Rootstocks with Different Vigor Influenced Scion–Water Relations and Stress Responses in AmbrosiaTM Apple Trees (Malus Domestica var. Ambrosia)

In recent years, awareness has been raised around the benefits of diversifying rootstocks, in order to enhance tree health and sustain apple fruit production under the influence of climate change. However, performances of many rootstocks under stresses remain unclear. This study aimed to set the first step towards a much-needed comprehensive evaluation on water relationships and stress responses of scion–rootstock systems for the popular apple cultivar AmbrosiaTM (Malus domestica var. Ambrosia), because its production and horticultural knowledge had been largely limited to the Malling 9 rootstock (M.9). Five rootstocks were evaluated in a greenhouse water deficit experiment and at the onset of heat stress in a field trial in Summerland, British Columbia, Canada. Multiple stress indicators of leaves and fruits were analyzed to elucidate water use strategies and drought resistance mechanisms. The rootstocks led to differences in scion vigor, and stomatal and photosynthetic characteristics. The largest semi-dwarfing Geneva 202 (G.202) demonstrated more water use and higher stress susceptibility. Large dwarfing Geneva 935 (G.935) and Malling 26 (M.26) showed more stringent stomatal control and reduced water use under stresses, typical of a drought-avoidance strategy. The smallest large dwarfing M.9NIC29® and the small dwarfing Budagovsky 9 (B.9) led to smaller and denser stomata. B.9 demonstrated the most stable water status and drought tolerance. The study suggested that scion stress responses were influenced by rootstock vigor and tree water use strategies. It implied the necessity of vigor-specific irrigation management for alleviating stresses and achieving production goals of different rootstocks.


Introduction
In apple horticulture, the use of rootstock is an effective approach to manage tree growth and yield, to improve resource use efficiency, and to confer scion with desirable traits of resilience against a variety of abiotic and biotic stressors [1]. Limited water transporting capacity in semi-dwarfing and dwarfing rootstocks are often associated with lower root vigor, lower root partitioning of dry matter, smaller numbers of coarse roots and fine roots [2], smaller vessels and lower root hydraulic conductance [3], and thicker root bark [4]. Water transport in the scion-rootstock system may be further restricted by xylem dysfunction and increased resistivity of the graft union [5]. Such limitations constrain scion hydraulics and tree growth, by the well-known mechanism of altering the root-to-shoot hydraulic signals. Other mechanisms, such as interfered systemic chemical communications via endogenous growth regulators, nutrients, RNAs and proteins, are less understood [6][7][8]. Abundant studies have shown that different rootstocks can lead to variations in scion-water relationships, stomatal development and regulation, and xylem hydraulic properties in a number of apple cultivars [3,[9][10][11][12][13]. Although rootstocks with different vigor and water transport capacity are assumed to alter the water demand,  (Table 1). The leaf gas exchanges were assessed during the soil water depletion. On Day 1, M.9, G.202 and G.935 demonstrated higher net photosynthetic rates (P n ) than M.26 and B.9 (Figure 1). On Day 4 and Day 7, M.26 showed lower P n than the other rootstocks. From Day 1 to Day 4, none of the rootstocks showed significant decreases in P n . From Day 1 to Day 7, the P n decreased in M.26 and G.202, but it did not change significantly in others. The maximum quantum efficiency of CO 2 assimilation (ΦCO 2 ) also declined the most in G.202 (∆ΦCO 2 at −5.37 mmol CO 2 mol −1 absorbed quanta), followed by M.26 (−2.59 mmol CO 2 mol −1 absorbed quanta), while the extent of decrease in other rootstocks was significantly less (Table 1). Rehydration led to a further decrease in P n in M.9 and G.935, but not in M.26, B.9 or G.202. Overall, during soil water depletion, M.26 demonstrated the lowest P n , whereas B.9 showed the most stable P n ( Figure 1) and the least decrease in ΦCO 2 (Table 1). Table 1. Changes in instantaneous water use efficiency (WUE i ), stomatal conductance (g s ) and maximum quantum efficiency of CO 2 assimilation (ΦCO 2 ) of Ambrosia TM scion on five rootstocks, along with the decline in soil volumetric water content (VWC) from Day 1 to Day 7 after irrigation in the greenhouse experiment. Transpiration rate (T r ) changed in a pattern similar to P n . On Day 1, G.202 and G.935 demonstrated a higher T r than M.9 and B.9, and M.26 had the lowest T r (Figure 2). From Day 1 to Day 4, T r slightly increased in all the rootstocks, however the change was not significant. From Day 1 to Day 7, T r declined in G.202, but did not show significant change in others ( Figure 2). Stomatal conductance (g s ) declined the most in G.202, followed by M.26, G. 935, and M.9. On the contrary, ∆g s suggested an increase in B.9 despite its statistical insignificance (Table 1). Rehydration led to a further decrease in T r in M.9; however, it did not have any significant impact on others. During the dehydration-rehydration cycle, T r in M.9, M.26 and G.935 significantly deceased from Day 4 to one day after rehydration.
Overall, M.26 demonstrated the lowest T r , suggesting the least water use, which was consistent with its least ∆VWC in soil (Table 1). B.9 showed a relatively low but the most stable T r during soil water depletion. By Day 7, instantaneous water use efficiency (WUE i ) increased the most in G.202, followed by G.935 and M.9 (Table 1). WUE i declined more in M.26 than in B.9. change significantly in others. The maximum quantum efficiency of CO2 assimilation (ΦCO2) also declined the most in G.202 (ΔΦCO2 at −5.37 mmol CO2 mol⁻¹ absorbed quanta), followed by M.26 (−2.59 mmol CO2 mol⁻¹ absorbed quanta), while the extent of decrease in other rootstocks was significantly less (Table 1). Rehydration led to a further decrease in Pn in M.9 and G.935, but not in M.26, B.9 or G.202. Overall, during soil water depletion, M.26 demonstrated the lowest Pn, whereas B.9 showed the most stable Pn (Figure 1) and the least decrease in ΦCO2 (Table 1).

