Assessment of Trunk Diameter Fluctuation-Derived Indices for Detecting Water Stress in Sweet Cherry Trees

Abstract

The continuous and reliable assessment of crop water status through water indicators enables the sustainable management of water resources, especially in arid or semi-arid climate scenarios exacerbated by climate change. Therefore, the main objective of this study is to determine and compare the sensitivity of indices derived from trunk diameter fluctuations for the accurate and automatic detection of changes in the water status of cherry trees. The water stress indicators examined are maximum daily trunk shrinkage (MDS), trunk growth rate (TGR), early daily trunk shrinkage (EDS), and late daily trunk shrinkage (LDS). During two growing seasons, ‘Lapins’ sweet cherry trees were subjected to different water stress levels: (i) a control treatment irrigated at 115% of crop evapotranspiration demand to ensure non-limiting water conditions, and (ii) a deficit irrigation treatment, with two irrigation withholding cycles. Vegetative growth was affected by water stress. Trunk daily growth rate and late daily trunk shrinkage exhibited a high variability and did not clearly show differences in plant water status. Both EDS and MDS showed a third-degree polynomial relationship with Ψ_stem. MDS had a lineal relationship with Ψ_stem of up to −1.4 MPa; however, further decreases in Ψ_stem did not necessarily lead to increased MDS. In contrast, EDS became non-linear at −1.8 MPa, making it a more useful plant water indicator than MDS for ‘Lapins’ sweet cherry trees when detecting severe water stress conditions. The frequencies of both MDS and EDS decreased from 85% to 35% when water stress increased. Therefore, the information provided by MDS and EDS frequencies, along with their daily values, could be useful as irrigation management tools for sweet cherry trees.

1. Introduction

In recent decades, the sweet cherry tree (Prunus avium L.) has become one of the most important fruit trees in the world, as it is a stone fruit that is highly valued by farmers due to its high production value and by consumers due to its organoleptic and nutritional properties [1,2]. However, the sweet cherry tree is sensitive to climatic and soil conditions, which could affect the crop’s water status and therefore its yield and fruit quality [3]. The increasing frequency and intensity of extreme weather events due to climate change poses a significant threat to agricultural productivity and global food security [4]. For example, droughts disrupt the critical balance between water demand and supply for crops, reducing yield and increasing crop losses. However, drought is not the only risk for this crop, as rain or hail close to the harvest period can lead to significant production losses. For this reason, many cherry growers around the world have started to introduce cherry trees under a protective cover, which is an important technological advancement in ensuring cherry production and quality, although it is currently associated with high production costs. Under these new growing conditions, precise control of all inputs, including irrigation water, has become increasingly important.

Currently, stem water potential (Ψ_stem) is being utilized as a reference indicator of crop water status, as it mirrors the water potential or energy status, which is traditionally used as the reference to compare against other water status indicators [5,6,7,8]. Although there has been progress in the automation of Ψ_stem measurements [5], the results need to be validated in the long term in several fruit trees.

The monitoring of crop water status has been made possible by advances in technology, digitalization, and automation, which is useful for managing irrigation systems. In this way, crop water status indicators derived from trunk diameter fluctuations (TDFs) have been used to schedule irrigation in many crops, such as almonds [9,10,11,12,13], olives [14,15], nectarines [16], and sweet cherry trees [17]. TDF measurements are related to dynamics of water storage and, therefore, with plant water status [18]. Thus, TDFs are the combination of four main trunk factors: (i) reversible dehydration/rehydration of living cells; (ii) expansion of dead conducting elements due to rising and falling internal stresses; (iii) irreversible radial growth; and (iv) thermal expansion and contraction [8,19,20,21,22]. Dendrometers are non-destructive instruments that allow for the continuous monitoring of TDFs [8,19,23]. Traditionally, linear variable displacement transducers (LVDTs) have been valuable in evaluating water stress in plants by measuring TDFs. They enable continuous and automated recording of maximum daily trunk shrinkage (MDS) and trunk growth rate (TGR), which have been considered as reliable indicators of water stress, and have been instrumental in developing automated irrigation scheduling [8,20,21].

TGR has been proposed as a sensitive plant water indicator in olive trees [24] and in young almond trees due to their vigorous growth [25]. TGR data vary throughout the season [12,16], the age of the plant [9,23], and crop load [26]. On the other hand, MDS has been proposed as an irrigation management tool due to its sensitivity to water stress [10,11,27]. In mature fruit trees, MDS increases with water deficit and is strongly correlated with Ψ_stem [19,20]. However, this relationship is non-linear, with maximum MDS values occurring at specific Ψ_stem crop-dependent values [19,20]. Additionally, the use of MDS for irrigation scheduling requires a baseline value that is dependent on fruit species, tree size, phenological stage, and fruit load [8,19,23]. Subsequently, de la Rosa et al. [16] proposed two indices derived from TDFs as alternative indicators to MDS in nectarine trees: early daily trunk shrinkage (EDS), which is the shrinkage that takes place before solar noon, and late daily trunk shrinkage (LDS), which occurs before midday. EDS was found to be more useful than MDS in nectarines, as it did not show a loss of linearity in its relationship with Ψ_stem.

