Location and grapevine growing conditions
All work described in paper was undertaken at the National Wine and Grape Industry Centre, Charles Sturt University in Wagga Wagga, New South Wales, Australia (35.06 °S, 147.36 °E), and utilized an existing population of large potted Shiraz grapevines (Vitis vinifera L., clone PT23) that were located in a bird-proof wire mesh enclosure. The vines were planted on own roots in 2008, and trained to single spur-pruned bilateral cordon on a fixed fruiting wire 80 cm from the ground. A second fixed wire at 120 cm above the ground provided support for the canopy. The pots were 52 L in volume (500 mm wide × 380 mm high), and arranged with 1 m spacing in rows of 5 vines in an E-W alignment with 3 m between rows. Pots were filled with a commercial bulk-composted potting mix, and fertilized manually with diluted liquid fertilizer (Megamix Plus®, RUTEC, Tamworth) and additional applications of magnesium sulphate and gypsum. Irrigation was provided via a 4 L h− 1 pressure compensated emitter installed on each side of the trunk to provide water to two sides of the root system. Irrigation was controlled by an automated timer, with the schedule and duration of three to four daily irrigations adjusted manually according to seasonal water requirements. Shoot numbers ranged from approximately 30 to 38 per vine, providing a sprawl canopy with similar structure and density to field-grown grapevines.
Gas exchange chambers
Six gas exchange chambers were constructed with transparent acrylic tops that incorporated a lightweight plastic and aluminum frame (Cubelok, Capral Limited, Parramatta, Australia), and a base made from polyethylene plastic panels and a galvanized steel frame. The chamber top and base were both designed in two halves that could be installed from each side of the trellis with minimal impact on the shape of the canopy (Fig. 1). The chamber top enclosed a volume of 1.2 m3 and was constructed from flat sheets of 3 mm thick acrylic on the sides parallel to the vine row, and 4 mm thick on the two sides perpendicular to the vine row. Under clear sky light transmission between 400 and 700 nm was 95%. An 85 mm gap between the top of the 3 mm sheets allowed outgoing air to exit the chamber, with an overlapping ridge cap supported 24 mm above the chamber sides to minimize ambient air intrusion during windy conditions. When installed, the maximum outside dimensions of the assembled chambers were 1620 mm perpendicular to the row, 1000 mm parallel to the row and 2000 mm high.
The external panels of the chamber base were constructed from 19 mm plastic boards made from a combination of high density polyethylene and polypropylene (UniboardECO, Dotmar Plastic Solutions, Sydney, Australia). On the top of each half of the chamber base, three 24 VDC blower fans (San Ace B97, Model 9BMB24P2G01, Sanyo Denki Co. Ltd., Tokyo, Japan) were mounted against a 64 mm diameter hole, and angled to mix and distribute air in a horizontal plane across the base of the chamber. An air-tight rectangular box made from a combination of 12 mm and 6 mm sheets of the same plastic was fixed to the underside of each chamber base half to enclose the air intake of the three fans. A port on the underside of each of the boxes was then connected to a common air intake made of 105 mm internal diameter (nominal size 100 mm) polyvinyl chloride (PVC) pipe and compatible fittings. This air intake could either be attached to a 3 m high chimney for ambient air, or to the outlet of the CO2 scrubber described in the next section. The air inside the chamber was then displaced vertically through the canopy and exited through the vent at the top of the chambers. Smoke tests (DATAX, BJÖRNAX AB, Nora, Sweden) were used to assess the uniformity of air flow through the chambers. Removable panels on the lower face of the acrylic chamber tops allowed access to fruit on both side of the vine for berry sampling. Adhesive tape was used to seal the gaps created by the trellis wires between the chamber halves, and adhesive foam weather strips were used between the panels of the chamber top and base.
The speed of the 6 fans was controlled by a pulse width modulation (PWM) signal from an Arduino® UNO compatible micro-controller mounted inside the intake box in one side of the chamber base. During the day, the PWM signal was programmed to a duty cycle of 25%, and an input signal from a manually operated switch was used to slow the PWM to a duty cycle of 10% at night. For the day speed, this equated to a flow rate of approximately 2.6 m3 per minute, or 2.2 chamber volumes per minute, and at night, 1.6 m3 per minute or 1.3 chamber volumes per minute.
