Trunk water potential (Ψ_trunk) measured using microtensiometers has been recently reported as a reliable indicator of tree water status in apples (Lakso et al., 2022). This indicator was strongly related to the stem water potential measured with the pressure chamber, the reference indicator for assessing the water status of fruit trees (McCutchan and Shackel, 1992; Naor, 2000; Shackel, 2011). However, microtensiometers are not as responsive as the pressure chamber to detect fast changes in tree water status and can show a time lag (Lakso et al., 2022; Blanco and Kalcsits, 2023). In this experiment, all the indicators derived from Ψ_trunk followed a similar pattern ( Figure 2A ; Figure 3A ; Figure 4A ). Ψ_trunk rapidly changed after each irrigation event and midday and minimum daily Ψ_trunk increased by >0.4 MPa the next day, by 0.2 MPa for predawn Ψ_trunk, and by 0.3 MPa for the daily mean. Recently, Gonzalez Nieto et al. (2023a) reported similar midday Ψ_trunk for ‘Gala’ apples under water restrictions in New York State. In contrast, the predawn values reported for this study in ‘Gala’ were not as equally affected by water stress as the midday or minimum values and they were never lower than -0.4 MPa. In the present study, predawn Ψ_trunk was clearly affected by the drought cycling with values lower than -1.0 MPa ( Figure 2A ) the days previous to the irrigation events. These differences might be attributed to the more demanding environmental conditions recorded in this experiment in the semi-arid climate of Washington State compared with more temperate conditions in New York State and to the cultivar used for this experiment. ‘Honeycrisp’ cultivar has shown lower water potentials than those reported for ‘Gala’ under water restrictions and fully irrigated (Valverdi et al., 2019).

The minimum values for Ψ_trunk on July 11^th, 2022, before the irrigation was applied, were recorded during the afternoon (1700 – 1800 h, Figure 3A ), a similar time of the day to those recorded in pears and vines under water stress in Washington State and Australia (Blanco and Kalcsits, 2021; Pagay, 2022) and slightly later than those recorded in nectarine trees in Spain (Conesa et al., 2023). Two days later on July 13^th, 2022, the pattern was similar with maximum values at dawn and minimum values during the afternoon but with fewer negative values ( Figure 3B ). This rapid recovery of the trunk and stem water potential has been also observed in ‘Golden Delicious’/MM106 and ‘Gala’/G.11, with rapid changes of midday and minimum stem and trunk water potentials of more than 0.5 MPa (Doltra et al., 2007; Gonzalez Nieto et al., 2023a).

SF sensors recorded tree water use, with values that ranged between 2 and 4 L day^-1 ( Figure 4B ). Similarly, Bhusal et al. (2019) reported sap flow rates for the medium maturing cultivar ‘Hongro’ on M.9 rates of 3 L day^-1 for fully irrigated trees. However, they reported SF of 0.75 L day^-1 for trees under no irrigation for 50 days and midday stem water potential below -2.5 MPa. In this experiment, for ‘Honeycrisp’ apples, the minimum value of SF measured was closer to 1.5 L day^-1. For the days with midday trunk water potential below -1.7 MPa, it was observed a reduction in the rates of sap flow from values of 3.2 L day^-1 to below 2.0 L day^-1. These values were higher than those reported in ‘Fuji’ and ‘Golden Delicious’ under water restrictions (Liu et al., 2012; Bhusal et al., 2019). Moreover, it was also observed the strong effect that environmental conditions such as VPD, radiation, and air temperature had on tree transpiration rates (Liu et al., 2012). Thus, on July 2^nd, 2022, although the minimum Ψ_trunk was -1.5 MPa, the low solar radiation (below 23 MJ m^-2) and low evaporative demand (air temperature and VPD below 30°C and 3 kPa, respectively) caused the SF to decrease below 2.0 L day^-1 ( Figures 1 , 2B ).

