Characterization of the Transpiration of a Vineyard under Different Irrigation Strategies Using Sap Flow Sensors

Abstract

Lysimeters are the reference method for determining ET_c, but they are expensive and complex, which limits their use. The first objective of this work was to adjust and evaluate the robustness of sap flow sensors in order to determine the transpiration of a vineyard and, together with an evaporation model, to calculate the ET_c of the vineyard. For this purpose, we compared water consumption data obtained from a vineyard weighing lysimeter (ET_cLys) with the sum of transpiration obtained from sap flow sensors (T_SF) and evaporation estimated empirically over four years (2012, 2013, 2014 and 2015). The second objective was to obtain the relationship between the vegetative growth and transpiration of the vines with different water availability (irrigation and rainfed treatments), as an alternative method for estimating vine water needs adjusted to their real development. The third and last objective was to evaluate the transpiration response of the vines when subjected to water stress. We carried out the work in an experimental vineyard which has a well-established weighing lysimeter. As a result, a good match was obtained between vine sap flow and transpiration (R² = 0.85) as well as a good relationship between vegetative growth and vine transpiration (FiPAR: R²_Irrigation = 0.34. R²_Rainfed = 0.54; LAI: R²_Irrigation = 0.68. R²_Rainfed = 0.53).

1. Introduction

The quantification of vineyard water needs constitutes a critical step in the adoption of successful irrigation strategies. Irrigation management is the major and most controllable determinant of grape quality in arid areas [1]. Introducing the water needs of the crop into a water balance allows the volume of irrigation water to be adjusted to the crop requirements [2]. For practical purposes, crop water needs are assimilated to the ET_c, which includes both plant transpiration and soil evaporation. They are obtained under standard conditions, corresponding to a disease-free and well-fertilized crop growing in a large field with optimal water content and reaching the maximum production for these agro-climatic conditions [3,4]. ET_c has an important seasonal (from budbreak to leaf fall) and interannual variation, as it changes according to weather conditions, cropping practices and irrigation scheduling and strategy [3,5,6,7]. However, both soil and climate are determinants of the agronomic behavior of grapevines [8], known as the terroir effect [9], and to which varietal aspects must be added.

Among the methods used in the determination of ET_c, the use of weighing lysimeters is the most direct methodology [10] and the one that offers the highest accuracy and time resolution [11]. It is considered the main reference method for the adjustment of other methodologies. There are very few weighing lysimeters for the study of grapevine; two of them are in Spain and one of these [6] was used in this study. This infrastructure is expensive and immovable. These limitations make it difficult to evaluate the influence that different agroecological aspects can have on vineyard water consumption with this system. However, other methods for determining ET_c which, although a priori may be more imprecise, are more affordable and versatile. Among these are the water balance, widely used in woody crops. [12,13,14,15,16], Bowen ratio techniques [17,18], which are based on the eddy correlation [19,20,21] and the surface renewal method [22]. Another possible approach is to calculate or measure the two components of ET_c separately [23].

Several techniques are currently available to measure or estimate water use in woody plants. Those that use basic measurements of water movement through the plant are based on the properties of heat dissipation through the flow of sap through the stem [24] and offer several advantages: direct measurement of water through the plant (or smaller sections of the pathway); continuous, long-term monitoring without interruption of the plant canopy or root environment; and automation of transpiration rate calculation. The three types available are: heat dissipation, heat balance, and heat pulse [23,25,26]. The sap flow methods allow estimation of water losses in the plant by transpiration (T), measuring the speed at which the sap flow moves through the xylem. Discontinuous heat pulse methods are the most suitable for monitoring water in the tree under field conditions since they have lower energy requirements [27]. Of the latter, the heat pulse compensation method is the most appropriate for low and moderate flows [28], being a reliable technique for the continuous measurement of sap velocity in woody crops [27,29,30,31,32]. Some authors have suggested that sap flow measurement can be helpful for irrigation management in vineyards [19,33,34,35], although there is some uncertainty given the different approaches used as to the accuracy with which actual plant water consumption is determined. In these cases, it is recommended to perform direct calibrations which, in field conditions, can be performed with lysimeters [24,36], with the log cutting method [30], with gas exchange measurement chambers in complete plants [37,38,39], with a water balance under conditions in which evaporation is eliminated or quantified separately, or with micrometeorological methods [40].

Another important application of sap flow measurements is to gain insight into the dynamics of transpiration in the plant [23,41], especially in deficit irrigation studies where the objective is to quantify the reduction in net transpiration related to the water stress endured by the plants [42].

Soil evaporation (E) can be a quantitatively important component of crop water requirements, varying substantially depending on the frequency and method of irrigation, the soil maintenance system, and crop cover. Therefore, determining it independently of T is of interest when proposing more efficient systems of water use by crops [43]. In this sense, the UN’s Food and Agriculture Organization (FAO) presents a dual crop coefficient approach, in which the effects of T and E on ET_c are determined separately using two coefficients: the basal crop coefficient (K_cb) to describe T and the evaporation coefficient (K_e) to describe E [3]. Different methodologies can be used to quantify E: empirical measurements or adapted models [44,45], using microlysimeters [46] or by covering cycles in a weighing lysimeter [5].