In recent years, Corell et al. [28] and Martín-Palomo et al. [15] proposed that the frequency of TGR within specific ranges can determine the water stress level in olive trees. This frequency approach consists of establishing, for a specific level of water stress, a range of TGR values that minimizes interannual variation, independent of phenology and evaporative demand [13,15,28]. However, the use of frequencies in MDS and EDS has not been studied in sweet cherry trees.

This study evaluates the usefulness of several plant water status indicators in sweet cherry trees under different water stress levels during the postharvest period. This phase has been considered by several authors to be suitable for the application of a controlled water deficit in commercial plantations of early cherry varieties [17,29,30,31]. The hypothesis of this research is that the indices derived from TDFs and their frequency data can provide sufficient information on the response of sweet cherry trees to water stress and to establish thresholds for deficit irrigation scheduling. The objective of this study was to determine if TDF measurements are useful for both deriving growth patterns and detecting the onset and severity of water stress in sweet cherry trees. Therefore, the specific objectives were (i) to assess the relationships between midday stem water potential and trunk diameter fluctuation-derived indices, and (ii) to determine ranges of frequencies and assess their use as an irrigation management tool in sweet cherry trees under conditions of water scarcity and occasional water excess due to changes in climatic conditions.

2. Materials and Methods

2.1. Experimental Site

The experiment was conducted in a plot located at the Agricultural Experimental Station of the Technical University of Cartagena (37°35′ N, 0°59′ W, Cartagena, SE Spain) during two growing seasons (2018 and 2019). The sweet cherry trees [Prunus avium L. ‘Lapins’, grafted onto ‘Mirabolano’ rootstock] were planted in 2015 at a distance of 3.5 m × 2.25 m. The trees were irrigated with a line of drip emitters (2.2 L h⁻¹) spaced 0.75 m apart. The irrigation water had a pH value of 8 and an electrical conductivity mean of 1.1 dS m⁻¹. The soil had a sandy clay loam texture (34.5% clay, 21.3% silt, and 44.2% sand) and a normal organic matter content (2.3%).

Climatic parameters were collected by an automatic meteorology station of the Agricultural Information System of Murcia (CA52, www.siam.es, accessed on 25 March 2024), located near to the experimental plot. The Penman–Monteith equation was used to calculate the daily reference crop evapotranspiration (ET₀) values, and air temperature and relative humidity data were used to calculate the daily mean vapor pressure deficit (VPD) [5]. The study area was typical of a semi-arid Mediterranean climate, characterized by mild winters and warm summers. The ET₀ value reached a daily maximum of about 7.7 mm d⁻¹ in summer, and its values decreased at the end of each year (Figure 1). Temperatures ranged from 7.5 to 29.2 °C, and vapor pressure deficit (VPD) reached a maximum of 2.5 and 2.6 kPa in 2018 and 2019, respectively. Cumulative ET₀ and rainfall were 709.7 and 193.8 mm, respectively, from June to October in both years (postharvest period), with the rainfall on days of year (DOYs) 254 and 257 in 2019 being particularly significant (202.8 mm).

2.2. Experimental Design and Treatments

The experimental design was a randomized complete block with three replicates per irrigation treatment. Each replicate consisted of a row of four trees, with two central trees monitored (n = 6). Two irrigation regimes were applied: (i) control treatment, CTL, irrigated daily to satisfy 115% of the crop evapotranspiration (ETc), to guarantee non-limiting soil water conditions; and (ii) drought stress treatment (DT), with two drought-recovery cycles. The drought periods ended when the midday stem water potential reached (Ψ_stem) −1.6 MPa for the first drought cycle in both experimental years and −2.2 MPa and −2.4 MPa for the second drought cycle in the first and second experimental year, respectively. Afterwards, a recovery irrigation period was applied in which the same CTL irrigation doses were applied until the Ψ_stem values of the deficit treatment reached values similar to the CTL. ETc was determined according to the indications proposed by Allen et al. [32] and Marsal et al. [29]. Further details of the field conditions, water, and crop characteristics have been previously described in Blaya-Ros et al. [33].

2.3. Soil Water Status

Two access tubes per replicate (n = 6) were installed at 23 cm from the central emitter located under the canopy projection to represent two trees each. Volumetric soil water content (θv), from 0 to 0.8 m every 0.10 m was measured every 2–5 days between 9:00 and 11:00 h (Universal Time, UT) using a capacitance probe (Diviner2000, Sentek Pty. Ltd., Stepney, South Australia). Soil water stock (SWS), from a depth ranging from 0 to 0.8 m, was calculated from the sum of the θv values of the different soil profiles.