Figure 1.
Leaf net photosynthetic rate of Ambrosia TM scion on five rootstocks in the greenhouse water deficit experiment. Bars and error bars are the mean and standard error (n = 6). Different uppercase letters in the same color indicate significant differences amongst rootstocks on each day (red, blue, and green for Day 1, Day 4 and Day 7 after irrigation, respectively; black for one day after rehydration) (P ≤ 0.05, ANOVA, Fisher LSD pairwise comparisons). Different lowercase letters in each rootstock indicate significant day-to-day variations (P ≤ 0.05, ANOVA, Fisher LSD Pairwise Comparisons).
Transpiration rate (Tr) changed in a pattern similar to Pn. On Day 1, G.202 and G.935 demonstrated a higher Tr than M.9 and B.9, and M.26 had the lowest Tr ( Figure 2). From Day 1 to Day 4, Tr slightly increased in all the rootstocks, however the change was not significant. From Day 1 to Day 7, Tr declined in G.202, but did not show significant change in others ( Figure 2). Stomatal conductance (gs) declined the most in G.202, followed by Figure 1. Leaf net photosynthetic rate of Ambrosia TM scion on five rootstocks in the greenhouse water deficit experiment. Bars and error bars are the mean and standard error (n = 6). Different uppercase letters in the same color indicate significant differences amongst rootstocks on each day (red, blue, and green for Day 1, Day 4 and Day 7 after irrigation, respectively; black for one day after rehydration) (P ≤ 0.05, ANOVA, Fisher LSD pairwise comparisons). Different lowercase letters in each rootstock indicate significant day-to-day variations (P ≤ 0.05, ANOVA, Fisher LSD Pairwise Comparisons). The stomatal density and size of Ambrosia TM leaves showed variations amongst rootstocks ( Figure 3). Stomata of higher density ( Figure 3A) and smaller size ( Figure 3B) were found in M.9 (341 counts per mm 2 of leaf area, 280.1 µm 2 in size on average) and G.935

Stomatal Characteristics
The stomatal density and size of Ambrosia TM leaves showed variations amongst rootstocks ( Figure 3). Stomata of higher density ( Figure 3A) and smaller size ( Figure 3B) were found in M.9 (341 counts per mm 2 of leaf area, 280.1 µm 2 in size on average) and G.935 (345 counts per mm 2 , 275.6 µm 2 ). In G.202, the stomata had the lowest density (306 counts per mm 2 ) and the largest size (308.3 µm 2 ). Compared to M.26 (323 counts per mm 2 , 293.3 µm 2 ), B.9 had similar medium-sized stomata (295.9 µm 2 ), but at a significantly higher density (347 counts per mm 2 ), which was approximate to M.9 and G.935. Area of stomata per leaf area was not significantly different amongst rootstocks, however the median of this parameter was the highest in B.9 (0.101 mm 2 /mm 2 ) and the lowest in M.26 (0.093 mm 2 /mm 2 ) ( Figure 3C).   The absence of letters means that there were no significant differences.

Leaf Defoliation, Chlorosis and Scion Trunk Growth
Soil water depletion caused defoliation. The intensity of defoliation varied amongst rootstocks. The most severe defoliation occurred in G.202, followed by M.26 and G.935; M.9 and B.9 defoliated the least ( Figure 4A). G.202 demonstrated higher leaf chlorophyll concentration in the remaining leaves, whereas M.26 showed larger tree-to-tree variation ( Figure 4B).

