IntroductionTop

Low prices for apples over the last years have increased the plantings of new pear (Pyrus communis L.) orchards in Europe (Vercammen, 2011b). Moreover, yield efficiency, fruit size and quality will need to be improved in order to justify those investments (Webster, 2002). The adoption of high-density orchards for pear production has resulted in a significant improvement in yield and fruit quality. However, early cropping is often not achieved, and remains one of the main challenges when planting a pear orchard (Webster, 2002). A positive correlation exists among yield, light interception and tree density (Palmer et al., 1992). While several studies have reported a positive relationship between tree density and yield (Vercammen, 1999; Kappel & Brownlee, 2001; Elkins & DeJong, 2002; Sansavini & Musacchi, 2002; Robinson, 2008); Wagenmakers & Tazelaar (1997) observed an increase about 2.3 t/ha over 8 years when light interception was increased by 1%. However, Lakso et al. (1989) and Musacchi et al. (2005) reported that if orchard efficiency is not maintained, an excessive yield increase can also reduce fruit quality.

Training systems, as a way to manage the tree canopy, can play a key role in order to take advantage of the tree productivity potential (Lakso & Robinson, 1997). Numerous studies on pear training systems have been carried out around the world (Deckers, 1992; Sansavini & Musacchi, 1993; Corelli-Grappadelli, 2000; Wertheim et al., 2001; Elkins & DeJong, 2002; Musacchi, 2008; Robinson, 2008; Sosna & Czaplicka, 2008; Turner et al., 2008; Monney & Evéquoz, 2009; Vercammen, 2014; Heijerman et al., 2015), but research shows that no system is optimum for all conditions (Barritt, 1987). Therefore, it is necessary to conduct exhaustive studies to find the best training system for each particular situation: cultivar, rootstock, climate, and economic conditions.

Italy and Spain are the most important pear producing countries in Europe (Deckers & Schoofs, 2008). While ‘Conference’ is the second most important cultivar grown in Italy (Deckers & Schoofs, 2008); it is the most important cultivar grown in Spain (Iglesias & Casals, 2013), and in Northern Europe, with 80% of the acreage in Netherlands (Heijerman et al., 2015), and 85% in Belgium (Vercammen, 2014). ‘Conference’ is a very fertile cultivar that tends to crop on spurs, but with significant smaller sizes when crop is bore on old branches (Sansavini & Musacchi, 1993). Therefore,paul kalkbrenner feed your head training system is key not only to increase yield, but also to increase profitability through bigger fruit sizes. Regarding that, Sansavini & Musacchi (1993) recommend that ‘Conference’ must be well pruned yearly, eliminating one-third of the fruiting spurs and their branches.

A good tree establishment after planting will help to achieve precocity (Heijerman et al., 2015). Regarding that, rootstocks are crucial for tree establishment but also to make trees more manageable through vigor control (Sansavini & Musacchi, 2002). Pear orchards in North America are mostly planted on Pyrus seedling rootstocks, as Quince (Cydonia oblonga Mill.) rootstocks routinely suffer from winter damage, fire blight (Erwinia amylovora Burill) infections and pear decline (Westwood & Lombard, 1983; Lind et al., 2003; Mitcham & Elkins, 2007; Robinson, 2011). However, clonal pear rootstocks generally delay cropping with respect to Quince (Sansavini & Musacchi, 2002). Therefore, while most of the studies done in North America are done with Pyrus and ‘Old Home’ × ‘Farmingdale’ (OH×F) rootstocks (Elkins & DeJong, 2002; Turner et al., 2008; Robinson & Dominguez, 2015), EM Quince A and C are the most widely planted pear rootstocks in Europe (Mitcham & Elkins, 2007). As a result, the better early cropping of Quince compared to pear clonal rootstocks, plus the milder winter temperatures in southern Europe like Italy and Spain compared to North America, make the use of quince EMC more suitable for such areas, ensuring optimum yield and fruit quality. This justifies why Quince Adams, MC, and Sydo are the most used rootstocks for trials in Europe (Deckers, 1992; Sansavini & Musacchi, 2002; Musacchi, 2008; Vercammen, 2014).