2.4. Plant Water Status

2.4.1. Midday Stem Water Potential

Midday stem water potential (Ψ_stem) was determined at solar noon (12:00–13:00 UT) every 2–5 days during the postharvest period using a Scholander-type pressure chamber (mod. SF-PRES-70, SolFranc Tecnologías, S.L., Tarragona, Spain), according to the recommendations by [34,35]. Ψ_stem was measured in 2 mature leaves per replicate (n = 6). Leaves near the trunk were covered with small black plastic bags and covered with silver foil for least 2 h before measurements. These measurements were used to quantify the water stress experienced by the trees for each week, using the mean of the water stress integral (SΨ_stem; MPa day⁻¹), following the methodology proposed by Myers [36]:

$${\mathrm{S}\mathsf{\Psi }}_{\mathrm{stem}}{(\mathrm{MPa} \mathrm{day}}^{-1})=\left|{\displaystyle \sum }_{\mathrm{i}=0}^{\mathrm{i}=\mathrm{t}}\left({\mathsf{\Psi }}_{\mathrm{m}}{ - \mathsf{\Psi }}_{\mathrm{c}}\right) \text{·} \mathrm{n}\right|$$

where t is the number of measurements of Ψ_stem, Ψ_m is the average Ψ_stem for each time interval, Ψ_c is the maximum Ψ_stem measured through each season (−0.44 MPa in both 2018 and 2019), and n is the number of days within each interval.

2.4.2. Trunk Diameter Fluctuations

Trunk diameter fluctuations (TDFs) were recorded by linear variable displacement transducers (LVDT sensor; model DF ± 2.5 mm, Solartron Metrology, Bognor Regis, UK) in 6 trees per treatment during the postharvest period. LVDT sensors were installed on the north side of each trunk, 20–30 cm above the graft, and mounted on holders built of aluminum and INVAR. Four TDF-derived indices were determined: maximum daily shrinkage (MDS), as the difference between the daily trunk maximum and the daily trunk minimum; trunk daily growth rate (TGR) [10], calculated as the difference between maximum daily trunk diameter of two consecutive days; early trunk shrinkage (EDS), which took place between the daily trunk maximum (between 6:00 and 9:00) and midday; and late daily trunk shrinkage (LDS), which occurred between midday and the daily trunk minimum (around 16:00) [16]. The LVDT sensors were connected to a datalogger (Model CR1000 with AM16/32B multiplexer, Campbell Scientific Ltd., Logan, UT, USA) programmed to take measurements every 30 s and report mean values every 10 min.

MDS and EDS were divided into five ranges [14]. The MDS frequencies were (i) “no stress”, below 150 µm day⁻¹; (ii) “mild stress”, between 150 and 250 µm day⁻¹; (iii) “moderate stress”, between 250 and 350 µm day⁻¹; (iv) “moderate–severe stress”, between 350 and 450 µm day⁻¹; and (v) “severe stress”, above 450 µm day⁻¹. The EDS frequencies were (i) “no stress”, below 100 µm day⁻¹; (ii) “mild stress”, between 100 and 150 µm day⁻¹; (iii) “moderate stress”, between 150 and 200 µm day⁻¹; (iv) “moderate–severe stress”, between 200 and 250 µm day⁻¹; and (v) “severe stress”, above 250 µm day⁻¹.

In addition, in the present work, four ranges of Ψ_stem were grouped according to recommendations by Blanco et al. [17], Marsal et al. [30], and González-Teruel et al. [37]. Thus, four water stress level were established: (i) no water stress conditions, with Ψ_stem values above −0.8 MPa; (ii) mild–moderate water stress, between −0.8 and −1.2 MPa; (iii) moderate–severe water stress, between −1.2 and −1.6 MPa; and (iv) severe water stress, values below −1.6 MPa.

Signal intensity (SI) for the different water status indicators was determined as the ratio between DT and CTL values, and sensitivity as the ratio between SI and coefficient of variations (CV; the ratio of the standard deviation to the mean) [10].

2.5. Vegetative Growth

At the beginning and end of each experimental period (early June and early October—postharvest period), the trunk diameter (TD) of the two central trees per replicate was measured 0.15 m above the grafting point using a tape measure (Pi meter MF612 A, Weiss, Erbendorf, Germany). The annual increase in TD (ΔTD) was determined by subtracting the initial TD from the final TD measured each year.