In most Spanish and worldwide wine-growing areas, irrigation is a crucial element to improve the water status of the vines, impacting on productivity and the characteristics of the grapes. The effect of environment on water status through stomatal conductance has been well interpreted in this species [47,48], as well as leaf water potential [49,50,51], stem water potential [52] and transpiration under water stress conditions [48].

Vegetative growth is one of the physiological processes most sensitive to water deficit in plants [53,54]. The increase in water available to the vineyard is usually accompanied by greater vigor, more vegetative development, and a higher leaf area [55]. A close relationship has been obtained between the water consumption and vegetative development of Tempranillo vines [6], as well as for cv. Thomson Seedless [7]. Relationships of this type have also been determined with the leaf area index (LAI) for clementines [56], and apple and pear trees [57], and with the radiation interception fraction for peach [2] and orange trees [58].

Although several studies have been published on the determination of vine water requirements using different methodologies [6,18,19,20,35,59], few studies have separated the two components of vineyard ET_c (E and T). The first objective of this study was to adjust and evaluate sap flow sensors with the help of a weighing lysimeter, determine vineyard transpiration and then, together with an evaporation model, the ET_c. The second objective was to determine the relationship between vegetative development and T of the vines with different water availability (irrigated and dry) as an alternative method for the evaluation of water consumption and, finally, to evaluate the effect of water stress on vine transpiration.

2. Materials and Methods

2.1. Location, Description of the Vineyard, and Weather Conditions

The trial was carried out during the 2012, 2013, 2014, and 2015 agricultural seasons in an experimental cv. Tempranillo vineyard of 1.7 ha located at “Finca La Orden” in Badajoz (38°51′ N, 6°40′ W, altitude 188 m), Spain. Planted in 2001, the vine was grafted on Richter 110 rootstock, trained on a trellis and formed in a double Royat cordon, leaving six spurs per plant and two buds per spur. The planting frame is 2.5 × 1.2 m, with a trunk height of 0.60 m and a trellis height of 1.50 m, in an east-west row orientation. In addition to winter pruning, the number of shoots per vine was adjusted in the spring (12 shoots per vine). We also carried out a clipping to contain the vegetation on the trellis.

The soil is alluvial, with a loam to sandy loam texture, slightly acidic, and lacking in organic matter. Soil depth is greater than 2.5 m and with low stone content. The area has a Mediterranean climate with a mild Atlantic influence, dry and hot summers, with high daily radiation and evaporative demand. We obtained agrometeorological data from a station close to the vineyard (100 m) with the characteristics described in [60]. We calculated the vapor pressure deficit (VPD) with the maximum and minimum temperatures and relative humidity [3].

2.2. Experimental Design

In the experimental vineyard, we established two treatments (Irrigated and Rainfed) with a randomized block design with four replications per treatment. Each experimental plot consisted of 6 rows with 18 vines each row (108 plants per experimental plot), with the two outer rows as borders, and the four remaining rows under study. The lysimeter is located in the irrigation plot of block III, in a central position in the vineyard. In this block, we selected four vines per treatment for the installation of sap flow sensors, with two of these being the vines of the weighing lysimeter.

2.3. Irrigation Scheduling

The vines were irrigated daily with a drip irrigation system with two self-compensating emitters per vine (4 l h⁻¹ for each dripper). Irrigation started when plants reached a midday stem water potential (Ψ_s^md) threshold value of −0.6 MPa in order to avoid water stress situations [61,62], and was maintained until the first half of October. At each irrigation, we applied a volume of water equal to the water consumed (100% ET_c), quantified with a weighing lysimeter, since the previous irrigation. The size of the weighing lysimeter was 2.67 × 2.25 m and 1.5 m deep. A detailed description of the weighing lysimeter can be found in [6]. We obtained the reference evapotranspiration (ET_o) from an agrometeorological station located 100 m from the experimental vineyard and run by the Irrigation Advisory Network of Extremadura (Spanish initials REDAREX; www.redarexplus.gobex.es/RedarexPlus, (daily access each year of study from April 1 to October 31) (accessed on 3 October 2021)), using the FAO-56 Penman-Monteith method [3].

2.4. Vegetative Growth, Water Status and Production

We characterized vegetative growth by determining the fraction of intercepted photosynthetically active radiation (FiPAR) and LAI. FiPAR, from two labelled vines per elementary plot, was measured at noon with a linear ceptometer (80 cm probe length; Accupar Linear PAR LP-80; Decagon Devices, Pullman, WA, USA). Measurements were taken every two weeks at solar noon on clear days from flowering to the onset of senescence, following the methodology described in [63]. We obtained FiPAR values in rainfed areas for 2012 and 2013 from the empirical relationship between LAI and FiPAR obtained during the four years under study, using the following linear equation:

(LAI = 9.1197 FiPAR − 0.0392; R² = 0.75; n = 32)

(1)

LAI was measured periodically along the crop cycle every year, except in 2015 when only two measurements were made, with an LAI-2200 canopy analyzer (Li-Cor Inc. Lincoln, NE, USA). A baseline measurement was taken above the vegetation, followed immediately by measuring eight points at ground level in the center of eight quadrants into which the planting frame had been divided (Scheme 1). We used the LAI measurements obtained to determine the leaf area (LA) of the vine, using a previously established empirical relationship in the same experimental vineyard [64]:

(LA = 3.1076 LAI + 0.0198; R² = 0.96)

(2)

We measured the Ψ_s^md on leaves placed at least 2 h previously in plasticized and aluminum-coated bags, from the north side and the lower part of the canopy close to the cordon [65]. Measurements were made with a Scholander pressure chamber (Soil Moisture Corp., Model 3500, Santa Barbara, CA, USA) on two leaves per experimental plot (1 per vine) at weekly intervals.

