Orchard Automation and Genes Relevant to Apple Tree Architecture Kenong Xu Horticulture Section, School of Integrative Plant Science, Cornell University, Geneva, NY

Keywords: robotics, harvester, columnar, weeping, apple genes

he Green Revolution is a well-known term that described Fully automated robotic apple fruit harvesters the drastic yield increase in field crops such as maize, rice Developing fully automated robotic fruit harvesting systems and wheat in the 1950s and 1960s. The central element is an effective way to replace human hands. Important progress T that made the has been made towards commercialization of high-end robotic Green Revolution fruit harvesters. Two companies, Fresh Fruit Robotics (FFRobot- At least two different prototypes of fully “ a reality is ics) and Abundant Robotics, appear to be the current front run- automated robotic fruit harvesters have the landmark ners in this race. The prototype by FFRobotics has 4–12 robotic been tested for commercial applications accomplishments arms, each with three-fingered grips to grab fruit and twist it by high-end users, but apple tree canopy in genetic from a branch (Figure 1a). The fruit are placed in the collection poses a major challenge to these robotic improvement bucket and then transported to the bins through a conveyor sys- systems. Important genes for apple of plant tem (Figure 1b). It is reported that the machine can pick up to architecture. In 10,000 apples an hour, and is capable of taking 85% to 95% of the columnar and weeping tree forms have apple orchards, fruit off the trees, depending upon tree canopy characteristics. been or are being identified.” pomologists and According to the manufacturer, the system 1) is a reliable and horticulturists robust harvester that emulates the hand harvesting process for have long been efficient, cost-effective and bruise-free fruit picking; 2) offers a improving orchard productivity through the development unique opportunity for growers to reduce costs significantly by of dwarfing rootstocks and tree pruning/training systems to supplementing or replacing human pickers; 3) picks ten times keep trees in optimal shape. Although orchard outputs have more usable fruit compared with an average worker; and 4) col- been steadily improved, the associated production costs have lects data for fruit picked per tree, acre, and orchard. also increased considerably. This is because common orchard The robotic picker by Abundant Robotics uses a vacuum to operational tasks, such as apple tree pruning, fruit thinning, and suck the apples off trees (Figure 2). The fruit picking process is fruit harvesting are still conducted manually, requiring a large guided by computer-aided cameras and sensors, which detect amount of expensive labor. Augmented by a shortage of seasonal fruit. The system can suck fruit off trees at one apple per sec- farm workers, demands for automation of those common orchard ond and deliver them into bins. The robot picker detects 95% operational tasks have become more intensified, especially of apples and isn’t bothered by leaves or new growth, but can be for fruit harvesting. Here I update the progress that has been obstructed by limbs. The company raised millions of investment made in developing fully automated robotic fruit harvesters, the from big-name companies such as Google. The Washington Tree challenges in such efforts, and the latest advances in apple tree Fruit Research Commission also granted $500,000 toward the architecture genomics that may aid orchard automation. development of Abundant Robotics’ picker. Abundant Robotics

A B

Figure 1. FFRobot, a robotic fruit harvester developed by Fresh Fruit Robotics. A. Robotic hand and fingers. B. Conveyor transport system.

FRUIT QUARTERLY . VOLUME 26 . NUMBER 3 . FALL 2018 29 Figure 2. A robotic fruit harvester developed by Abundant Robotics. A. Overview of the robotic harvester. B. Close-up view of the vacuum end for fruit picking (source: Good Fruit Grower/YouTube). plans to have its robotic fruit pickers commercially available in fall 2019. In addition to the high-end robotic fruit harvesters, proto- types for low-cost robotic harvesters working in existing orchard conditions are also being tested. One example is the system developed by Washington State University (Figure 3) (Silwal et al. 2017). Tests of the system in a commercial apple orchard indicated that the machine’s vision system was accurate and had an average localization time of 1.5 seconds per fruit. Among the 150 fruit attempted, 127 (84%) were harvested with an average picking time of 6.0 seconds per fruit (Silwal et al. 2017). Ap- parently, further improvement is needed before such low-end robotic fruit harvesting systems can be used by growers.