2.6. Statistical Analysis

An ANOVA and Duncan’s multiple range test (p ≤ 0.05) were used to identify significant differences among treatments and water stress indicators. IBM SPSS Statistics (SPSS Inc., 24.0 Statistical package, Chicago, IL, USA) was used for data analysis. The RStudio package v.25 (RStudio Inc., Boston, MA, USA) and Sigmaplot Plus (Systat Software, San Jose, CA, USA) were used to perform both linear and non-linear regression analyses between water indicators. In addition, a principal component analysis (PCA) was performed on the responses of the water stress indicators using irrigation treatment as the attribute for comparison (factoextra: fviz_pca_ind). Both analyses serve the purpose of testing whether soil and plant water indicators from the irrigation treatments were comparable.

3. Results

Soil water stock (SWS) at a depth of 0.8 m showed different patterns throughout the experimental period in both irrigation treatments (Figure 2a,b). SWS significantly affected plant water status, as determined by stem water potential (Ψ_stem; Figure 2c,d). During the pre-drought period, there were no significant differences between treatments. SWS was about 297.5 mm at a depth of 0.8 m, and Ψ_stem was around −0.5 MPa in both irrigation treatments. However, when the first drought period started in the DT, on day of the year (DOY) 187 in 2018 and on DOY 176 in 2019, the water stress caused a continuous decrease in both variables, until reaching values of approximately 228.4 (DOY 202) and 219.5 mm (DOY 197) in SWS and −1.5 and −1.6 MPa in Ψ_stem during the first water deficit period in 2018 and 2019, respectively. Similarly, SWS at the first 0.8 m in depth was approximately 220.2 (DOY 237) and 210.2 mm (DOY 235) in the second drought cycle. SWS values in the drought stress treatment (DT) were slightly higher in 2018 than in 2019, which would be consistent with Ψ_stem values being lower during the second experimental year, reaching values of −2.1 and −2.5 MPa during the second drought cycle in 2018 and 2019, respectively. After irrigation was resumed, SWS and Ψ_stem values increased to values similar to the CTL trees. Nevertheless, the CTL maintained an average of 297.7 mm and −0.65 MPa for SWS and Ψ_stem, respectively, in both experimental years.

The CTL maintained constant soil water content (θ_v) values at all depths during both experimental years (Figure 3). θ_v in the CTL was above 35% in all layers. The DT exhibited similar values to those found in the CTL (no significant difference) during the pre-drought periods. However, θ_v values decreased during the drought periods when the irrigation was withheld. Thus, θ_v showed significant differences between irrigation treatments, especially in the first 0.3–0.4 m in depth, where θ_v showed a greater reduction than below a depth of 0.4 m. In all the recovery periods, water reserves were restored to levels comparable to those of the CTL.

The stress integral (SΨ_stem) data exhibited significant differences between treatments and seasons (Figure 4a). The CTL presented a mean SΨ_stem value of 17.75 MPa d⁻¹ in both experimental years. However, the DT showed a significant increase in SΨ_stem values as compared to the CTL trees, and showed a significantly higher value in the second year, accumulating to 56.7 and 82.9 MPa d⁻¹ in 2018 and 2019, respectively. Water stress significantly affected vegetative growth, determined by the trunk diameter differential (ΔTD). In this study, significant differences were observed between treatments in both years and between years in the same treatment (p < 0.05). The TD of the trees showed less trunk diameter growth throughout the postharvest period, resulting in ΔTD values that were 27.1 and 40.6% lower than the CTL trees for the first and second experimental year, respectively.

Regarding trunk diameter fluctuation (TDF) parameters, the maximum daily trunk shrinkage (MDS) and trunk growth rate (TGR) exhibited a day-to-day variability throughout the experimental period (Figure 5a,b). During the pre-drought period, the DT trees had similar maximum daily trunk shrinkage (MDS) values to the CTL trees, but they became higher when water withholding was imposed. MDS values increased as water stress progressed, and the differences increased at the end of each deficit period. During both drought periods, significant differences in MDS rapidly developed when water stress was imposed in both seasons. Thus, the maximum MDS values were recorded during the drought periods. Furthermore, lower MDS values were observed in 2018 (255 μm) than in 2019 (440 μm). The mean MDS values were around 110 μm in the CTL trees during both experimental years. In 2019, the rainfall (202.8 mm; between DOY 254 and 257; Figure 1) caused the loss of a week’s worth of TDF data. Trunk growth rate (TGR) is another water indicator that can be used as an indicator of plant water status. Maximum values of TGR, near 302 μm day⁻¹, were reached at the beginning of the experimental period of each year (June–July), which declined thereafter until growth slowed in late September. The TGR values in CTL trees were higher than in DT trees. The DT trees showed an effect of accumulated water stress during the recovery periods, as the trees were exposed to water restriction (Figure 2 and Figure 3), which affected stem growth even after irrigation recovery. In addition, TGR showed a higher coefficient of variation (CV) than MDS. The DT trees showed TGR values close to 0 μm, corresponding to no stem growth (Figure 4b). This is reflected in the cumulative trunk growth rate (ΣTGR) values of the CTL trees, where growth was higher than in the DT trees (Figure 5e,f). In both years, ΣTGR showed significant differences between irrigation treatments about 20 days after the irrigation was withdrawn in the DT (Figure 2). The CTL trees recorded two slopes in the ΣTGR trend: (i) a steep slope at the beginning of the season, from DOY 178 to 210 in 2018 and DOY 166 to 215 in 2019; and (ii) a second slope, less abrupt than the first one, which lasted until the end of the measurement season. This indicates a stronger growth of sweet cherry trees until the beginning of August. Growth then slowed down slightly as the season progressed. Similarly, ΣTGR values in the DT trees were significantly affected by water stress. The increase in TGR values resulted in a temporary increase in ΣTGR. However, the short recovery period limited the self-compensating growth that typically occurs during periods of increasing ΣTGR slope. Therefore, at the end of the experimental period, the increase in trunk diameter of CTL trees was 34 and 65% greater than the DT trees in 2018 and 2019, respectively. This higher growth of the CTL trees as measured by the LVDT sensors is consistent with the data collected by the tape measure (Figure 4b).