SF was the highest in the early morning with no differences observed at dawn on days before and after irrigation (0.2 L h^-1). However, differences were first evident during the morning as the environmental conditions became more demanding. In the days before the irrigation, from dawn onwards, SF slowly decreased ( Figure 3C ). On the other hand, for the same tree in the days after the irrigation was applied the maximum daily values occurred between 0900 and 1100 h and then slowly decreased showing a sawtooth trend ( Figure 3D ). This pattern is similar to those reported in the cultivars ‘Braeburn’ (Green et al., 2003), ‘Fuji’ (Fernandez et al., 2008), ‘Nicoter’ (Ben Abdelkader et al., 2022), and ‘Mutsu’ (De Swaef et al., 2009). Rootstocks have also been reported to influence transpiration and SF (Cohen and Naor, 2002). Other apple cultivars grafted onto M.9, such as ‘Golden Delicious’ had similar SF values to those reported here, between 0.2 and 0.5 L h^-1, while for the same cultivar on a more vigorous rootstock, SF was consistently higher (Li et al., 2002).

Canopy temperature (T_c) is strongly affected by environmental conditions, so its use as an absolute value for assessing tree water status is not recommended (Idso et al., 1981; Figure 2C ). However, when accounting for ambient conditions (T_a), the use of thermal-based indices, such as the difference between canopy and air temperature (T_c-T_a) and the crop water stress index, can be reliable for assessing water stress in fruit trees (Gonzalez-Dugo et al., 2013; Ramirez-Cuesta et al., 2022; Blanco et al., 2023 The continuous monitoring of the T_c-T_a has been described as the “heartbeat” of the tree water status (Mira-Garcia et al., 2022). In apple trees, Gómez-Candón et al. (2022) stated that T_c-T_a is a sensitive indicator for many apple cultivars, and values between 1 and 2°C were related to midday stem water potentials between -1.6 and -1.8 MPa. Our results for ‘Honeycrisp’ followed a slightly different trend with T_c-T_a at midday between 0.5 and 2.5°C ( Figure 4C ) when the environmental conditions were highly demanding and those days previous to the irrigation events ( Figure 3E ) but with negative values of T_c-T_a at midday on the days after the trees were watered ( Figure 3F ) and on cloudy days with low evaporative demand, such as on July 2^nd when total solar radiation was below 25 MJ m ^-2 ( Figure 1 ). These results differ from those described for ‘Inored’ apples, in which positive values of T_c-T_a were recorded in trees with soil water content values close to field capacity and midday stem water potential of -1.2 MPa (Gómez-Candón et al., 2022). Midday T_c-T_a values higher than 2.5°C in ‘Honeycrisp’ apples occurred on days previous to the irrigation events, with minimum Ψ_trunk values similar to -2 MPa, and air temperature and VPD higher than 35°C and 4 kPa, respectively (July 19^th and 20^th, 2022; Figure 4C ). Similarly, the daily evolution of the T_c-T_a was affected by environmental demand and soil water availability. On July 11^th, 2022, (the day before the second irrigation event) the maximum positive difference between the canopy and air temperature was 2.9°C at midday and the most negative difference was -3.4°C during the night ( Figure 3E ). On July 13^th, 2022, the day after the second irrigation event, all the values of T_c-T_a recorded were negative, with the smallest difference at 1000 h, -0.4°C, and the largest at 1800h, - 5.6°C ( Figure 3F ). This pattern of negative T_c-T_a values during the complete day agrees with those evolutions reported in ‘Fuji’ apples under full irrigation and mild water stress by Osroosh et al. (2015) with values similar to zero at midday and a second peak during the afternoon. The difference between T_c and T_a at 1800 h should be related to Tőkei and Dunkel (2005) in apple trees, where increases in transpiration rates during the afternoon occurred as a result of the decrease of water stress and opening of stomata.