Harvesting was performed in each plot when the concentration of soluble solids was close to 23 °Brix (mean value of each treatment). We weighed and counted all the bunches of 10 previously selected plants from each experimental plot at harvest time, determining the average yield for each plot.

2.5. Water Consumption

We measured the ET_c directly in the weighing lysimeter (Figure 1), where we reproduced the growing conditions of two vines in the experimental vineyard. We calculated this ET_c from the difference in weight in the lysimeter between two consecutive measurements, adjusted to the equivalent area of an individual vine [6]. In 2013, it was impossible to obtain the complete evolution of ET_c due to technical problems with the lysimeter during that season.

The T of the vines was quantified using sap flow sensors developed and assembled by the IAS-CSIC (Córdoba, Spain). The data was processed using the Prosap 3.1.3. program and using the heat pulse compensation (HPC) method plus the average gradient technique [28], as indicated for sap velocities lower than 12 cm h⁻¹. Eight sap flow sensors (one per vine) were installed in an area of the trunk of the vine free of irregularities at a height of 20 cm from the ground, with prior debarking of the chosen area. We monitored four plants in the experimental irrigation plot in block III, two of them on the vines located in the weighing lysimeter. Furthermore, we installed four sensors on vines in the rainfed treatment in block III.

Each sensor consists of three stainless steel tubes of 2 mm diameter, identified by colors; two are thermocouples, and the third is the heater. The thermocouples are located 10 mm above and 5 mm below the heater (Figure 2). The thermocouples, in turn, consist of 4 measuring segments situated at 5, 15, 25, and 35 mm from the end inserted into the cambium. The heater contains a continuous heating wire and is provided with a precision regulator to stabilize the energy released into the xylem, making it independent of possible voltage fluctuations of the batteries. The system is fitted with a CR1000 datalogger (Campbell Scientific Inc., Logan, UT, USA), programmed to run a measurement cycle every 15 min. Heat pulse velocity values due to the wound effect [32] were corrected considering 2.6 mm wound diameter according to [66], subsequently converted to sap velocity and integrated first along the trunk radius (using the radial velocity curve given by the probe) and then around the azimuth angle [67].

Each month, we determined the diameter of each of the vines at the height of the probe insertion using digital calipers, implementing these values in cm in one of the processing files named DIAM, with the values obtained increased by multiplying them by a factor α = 3, based on an adjustment made during a covering cycle in the weighing lysimeter in 2014. The obtained adjustment of the transpiration responded to the linear equation:

(y =2.6273x − 4.2398; R² = 0.95; n = 6)

(3)

The values obtained were processed using the Prosap 3.1.3 program and averaged to then obtain, from sap flow values in (l h⁻¹), values of daily transpiration (T_SF) in mm day⁻¹, or in l vine⁻¹ day⁻¹.

2.6. Soil Evaporation

Soil evaporation was obtained directly from the lysimeter by several covering cycles from May to October during the 2010 season on rain-free days. This covering was done with black plastic to avoid water evaporation from the soil surface, obtain T for those days, and calculate the T/ET_o ratio. E was determined using K_e:

K_e = (ET_c/ET_o) − (T/ET_o) = E/ET_o

(4)

We obtained ET_c as the average value of the two days before and after the day the lysimeter was covered, as described in [68], and subsequently established a correlation between the K_e and the ET_c of the lysimeter responding to the linear equation:

(K_e = 0.0606 ET_c − 0.0689; R² = 0.65; n = 7)

(5)

Thus, crop evapotranspiration was determined by sap flow (ET_cSF) as the sum of the T_SF and E components.

2.7. Statistical Analysis

The yield results were subjected to a one-factor analysis of variance (ANOVA) for each year, using SPSS version 20 (IBM, Armonk, NY, USA). We performed the correlations using linear regression analysis.

The procedure followed in this study was divided into several stages (Data Collection, Adjustment and Validation), as shown in Scheme 2.

3. Results

3.1. Weather Conditions

The average annual rainfall was 441 mm, with 2013 being the wettest year with 585 mm and 2015 the driest with 312 mm. Despite these yearly values, the least rainy year during the vegetative period (April–September) was 2013, with 64 mm. However, rainfall was abundant in March, before budbreak (181 mm), and in October (135 mm), close to leaf fall. The years 2012 and 2014 were the wettest years between budbreak and leaf fall, with the same rainfall of 136 mm (Figure 3).