Scion Trunk Growth, Yield and Fruit Quality
By the end of the growing season in the field in 2020, G.202 showed the largest TCSA and the highest scion vigor, followed by G.935, M.26 and M.9 which demonstrated similar scion vigor. The TCSA in B.9 was significantly smaller than others ( Figure 5B). Except for B.9, the rootstocks had larger TCSA after the growing season in the field than after fourmonths growth in the greenhouse ( Figure 5A,B). Yield efficiency was the highest in B.9, followed by M.9, M.26 and G.935, and was the lowest in G.202 (Table 2). Leaf chlorophyll concentration was also lower in G.202 than in others (Table 2). There was no symptomatic defoliation in any rootstock.
Rootstocks did not lead to significant differences in fruit weight ( Figure 6A) or dry matter weight ( Figure 6B) (P ≤ 0.05). However, fruit weight in G.202 was marginally lower than in G.935 and M.26 ( Figure 6A); G.935 had marginally higher fruit dry matter weight

Scion Trunk Growth, Yield and Fruit Quality
By the end of the growing season in the field in 2020, G.202 showed the largest TCSA and the highest scion vigor, followed by G.935, M.26 and M.9 which demonstrated similar scion vigor. The TCSA in B.9 was significantly smaller than others ( Figure 5B). Except for B.9, the rootstocks had larger TCSA after the growing season in the field than after four-months growth in the greenhouse ( Figure 5A,B). Yield efficiency was the highest in B.9, followed by M.9, M.26 and G.935, and was the lowest in G.202 (Table 2). Leaf chlorophyll concentration was also lower in G.202 than in others (Table 2). There was no symptomatic defoliation in any rootstock. Rootstocks did not lead to significant differences in fruit weight ( Figure 6A) or dry matter weight ( Figure 6B) (P ≤ 0.05). However, fruit weight in G.202 was marginally lower than in G.935 and M.26 ( Figure 6A); G.935 had marginally higher fruit dry matter weight with more fruit-to-fruit consistency ( Figure 6B). After one month of air storage at 4 • C, the fruits of M.26, G.202 and G.935 had higher soluble solid contents (SSC, • Brix %) than M.9 and B.9 ( Figure 6C). Sugar: acidity ratio was the highest in G.935, followed by M.26, G.202 and B.9, and was the lowest in M.9 ( Figure 6D). Tissue water potential of fruit hypanthium (Ψ fruit ) was higher in M.9 and B.9 than in others ( Figure 6E). Fruit firmness was also higher in M.9 and B.9, followed by G.202 and G.935, and was the lowest in M.26 ( Figure 6F). Fruit mineral nutrients varied amongst rootstocks, except zinc ( Table 2). Fruits of G.935 had higher nitrogen, phosphorous, and boron, but lower potassium, magnesium, and calcium, whereas contrasting compositions were found in the fruits of B.9 (Table 2).

Water Relationships in Response to the Onset of Heat Stress
At the onset of heat stress in August 2020, G.202 was the lowest in P n , electron transport rate (ETR) ( Figure 7A), and variable fluorescence: maximum fluorescence ratio at steady-state photosynthesis lighting conditions (Fv'/Fm') ( Figure 7B). WUE i was not significantly different amongst rootstocks (P ≤ 0.05), however G.202 showed slightly higher WUE i with large tree-to-tree variations ( Figure 7C). Leaf P n ( Figure 7A) and g s ( Figure 7D) were the highest in M.9 and B.9, followed by G.935 and M.26. Except for G.202, rootstocks showed higher P n in the field than in the greenhouse (Figure 1), as a consequence of adapting to more abundant ambient photosynthetically active radiation (PAR).
Leaves on vegetative branches had higher T r than those on fruiting spurs ( Figure 8A). Leaf T r on vegetative branches was the highest in M.9, followed by B.9 and G.935, and was the lowest in M.26 and G.202; higher leaf T r corresponded to higher Ψ fruit ( Figure 8B, r 2 = 0.89). On fruiting spurs, higher leaf T r corresponded to a lower surface temperature (T surface ) of adjacent sunlit fruits ( Figure 8C, r 2 = 0.74); however, rootstock difference was not significant. Average leaf T surface was correlated with fruit T surface ( Figure 8D, r 2 = 0.94); leaf T surface was lower in B.9 than in M.26 (P = 0.04) and G.202 (P = 0.06).
with more fruit-to-fruit consistency ( Figure 6B). After one month of air storage at 4 °C, the fruits of M.26, G.202 and G.935 had higher soluble solid contents (SSC, °Brix %) than M.9 and B.9 ( Figure 6C). Sugar: acidity ratio was the highest in G.935, followed by M.26, G.202 and B.9, and was the lowest in M.9 ( Figure 6D). Tissue water potential of fruit hypanthium (Ψfruit) was higher in M.9 and B.9 than in others ( Figure 6E). Fruit firmness was also higher in M.9 and B.9, followed by G.202 and G.935, and was the lowest in M.26 ( Figure 6F). Fruit mineral nutrients varied amongst rootstocks, except zinc ( Table 2). Fruits of G.935 had higher nitrogen, phosphorous, and boron, but lower potassium, magnesium, and calcium, whereas contrasting compositions were found in the fruits of B.9 (Table 2).