As expected, daily trunk diameter fluctuations followed a similar pattern to Ψ_trunk and were responsive to changes in both soil water content and environmental conditions ( Figure 2D ). MDS increased when the trees were under water stress going from 120 µm to 250 µm during the drought cycles and decreased after each irrigation event ( Figures 3C and 4D ). However, after reaching midday Ψ_trunk values of -1.8 MPa, MDS values did not continue to increase. Moreover, MDS values were also highly dependent on the environmental conditions so for consecutive days with similar values of minimum Ψ_trunk, MDS varied by more than 25% ( Figure 4D ). Similarly, Du et al. (2017) reported for ‘Golden Delicious’ apple trees that MDS was strongly affected by environmental conditions. Thus, these both factors highlight the limitations of using absolute MDS values as a unique tree water status indicator. That is why, to decrease the variability of the MDS it has been recommended to express it relative to the MDS of a reference tree (non-stressed) which under commercial conditions might not be always suitable (Naor and Cohen, 2003).

For the TGR, negative values were recorded when the trees were under mild to severe water stress ( Figure 2D ) (midday Ψ_trunk values ranging from -1.1 to -1.7 MPa) and, as such, this indicator is not suitable for quantifying water stress in mature ‘Honeycrisp’ apple trees. However, TGR was sensitive to identifying the irrigation events applied, showing increases in trunk diameter of more than 40 µm for the days following irrigation ( Figure 4D ). Similar results were also reported in young ‘Cox Orange Pippin’ apple trees (De Swaef et al., 2009). Blanco and Kalcsits (2023), reported that variations in trunk diameter immediately followed changes in trunk water potential in pear trees. Increases in trunk diameter observed immediately after irrigation match the results of Bonany et al. (2000) in potted trees from the combination ‘Golden Delicious’/M.9. They suggested that young trees and trees on dwarfing rootstocks rapidly use and refill water stored in the trunk as a water source to maintain transpiration and fruit growth depending on the soil water content. In contrast, mature vigorous trees with a greater root volume such as ‘Golden Delicious’/MM106 do not follow this behavior (Doltra et al., 2007). As with the MDS, TGR values also showed a high variability on days with similar values of Ψ_trunk ( Figure 4D ), and it has been reported that TGR can vary significantly between cultivars, and rootstocks, and depend on factors such as tree vigor, crop load and age (Ortuño et al., 2010).

Daily FRGR was positive for most of the days of the experiment. Greater rates were recorded after irrigation, and when the evaporative demand was not excessively high (VPD < 3 kPa), while values close to 0 mm were recorded on days with ET₀ values higher than 6 mm d^-1. The daily FDV equaled the daily absolute growth since the maximum fruit diameter of the day generally matched the minimum fruit diameter of the next day ( Figures 3I, J ). Both, FRGR and FDV showed a similar trend to the TGR, emphasizing how both fruit growth and trunk water storage recovered when water supply increased ( Figure 4E ). The decrease in the daily FRGR and the FDV could be an early water status indicator to detect water deficit conditions for the trees ( Figure 2E ). Similar sensitivity of apple fruit diameter to water deficit has been reported in several cultivar and rootstock combinations as well as a range of environmental conditions (Morandi et al., 2017; Gonzalez Nieto et al., 2023a). Fruit growth occurred between evening and early morning for days before and after irrigation. No growth or a slight shrinking of the diameter was detected during midday or afternoon. These results agree with those reported by Boini et al. (2019) for ‘Imperial Gala’. When changes in fruit diameter were transformed into changes in mass (for ‘Honeycrisp’ apples, Diameter > 25 mm; Fruit Mass (g) = 0.07[Fruit Diameter (mm)]² – 3.63[Fruit Diameter (mm)] + 59.41; Kalcsits et al., 2017), it was observed that the fruit growth was almost twice higher the day after the irrigation (0.7 mm, which equals ≈ 1.36 g fruit day^-1; Figure 3J ) than the day before (0.4 mm ≈ 0.69 g fruit day^-1; Figure 3I ).

For ‘Honeycrisp’ apples, Ψ_trunk daily range showed the strongest and the most significant relationship with the SF, T_c-T_a, and MDS (based on a correlation analysis) of all the water stress indicators derived from Ψ_trunk, followed by minimum and midday Ψ_trunk which had similar results ( Figure 5 ).