As shown in Figure 3, the seasonal pattern of solar radiation (SR) was similar in the four years. The month with the highest SR was July, with average values ranging from 29.5 MJ m⁻² day⁻¹ in 2012 to 27.7 MJ m⁻² day⁻¹ in 2014. However, the VPD had the maximum values shifted towards August, except in 2015 when they were reached in July. These maximum values ranged between 3.5 and 3.1 kPa day⁻¹. For the vegetative period of the vines, the minimum values were in April, ranging between 0.7 and 1.0 kPa day⁻¹ (Figure 3). Mean daily temperature (Tm) had a similar pattern to SR, with the highest values ranging from 30.5 °C in 2015 to 28.9 °C in 2014 (Figure 3).

3.2. Vegetative Growth, Water Status, and Yield

The seasonal evolution of LAI, FiPAR and Ψ_s^md for the two treatments and the four years of study is shown in Figure 4.

The LAI of the irrigation treatment increased after budbreak until clipping, which caused a sharp decrease to recover the increasing trend until veraison, after which it decreased in 2012 and 2014 (Figure 4A–C). The highest LAI value was observed in 2012 in the irrigation treatment before veraison. Around harvest, the LAI of this treatment reached values close to 3.70. In the rainfed treatment, the maximum was 3.12 in 2013, with values at harvest between 1.76 and 2.26 (Figure 4A–C).

Like the LAI, FiPAR increased from budbreak until a sharp decline after clipping, with a subsequent recovery until the last measurements made in 2012 and 2013. In 2014 and 2015, there was a slight recovery of FiPAR post-clipping followed by a stabilization and final drop with leaf senescence. The maximum FiPAR value was reached in the irrigation treatment in 2012 with 0.53 at the end of the season (Figure 4E). In the rainfed treatment, the FiPAR increase was slower than in the irrigation treatment from early stages in 2012 and 2015, reaching stable values before the clipping which were maintained until leaf senescence at the end of the cycle (Figure 4E,H). In 2013 and 2014, the differences between treatments were established later, at day of year (DOY) 192 and 178 in 2013 and 2014, respectively, being lower than in previous years from this point onwards. Before the clipping, the maximum FiPAR rainfed treatment values ranged between 0.20 and 0.37 in the years under study, with this treatment reaching values of around 0.25 at the end of the measurement period (Figure 4E–H).

The most noteworthy differences between treatments in terms of vegetative development correspond to 2012, both in LAI and FiPAR, while the least differences were observed in 2013.

Regarding water status, the irrigation treatment vines maintained constant Ψ_s^md values throughout the cycle, around −0.6 MPa, except in 2014 when we observed a slight initial decrease, stabilizing after 150 DOY at around −0.7 MPa (Figure 4I–L). Although the rainfed treatment achieved significantly lower Ψ_s^md values in all years than the irrigation treatment, there were notable differences between years, with minimum values ranging from −1.0 MPa in 2013 to −1.8 MPa in 2015. We observed the most significant differences between treatments in the 2012 and 2015 seasons, which began to separate at around DOY 150. The minimum values of the rainfed treatment were reached around veraison when we observed the greatest differences with the irrigation treatment. A subsequent recovery brought the Ψ_s^md of both treatments closer but without reaching the same level (Figure 4I,L). In 2013 and 2014, there were no differences between treatments up to DOY 191 and DOY 178, respectively. In 2013, there were constant differences between the irrigated and rainfed treatments from veraison onwards. In 2014, the Ψ_s^md recovery of the rainfed treatment at the end of the cycle again equaled the potentials of both treatments (Figure 4J,K).

Yields in the irrigation treatment ranged between 10,864 and 28,689 kg ha⁻¹ in 2014 and 2013, respectively, and in the rainfed treatment between 8243 kg ha⁻¹ in 2015 and 17,300 kg ha⁻¹ in 2013 (

Table 1. Yield in kg ha⁻¹ for the different irrigation managements during the years of study 2012, 2013, 2014, and 2015.

Treatment	Year
	2012	2013	2014	2015
Irrigation	13,567	28,689	10,864	13,744
Rainfed	8753	17,300	9830	8243
Significance	***	***	ns	***

*** indicates that the irrigation factor was significant at p < 0.001; ns, not significant.

). The year 2013 was the most productive in both treatments. In 2014, both treatments had similar yield and this was the only year in which no significant effect of irrigation management was found (Table 1).

3.3. Evaporation, Transpiration, and Evapotranspiration

The seasonal ET_o was different in every year, with the highest values obtained in 2012, mainly up to veraison (Figure 5A–D). In all years there was a similar seasonal trend between ET_o, ET_cLys, and T_SF, but in the last two years the separation between ET_o and ET_cLys and T_SF was greater than in the other years. The seasonal evolution of T_SF in the lysimeter plants, under non-water-limiting conditions, had the same tendency as ET_cLys, but with lower values (Figure 5A–D). Both processes had close values prior to the first irrigation, mainly in 2015 until DOY 133 (Figure 5D). Both parameters decreased with the clippings and increased until veraison, with a subsequent plateau phase. The decreasing trend started before ripening and continued to leaf fall (Figure 5A–D). T_SF was noticeably lower in rainfed plants than irrigated ones, with values between 0 and 1 mm/day during most of the season in 2013 and 2014, and higher values in 2012 and 2015. Differences between treatments started early in the season around the beginning of flowering (DOY 130), and even before the first irrigation in 2013 and 2014 (Figure 5A–D) and without having yet produced a differentiation in Ψ_s^md values due to irrigation in these years (Figure 4I–L).