Rootstock Performances and Water Use Strategies
Semi-dwarfing and dwarfing rootstocks are commonly used to control scion vigor and alter yield efficiency in high-density plantings. Semi-dwarfing and dwarfing rootstocks usually produce apple trees that are about 60−90% and 30% to 60% of the size of seeded trees, respectively. Based on final trunk diameter and tree height, dwarfing rootstocks are further categorized into large, moderate, and small. In the field trial, the examined rootstocks rendered different scion trunk growth in Ambrosia TM apple, as expected for their vigor classifications ( Figure 5B). The semi-dwarfing G.202 had the largest TCSA, Larger rootstocks lead to lower yield efficiency, and vice versa. In the field trial, the semi-dwarfing G.202 led to the lowest yield efficiency, the lowest leaf chlorophyll concentration, and lower photosynthetic capacity ( Figure 7A) and g s ( Figure 7D, Table 2). Firstly, it could indicate that as more resources were allocated to trunk growth, depleted nutrients might impair leaf function in chlorophyll synthesis and stomatal regulation. Secondly, lower leaf gas exchanges could suggest that its scion was exposed to water deficit, possibly caused by a hydraulic imbalance between the scion and the rootstock. Such imbalance often becomes more significant when the discrepancy between the scion demand and the rootstock supply aggravates. In addition, G.202 produced fruits with lower weight ( Figure 6A), higher soluble solid content ( Figure 6C), and lower Ψ fruit ( Figure 6E), which implied seasonal water deficit. Higher stomatal density was reported in the leaves of more vigorous rootstocks [20,21]; however, in the greenhouse experiment, G.202 led to the largest stomatal size but the lowest density in Ambrosia TM scion leaves. This suggested that while higher stomatal density of rootstock leaves may be a good indicator for stronger vigor of rootstock itself, stomatal density of scion leaves may not show such correlation. Interestingly, the stomatal characteristics of G.202 were associated with the greatest decrease in ΦCO 2 and g s , the greatest increase in WUE i , and the most severe defoliation by the end of the water deficit cycle (Table 1). Larger stomata might be prone to stomatal surplus under the medium level of photosynthetically active radiation (PAR) and in well-watered conditions, when the excessive increase in stomatal aperture and transpirational water loss were no longer coupled with photosynthetic gain. This could explain the concurrence of the decrease in g s and the increase in WUE i in G.202 when the soil water was depleted (Table 1). G.202 had similar P n under PAR = 800 µmol photon m −2 s −1 in the greenhouse and under PAR = 1600 µmol photon m −2 s −1 in the field, suggesting that following defoliation, the remaining leaves photosynthesized at high capacity under relatively low PAR in the greenhouse. The efficiency of such adjustment is debatable, because maintaining older leaves as a photosynthate source during stress recovery may be a more valuable strategy rather than mobilizing nutrients into sink tissue during stress [22].

Large Dwarfing G.935, M.26 and M.9
In the three large dwarfing rootstocks with intermediate trunk growth and yield efficiency, M.26 demonstrated the lowest leaf gas exchange rates (Figures 7A and 8A) and the lowest Ψ fruit in the field trial ( Figure 6E). This was consistent with its low P n and T r under both well-watered and water deficit conditions, and its least soil water depletion by the end of dehydration cycle in the greenhouse experiment (Figures 1 and 2, Table 1). It suggested that compared to M.9 and G.935, M.26 had lower water use, possibly attributed to its lower stomatal density and smaller stomatal area per leaf area ( Figure 3A,C). By the end of the water deficit cycle, ΦCO 2 , g s and WUE i declined the most in M.26 amongst the three large dwarfing rootstocks (Table 1). There was also significant defoliation, suggesting severe water stress ( Figure 4A).
G.935 demonstrated modest water status and was slightly more vigorous than M.26 and M.9. Its fruits demonstrated several desirable quality attributes ( Figure 6), such as higher and more consistent fruit weight and dry matter, and a higher sugar: acidity ratio-a parameter relevant to the sensory perception of apples [23,24]. Moderate scion growth and good fruit attributes suggested a well-balanced carbohydrate allocation between vegetative growth and fruit development.
M.9 was a slightly less vigorous large dwarfing rootstock, although produced significantly larger trees than the small dwarfing B.9 ( Figure 5B). Similar to B.9, the leaves on M.9 had stomata of a smaller size and higher density in the greenhouse (Figure 3). This could be attributed to the restricted rootstock water supply. Similarly, a study on the impact of deficit irrigation showed that water limitation led to higher stomatal density in Golden Delicious on EM-II rootstock [25]. In the field trial, leaf gas exchanges as well as the T surface of leaves and fruits (Figures 7 and 8) at the onset of heat suggested a better water status in M.9 than in the other large dwarfing rootstocks. Higher Ψ fruit , and moderate fruit size and dry matter (Figure 6), indicated that M.9 was likely exposed to less seasonal water stress.