During the experimental period, EDS data (Figure 6) exhibited a behavior similar to the MDS (Figure 6a,b). In both experimental years, mean EDS values in the CTL trees ranged from 0 to 139 μm. As was the case for MDS, EDS showed significant differences between treatments during the drought cycles in both years. Subsequently, when the irrigation was withdrawn, the differences in EDS values between the irrigation treatments increased. Thus, the DT trees reached maximum values of around 200 µm in 2018 and 300 µm in 2019 during the water stress periods. These differences persisted until irrigation was resumed, at which point EDS values rapidly decreased to levels similar to the CTL trees. Notably, LDS values were consistently lower than EDS values across both irrigation treatments and years. LDS values in DT trees were significantly higher than those in CTL trees for a limited portion of the water deficit periods in 2019. EDS showed a similar tree-to-tree variability than MDS and was lower than LDS and TGR.

Signal intensity (SI) was affected by water stress in both seasons (Figure 7a,b). SI values of SWS, Ψ_stem, MDS, EDS, and LDS increased, while SI values of TGR decreased as expected. EDS had the highest SI values (3.5–4.0) in both years, followed by Ψ_stem, MDS, LDS, SWS, and TGR. TGR showed SI values below 1 due to the lower growth of DT trees compared to CTL trees (Figure 5c,d). SWS in both years and LDS in 2018 showed SI values close to 1. Ψ_stem and SWS had a high sensitivity (S; Figure 7c,d), mainly because both indicators had a very low coefficient of variation (CV), 6.7% and 5.2%, respectively. MDS and EDS had similar S in 2018; however, EDS had higher S than MDS in 2019. TGR was the water indicator with the lowest S in both seasons as a result of its low SI.

Figure 8 shows the frequency patterns of different ranges of maximum daily trunk shrinkage (MDS; Figure 5a,b) and early daily trunk shrinkage (EDS; Figure 6a,b) for four water stress levels during both seasons (2018 and 2019). These water stress levels were determined based on midday stem water potential (Ψ_stem) values (Figure 2c,d). The MDS and EDS frequencies of the low range (<150 for MDS and <100 μm for EDS) were highest under no water stress conditions (Ψ_stem > −0.8 MPa) in both seasons. The frequency values changed with increasing plant water stress. The low frequencies (<150 for MDS and <100 μm for EDS) decreased sharply in both seasons, from almost 80–90% to about 35% when the moderate–severe stress level (−1.2 MPa < Ψ_stem < −1.6 MPa) was reached. Furthermore, under moderate–severe water stress level (−1.2 MPa < Ψ_stem < −1.6 MPa), a high variation was exhibited in the range between 250 and 450 μm for MDS and between 150 and 250 μm for EDS. On the other hand, during the severe water stress periods, the range of frequency >450 μm for MDS and >250 μm for EDS showed a slight decrease, from 10 to 20% to about 5 to 10%. Conversely, the frequency of MDS values below 350 μm and EDS values below 200 μm increased from less than 10% to about 30%, especially in the frequencies of MDS (Figure 8a,b).

To analyze the effect of water stress on the relationships between soil and plant water indicators in sweet cherry trees, Spearman’s correlation analysis was performed (Figure 9). The correlation matrix revealed strong associations between the water indicators. The soil water status, as measured by soil water stock (SWS), had a strong influence on midday stem water potential (Ψ_stem).