The cumulative T_SF (Figure 5E–H) for the irrigation treatment increased after budbreak in all years and remained with the same slope until harvest. From this point on, in 2012 and 2015 the trend was maintained, while in 2013 and 2014 a change in the slope can be observed with a drop in T_SF. In the rainfed treatment, the slope was lower and with no drastic changes in the trend along the season. Once again, the years 2012 and 2015 differed from 2013 and 2014, with lower cumulative transpiration in the two central trial years (Figure 5E–H).

Table 2. Monthly average daily values, from April to October, of the reference evapotranspiration (ET_o PM), crop evapotranspiration in the lysimeter (ET_cLys), crop evapotranspiration obtained with the sap flow sensors (ET_cSF) in the irrigation treatment, crop coefficient obtained with the lysimeter (K_cLys), crop coefficient obtained with T_SF + E (K_cSF), evaporation (E), transpiration obtained with the sap flow sensors in the irrigation treatment (T_SFirrigation) and transpiration obtained with the sap flow sensors in the rainfed treatment (T_SFrainfed) during the years 2012, 2013, 2014 and 2015.

Months	2012
	ETo PM	ETcLys	ETcSF	KcLys	KcSF	E	TSFirrigation	TSFrainfed
	(mm day−1)	(mm day−1)	(mm day−1)			(mm day−1)	(mm day−1)	(mm day−1)
April	3.47	0.47	--	0.14	--	0.00	---	---
May	5.54	3.11	--	0.56	--	0.78	---	---
June	6.79	5.87	---	0.86	--	1.95	---	2.42
July	7.26	6.23	7.41	0.86	1.02	2.25	5.16	1.65
August	6.29	6.96	7.53	1.11	1.20	2.29	5.24	1.28
September	4.23	5.31	5.69	1.26	1.35	1.29	4.40	1.15
October	2.49	3.03	2.88	1.22	1.16	0.37	2.51	0.74
Months	2013
April	3.55	0.89	1.22	0.25	0.34	0.00	1.21	0.66
May	4.95	2.06	2.41	0.42	0.49	0.33	1.85	0.91
June	5.88	4.83	6.24	0.82	1.06	1.40	4.67	0.70
July	6.37	--	--	--	--	--	4.80	0.17
August	5.72	--	--	--	--	--	4.07	0.09
September	4.10	--	--	--	--	--	3.42	0.18
October	2.14	--	--	--	--	--	1.34	0.14
Months	2014
April	3.42	2.07	--	0.61	--	0.25	---	---
May	5.49	2.80	2.40	0.51	0.44	0.58	1.82	0.66
June	5.85	3.77	3.69	0.64	0.63	0.99	2.70	0.54
July	6.20	5.25	4.89	0.85	0.79	1.63	3.26	0.37
August	6.10	4.21	4.29	0.69	0.70	1.15	3.14	0.14
September	3.60	1.56	0.61	0.43	0.17	0.13	0.48	0.28
October	2.19	1.49	0.28	0.68	0.13	0.05	0.23	0.45
Months	2015
April	3.78	1.19	1.35	0.31	0.36	0.06	1.29	0.48
May	5.85	3.34	3.46	0.57	0.59	0.83	2.63	1.07
June	6.26	4.61	4.78	0.74	0.76	1.38	3.40	0.95
July	7.05	4.56	4.11	0.65	0.58	1.49	2.62	0.70
August	5.76	4.34	3.17	0.75	0.55	1.14	2.03	1.25
September	4.07	3.26	2.37	0.80	0.58	0.59	1.78	0.75
October	2.22	1.73	1.92	0.78	0.87	0.10	1.82	0.61

shows the average daily values of different parameters for the months from April to October each year. In the first two years, part of the data were lost (Table 2). In the first two years, ET_cSF was higher than ET_cLys, while in the last two years ET_cLys had slightly higher values in May to October in 2014 and July to September in 2015 (Table 2). Considering the average of the four years, the ET_c obtained by both procedures was very similar (

Table 3. Average of the four years of study of the mean daily values for each month of reference evapotranspiration (ET_o PM), crop evapotranspiration obtained in the lysimeter (ET_cLys), crop evapotranspiration obtained with the sap flow sensors (ET_cSF) in the irrigation treatment, crop coefficient obtained from the lysimeter (K_cLys), crop coefficient obtained as T_SF + E (K_cSF), evaporation (E), transpiration obtained with the sap flow sensors in the irrigation treatment (T_SFirrigation) and transpiration obtained with the sap flow sensors in the rainfed treatment (T_SFrainfed).

Months	2012–2015
	ETo PM	ETcLys	ETcSF	KcLys	KcSF	E	TSFirrigation	TSFrainfed
	(mm day−1)	(mm day−1)	(mm day−1)			(mm day−1)	(mm day−1)	(mm day−1)
April	3.55	1.16	1.29	0.33	0.35	0.16	1.25	0.57
May	5.46	2.83	2.76	0.51	0.51	0.63	2.10	0.88
June	6.19	4.77	4.90	0.77	0.82	1.43	3.59	1.15
July	6.72	5.35	5.47	0.78	0.80	1.79	3.96	0.72
August	5.97	5.17	5.00	0.85	0.82	1.53	3.62	0.69
September	4.00	3.38	2.89	0.83	0.70	0.67	2.52	0.59
October	2.26	2.08	1.69	0.89	0.72	0.17	1.48	0.49

). Consequently, the crop coefficient (K_c) values obtained in both cases were similar when considering the four years as a whole. The maximum values of 0.85 and 0.82 for ET_cLys and ET_cSF, respectively, were in August (Table 3).