Small Dwarfing B.9
The small dwarfing B.9 led to the slowest scion growth ( Figure 5). During the greenhouse water-deficit cycle, it demonstrated the most stable leaf gas exchanges (Figures 1  and 2) and the least defoliation (Figure 4), indicating more drought tolerance. This could be attributed to a lower hydraulic demand by a smaller scion accompanied with high stomatal density and moderate stomatal size (Figure 3). In the field trial, similar to M.9, B.9 showed high leaf gas exchanges as well as a low T surface of leaves and fruits at the onset of heat stress (Figures 7 and 8), and higher Ψ fruit after storage, which suggested a good tree water status. Fruits on the two smallest rootstocks had higher firmness and a significantly lower sugar: acidity ratio, indicating the delayed maturation.

Drought Susceptibility versus Drought Resistance
Rootstocks influenced scion-water relationships, nutrient levels, and yield efficiency in Ambrosia TM apple. This suggested that dwarfing intensities led to different water use strategies. Larger rootstocks, such as G.202, support more vigorous scion growth and higher yield potential, at a cost of more water use. This water use strategy is associated with wider scion xylem vessels and larger stomata. Such hydraulic characteristics may make the trees more susceptible to water deficit and heat stress. Under stresses, the overall leaf function declines, shown as chlorosis, reduction in g s and leaf gas exchanges, or the maintenance of P n and chlorophyll concentration in limited leaves at the cost of defoliation, as observed in G.202 (Figures 1 and 4). Therefore, good water management needs to be implemented to ensure tree health and fruit quality for such rootstocks. On the contrary, smaller rootstocks impose more hydraulic restriction on scions, leading to slower scion growth and lower yield potential; these rootstocks have less water demand, and consequently, can be more resistant to water deficit.

Drought Tolerance and Avoidance
The photosynthesis and transpiration of the studied rootstocks responded differently to soil water depletion and to the onset of heat stress, which indicated different mechanisms of stress resistance, i.e., tolerance and avoidance [26]. Drought tolerance is often associated with more resilient xylem transport and more flexible stomatal control; at the onset of stress, trees are able to maintain moderate and stable water use. M.9 and B.9 demonstrated such characteristics in this study. Higher stomatal density and smaller stomatal size enabled the leaves to fine-tune the stomatal regulation and to maintain g s at a certain level for carbon assimilation and heat dissipation during stress (Figures 7 and 8). This pinpointed the importance of understanding stomatal characteristics for predicting water use strategies and stress responses [27,28].
Drought-avoidant trees tend to avoid stress by decreasing water demand, via lowering g s or reducing leaf area. This mechanism was observed in M.26 and G.935, shown as a reduction in leaf gas exchanges and g s , defoliation, lower stomatal area per leaf area in the water deficit experiment (Figures 1-4), and relatively low photosynthesis, transpiration, g s and Ψ fruit in the field trial (Figures 6-8). Due to the decrease in transpirational cooling, fruits on these rootstocks could have higher T surface (Figure 8) and be exposed to more heat injury. The knowledge of different water use strategies and stress response mechanisms can facilitate the decision-making in rootstock selection and in irrigation management in accordance with the water use of specific rootstocks for stress alleviation and yield potential attainment.