Predawn Ψ_trunk did not show any significant relationship with SF, T_c-T_a, or MDS (p-values > 0.05). Several authors have also reported poor relationships between the predawn stem water potential and SF or MDS in apples, walnuts, and vines (Améglio et al., 1999; Intrigliolo and Castel, 2007; Liu et al., 2011). However, strong relationships have been reported in other tree fruit such as peaches and sweet cherries (Fereres et al., 1999; Livellara et al., 2011).

Based on linear regression analysis ( Figure 5 ), the relationship between the indicators derived from the trunk water potential and MDS, SF and T_c-T_a followed a polynomial pattern ( Table 1 ). Thus, MDS reached its maximum values when midday and minimum Ψ_trunk were -1.8 and -2.0 MPa, respectively for ‘Honeycrisp’ apples. After that point, more negative tree water potentials were not related to greater MDS ( Figure 5 ). This limitation has been previously reported in other fruit trees such as citrus, olive, stone fruits, and vines (Goldhamer et al., 1999; Ortuño et al., 2010; Blanco et al., 2018).

T_c-T_a was more closely related to the daily range of Ψ_trunk than to the minimum and midday Ψ_trunk ( Table 1 ). Ψ_trunk was the lowest during the afternoon when T_a was the highest. However, the greatest difference between T_c and T_a was often recorded at midday. In almond trees, Gonzalez-Dugo et al. (2012) also reported a strong relationship, which followed a second-degree polynomial function, between midday stem water potential and T_c-T_a measured in the afternoon. A similar second-degree relationship between the stem water potential and thermal-based indicators has been reported in pear trees under similar environmental conditions (Blanco et al., 2023).

The relationship between midday Ψ_trunk and SF was parabolic ( Table 1 ) with maximum values, above 4 L day^-1, in the range between -1.2 and -1.7 MPa, and with values below 2 L day^-1 when the trees were under severe water stress (< -2.0 MPa) or under nondemanding atmospheric conditions (> -0.9 MPa). De Swaef et al. (2009) reported a similar trend for ‘Mutsu’ apples, however the threshold value observed was -1.4 MPa, higher than that in this study. In grapevines, Patakas et al. (2005) reported that SF was 50% lower when the vines were under slight water stress.

Among the widely studied continuous indicators considered, MDS showed the strongest and most statistically significant relationship with Ψ_trunk, T_c-T_a ranked second, and SF was last ( Figure 5 ). The strong relationship between Ψ_trunk and MDS might be due to the ability of both indicators to rapidly detect changes in the tree water status (Conejero et al., 2007; De Swaef et al., 2009). However, when the tree was under mid- to severe water deficit (< -1.8 MPa), Ψ_trunk was still able to detect and quantify water stress (until values below -2.5 MPa) while the MDS could not identify a situation of severe water stress. Ψ_trunk overcame the limitations of MDS and did not show any threshold limit to detect severe water stress in the range between -0.2 and -2.5 MPa. However, some limitations that have been reported to affect MDS such as the age of the tree, the crop load, or the phenological stage of the tree (Fernandez and Cuevas, 2010), might also affect Ψ_trunk. Similarly, T_c-T_a and SF have been reported to be strongly related to environmental conditions and were not recommended to detect slight water deficits or rapid physiological changes in response to water deficits (Ortuño et al., 2006; Mira-Garcia et al., 2022; Blanco et al., 2023).

The similar results obtained with the relationships between the SF, T_c-T_a, MDS, and TGR, and the midday and the minimum Ψ_trunk are explained since both (midday Ψ_trunk and minimum Ψ_trunk) were strongly related (Ψ_trunk min = 1.18Ψ_trunk md + 0.06; R² = 0.88). Predawn and midday Ψ_trunk were also strongly correlated (Ψ_trunk md = 1.07Ψ_trunk pd - 0.94; R² = 0.74). Daily maximum and minimum values of tree water potential, which correspond to predawn and midday -early afternoon respectively, have traditionally been used to determine tree water status in fruit trees (baselines) and threshold values to manage irrigation (Améglio et al., 1999; Naor et al, 1995; Shackel et al., 2021). However, the ability to continuously monitor Ψ_trunk with the microtensiometers provides an opportunity to explore other water potential-based indicators such as the daily mean and the daily range of Ψ_trunk that can integrate the tree water status for the whole day. These two indicators have not been widely explored when measuring the stem or leaf water potentials, but have the potential to be included in automated irrigation systems and are not as vulnerable as the midday Ψ_trunk to be affected by time lag problems. Both indicators, the daily mean and the daily range of Ψ_trunk, were more related to the midday and minimum Ψ_trunk than to the predawn Ψ_trunk ( Figure 6 ). The present work is the first report that assesses their adequacy as tree water status indicators in apple trees.