E was higher in 2012 than in subsequent years, reaching 2.29 mm day⁻¹ as the mean value for August (30% and 33% of ET_cLys and ET_cSF, respectively) (Table 2). The T_SF in the irrigation treatment was variable, with the highest daily values reached in all years in July but with notable differences between the first two years close to 5 mm day⁻¹, while in the latter years the maximum barely exceeded 3 mm day⁻¹ (Table 2). The T_SF in the rainfed treatment was much lower than in the irrigated treatment in all cases, but without a clearly defined seasonal pattern.

Figure 6 shows the daily evolution of T_SF per unit of leaf area for different days in the months under study in 2012. As can be seen, there is a certain parallelism between the daily evolution of T_SF and SR in the early morning hours, with a certain shift of the maximum values of T_SF towards the afternoon. The number of hours of sunshine affects T_SF, so that in June the daily transpiration period was 14 h and 30 min (Figure 6A), while in October it was reduced to 10 h (Figure 6E). From the first measurements in June 2012 (Figure 6A), the rainfed treatment had a lower T_SF per unit leaf area than the irrigation treatment. However, the daily evolution followed the same trend in both treatments in all months, with no abrupt restriction of T_SF detected in the rainfed treatment, even with Ψ_s^md of −1.3 MPa (Figure 6B), as was observed in 2013 and 2014 (Figure 5B,C).

3.4. Relationship between Methodologies for Determining ET_c

The relationship between ET_cSF obtained from the sum of its components T_SF and E and ET_cLys obtained directly from a weighing lysimeter for the four years of the trial is presented in Figure 7. The linear regression was highly significant with R² = 0.85 but did not coincide with the 1:1 line. With low values, ET_cLys was higher than ET_cSF, while the situation was reversed for high values. As seen previously in Table 2, in the first two years ET_cSF overestimated ET_c relative to the lysimeter measurements.

The year 2012 was the year in which the greatest amplitude of values was obtained, with maximum values higher than 9.0 mm, while in the other years the maximum values were between 6.0 and 7.0 mm.

With all this, the validation of the adjustments made was satisfactory, as it was possible to reproduce ET_cLys by means of the sap flow sensor methodology and evaporation modeling.

3.5. Relationship between Vegetative Growth and T_SF

The relationship between T_SF and the measured vegetative growth indexes, FiPAR and LAI, for the two treatments in the four years of the trial is presented in Figure 8. The increase in FiPAR and LAI in the irrigation treatment was accompanied by a rise in T_SF. This relationship was more robust with LAI with an R² = 0.68 (Figure 8B) than for FiPAR R² = 0.34 (Figure 8A). In the rainfed treatment, the T_SF decreased as the vegetation on the vines increased. The correlation coefficients obtained for FiPAR and LAI were similar: R² = 0.54 and 0.53 for FiPAR and LAI, respectively (Figure 8A,B).

4. Discussion

Although calculations of vineyard water requirements consider both evaporation from the soil surface and transpiration from the vines and living plant covers, only the water transpired by the vines influences vine productivity so that efficient irrigation is achieved when most of the water applied is consumed as transpiration [69]. Therefore, the quantification of transpiration provides the necessary information to improve water use both in high and low water availability conditions. Although sap flow sensors determine plant water consumption, it is necessary to adjust this methodology to obtain accurate values, especially when the aim is to evaluate plant response to different factors. The regulation of vine water relations is an essential tool for managing vineyard quality, which depends on the availability of water and the plant mechanisms involved in regulating gas exchange in the plant canopy.

In this work, we evaluated the use of sap flow measurements in vineyards, both for the accurate assessment of water consumption, with the support of a weighing lysimeter as a reference, and to analyze the effect of water stress on the vines on the daily and seasonal patterns of transpiration.

4.1. Adjustment and Evaluation of Sap Flow Sensors to Determine Vine’s Transpiration

Although different authors have proposed the use of sap flow measurements to quantify water losses by transpiration in grapevines [19,33,34,35], as in other woody species [27,29,30,31,32,66], direct calibrations give reliability to the measures. In this work, by having a weighing lysimeter it was possible to establish a calibration coefficient of α = 3. Although we obtained this coefficient on the two lysimeter vines, the values and trends of the sap flow rates of the four irrigated vines, together with the rainfed vines, suggest that the fit was acceptable for all of them. Other authors [70] indicated the need to establish a calibration coefficient (α) for each year and sensor, which may have reduced the adjustment obtained in this work by establishing a single coefficient for all years and sensors.