Stress Indicators for Ambrosia TM Scion-Rootstock System
To reveal water use strategies and stress response mechanisms of different scionrootstock systems, it often requires a comprehensive evaluation of tree-water relationships under a gradient of stresses, which involves the interpretation of multiple physiological parameters. In this study, stress indicators of leaves and fruits were explored. Their sensitivity to rootstocks and to the onset of stresses were assessed, based on their responsiveness and on their causal relationships (Table 3). Photosynthetic and transpirational capacities varied in rootstocks and in responses to stresses. Both leaf gas exchange rates and g s , and ETR and Fv'/Fm', sensitively detected the differential rootstock responses to the onset of heat stress. Strong correlations existed amongst P n , ETR, Fv'/Fm', T r and g s . The ∆ΦCO 2 was highly correlated with ∆g s in the five examined rootstocks (Table 1; r 2 = 0.98). This showed that the decrease in g s under water deficit became a restricting factor for the quantum efficiency of CO 2 assimilation in Ambrosia TM apple leaves. These measurements can generate reliable information quickly and non-destructively, for early assessment on rootstock water use strategies: (1) large rootstocks that demand higher water use tend to demonstrate high leaf gas exchange rates under good water conditions, however their photosynthetic capacity and g s rapidly decline to a much lower, symptomatic level under water deficit, which suggests drought susceptibility. Such susceptibility can expose the trees to more subsequent heat stress, due to the compromised transpirational cooling effect, which is often observed as a concurrence of decreased T r and increased average tissue T surface ; (2) drought-avoidant scion-rootstock systems typically reduce g s at the onset of water deficit, to conserve water and to continue leaf gas exchanges at a lower level; (3) drought-tolerant rootstocks support less vigor and demand less water. They are capable of maintaining leaf gas exchanges at a moderate and steady level; their relatively high ETR and Fv'/Fm' indicate good water status at the onset of presumed stresses. Other traits of tolerance include the lower T surface of leaves and fruits, higher Ψ fruit , and less extent of leaf chlorosis and defoliation. Lower scion hydraulic conductance and elevated abscisic acid may be involved in the better adaptation of smaller scions to water-limited environments [29].
Calculated based on instantaneous carbon gain P n against water loss T r , WUE i should be interpreted carefully with reference to other parameters such as P n , g s , and longer-term WUE [30]. Increased WUE can be a consequence of either enhanced photosynthetic carbon gain, or reduced water loss caused by various factors, and therefore, it does not necessarily correspond to higher yield potential. Increased WUE at the cost of significantly reduced water use can be associated with decreased yield [31], as in drought-susceptible and drought-avoidant crops under water-limited conditions. On the contrary, scion-rootstock systems with drought tolerance tend to maximize soil moisture use; higher g s and lower WUE may associate with higher carbon gain and higher yield potential, given that soil moisture stays away from extreme deficits.
Leaf T r on vegetative branches was higher than that of on fruiting spurs, likely due to less water availability in fruiting spur leaves due to fruit-leaf competition for water. The leaves on vegetative branches were more responsive to the onset of heat stress ( Figure 8A); average T surface showed more variation amongst rootstocks than T surface of directly sunlit tissues did. This suggested that the average T surface of leaves on vegetative branches could be a more accurate indicator for the tree water status. However, it showed less sensitivity and more leaf-to-leaf variation than leaf gas exchanges, ETR and Fv'/Fm', which made it a less effective stress indicator. A similar insensitivity of leaf temperature was also reported in a plant-water relationship study on young peach trees [32].

Water Potential and Quality Attributes of Fruits
Mid-day stem water potential has been widely used in plant water status assessment since the invention of the hydrostatic pressure method [33]. In fruit trees, it is well correlated with leaf transpiration [34]. However, the data collection involves destructive sampling and subjective judgement by the operator of the pressure chamber. The optimal time window for measurement is narrow, due to its strong diurnal pattern under the impact of leaf transpirational water loss. In comparison, Ψ fruit varies less diurnally because of the relatively low surface conductance and transpirational rate in fruits. Its main component, the osmotic potential, is determined by the number of soluble solids and water availability in fruits. Previous studies showed it was an important parameter in seasonal fruit water relations, which could demonstrate the effects of irrigation and crop load on fruit and tree water status [35,36]. In this study, Ψ fruit was negatively correlated with fruit SSC, and positively correlated with leaf T r ( Figure 8B), which suggested that it could be an effective water status indicator. Fruit weight, size, and dry matter accumulation are largely determined by fruit expansion and carbohydrate allocation, both impacted by water availability [37]. In turn, the variations in fruit weight and dry matter can reflect the changes in seasonal water availability. The concurrence of lower fruit weight and higher DM% or SSC% (i.e., less water content %), often indicates the reduction in fruit expansion and carbohydrate accumulation caused by seasonal water deficit (Table 3).

