The experimental orchard was established in the spring and early summer of 2010, on a north/south orientation, at the Parma Research and Extension Center, University of Idaho (43.7853˚N, 116.9422˚W) on a sandy loam soil with an approximate pH of 7.3. The experimental site is a semi-arid climate with an annual precipitation of approx. 297 mm. Climate conditions of this experimental site were almost similar to those in the Pacific Northwest [46]. General orchard cultural practices such as irrigation and thinning were reported in a different aspect of this comprehensive study elsewhere [5] [47] [48].

Total nitrogen of 60 g/tree as UAN 32 (urea and ammonium nitrate, 32% N) was applied through the driplines. Each year, two fertigations of 30 g/tree N were made: one in late-May and second one two weeks later. Potassium was applied as 30 g K/tree as potassium oxide, via fertigation, once a year in late-May. Phosphorous, as monoammonium phosphate (61% P₂O₅), was applied at the rate of 150 g of formulation to each tree-planting hole, only once at the time of planting. Micronutrients, particularly, iron and zinc, were sprayed twice in spring and once in early summer each year. Calcium (Ca) was sprayed three times with cover sprays during spring every year.

“Aztec Fuji” trees on “M.9”, RN 29 (“Nic 29”) and Budagovsky 9 (“Bud 9”) rootstocks (C & O Nursery, Wenatchee, WA) were planted at 0.9 × 3.90 m spacing with a north-south row orientation. “Snow Drift” crab apple (Malus x “Snowdrift”) on “M.26” EMLA rootstock (C & O Nursery, Wenatchee, WA) was interplanted in each row as a pollinizer after every 10th “Aztec Fuji” tree. Trees were trained to either central leader (CL) or tall spindle (TS) during the dormant season in early-March each year. In trees with a CL training system, four main scaffolds were kept permanently and the other side beaches above these scaffolds were kept shorter at about 45˚ angle to form a “Christmas-tree shape”. Trees with a TS system were trained to have between 14 to 18 side branches, which were trained to 110˚ from the vertical axis with tree ties. After the first year, only side branches with thicker than 1/2 to 2/3 diameter of the main leader were removed, leaving a 12cm-stub for regrowth. Tree leaders in both CL and TS were maintained at approx. 3.75 m in height.

Each rootstock block contained 12 trees, where 6 of the adjacent trees in a row received TS training while 6 trees on the adjacent row received CL training. In 2015, trees were in full bloom on April 11, and girdling treatments were applied on April 27^th. In 2015, paired trees of these 6 tree-plots received one of the three girdling treatments as follows: 1) Control or No Girdling (NOGD), where no girdling was applied; 2) Bark Girdling (BG15) at 30 cm above the graft union, where a complete 3 - 4 mm ring of bark with cambium was removed by cutting with a sharp knife so that xylem wood tissue was exposed; 3) Score Girdling (SG), where the bark of tree trunk was scored in a spiral pattern, going one and half turns around the tree trunk, from 30 cm above the bud union with a sharp linoleum utility knife. In 2016, trees were in full bloom on April 15 and girdling treatments were applied on May 16^th. In 2016, the experimental arrangement was similar 2015, except for the fact that the second of each paired tree of the girdled treatments was not girdled or scored respectively. Where trees were girdled in 2016, this took place at approximately 6 cm above the 2015 bark girdling (BG15) site. This treatment was designated BG1516. Where trees were scored in 2016, this also took place approximately 6 cm above the 2015 score girdling site (SG15). Trees scored in both years were designated SG1516.

To determine chlorophyll index (CIs), four random terminal and side branches in each experimental tree were tagged and their final growth was measured in August, September and October 2015 and also June and October 2016, using a SPAD meter. CIs between 10 am and 2 pm were determined using equipment based on a new technology (Field Scout Chlorophyll Meter, Model CM 1000, Spectrum Technologies, Aurora, IL, USA) and determined according to the method described by Mahdavi et al. [49] [50]. The equipment uses the ambient and reflected 700 and 840 nm lights to calculate the relative CI. It measures conical viewing areas between 30 and 180 cm from lens. The equipment measures index of relative chlorophyll content ranges from 0 to 999.

Thirty leaves were sampled from each tree at random from the middle of the current-season shoots in mid-August each year. Leaves were placed in a cooler and taken to the University of Idaho Pomology and Viticulture laboratories each year for analysis. Leaves were washed in a mild solution, containing 1% Liqui-Nox anionic detergent (AlcoNox Inc., White Plains, NY, USA), rinsed in four different 25-L containers of distilled water, and dried in a forced-air oven at 65˚C. Dried leaves were ground to pass through a 40-mesh screen using a Cyclotec Sample Mill (Model 1093; Tecator, Hoganas, Sweden). Leaf N concentration was determined by combusting the dry leaf tissues using a LECO Protein/Nitrogen Analyser (Model FP-528; LECO Corp., St. Joseph, MI, USA).

Twenty fruits per tree were randomly sampled for quality evaluation and yield was recorded from each tree on October 17-20 during 2015-2016.