CO2 scrubbers
A portable CO2 scrubber was designed so that it could be attached to a chamber, in place of the ambient air intake chimney and reduce the CO2 concentration of supply air by approximately 200 ppm compared to ambient air (Fig. 2). Three scrubbers were constructed from a combination of 15 mm plywood and 100 mm nominal diameter (105 mm ID) PVC pipe and fittings that completely scrubbed all CO2 from incoming air and then mixed this back in with ambient air to obtain the desired CO2 concentration. Each scrubber contained a total of 27 kg of soda-lime (Sofnolime®, Molecular Products Limited, Harlow, UK), divided equally across eight tubes of 100 mm diameter and 480 mm depth that were connected in parallel. Air was pulled through the soda-lime beds, and then an open-cell foam filter under vacuum using a single 24 VDC centrifugal fan (SanAce C175, Model 9TG24P0G01, Sanyo Denki Co. Ltd). A manually operated valve was used to allow some air from an ambient inlet chimney at 3 m to by-pass the scrubbing beds on the upstream side of the fan. Fan speed was controlled by a PWM signal in the same manner as those in the chamber bases, but a potentiometer was used instead of a switch to provide variable control over the fan speed. Through a combination of ambient air bypass and fan speed control, the design allowed variable control over the airflow rate and CO2 concentration range. To minimize temperature differences between the ambient and low CO2 treatments, scrubbers were covered with reflective foil to reduce solar heating, and ice placed on top of scrubbers and replaced during the day as required.
To test the full operating range of the scrubbers, a series of measurements were made between the minimum and maximum fan speeds with the bypass valve closed progressively until the maximum amount of air possible was forced through the scrubbing beds. A portable infrared gas analyser (Li400XT, LiCOR Biosciences, Lincoln, Nebraska, USA) was used to measure the concentration of outgoing CO2, and the process was repeated with a second scrubber.
Gas exchange measurements
When assembled and running as a complete system with six chambers, the CO2 and H2O concentrations of incoming and outgoing air were recorded on a 30 min cycle, providing 48 measurement points in a 24 h period. During the measurement period for each chamber, air was sampled simultaneously from the chamber inlet and outlet at a rate of approximately 2 L min− 1 via 10 m of 6 mm external diameter polyethylene tubing using a pair of diaphragm pumps (SP550 EC-BLa, Schwarzer Precision GmbH + Co. KG, Essen, Germany) connected in parallel on each line. As shown in Fig. 1b, the inlet air sample was drawn from the PVC air intake tube on the downstream side of the intake fans. The outlet sample was taken from two points within the chamber at 20 cm below the air outlet and 30 cm apart to provide an average air sample. For both inlet and outlet lines, the two pumps were connected in parallel across the outlets of six individually switchable solenoid valves (V2 miniature pneumatic solenoid valve, Parker Hannifin Corp., Cleveland, Ohio, USA) to avoid dead-air volumes and maximize flushing between cycles. These pumps then delivered the air to a 50 mL buffer container that was then sub-sampled via a switchable solenoid valve by a single pump (SP550 EC-BLa) at a rate of approximately 800 mL min− 1 for subsequent CO2 and H2O measurements. Excess air from the 50 mL buffer volumes vented to atmosphere.
Carbon dioxide and H2O vapour concentrations were measured at 5 s intervals using an infrared gas analyser (Li840A, LiCOR Biosciences, Lincoln, Nebraska, USA) set to 50 °C and the average value recorded at 30 s intervals with an external data logger (CR1000, Campbell Scientific, Logan, Utah, USA). For the inlet air, these measurements were made for 2 min, and for the outlet air sample, 3 min. A relay controller (SDM-CD16AC, Campbell Scientific) was used to drive the progressive switching of the six pairs of solenoid valves at 5 min intervals, and within that period, to switch the single sub-sampling solenoid from the chamber inlet to outlet after 2 min. Data points recorded during transition periods were removed during subsequent processing, leaving two inlet measurements and four outlet measurements to average for subsequent gas exchange calculations.
Chamber and external environmental parameters were recorded with a second data logger and multiplexer (CR1000 and AM2 5 T, Campbell Scientific, Logan, Utah). Air inlet temperature was measured with T-type thermocouples (24 AWG) at a single point inside the inlet tube under the chamber base, and at two outlet points on each side of the canopy (wired in parallel for average value) from under white plastic radiation shields attached to the fixed foliage wire. Air velocity was recorded in the centre of the inlet tube with a hot-film element sensor (EE576 or EE671, E + E Elektronik, Engerwitzdorf, Austria). Photosynthetically active radiation outside the chamber was recorded with Quantum sensor (LI-190R, LI-COR Biosciences), and temperature and relative humidity (HMP-50, Vaisala, Helsinki, Finland) inside the bird-proof enclosure were recorded at the same intervals as the chamber measurements.