Ψ_stem exhibited positive correlations with SWS and trunk diameter growth rate (TGR) but negative correlations with maximum daily trunk diameter shrinkage (MDS), early daily trunk diameter shrinkage (EDS), and late daily trunk diameter shrinkage (LDS). Notably, most plant indicators showed correlations greater than 0.5, except for LDS and TGR, which showed no significant correlations with Ψ_stem, the traditional reference water indicator.

On the other hand, Figure 10 shows the relationships between SWS, MDS, EDS, and LDS with Ψ_stem. SWS showed a high third-degree polynomial relationship with Ψ_stem in 2018 (R² = 0.82), 2019 (R² = 0.91) and both experimental years (R² = 0.85) (Figure 10a). Regarding the relationships between plant indicators, MDS and EDS showed third-degree polynomial relationships with Ψ_stem in all the years studied (Figure 9b,c). Both TDF indices increased as Ψ_stem dropped, from −0.5 MPa to a threshold value of around −1.4 MPa in MDS (Figure 9b), and −1.8 MPa in EDS (Figure 10c). In this experiment, values below these thresholds indicated that Ψ_stem decreases were not directly related to higher EDS and MDS values, showing a loss of slope as stress increased. The relationship between EDS and Ψ_stem was better than that found with MDS, reaching the high correlation coefficient of 0.83 in 2019. On the other hand, there was no significant relationship between LDS and Ψ_stem in any year of study (Figure 10d). However, EDS and MDS showed an interannual variability due to the higher values observed in 2019.

A principal component analysis (PCA) was performed on the water indicators for the different irrigation treatments (

Table 1. Principal component analysis with extracted components and total variance explained.

Component	Total	Initial Eigenvalues		Extraction Sums of Squared Loadings
		% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	3.954	65.908	65.908	3.954	65.908	65.908
2	0.941	15.677	81.585
3	0.659	10.983	92.567
4	0.209	3.475	96.042
5	0.165	2.753	98.795
6	0.072	1.205	100.000

and Figure 11). The effects of the irrigation treatments on the soil and plant water indicators during both experimental years are summarized by the projections in the two principal component coordinates, principal component 1 (PC1) and principal component 2 (PC2). According to the PCA, PC1 and PC2 accounted for 65.9 and 15.7% of the total variance, respectively, and explained 81.6% of the variability of the data. The principal component analysis showed that the water indicators Ψ_stem and SWS provided the greatest contribution to PC1, followed by EDS and MDS. In contrast, PC2 was mainly driven by the water indicators derived from trunk diameter fluctuations, except for MDS. Thus, LDS was the water indicator with the largest contribution in PC2, followed by EDS and TGR. Thus, it can be observed that EDS contributes to both principal components. On the one hand, in the CTL (red ellipse), the PCA showed a strong interaction with TGR, but this only contributed only about 10% of the total variance. On the other hand, in the DT (blue ellipse), there were strong contributions from the indicators studied.

4. Discussion

Soil water status, as determined by soil water content measurements, had a significant effect on plant water status. Midday stem water potential (Ψ_stem) is a well-established indicator of plant water stress in fruit trees [6,17,35]. The CTL trees were irrigated to satisfy 115% of crop evapotranspiration, ensuring constant soil water content values (Figure 2a,b) throughout the experimental period and the soil profile (Figure 3). This approach successfully maintained Ψ_stem values within the range of non-stressed conditions for sweet cherry trees [17,30,37]. In contrast, the drought stress treatment (DT), which included two drought stress cycles, showed a greater water depletion (Figure 2a,b and Figure 3), and consequently a decrease in Ψ_stem (Figure 2c,d and Figure 4). A strong relationship between soil water stock (SWS) and Ψ_stem was observed in the present work, supporting the dependence of plant water status on soil water status. The minimum Ψ_stem values reached by the DT trees in both seasons were below the threshold values recommended by Marsal et al. [30] and Blanco et al. [17] for ‘Summit’ and ‘Prime Giant’ sweet cherry trees, which resulted in a decreased canopy volume and significant leaf area loss. Although SWS had a high sensitivity (S), its low signal intensity (SI) limits its use for IS-based irrigation scheduling. Contrary, in Blanco et al. [17], the soil matric potential had a high SI and S despite its high coefficient of variation (48%). The different hydrophysical properties of the soil in the two experiments could be the reason for this distinct behavior between the two related variables. In our case, it is a soil with high bulk density, resulting in a compacted soil in which small changes in SWS are reflected in significant reductions in Ψ_stem.