Free proline concentrations were determined according to the method described by Bates et al. [51]. Also, chlorophyll a, b and a + b concentrations were measured according to the method described by Arnon [52]. Total leaf anthocyanin and carotenoid concentrations were measured determined according to the method described by Sims & Gamon [53].

Other than a case between rootstock and tree architecture for yield per tree in 2016, no significant interaction was observed between rootstock-training or rootstock-training-girdling treatments for any measurements in this study. Thus, only direct effects of rootstock, tree architectures, and type of girdling are reported in Tables 1-3.

^zAbbreviations: NOGD = no girdling (control); BG15 = bark girdling in 2015; SG15 = score girdling in 2015; BG1516 = bark girdling in 2015 and 2016; SG1516 = score girdling in 2015 and 2016. ^xMean separation within columns in each group of treatments by LSD at 5% level.

^zAbbreviations: NOGD = no girdling (control); BG15 = bark girdling in 2015; SG15 = score girdling in 2015; BG1516 = bark girdling in 2015 and 2016; SG1516 = score girdling in 2015 and 2016. ^xMean separation within columns in each group of treatments by LSD at 5% level. No interaction existed among any treatments.

^zAbbreviations: NOGD = no girdling (control); BG15 = bark girdling in 2015; SG15 = score girdling in 2015; BG1516 = bark girdling in 2015 and 2016; SG1516 = score girdling in 2015 and 2016. ^ySignificance: Significant at 1% level if shown by **, at 5% level if shown by *, and no significant if shown by NS. ^xMean separation within columns in each group of treatments by LSD at 5% level or less.

Leaves from terminal and side branches of trees on both “Bud 9” and “Nic 29” rootstocks had lower CI during early growing season (August 15, 2015 and June 30, 2016) than those in late season (October 16, 2015 and October 20, 2016) (Table 1), due to the declining trend of seasonal leaf nitrogen [8] and perhaps chlorophyll content. Side branches of trees on “Bud 9” had significantly higher CI (up to 5%) than those on “Nic 29” rootstock during all sampling times in 2015 and 2016 (Table 1). In contrast, terminal branches of trees on “Nic 29” rootstock tended to have up to 6% higher CI than those on “Bud 9” in both 2015 and 2016. Terminal buds of trees on dwarf rootstocks, such as “Bud 9”, grows slower and cease before those on more vigorous rootstocks such as “Nic 29” [3] , perhaps leading to a higher distribution of nitrogen and chlorophyll in the side branches in trees on “Bud 9”. Leaf samples were taken from terminal shoots, and leaves from trees on “Nic 29” had significantly (up to 5%) higher concentration of N than those on “Bud 9” (Table 3).

Leaves from terminal shoots of trees on “Nic 29” rootstock had 54% higher proline, 14% higher chlorophyll b, and about 4% higher total chlorophyll (a + b) those did those on “Bud 9” rootstock, perhaps due to their higher N concentrations (Table 2). The impact of rootstock on the actual total chlorophyll content of terminal shoots (Table 2) parallels the results found in their non-destructive chlorophyll index measurements (Table 1). However, there was no significant difference between two rootstocks for the levels of carotenoids and anthocyanins.

Trees on “Nic 29” also had up to 7% larger average (heavier) fruit than those on “Bud 9” in both 2015 and 2016, despite their 83% higher yield in 2015 (Table 3). Usually heavier yield is associated with smaller fruit size. However, trees in all treatments were thinned at about 15 cm spacing at the time of hand thinning. Therefore, higher yield and larger fruit size in trees on “Nic 29” compared to those on “Bud 9” is due to their larger canopy, leading to a higher leaf/fruit ratio in trees on “Nic 29”. It is extremely difficult to determine true effects of rootstock on the scion cultivar fruit quality and other characteristics. This is mainly because confounding effects such as rate of light penetration, crop load, and tendency of biennial bearing can confuse the results. Similar to our assessment, Autio [54] experienced the same challenge in evaluating the impact of six rootstocks on fruit quality attributes of “Delicious” apple. However, he concluded that fruit size consistently was largest from trees on “M.9” EMLA and smallest from trees on “OAR 1” rootstocks.

Trees with CL architecture had up to 6% higher terminal shoot chlorophyll index than did those with a TS system in both 2015 and 2016 (Table 1). However, side branch chlorophyll indices in these two training systems were statistically similar (Table 1). Also, Trees with a CL system had 43% higher proline content and 4% higher chlorophyll a than did those with a TS training system in 2015 (Table 2). While, there were no significant differences between two training systems, nor between carotenoids and anthocyanin concentrations. The impact of training system on the actual total chlorophyll content of terminal shoots (Table 2) parallels the results found in their non-destructive chlorophyll index measurements (Table 1). In addition, Trees with a CL system in 2015 measurement had 4% higher leaf N as compared to trees with TS system. Nevertheless, trees with a TS training had 13% higher yield than those with a CL training (49.1 vs. 43.3 kg/tree), while fruit weight was similar in both tree training systems (Table 3). There was a significant interaction between rootstock and tree architecture for yield per tree in 2016 (Table 3). In this interaction, trees on “Nic 29” with a CL training had significantly lower yield (39.9 kg/tree) than those on “Nic 29” with a TS tree training that had 53.9 kg/tree in 2016. Trees on “Nic 29” with a CL training had denser and darker canopies (personal observation), resulting in excessive shading and lower fruit bud formation, and thus lower yield than those on “Nic 29” with a TS training.