Aim of this study was to evaluate five training systems on a Quince EMC rootstock with ‘Conference’ that involved the use of feathered trees and increased densities, which thereby permitted more intensive production. Assessment of yearly and early cropping, labor requirements and fruit quality were studied. In addition, orchard profitability through different economic factors was also evaluated.

Material and methodsTop

A field trial was planted at the experimental station of IRTA (Institute of Research and Technology, Food and Agriculture) in Mollerussa, Spain (41°36'51.13''N; 0°52' 22.75''E) in 1999. The experiment was organized in a randomized complete block design with three replications. Training system was the main plot factor with each main plot consisting of 2 rows 10 m long. Five different training systems on Quince EMC rootstock and ‘Conference’ as the scion cultivar were compared. Training systems descriptions are given in Table 1. Foliar GA3 sprays (1.5 g/ha) to promote early cropping were applied at full bloom during the first three years. Trees were drip-irrigated (climate is semi-arid Mediterranean, with a mean annual rainfall of 350 mm), and received 100 kg N/ha, 40 kg P, and 120 kg K2O each year.

Table 1. Training systems.

Axis 1 and 2 were supported by a 4-wire trellis (2.5 m), whereas Tatura systems were supported by 4-wire trellis (0.5 and 2 m) with 2 wires on each side.

The Axis 1 system was the standard system grown by the farmers. A non-preformed tree without feathers (whip) was used in this case (Table 1), heading the leader at 80 cm right after planting. A strong vertical shoot arising near the heading cut was tied to the wires and trained as the leader. The remaining shoots were selected as scaffold branches and tied to 40° above the horizontal, encouraging light penetration. Due to the branching ability of ‘Conference’, most of the branches were kept uncut during the early years after planting, to encourage formation of laterals along the axis and discourage apical dominance.

A preformed tree (two-year-old tree with feathers) was used for the Axis 2 system (Table 1). Trees were developed by leaving the leader un-headed at planting and selecting the more vertical feather as the leader. Then, similarly to Axis 1, the remaining feathers were tied to 40° above the horizontal and kept uncut to encourage the growth of more laterals along the axis.

Two-year-old preformed trees with 4 or 2 tiers were used for Tatura 4 and Tatura 2, respectively (Table 1). Tatura 2 and Tatura 4 systems were developed by tying at planting each axis (2 or 4) to the wires of the trellis (as V). These systems were trained to short limbs and spurs that were periodically renewed.

A 1-year-old preformed tree was used for the Tatura 1 (Table 1). With this system, trees were tied to a tilted structure to a 15° angle from vertical.

Overall for the different systems, during the early years tree training was based on encouraging development of new branches to quickly fill the space assigned to the trees. In the first through the second year, dormant pruning was minimal for the preformed trees (Axis 2 and Taturas), promoting and keeping all the fruiting structures that were developing. Large diameter limbs (> 3 cm) were removed back to the trunk with an angled cut to grow replacement limbs. In the case of Axis 1, the main goal for the first 3 years was to establish the central axis structure.

For all the systems, once trees filled the allotted space, a balance pruning to promote fruiting wood was developed. Cuts were made on >1-year-old wood, promoting fruiting spurs. Leaders were cut to a side shoot when they reached its maximum height of 3.3 m. Once a branch diameter exceeded 3 cm it was cut back to its point of origin and renewed. The rest of the cuts were made to a side branch to keep growth balance and fruit quality.