Since fruit is the real target for growers, direct, continuous monitoring of fruit growth should be compared with the proposed tree water status indicators. The variability found among fruits for FRGR and FDV was similar to the variability reported in apple size within fruits from the same tree (Kalcsits et al., 2019).

The indices derived from the fruit growth were more related to the Ψ_trunk than to other tree water status indicators such as the MDS, SF, or T_c-T_a. According to our results, the daily mean of Ψ_trunk followed by the midday and minimum Ψ_trunk were the indicators derived from the trunk water potential that most closely corresponded to fruit growth rates ( Table 2 ). Similarly, Boini et al. (2019) reported a strong relationship between midday stem water potential and changes in the FGR of ‘Gala’ apples. In this sense, recently, Gonzalez Nieto et al. (2023a) have developed the first logistic model that successfully relates Ψ_trunk with fruit growth for the cultivar ‘Gala’ in New York State.

When trunk and fruit diameters were compared, there was a stronger relationship with the TGR than with the MDS, although there was a trend of lower FRGR when MDS increased. Bonany et al. (2000) described a linear relationship between MDS and FGR of young ‘Golden Delicious’ apples. However, that equation cannot be applied to our experiment because although the water potentials recorded in both experiments were similar, daily minimum values in the range between -0.7 and -2.2 MPa, MDS was larger, and RFGR were smaller. More work is needed to tune the relationship between the tree water status indicators and fruit growth during the full season and to assess for different cultivar/rootstock combinations how continuous monitoring of trunk water status can improve fruit yield and enhance fruit quality.

Table 3 reports the sensitivity analysis of continuous tree water status indicators that were assessed. The index with the highest SI was T_c-T_a (SI = 2.77) but the high CV decreased its sensitivity. Midday Ψ_trunk and the daily range of Ψ_trunk were the tree water status indicators with the highest sensitivity (S > 20) followed by the daily mean of Ψ_trunk (S = 17.53). Among the indicators derived from Ψ_trunk, all showed similar SI, with the highest value observed for the daily mean Ψ_trunk. However, the lowest CV was observed for midday Ψ_trunk which consequently increased its S. MDS had a high S, similar to the minimum Ψ_trunk, while the highest CV and the lowest S were calculated for the TGR and the FRGR (S < 1). The CV values obtained for the MDS and the SF in ‘Honeycrisp’ apples were slightly lower than those reported by Wheeler et al. (2023) for the same indicators in ‘Aztec Fuji’ and ‘Escilate’, however, they followed a similar trend with the highest variability for SF, followed by the MDS and with the lowest variability for the stem water potential (0.09 – 0.11). Regarding the trunk water potential, similar results were reported by Conesa et al. (2023) who also highlighted the high sensitivity and low CV of Ψ_trunk.

Therefore, the order proposed for the indicators according to their sensitivity was: midday Ψ_trunk = daily range of Ψ_trunk > daily mean Ψ_trunk > MDS = minimum Ψ_trunk > predawn Ψ_trunk > T_c-T_a = SF > FDV > FRGR = TGR. Regarding the order of the indices derived from Ψ_trunk based on their relationship with the widely studied, traditional, continuous indicators, the general trend followed: daily range of Ψ_trunk > minimum Ψ_trunk = midday Ψ_trunk > daily mean Ψ_trunk > predawn Ψ_trunk ( Figure 7 ). On the other hand, the daily mean Ψ_trunk and the minimum value of Ψ_trunk were the indicators more related to those indices derived from fruit growth.