Trees on “Nic 29”, with either a CL or TS training system, needed more time for training and pruning, and this issue should be considered when choosing this rootstock. “Aztec Fuji” trees on “Nic 29” rootstock can be suitable for planting at farther than 0.9 m spacing between trees in a row and in the lighter soil in the region. “Bud 9” is also a suitable rootstock for “Aztec Fuji” apple if extra-large fruit size is not the main objective of apple production.

Bark girdling in 2015 (BG15) significantly reduced CI in terminal branches than NOGD (up to 7%). Also, bark girdling in 2015, repeated in 2016 (BG1516) treatments reduced leaf N concentrations than all other treatments (up to 11%) (Table 2). The total leaf chlorophyll content and chlorophyll b in bark-girdled trees was also lower than those of non-girdled (up to 12% and 20% in Chl a + b and Chl b, respectively) (Table 2). Reduction of leaf chlorophyll as a result of bark girdling is in general agreement with previous reports [20] [30] [41] [42] [44] [45]. Girdling reduced leaf chlorophyll content in citrus trees, perhaps because girdling increases carbohydrates, resulting in changes to PSII system [42]. In addition, our results revealed that bark girdling (BG15) decreased proline content of leaves as compared to those on non-girdled (56%) and score-girdled (27%) trees (Table 2). Trees receiving BG15 or BG1516 had significantly lower leaf N than other treatments (up to 6.5% and 11% for BG15 and BG1516, respectively) (Table 2) which resulted in improved fruit color in these trees (data not shown). The results are in agreement with Schechter et al. [28] who reported that nutrient concentrations; including N in leaves on girdled non-fruiting limbs were generally lower than those in the other treatments. The impact of girdling in their non-fruiting “Sturdeespur Delicious” agreed with our results in fruiting “Aztec Fuji” apples. Impacts of girdling on nutrient partitioning in apple fruit tissue on different rootstocks warrant further investigation. Also, better color in the fruit from apple trees with lower leaf N was previously reported by Fallahi et al. [19] , which can be due to lower chlorophyll content in the skin of these fruit. Girdling can reduce resistance to cold stress in grape vines [12] and tree fruit (Fallahi, data not published). Thus, lower proline content of trees receiving a bark girdling treatment in 2015—girdled trees may make these girdled trees susceptible to extreme sub-freezing temperatures, and the authors have repeatedly observed this situation (data not shown).

Girdling treatments in 2015 did not affect yield per tree in 2015. This result was expected because flowers of 2015 crop were initiated, and fruit of this season were already set before we applied our girdling treatments in May 2015. However, trees with BG15 or SG1516 treatments had significantly (21% to 25%) higher yield in 2016 compared to those with NOGD treatment (Table 3), because girdling or scoring in 2015 resulted in more fruit initiation and fruit set, leading to higher crop in 2016. Enhancing yield/tree by girdling is in general agreement with the result of Ibrahim et al. [41] on Washington navel orange trees.

Fruit weight in 2015 and 2016 was affected by girdling treatments. In 2015, both bark and score girdled trees (BG15 and SG15) had 14% and 12.5% (respectively) higher fruit weight than non-girdled trees (Table 3). However, in 2016 only trees receiving repeated of score girdling on the same tree (SG1516) had about 12% higher fruit weight than NOGD (Table 3). These results suggest that carbohydrates accumulation is increased in the girdled trees. Similar to our results, Nguyen & Chung [55] reported that a simple S-shape scoring increased fruit weight and size in “Wax” apple. Based on our two-year data, we conclude that score and bark girdling in one year is sufficient to increase fruit weight of the current year and the yield in the following year. However, if both fruit weight and higher yield are the critical objectives of fruit production, annual bark scoring should be practiced in each year. Similarly, Mostafa & Saleh [36] found that girdling increased fruit weight in “Balady” mandarin.

Trees, which are bark girdled, needed about 6 weeks to heal. Trees, which are score girdled need about 3 weeks to heal. Many apple growers, particularly in the humid regions, are concerned that girdling or scoring during early growing season might introduce fireblight bacteria through the wounded areas. Although we did not observe any sign of fireblight incident in theses experimental trees, one can postpone the girdling or scoring to late June or early July when fire blight inoculum is reduced in the orchard. This practice must be tested in each apple growing region before it is widely recommended to growers. It is extremely important to allow enough time to the girdling or scoring wounds to heal to avoid possible freeze damage. In general, reduced leaf N concentrations of girdled trees may reduce photosynthesis in the long-term. The positive effects of girdling on yield and fruit weight may encourage some orchardists to use this method in fruit crop trees; however they should be aware of possible cold damage if temperatures plunge down to temperatures below which cells are acclimated.