Sub-ambient CO2 experiment
At the first visual indication of berry colour change in late December 2013 the chambers were installed on the canopies of six grapevines selected from the larger population in the bird-proof enclosure. For a period of 10 days from December 26 three chambers were supplied with ambient air, and three chambers connected to the CO2 scrubbers and supplied with air at a target of 200 ppm below ambient during daylight hours. While no pre-treatment comparison was able to be made due to an earlier expected onset of véraison, from January 5 all chambers were run at ambient CO2 for an additional 4 days to compare them in the post-treatment period. When the chambers were removed, the length of every leaf from each vine was recorded. Using a regression established between leaf area, measured (LI-3100C, LI-COR Biosciences), and length, with leaves sampled destructively from vines that had not been used in the chambers, the total leaf area of each vine was then calculated.
Berry sampling and analysis
Berry samples collected on the first day of the 10 day scrubbing period, and then on the morning of the eleventh day, were used to assess the effect of the sub-ambient treatment on berry sugar accumulation. Accessed via removable panels on the side of the chambers, 20 random berries were collected from each side of the vine to provide a 40 berry sample. The samples were immediately weighed, separated into skin and seeds over ice and then frozen in liquid nitrogen. The pulp was manually homogenised while cooled over ice, and juice separated and 1 mL sub-samples frozen in liquid nitrogen. Remaining juice was used to determine soluble solids. All samples were stored at − 80 °C prior to subsequent analysis. Juice samples were thawed, vortexed to mix and filtered to 0.22 μm and fructose and glucose concentrations determined by HPLC-RI on a 300 mm × 7.8 mm Aminex HPXM87H ion exclusion column (BioMRad Laboratories, Berkeley, USA) using the method of Frayne [42].
To estimate the total amount of sugar accumulated by each treatment in the 10 days between the first and last berry sampling, sugar content per berry was calculated using the weight to volume relationship for Shiraz described by Gray and Coombe [43]. Fruit from all vines was harvested and weighed on February 12, and using harvest sampling berry weights, the total number of berries per vine was calculated. Allowing for the removal of berries at each sampling date, an estimate of total sugar concentration per vine could be made based on total berry volume and juice sugar concentration.
Gas exchange calculations
Volumetric air flow was calculated from the inlet air velocity measurements for the 5 min period of gas exchange for each chamber and converted to molar air flow (Eq. 1) as per Long and Hallgren [44], where ue = molar flow of air (mol s− 1), fv = volumetric air flow (cm3 min− 1), 22.4 = volume (dm3) of one mole of air at standard temperature and pressure of 273.15 K and 101.3 kPa respectively, T = air temperature (°C), and P = atmospheric pressure (kPa). The inlet thermocouple for each chamber was used for the air temperature, while atmospheric pressure was obtained from the Australian Bureau of Meteorology Wagga Wagga airport weather station (35.16 °S, 147.46 °E), situated approximately 15 km south east of the experiment location.
$$ {u}_e=\frac{f_v}{1000}.\frac{1}{22.4}.\frac{273.15}{\left(273.15+T\right)}.\frac{P}{101.3}.\frac{1}{60}\kern5.25em (1) $$
Average canopy transpiration (E; Eq. 2), photosynthesis (A; Eq. 3) and rates were calculated using the following equations [45], were s = leaf area (m2), we and wo = inlet and outlet vapor concentration (mol mol− 1) respectively, Δ = difference between inlet and outlet CO2 concentrations (mol mol− 1) and ce = inlet CO2 concentration.
$$ \kern0.75em E=\frac{u_e}{s}.\frac{\left({w}_o-{w}_e\right)}{\left(1-{w}_o\right)}\kern5.5em (2) $$
$$ \kern0.5em A=\frac{u_e}{s}.\frac{\left(1-{w}_e\right)}{\left(1-{w}_o\right)}.\varDelta -E.{C}_e\kern1.5em (3) $$
Outlier values, screened based on the ratio of the inlet and outlet CO2 concentrations and to allow daily sums to be calculated, were replaced with the average of two adjacent readings for that chamber. An outlier or potentially incorrect reading was defined as outlet concentration that was more than 10% lower than the inlet during the day, or an outlet concentration more than 2% higher at night. Only 25 points from 4032 measurement points over 14 days required replacement with this method, and in most cases these could be explained by berry sampling, or the connection or disconnection of the CO2 scrubbers inadvertently coinciding with the 5 min monitoring period for each chamber.
Statistical analysis
For the gas exchange data, the three chambers in each treatment were blocked according to the leaf area of each vine and analysed as a split plot experiment with treatment as a main plot and time as the subplot factor using Genstat v18 (VSNI, Hemel Hempstead, England, UK). Mean chamber temperature inside the chamber was used as a covariate for comparison of night respiration rates. The 10 day period with the ambient and sub-ambient treatments was analysed separately from the 4 day period. Treatment means for single measurements were compared using Student’s T-test.