Summary and Future Perspectives
Ambrosia TM apples on semi-dwarfing, large dwarfing, and small dwarfing rootstocks performed differentially under greenhouse and field conditions, also responding differentially to water deficit and heat stress. Plant-water relationships suggested different water use and stress response strategies in the rootstocks of different vigor levels. The semi-dwarfing G.202 led to the largest scion growth and demonstrated more water demand and more susceptibility to stresses. Large dwarfing rootstock G.935 and M.26 showed more stringent stomatal control and reduced water use at the onset of stresses, to avoid water loss. The smallest large dwarfing M.9 and the small dwarfing B.9 appeared to be more drought tolerant, and maintained leaf functions at a moderate but stable level. Future studies should be extended to other apple cultivars, to investigate whether these rootstocks would influence scion-water relationships in a way similar to Ambrosia TM . More knowledge about scion-rootstock water use strategies can facilitate the decision-making on irrigation management, to alleviate stresses and achieve production goals for specific rootstock vigor categories.
In the Ambrosia-rootstock systems, the sensitivity of stress indicators from high to low was leaf gas exchanges ≈ ETR ≈ Fv'/Fm' > leaf T surface > chlorosis and defoliation in leaves, and Ψ fruit ≈ SSC% > fruit T surface > fruit weight in fruits (Figures 6-8, Table 3). Strong correlations existed between leaf photosynthetic parameters, leaf transpirational rate, and Ψ fruit (Figures 7 and 8).
This study set the first small step towards a comprehensive understanding of water relationships in Ambrosia TM scion-rootstock systems, and of the sensitivity and relationship of multiple stress indicators. In future studies, developmental traits such as stomatal size and density, leaf area, and trunk increment should be re-evaluated under varied field conditions and at different tree ages; controlled field experiments should be conducted under aggravating effects of water deficit, heat stress, and heavy crop load. Multifaceted physiological studies on root-shoot-fruit dry matter partitioning, shoot-grafted union-root hydraulic conductivity, as well as scion-rootstock differentiations in xylem vessels and symplastic water transport, will provide more conclusive evidence for the speculated rootstock-specific water use and stress response strategies. Trees were irrigated to full water holding capacity every three days, and the fertilizer 20-20-20 was applied monthly at a rate of 1 g/1 L. The water deficit experiment was conducted in January 2019. During the dehydration cycle, trees were not irrigated for one week to allow soil water depletion, and then the pots were recovered to full water-holding capacity. Volumetric water content (VWC) was monitored in the topsoil of 5 cm depth using 5TM soil sensors and was recorded at 30 min interval using an EM50 data logger (Meter Environment, Pullman, WA, USA). The variation in VWC in the beginning and at the end of the dehydration cycle (∆ VWC Day 7-Day 1 , m 3 /m 3 ) was analyzed to demonstrate the change in soil moisture in the pots of each rootstock (n = 3 pots).

Field Trial Conditions
Three-year-old Ambrosia TM trees grafted on M.9, M.26, B.9, G.202 and G.935 were grown in silt-loam soil at the experiment farm (49 • 33 45" N, 119 • 38 55" W, elevation 454 m) in complete randomized block design (n = 6 plots for each rootstock, 3 trees per plot). Under the typical South Okanagan climate, the growing season starts with a cool and moist spring, followed by a hot and dry summer, with daily maximum temperature of 28.4 • C in July and 28 • C in August (Environment Canada), and historical average moisture deficit of 157 mm in July and 133 mm in August (data acquired from www.farmwest.com, BC-Okanagan South-Summerland EC; Access date: 27 December 2020). In 2020, precipitation and average moisture deficit were 15 mm and 155 mm in July, and 11 mm and 140 mm in August, respectively. Irrigation was applied through drip lines from 08:00 to 10:00 a.m. every three days from May to early October. Trees were trained to Tall Spindle Axe structures in high-density planting. Blossoms were hand-thinned in May 2020, followed by fruitlet thinning in June, to keep five fruits per tree. Stress responses were evaluated under low VWC and high air temperature in August. Yield was assessed at harvest in late September.

Leaf Photosynthesis and Transpiration
Leaf photosynthetic rate (P n ), transpirational rate (T r ) and stomatal conductance (g s ) were measured using the infrared gas analyzer of LICOR-6800 Portable Photosynthesis System (LICOR, Lincoln, NE, USA). Instantaneous water use efficiency (WUE i ) [30] was calculated as P n divided by T r , both parameters being measured simultaneously at a stable photosynthetically active radiation level (PAR). Relative electron transport rate (ETR), maximum quantum efficiency of CO 2 assimilation (ΦCO 2 ), and the variable fluorescence: maximum fluorescence ratio at steady-state photosynthesis lighting conditions (Fv'/Fm'), were measured using the chlorophyll fluorometer of LICOR-6800, to represent Y(II), i.e., the relative quantum efficiency of Photosystem II (PSII) [38]. To simulate the ambient mid-day radiation, PAR in the LICOR-6800 sample chamber was set at 800 µmol photon m −2 s −1 in the greenhouse experiment, and at 1600 µmol photon m −2 s −1 in the field trial. Other environmental parameters were set as flow rates at 700 µmol s −1 , chamber relative hu-midity at 50%, and the CO 2 concentration of sample chamber at 400 µmol CO 2 m −2 s −1 . Temperature in the sample chamber was set as equal to the ambient. In the greenhouse experiment, leaf gas exchange measurements were conducted at 24 h (Day 1), 96 h (Day 4) and 168 h (Day 7) after irrigation (dehydration cycle), and then at 24 h after irrigation was resumed (rehydration). In the field trial, leaf gas exchanges and Y(II) parameters were measured 72 h after irrigation on 2 days in August when VWC declined below 0.15 m 3 /m 3 and the maximum air temperature was 37.1-38.3 • C. The measurements were scheduled between 08:00 a.m. and 01:00 p.m., with the sequence of measurements randomized amongst rootstocks (n = 6).