Yield, fruit size, fruit quality (flesh firmness, soluble solids, and acidity), and required time to prune (dormant and summer) and harvest were recorded each year. A =50 kg-fruit sample from each elemental plot was collected for fruit quality, fruit size, and caliper distribution assessments. The sample was graded for fruit size and caliper distribution by a weight sizer machine (MAF RODA Iberica, Alzira, Spain). From this data we calculated a simulated packout. Firmness was measured at two opposite sides on the fruit equator using a digital firmness tester (Penefel; Ctifl, France). Soluble solid content (°Brix) and titratable acidity (malic acid g/L) were determined using the freshly prepared juice of the whole subsample. Soluble solid content was measured using a digital temperature compensated refractometer (model PR-101, Atago Co. Tokyo Japan), and titratable acidity (expressed as malic acid) was determined by titrating 10 mL of juice with 1.0 M NaOH to pH 8.2 (Torres et al., 2017). Crop value and economic return were calculated using 100% packout (€/ha) predicted from the average fruit price and the fruit size distribution recorded during the 10 years of the trial. Costs included planting (soil preparation, trees, trellis, fertilization, annual interest of capital, and labor); drip irrigation and fertigation installation; yield protection insurance, equipment rentals (mechanized machine for pruning and harvest, 4 €/h), administration, taxes, and land lease. Labor cost was categorized for unskilled (7.5 €/h), and skilled (pruning, 9.5 €/h) tasks. Net present value (NPV), internal rate of return (IRR) and break-even year to cash flow (BYCF) for each system over 10 years were calculated (Casler et al., 1993). NPV is the sum of discounted annual cash flows over 10 years using a fixed discount rate. The discount rate is determined by subtracting the rate of inflation from the current interest rate to arrive at a real rate of interest. IRR is the return on the cash flow stream generated over a certain number of years (in this case 10) (Casler et al., 1993; Robinson, 2011). We have used an interest rate of 5.5%, and a 4.5% rate discount for our basic comparisons. BYCF is the year when the accumulated NPV reaches zero, which equals to the year in which the investment has been recouped with interest. It can also be considered the year that an orchard can be removed or replanted in our case. A repeated-measures MANOVA was used to analyze yield and labor cost evolution along the seasons. Contrast tests were used to compare among training systems. Linear mixed models including training system as fixed factor and block as a random factor were built to separate treatment effects for the cumulative labor requirements, firmness, soluble solids, acidity, fruit weight, and caliper distribution at harvest. A linear mixed model including training system as fixed factor and block and year as random factors was built to separate treatment effects for the average efficiency rates. With all the models, the Tukey Honestly Significant Difference (THSD) post hoc test was used to compare training systems. Statistical significance was set at p = 0.05. Data were analyzed using the JMP statistical software package (vers 11; SAS Inst. Inc., Cary, NC, USA).

ResultsTop

Training system, time and the interaction of both was highly significant regarding yield and labor cost over the years (Fig. 1). Axis 1 had the lowest yields, followed by Axis 2 and then the Taturas. No significant differences were observed within the three different Tatura systems and with the Axis 2 (contrast tests, p > 0.05). Axis 1 had significantly lower yields than Axis 2 (contrast test, p < 0.028), and the Taturas (contrast tests, p < 0.05).

Labor cost followed a similar pattern as the yield, with lowest values for Axis 1, followed by Axis 2, and Tatura 2 as the highest-labor-requirement system (Fig. 1). Significant differences were observed between Tatura 1 and Tatura 2 (contrast test, p < 0.0373), whereas no differences were observed between Tatura 2 and Tatura 4, and Tatura 4 vs Tatura 1 (contrast tests, p > 0.05). Axis 2 required significantly less labor cost than the three different Tatura systems (contrast tests, p < 0.05), and higher than Axis 1 (contrast test, p < 0.0085).

Along the 10 years, Tatura 2 with over 1,200 h/ha, was the system that required more dormant pruning, followed by Tatura 1 and 4, Axis 2, and Axis 1 with less than 500 h/ha (Fig. 2). Summer pruning was mainly important on Tatura 2, with about 70 h/ha. Significantly higher amount of harvest time was required for the Tatura compared with the Axis systems.

Efficiency rates about 150 kg/h were observed for the Axis systems, whereas Taturas were significantly lower with values about 130 kg/h (Fig. 2).

Axis systems tended to have higher fruit firmness (0.1 kg) and sugar content than Taturas; however, no significant differences among systems were observed (Table 2). Average firmness was 5.7 kg, 15.4 °Brix, and 1.6 g/L of acidity. No significant differences among training systems were observed either for fruit acidity, fruit size, or caliper distribution. With an overall fruit size of 190 g, about 70% of the harvest had 65 mm or more for all the different systems.

