results and Discussion
There was a close linear relationship between vine transpiration and water supply for the 40 days of measurement data (Figure 3). Deviation from the 1:1 relationship was between 3 and 6%, and root mean square error was 72 mL for WW and 127 mL for WS. Such a close relationship confirms the reliability of the system for supplying vines with a volume of water that closely tracks measured transpiration. As the severity of the water stress increased from 100% to 70%, 50%, and 30% of WW, the actual fraction of water supplied to WS vines was 102% (average of prestress and rewatering), 67%, 46%, and 26%, respectively; these rates indicated that under water deficit, the system tended to deliver slightly less water than the defined threshold. This suggests that at the low daily transpiration rates recorded for WS plants (~300 to 1000 mL/vine), a slight reduction of the cylinder tank operating volume would be advisable. However, as shown in Figure 3, the defined threshold of 265 mL per watering event was well suited to the 400 to 2300-mL range of vine transpiration values. These metering rates are low compared to those measured in the field (Williams et al. 2003) because of the small canopy size of our vines, which had a final leaf area of 1.55 ± 0.107 m2 (WW) and 1.48 ± 0.077 m2 (WS). Specific transpiration rates calculated over the rewatering period when all vine canopies had reached full size were ~0.8 L/m2·day, a value close to those measured by Poni et al. (2014) and Palliotti et al. (2014) under comparable conditions.
Relationship between daily canopy transpiration measured by the chambers and daily vine water supply for the well-watered (•) or water stressed (○) treatments. Data are single daily values averaged from dawn to dusk during the chamber operating period (DOY 171–209). Linear equations: WW, y = 1.068x, R2 = 0.95; WS, y = 0.974, R2 = 0.94. Slopes of single regressions of WW and WS data did not differ according to the test of equality of slopes (p = 0.05). The dashed line indicates the 1:1 relationship.
Clearly, reliability of the system also depends upon accurate assessment of canopy transpiration, its driving factor. Close correlations have also been found between gravimetric daily vine water loss and canopy transpiration calculated by gas exchange on vines with final leaf area ranging from 3 to 14 m2 (Poni et al. 1999, 2014). Given that whole-canopy gas-exchange systems are sophisticated and specialized, the automated water-supply device easily fits with other methods, including gravimetric, heat-balance sap flow, and trunk diameter measurements.
Predawn leaf water potential (Ψpd) progressively decreased with the gradual reduction of the Tc fraction supplied and reached its lowest value (≅ −0.8 MPa) toward the end of the 30% Tc replenishment (data not shown). A steep decrease in Ψpd occurred when the water supply in WS was lowered from 50 to 30% of WW. One week after rewatering on DOY 200, Ψpd of previously stressed vines promptly recovered to the same level as that of the WW vines (−0.1 MPa).
The experimental period (DOY 171–209) had notably variable weather, with large fluctuations in average daily PAR and VPD (Figure 4A). Overall, stable weather with conditions conducive to high evaporative demand occurred only at the end of the 50% Tc deficit period and for 3 days during 30% Tc. The seasonal trends of canopy transpiration measured in WW and WS reflected these fluctuations, and average Tc during the prestress, 70, 50, 30%, and rewatering supply levels was 103, 74, 48, 28, and 93% of the Tc rates recorded in WW, respectively. Thus, despite day-to-day variability, the restriction in water use was fairly proportional to the severity of the imposed shortage, indicating that the system worked properly. The Ψpd value (−0.8 MPa) reached at the end of the stress cycle confirms that the system is suitable for short-term water deficit experiments in which progressive water stress is applied to vines.
(A) Seasonal trends of daily mean air vapor pressure deficit (VPD) and incoming photosynthetically active radiation (PAR); (B) whole-canopy transpiration (Tc) measured on Sangiovese grapevines that were well-watered (•) or water stressed (○). In panel B, the duration of each water deficit level is shown by the horizontal bar (left to right: 70, 50, and 30% of vine transpiration). Vertical bars indicate SE (n = 6). Arrow indicates rewatering.
To examine the accuracy of the system over a wider range of instantaneous Tc, diurnal trends in Tc are shown for DOY 184, the first day of 50% water deficit, during which there were clear-sky conditions and high evaporative demand (Figure 5A, B). Well-watered vines showed a bell-shaped Tc trend that closely tracked incoming PAR. There were 7.5 watering events that yielded a total of 1987 mL water supply versus the cumulative daily Tc value of 1897 mL. In WS, the Tc trend was slightly more variable throughout the day, during which 3.5 watering events provided 927 mL of total supplied water versus a cumulative Tc value of 975 mL.
(A) Diurnal trends (dawn to dusk) of air vapor pressure deficit (VPD) (⋄) and incoming photosynthetically active radiation (PAR) (•); (B) whole-canopy transpiration (Tc) measured on DOY 184 (first day of 50% water deficit in WS) in well-watered (•) or water stressed (○) Sangiovese grapevines. Vertical bars indicate SE (n = 6).
Conclusions
A new system designed to perform automated, unattended water-deficit experiments on potted plants was successfully tested. In addition to exempting operators from laborious and time-consuming manual irrigation, the system has the ability to calibrate water supply according to actual water use measured concurrently using a vine-enclosure system, and it can customize the water supply according to plant size and transpiration potential. If a whole-canopy gas-exchange apparatus cannot be put in place, the system is preconfigured to log signals from a set of scales that can measure gravimetric vine water loss. This equipment would be valuable for automating trials examining genotype response to drought, a subject for which data are needed given the effects of global warming on grapevine water relations.
Footnotes
- Received September 2014.
- Revision received November 2014.
- Accepted November 2014.
- ©2015 by the American Society for Enology and Viticulture
Vol 66 Issue 2