Vegetative growth, as assessed by the increase in stem diameter (ΔDT), was sensitive to water deficit, with differences observed between irrigation treatments at the end of each experimental season (Figure 4b and Figure 5e,f). The ΔDT values of the DT trees were lower than in the CTL trees, confirming the cumulative effect of water stress [38]. This is in agreement with the results obtained by Livellara et al. [39] on ‘Brooks’ and Blanco et al. [3] in ‘Prime Giant’ sweet cherry trees, who reported a reduction in upper crown diameter and shoot growth in trees under deficit irrigation during the postharvest period. According to the Ψ_stem values (Figure 2c,d), the DT trees recovered, and compensatory vegetative growth was observed (Figure 5e,f), maintaining the mean values of TGR slightly higher than the CTL trees at the beginning of the recovery period. However, this was not able to mitigate the lower vegetative growth in the DT trees. Although the managing of vegetative growth can be beneficial both on yield and labor cost reduction in mature trees [40,41], severe and prolonged water stress can reduce the number of fruits due to smaller canopies, and can also negatively affect future yields [39,42].

The use of trunk diameter fluctuation (TDF)-derived indices has been extensively studied for plant water status assessment and as a tool for irrigation programming in several fruit trees, including almond [11], plum [26], pear [7], nectarine [16], and sweet cherry trees [17]. In addition, TDF indices have been employed for irrigation scheduling based on signal intensity (SI), allowing for the adjustment of the amount of irrigation applied to the crop [10,11], especially in areas where climate change may alter previously established crop coefficients [43]. However, this process requires reference values for accurate irrigation management [19,20,27]. Currently, two methods have been used to establish reference values or thresholds: (i) maintaining trees without water stress, which requires more water and a greater management complexity; and (ii) deriving reference equations based on agroclimatic parameters [9,19,20]. However, the first option is not commercially viable, as it implies an increase in investment costs [19], and the second may introduce issues with crop-related factors, such as phenology, crop management [20], tree size and age [19], crop load [26], cultivar, and rootstock [44].

Trunk growth rate (TGR) and late daily trunk diameter shrinkage (LDS) showed significant differences with respect to the CTL trees; however, both indicators did not show a clear discernible pattern in response to drought water cycles due to the very high CV (≈49%). Similar CV values have been observed in pomegranate trees [45]. De la Rosa et al. [16] and Fernández and Cuevas [20] reported that the S of TGR is affected by a high CV, as observed in the present study. Thus, TGR and LDS data prevented the detection of patterns according to the crop water status. In addition, the high variability observed in both water indicators would suggest that a large number of instrumented trees would be required to adequately determine the water status of an orchard, which would further imply higher operational costs [46]. However, this problem could be mitigated by integrating other plant-based techniques to help identify zones of similar characteristics within heterogeneous orchards [20]. This is also confirmed by the low-significance relationships between TGR and LDS with Ψ_stem (Figure 9), which makes this water indicator irrelevant for assessing the plant water status. In nectarines, de la Rosa et al. [42] observed that TGR values increased early in the season until reaching maximum values in June–July, after which they were reduced significantly until reaching values of around 0 μm in September. The decrease in TGR values observed in this work coincides with the summer cessation of vegetative growth in sweet cherry trees [47].

Maximum daily trunk shrinkage (MDS) and early daily trunk shrinkage (EDS) showed lower tree-to-tree variability than LDS and TGR. In this experiment, the CV values were slightly lower than those reported in pomegranate trees [45] and higher than those observed in almond trees [10]. EDS and MDS had higher S and SI values than TGR and LDS, and can be considered as the most appropriate stress indicators of the four studied, derived from trunk diameter fluctuations. In this study, the SI values for MDS and EDS were greater than those reported by Blanco et al. [17] and de la Rosa et al. [16] under regulated deficit irrigation in mature sweet cherry and in nectarine, respectively. This is consistent with the higher level of water stress reached in our experiment. For this reason, MDS values are the most widely used automated plant water stress indicators in several fruit trees [19,20], demonstrating their usefulness in irrigation scheduling [10,11,20]. Similar to what was observed in the present work, EDS was found to be more useful than MDS for assessing plant water status in nectarine trees under severe stress conditions [16]. Although the CV values of MDS and EDS were higher than those shown for Ψ_stem (CV ≈ 7%), they detected the irrigation changes, both recovery and drought periods, earlier than Ψ_stem. Thus, Ψ_stem needed about 7 and 14 days for the first and the second drought periods, respectively, to reach similar values to the CTL trees. On the contrary, MDS needed only one day.