4.2. Quantification of ET_c in Vineyards by Sap Flow Measurements

The ET_cSF had a good correlation with ET_cLys (R² = 0.85) (Figure 7). The most remarkable feature is the underestimation of this methodology for values lower than 4.5 mm day⁻¹. This may be due to a low estimation of evaporation in periods when rainfall is common, the initial and final stages of cultivation (Figure 3), wetting the entire soil surface and, therefore, increasing the evaporative component within the water consumption of the vineyard [5]. This situation is not included in the proposed evaporation model. As shown in Figure 7, we obtained the best fit between the two ET_c determination methodologies in 2012, which was the year in which there were fewer rainfall events during cultivation and when we had no sap flow data until July, which resulted in an R² value of 0.93 (data not shown). After budbreak in 2013, a year characterized with low crop water consumption and high precipitation during spring, ET_cSF was proportionally higher than ET_cLys. These observations point to errors in the quantification of E, both due to overestimation and underestimation, and whose relative importance depends on factors such as rainfall and soil wetting and exposure to solar radiation. Therefore, they highlight the need to improve the quantification of E and its contribution to ET_c under different circumstances, as already pointed out by [6] under the same conditions.

The highest values of the transpiration coefficient (K_T) (calculated as the ratio of T_SF to ET_o) obtained in this work were significantly lower than the highest basal coefficient (K_cb) obtained by [6] (K_T between 0.63 and 0.65, compared to 1 for K_cb), except for 2012 when similar values were reached. K_cb is usually calculated from lysimeter ET_c measurements when the top soil layer is dry (ET_c ≈ ET_cb, zero evaporation), a condition that is hard to reach with drip irrigation due to the high irrigation frequency.

T_SF values were similar to those obtained by [71] for cv. Shiraz in the summer months with maximum values around 3.0 mm day⁻¹ versus the 4.0 and 3.6 mm day⁻¹ for July and August, respectively, obtained in our study (Table 3). However, our results were appreciably higher than those obtained by [19] for a set of varieties and water status with values ranging between 0.3 and 2.0 mm day⁻¹ versus the 0.1 and 5.2 mm day⁻¹ obtained in our study (Table 2). These higher values could be due to the absence of water stress observed in part of the vines monitored in our study.

The evaluated water consumption parameters followed a seasonal pattern similar to other studies [6,72]. The difference in values from other studies might be due to the increased vegetative growth and consequently the increased water consumption during July and August (K_c = 0.78–0.85) (Table 3). In the summer months (DOY 152–243), the greatest differences between ET_c and T_SF were observed (Figure 5A–D), coinciding with the application of irrigation, which increased evaporation due to the increase in the wetted soil surface. This gap between ET_c and T_SF was because these months were those with the highest evaporative component (Table 2 and Table 3), with this evaporation reaching a maximum in July and August (DOY 182–243). The maximum values of solar radiation and air temperature, determining factors for evaporation, were recorded in these months (Figure 3). These results are similar to those obtained for a vineyard of the same variety by [5], although in our case we obtained lower evaporation rates in the early stages of crop development period.

The maximum values of E in the irrigation months (Table 2 and Table 3) show the inefficiency of high irrigation frequencies in semi-arid climates as they lead to significant evaporation losses [5], with maximum values of 30% and 33% of the total value of ET_c in June, July, and August, in accordance with the findings of [5] for high irrigation volumes.

Moreover, we observed the effect of pruning on transpiration, both in the weighing lysimeter and in the data obtained using sap flow probes (Figure 5A–D). There was a decrease proportional to the leaf area removed by green pruning, which agrees with previous observations in vines [6] and other woody crops [2,57,67,73].

4.3. Relationship between Transpiration and Vegetative Growth of Grapevines

Crop transpiration is closely related to canopy development under non-water-limiting conditions [2,6,57,67,73], with a maximum stomatal aperture under these conditions with full illumination and high relative humidity [74]. In this study, T_SF in the irrigation treatment increased from budbreak with increasing LAI and FiPAR and decreased sharply with the clippings (Figure 4A–H). Comparing relationships between LAI and FiPAR with T_SF in the irrigation treatment with those obtained by [6] with K_cb, in both cases there is a better relationship with LAI than with FiPAR. In the rainfed treatment, dispersion is lower since, when stomatal closure occurs, environmental conditions do not have such a strong influence on vine transpiration. Contrary to the irrigation treatment, in the rainfed treatment, the greater development of the canopy reduces T_SF (Figure 8A,B), as stomatal closure is necessary for the plant to defend itself against water limitation [75]. Under conditions of low water availability, a larger canopy size means higher water demand and faster consumption of reserves, leading to more severe stress. Under these conditions, a series of plant mechanisms that control transpiration through stomatal closure comes to the fore [76].