Table 2. Average fruit quality variables (flesh firmness, soluble solids (SS), and acidity), fruit size and caliper distribution at harvest for each training system. No significant differences among training systems were observed at p value ≤ 0.05.

Great differences regarding the establishment cost (total investment at the end of year 1) were observed among systems (Table 3 and Fig. 3). Tatura 2 with 47,000 €/ha was the most expensive system, followed by the other two Taturas (4 and 1) (~35,600 €/ha), Axis 2 (19,149 €/ha), and Axis 1 (15,040 €/ha) as the cheapest option at planting (Table 3 and Fig. 3). Differences among systems were also observed regarding the yearly and cumulative cash flow, especially in how negative the cumulative cash flow curve dipped and became positive (Fig. 3). Axis 2 had the highest cumulated cash flow, near to 80,000 €/ha, followed by the Tatura 4 and 2 (~62,000 €/ha), Tatura 1 (53,000 €/ha), and Axis 1 (41,000 €/ha) (Fig. 3). With 5 years, Axis 2 was the system to reach first the break-even point, followed by Tatura 2 and 4 (6 years), and Tatura 1 and Axis 1 (7 years) (Table 3 and Fig. 3).

Table 3. Planting cost, average annual balance (years 6-10), break-even year to cash flow (BYCF), internal rate of return (IRR), and net present value (NPV) for each system.

All of the assessed training systems in this experiment had a positive IRR and NPV values after 10 years (Table 3 and Fig. 3). With a value near to 30%, Axis 2 had the highest IRR, doubling the one observed for the Taturas overall (~15%) (Table 3). The highest NPV (50,437 €) was also observed for Axis 2. Conversely, Axis 1 showed the second best IRR (18.4%), but the lowest net present value (24,250 €) (Table 3). Regarding the Tatura systems, Tatura 4 had the highest IRR (16.7%) and NPV (35,699 €) values (Table 3). On the other hand, Tatura 2 had higher NPV than Tatura 1 (32,926 € vs 28,930 €), but for the IRR value the opposite was observed (13.4% vs 14.4%) (Table 3).

DiscussionTop

Time to reach full crop production is key on the profitability of each training system. Yields at 2nd leaf were very low (1.5 t/ha) on Axis 1, compared to the other Axis 2 and Tatura systems, with yields about 10 t/ha. Similarly, at 3rd and 4th leaf, production of Axis 1 was 63% lower than its average production (5th to 10th leaf). Among all the systems tested, Axis 1 was the one with the lowest planting density (2,133 trees/ha). The positive relationship between tree density and crop value through the cumulative yield have also been reported in several studies comparing rootstocks and training systems for pear (Vercammen, 1999; Kappel & Brownlee, 2001; Elkins & DeJong, 2002; Sansavini & Musacchi, 2002; Elkins et al., 2008; Robinson, 2008, 2011; Robinson & Dominguez, 2015). Non-preformed trees (one-year-old whips) were used for Axis 1. In addition, tree training for that system consisted of heading the leader at 80 cm right after planting, starting the new tree structure the following year. This management technique is clearly a delay in regard to two-year-old preformed trees (Axis 2, Tatura 2-4), or even to one-year-old preformed trees with feathers but without heading the leader (Tatura 1). Sansavini et al. (2007) and Heijerman et al. (2015), already reported the importance of using highly feathered trees and reducing heading cuts in order to achieve high precocity to increase orchard profitability. However, low yields and delay to reach full production on Axis 1 was not only observed at the 2nd leaf, but it kept up to the 3rd and 4th.