Curvilinear relationships were observed between MDS and EDS with Ψ_stem (Figure 10a,b). According to these values, a decrease in Ψ_stem values does not lead to an increase in MDS or EDS values. Thus, when plant water stress becomes severe and exceeds a critical threshold, it can deplete water reserves in phloem tissues and eventually decrease water uptake [18]. Several factors may contribute to the observed changes in MDS behavior [19]: (i) increased transversal resistance to water movement from bark to xylem [48]; (ii) decreased stomatal conductance and, consequently, transpiration [49]; and (iii) reduced water uptake from the bark, including the phloem and cambium, as stored trunk water decreases [50,51,52]. Blanco et al. [17] observed a similar threshold of MDS data in ‘Prime Giant’ sweet cherry trees, where Ψ_stem values below −1.3 MPa caused a plateau in the relationship between MDS and Ψ_stem. As in the case of EDS, and contrary to what was observed by de la Rosa et al. [16] in nectarine during the postharvest period, in this work, a threshold was observed where EDS ceased to be linear with Ψ_stem. The higher threshold reached by EDS, as compared to MDS, is due to hourly stem shrinkage and ∆Ψstem showing a linear relationship before noon in trees under water stress conditions. In the afternoon (12:00–6:00 pm), the increase in stem shrinkage is similar to that of trees without water stress due to stomatal closure [16]. However, due to trunk growth (Figure 4b and Figure 5e,f), an interannual variation was observed, causing an increase in the values of MDS and EDS, which may lead to a change in the relationship of both water indicators with Ψ_stem.

The use of TGR frequency data has been recommended in olive trees [12,14,15] for evaluating irrigation strategies as a tool for irrigation scheduling. However, as reported in several crops [10,11,16,17,27,53], EDS and MDS were more appropriate indicators to assess the response of sweet cherry ‘Lapins’ than TGR and LDS due to their higher SI and S (Figure 8). In addition, in almond trees, TGR frequency data were ineffective for assessing plant water status because tree size and load may affect the extreme values of TGR (<−0.1 mm day⁻¹ and >0.3 mm day⁻¹), which influence TGR frequencies [13]. Therefore, in this experiment, MDS and EDS frequencies were grouped into four different water stress levels. As shown in Figure 8, the patterns of MDS and EDS frequencies were affected by water stress and were able to quickly show the effect of water stress. The MDS and EDS frequencies showed several physiological processes that could have different drought sensitivity levels and improve the water stress management. However, seasonal changes were observed in this experiment, which would make it difficult to classify the frequency values due to an increase in tree size and crop load [13].

The DT provided a wider range of values for all indicators (Figure 11), allowing for the selection of the water indicator most strongly related to Ψ_stem and irrigation management. Thus, the PCA showed, similar to the relationships between soil and plant water indicators (Figure 9 and Figure 10), that the plant water status is largely determined by the soil water status. In addition, EDS and MDS, the indicators derived from trunk diameter fluctuations, obtained the highest correlation coefficient with Ψ_stem, and contributed to principal component 1 (PC1), which could be useful for assessing plant water status. This indicated a weak relationship with the other water indicators, as previously reported by de la Rosa et al. [16] in nectarine.

Overall, the pattern described by both MDS and EDS in both seasons was consistent with soil water depletion, with a clear and continuous increase in both water stress indices with decreasing SWS and plant water energy status, Ψ_stem (Figure 5 and Figure 6). Similarly, the high SI values reached in both drought cycles corresponded to a significant decrease in the DT trees and an increase in the water stress integral (Figure 4 and Figure 5). Moreover, these findings were reflected on the high correlations obtained with Ψ_stem, an indicator considered key in the study of water relations (Figure 7) [53].

In light of all the results obtained, and considering that a severe water stress that persists during the postharvest period affects the fruit set of the following year, as a consequence of the reduction in the starch concentration in the reserve organs [30], it would be useful to study the effect of different irrigation treatments that are programmed on the basis of the SI of MDS and EDS on yield and fruit quality, in order to be able to recommend the most appropriate irrigation strategy for the postharvest period of early sweet cherry varieties.

5. Conclusions

Trunk growth rate (TGR) and late daily trunk shrinkage (LDS) were not clear indicators of water stress in ‘Lapins’ sweet cherry trees, due to the high tree-to-tree variability and their low relationship with stem water potential (Ψ_stem), the reference indicator of water status. In contrast, MDS and EDS showed a strong correlation with Ψ_stem. Furthermore, water stress significantly affected maximum daily trunk shrinkage (MDS) and early daily trunk shrinkage (EDS), which detected irrigation change faster than Ψ_stem. However, interannual variability was observed in both water indicators. EDS data showed a greater threshold value of Ψ_stem (−1.8 MPa) in ‘Lapins’ sweet cherry trees than MDS (−1.4 MPa), which is more useful under broader water stress conditions and allows it to be a reliable indicator over a longer period, making it more useful for irrigation management.

Therefore, the combined use of daily MDS, and especially EDS values and their frequencies, which minimize the variability of water indicators, could be useful for evaluating the continuous monitoring of plant water status and establishing ranges that allow the development of deficit irrigation protocols and the optimization of water resources. However, the occurrence of ‘false positives’ during periods of severe water stress or changes in values from year to year should be considered as this may affect the interpretation of the data obtained.