Tissue Surface Temperature and Fruit Water Potential, Ψ fruit
In the field trial, thermal images of sunlit apple fruits, and of sunlit and shaded leaves, were captured using FLIR E8 Infrared camera (FLIR ® Systems Inc., Wilsonville, OR, USA; 7 mm focal lens, 320 × 240 IR resolution), immediately before the photosynthesis measurements were conducted on the adjacent leaves. Surface temperature (T surface ) of fruits and leaves was automatically identified on the thermal images by the software FLIR Tools (V. 6.4, FLIR ® Systems Inc.) (n = 6). Average fruit T surface was calculated as the mean temperature of sunlit skin and shaded skin of the fruit. Average leaf T surface was calculated as the mean temperature of the sunlit and shaded leaves.
Tissue water potential of fruit hypanthium (Ψ fruit ) was measured using a WP4C potentiameter (Meter Environment). Fruits were collected at optimal harvest time, stored in air at 4 • C for four weeks, and recovered to 23-24 • C in laboratory conditions prior to measurement (n = 24 fruits per rootstock). The instrument was calibrated using −2.19 MPa standard KCl solution. One slice of fruit hypanthium tissue of 3-4 mm thick was cut out vertically from the fruit, and then a stainless sample cup was used to excise one disc of sample from the centre of the hypanthium slice, with the rim of sample cup at about 1 cm apart from the fruit skin. The disc of sample was gently pushed down to the bottom of the sample cup to half-fill the cup space. Fruit tissue residual was quickly and carefully wiped from the rim and the outside of the cup. When possible, the sample was placed in WP4C immediately after being excised. When a waiting time was inevitable, the cup lid was capped tightly to restrict water transfer until the instrument was ready to load sample. Measurements were automatically completed after 6-10 min equilibration in the fast mode.

Leaf Chlorophyll Concentration and Stomatal Characteristics
Leaf chlorophyll concentration was assessed in absolute units of µmol of chlorophyll per m 2 of leaf area, using the apple setting in the MC-100 Chlorophyll Meter (Apogee Instruments Inc., Logan, UT, USA) [39], on the same days of photosynthesis measurements. Five healthy, fully developed, and sunlit leaves per tree were randomly sampled nondestructively, to represent the average leaf chlorophyll concentration for each tree (n = 6). By the end of the greenhouse water deficit experiment, leaf area of defoliated leaves per tree was estimated by multiplying the dry weight of fallen leaves with 264.8 cm 2 /g, i.e., the coefficient of leaf area: leaf dry weight for Ambrosia TM apples.
For stomatal density and size analysis, a leaf imprint was made by painting a 5 mm strip of transparent nail polish on the interveinal area on the abaxial side of leaves. The completely dried nail polish film was peeled off and mounted in a drop of water on the microscopic slide. Five imprints were prepared per leaf, and 4 leaves were sampled from each of 6 trees per rootstock. The leaf imprints were observed in the mode of transmitted differential interference contrast under the 10× objective lens of Zeiss Axio microscope (Zeiss, Oberkochen, Germany). Images were acquired using AxioCam MR R3 camera (Zeiss) and software Zen Pro 2012 (Zeiss). The number of stomata per mm 2 of leaf area was counted in the software ImageJ (National Institutes of Health, V.1.53) [40] to represent stomatal density. The lengths and widths of leaf stomata were measured on 10 random stomata per imprint using the same software. The size of the stomata was estimated as the area of the ellipse, i.e., π × 1/2 length × 1/2 width. Stomatal area per leaf area was calculated by multiplying stomatal density with the average size of stomata per unit of leaf area.

Nutrient Use and Yield Efficiency
Trunk cross-sectional diameter was measured at 30 cm above the grafted union in north-south direction and east-west direction, using a Digimatic 8 ABS digital caliper with DP-1VA Digimatic data logger (Mitutoyo America Corporation, Renton, WA, USA). Trunk cross-sectional area (TCSA) was calculated in an approximation of a round disc, to represent the vigor of scion growth (n = 10 trees per rootstock in the greenhouse experiment; n = 6 plots per rootstock in the field trial, with 3 trees per plot). In the field trial, yield efficiency was calculated as the kilograms of fruits at harvest per cm 2 of TCSA. Fruits without any disorders were sampled at harvest (n = 3; each sample consisted of 2 slices per fruit, 3 fruits per tree, 2 trees per rootstock), completely dehydrated at 80 • C in an oven for one week, and sent to A&L Canada Laboratories Inc. (London, ON, Canada) for the complete mineral nutrient analysis.