Although differences in planting density ranged from 2,667 trees/ha on Axis 2 and Tatura 4, vs 5,333 trees/ha on Tatura 2-1, no significant differences regarding yield over the 10 years were observed among those systems. Other studies about training systems with ‘Conference’ have been conducted by Vercammen (2002, 2005, 2011a, 2014) reporting the V-system, with 367 t/ha after 10 years, as one of the most productive. Comparable to the Tatura 2 system that we tested, Musacchi et al. (2005) observed that a V-shape with 5,555 trees/ha was the highest productive system (181 t/ha, over 7 years). However, it is hard to make comparisons with the Axis system that we tested. For instance, Musacchi et al. (2005) used central leader trees but with different planting densities (Vertical Axis 7,936 trees/ha; Slender Spindle 3,968 trees/ha) and even with different rootstocks (Spindle bush 1,984 trees/ha – Sydo). Whereas Vercammen (2014) used a Long Pruning and Bush Spindle systems (1,714 trees/ha), and Spindle training (5,625 trees/ha). After 10 years, the cumulated yield of our Axis 2 system was 385 t/ha. Similarly, Robinson & Dominguez (2015) tested a Tall Spindle system with 2,243 trees/ha, reporting after 11 years, cumulated yields of 299 and 341 t/ha with ‘Bosc’, and ‘Barlett’ respectively.

With a nice fruit size of 190 g on average, no significant differences regarding size and fruit quality among systems were observed in our experiment. On the other hand, differences in fruit size among systems have been observed by other authors. For instance, larger pears were harvested by Vercammen (2014) on the V-system; and on the Spindle Bush by Musacchi et al. (2005). Most likely, fruit size variances can be explained by the differences in yield depending on the system (Robinson, 2008). In addition, different systems can induce a greater intensity of light that could affect the quality and size of the fruit. Hence, since light interception is more limiting in Belgium than in Spain, this may explain why we did not see differences among systems, while V-systems were reported to have larger fruits in Belgium (Vercammen, 2014). V-systems are reported to intercept more light than conic shapes in northern North America (Robinson & Lakso, 1989; Robinson, 2007).

Reasonably, time devoted to harvest showed a direct relationship with the production of each system. It is important to examine to what point one system is more or less efficient than another, as expressed in terms of yield harvested per working hour (harvest, dormant and summer pruning). Differences among the efficiency rate were observed in our trial. The Axis systems had an average of 148 kg/h, whereas the Tatura systems had a lower rate of 130 kg/h. Higher differences were observed regarding the pruning time required for each system. Tatura systems required considerably more hours of pruning, especially Tatura 2, which tripled the requirements of the Axis 1. In addition, the open structure of the Axis system facilitates harvest labor. Similar results were observed by Vercammen (2005), where after ten years the V-systems were the most labor intensive compared to Spindle and Long Pruning systems. Management and training differences among single-stem systems as the Axis, and the V-systems (Tatura), were also reported by Sansavini & Musacchi (2002).

Great differences were observed for planting cost, which was highly influenced by the tree type, density and trellis used for each system. The yearly activity costs showed that the most expensive systems at planting required higher annual cost of activity as well, mainly due to an increased need in the time of harvest, but also for pruning. If we examine the overall results, systems based on high planting densities and double production plan as the Tatura, have been able to achieve great yields. Nevertheless, its strong cost at planting hinders a quick investment pay off. That is, even in the case of great fruit prices, Tatura cannot beat the Axis 2, which had similar yields. As Robinson et al. (2007) pointed out, the greater the level of initial investment, the greater the risk in achieving expected profits. Thus, if two systems produce about the same NPV but one has much lower investment requirements, it is the preferred investment. Regarding that, Vercammen (2005) also suggested that a system that requires high management and planting cost as the Tatura is only an option if adequate reserves are available.

From a theoretical point of view, profitability of the Tatura could be improved by reducing the planting cost and/or increasing the production. In the first case, and based on the fact that the tree density should not be amended, there would only be two ways: (1) simplify the trellis; (2) reduce the labor cost. The other option (yield increase) could achieve economic returns similar to Axis 2, however, it does not seem that this may be continuously achievable year after year.

The results of this study show that the use of two-year-old preformed highly feathered trees is an improvement for both, early cropping and profitability. Within intensive systems, Axis 2 was the most profitable. Planting cost, trellis, and labor requirements had a large impact on the economic viability of each system. In addition, the low prices that growers are getting for the fruit, plus the increase of labor costs, enhance the need for simple training systems, with low labor input and highly productive. Tatura systems produced high yields, but the strong initial investment that needs to be done at planting makes these systems a risky investment. Axis 2 seems to be the most suitable system for early cropping while maintaining intermediate plantation costs and an appropriate level of production efficiency.