4.4. Evaluation of Vineyard T Response under Water-Limiting and Non-Water-Limiting Conditions

The conditions in which the strains develop (environmental conditions and water availability) determine the canopy size (number of leaves, size, and arrangement) and characteristics (number and size of stomata) [77], as well as the hydraulic system of the plant [78], and therefore, the transpiration potential at a given time. In grapevines, stomatal conductance depends on environmental conditions during the early stages of leaf development, as it sets the upper limit of conductance by modulating the density and size of the stomata, the current conditions, and the source-to-sink ratio [77]. When vines develop under high-temperature conditions in the absence of water deficit, they can increase canopy growth with increased transpiration [79]. In the year 2012, with higher temperature and radiation after budbreak, the irrigation treatment had the highest LAI and FiPAR values relative to subsequent years and possibly explains why the highest T_SF values were reached in this year (Figure 5E–H). The scarcity of previous rains in 2012 increased the differences between treatments. However, in 2013, although the temperature was also high in the initial stages, radiation was lower, with a higher number of cloudy days, and rainfall was higher, which reduced the differences in canopy growth between the two treatments. As mentioned above, a larger canopy size under stress conditions resulted in tighter control of transpiration [79]. Differences in T_SF were observed among the treatments studied in all years (Figure 5A–D), reaching the highest values in 2012 (Figure 5A).The rainfed treatment presented an anhisohydric behavior in 2012 and 2015, while in 2013 and 2014 its behavior was isohydric, with a drastic decrease in transpiration to reduce water losses in the vines (Figure 5A–D). These results further the controversy of whether the grapevine (Vitis vinifera L.) presents an isohydric or anisohydric behavior [80,81,82]. This different behavior found in our study could be due to environmental conditions [83,84] and the stage of crop development at which adverse conditions occur [85].

The daily change in transpiration agrees with the daily change in SR for irrigation treatment (Figure 6). Stomatic conductance is of particular importance in the control of transpiration in grapevine, which, under non-water-limiting conditions, has a good coupling with the atmosphere [86]. Transpiration intensity in grapevine depends on water vapor pressure differences between the atmosphere and the sub-stomatic chamber. T in irrigated vines is influenced by the VPD [79,87,88]. This fact was confirmed in this study, obtaining the highest transpiration rates in the irrigated vines when the highest VPD were recorded and decreasing as the VPD decreased (Figure 3A–D and Figure 5A–D).

We also noted, coinciding with the study of [89], that transpiration is related to illumination since it is proportional to the radiation received, increasing in the central hours of the day [74]. T increased from early in the morning in the irrigated vines, reaching a maximum value at around 14:00 h, before decreasing from 16:00 h onwards (Figure 6), which agrees with what was observed by [90]. On the other hand, in the rainfed vines there was no significant decrease after 8:00 h, as also indicated by [90], but showed a different behavior depending on the year, with anisohydric behavior in 2012 (Figure 6) and isohydric behavior in 2013 and 2014 when a significant decrease in transpiration was observed in rainfed vines (Figure 5).

Under irrigated conditions, the Ψ_s^md was maintained at around −0.6 MPa, except in 2014 when there was a slight seasonal decrease (Figure 4I–L). Despite maintaining high values of Ψ_s^md and similarity between years, the T_SF of the irrigation treatment was different from year to year, both due to differences in plant canopy size (Figure 4A–H) and consumption per unit leaf area (Figure 4). In the rainfed treatment, interannual differences in the Ψ_s^md of the vines must also be added (Figure 4I–L). Circumstances, which have already been highlighted in previous works on this same vineyard [64], highlight the importance of considering weather conditions when planning the agronomic management of the vineyard. In the years 2012 and 2015, in the rainfed vines, there came a time after veraison water stress fell with decreasing evaporative demand (Figure 5A,D), improving the Ψ_s^md of the vines (Figure 4I,L),which indicates a coupling of the vines with the atmosphere, as found for deciduous trees by some authors [91], with the atmosphere conditioning the water status of the rainfed vines. This fact was less palpable in 2014 and inappreciable in 2013 (Figure 4J–K), manifesting an isohydric behavior.

In other fruit crops, some authors have observed that crop load affects stomatal aperture [57,92,93,94,95], also affecting transpiration in woody crops [66,86]. This effect was not observed in this work since the yield of 2013 was higher than the other three years (which had similar yield values), whereas the highest transpiration values corresponded to 2012. On the other hand, a progressive decrease in transpiration was observed after the harvest, as was found for this same vineyard by [6], a decrease that in some years began even before the harvest. Therefore, there does not seem to be a stomatal closure after the harvest as has been observed in fruit trees such as peach, apple, pear and plum [57,92,93,94,95].

The main insecurity of the methodology proposed for the calculation of ET_c in vineyards comes, on the one hand, from the lack of reliability of the evaporation model at certain times of the crop cycle, and on the other hand, from the convenience of establishing annual calibrations for each sensor individually.

5. Conclusions

1^ª. The use of a single “α” for all years and sensors gave satisfactory results, but with a deviation that could perhaps have been reduced by individualizing α for each sensor and year.

2^ª. The results obtained in the four years of study show that the sap flow sensor methodology, along with the HPC methodology plus the average gradient technique, are useful for quantifying vineyard water consumption, as they provide a good fit with the data obtained directly from a weighing lysimeter.

3^ª. Evaporation is an essential component of vineyard water consumption. It is necessary to develop efficient models for its determination and irrigation management to minimize its weight in vineyard water consumption.

4^ª. A close positive relationship between vegetative development and water consumption, in the absence of water stress, has been demonstrated. In contrast, under rainfed conditions, this relationship is negative.

5^ª. The isohydric or anisohydric behavior of the vineyard is greatly influenced by irrigation management and the environmental conditions of each year. In high initial productive and vegetative expectations, if water inputs do not support these expectations during the season the grapevine (Vitis vinifera L.) shows an isohydric behavior.