Apple tree forms and challenges in orchard automation Figure 3. A low-cost robotic fruit harvester prototype devel- It is clear that substantial progress has been made in devel- oped by Washington State University (Silval et al. oping fully automated fruit harvesting systems. However, many 2017). believe that it is still many years until widespread application of such robotic harvesters, although they may be used by some growers who have a massive operation within the next 5–10 years. The reliability and cost of such robotic systems will ultimately ing” used here refers to tree form characterized by downward determine their applicable range. The major challenge appears growth of branches (Figure 4), which differs from the same term to be the complex canopy associated with the existing widely used for “ideotype IV” that describes the growth habit and fruit grown cultivars. As mentioned previously, large tree limbs can bearing type represented by ‘Granny Smith’ and ‘Rome Beauty,’ be obstructive to these robots. To help the robots see and pick where the branch bending is caused by the weight of tip-bearing the fruit, modification of tree architecture through pruning and fruit (Lespinasse and Delort 1986; De Wit et al. 2004). From the training is necessary. This also means that new trellis systems viewpoint of genomics, columnar and weeping tree forms offer must be installed to make the orchard “robot-ready”. rare and invaluable opportunities to investigate how apple tree forms are controlled genetically. Although tree pruning and training can make and maintain robot-ready canopies, this would require a substantial amount of labor. Developing apple cultivars that naturally develop into Genes relevant to columnar tree form a tree architecture form having a simple and narrow canopy Columnar growth habit in apple was originally discovered as would be a more desirable solution to address the challenges in a somatic mutation from ‘McIntosh’, called ‘Wijcik McIntosh’ in orchard automation. To do so, it is necessary and important to the 1960s (Lapins 1969). The most useful character of columnar understand the genetic control of apple tree forms. tree form appears to be its simple and narrow canopy, which is There are at least three distinct tree architecture forms in due to both a reduced number of branches and vertically growing apple: standard, columnar and weeping (Figure 4). Standard tree branches. Another desirable character is that fruit are set on spurs form is most common, and nearly all widely grown apple cultivars growing on “old wood” such as the main trunk or primary limbs, belong to this group. Columnar and weeping are two mutation requiring little pruning. For these reasons, “Wijcik McIntosh’ has tree forms derived from standard. Note that the term “weep- been used in many breeding programs to develop new columnar

30 NEW YORK STATE HORTICULTURAL SOCIETY apple cultivars (Tobutt 1984; Moriya et al. 2009). Genetic investigation into the columnar trait has been con- ducted intensively. Early inheritance studies indicated that the columnar trait is largely controlled by a dominant gene, called Co (columnar) (Lapins 1976). Later on, the Co gene was mapped by DNA markers to linkage group 10 in the apple genome (Conner et al. 1997; Bai et al. 2012). Recently, several research groups reported that an 8.2-kb DNA insertion (a long terminal repeat retroposon) in an inter-genic region at the Co locus is the genetic cause for columnar (Wolters et al. 2013; Otto et al. 2014). The consequence of the insertion appeared to be that it increased the expression of a nearby gene, called Co (encoding a 2-oxoglutarate and Fe(II)-dependent oxygenase). More recently, transgenic apples over-expressing the Co gene seemed to transform a stan- dard apple into a columnar-like apple tree (Figure 5) (Okada et al. 2016). So the increased expression of the Co gene is likely the biological cause for the columnar phenotype, elucidating the genetic control of columnar tree form. For decades it has been suggested that there are other genes involved in the columnar phenotype, as columnar progeny are often counted notably less than the expected numbers in breeding populations segregating for the trait. We have initi- ated a study to find such assumed genes. Our preliminary data indicate that at least two more genes are genetically required for the columnar trait. Efforts are under way to reveal the identity Figure 4. Apple tree forms. The three apple tree forms stan- of these columnar required genes. If successful, we would have dard, columnar and weeping are indicated. knowledge to deal with the variations controlled by these genes, which would greatly aid the effort to develop apple cultivars of an ideal tree architecture for both fruit production and orchard mechanization.

Genes controlling weeping tree form Weeping tree form is rare in domestic apples, despite its existence. Due to minimal usage of weeping tree form in apple production, very little research work was conducted in the past. However, weeping tree form is common in crabapples such as ‘Cheals Weeping’ and ‘Red Jade’ for ornamental purposes. Based on an inheritance study in the 1960s, weeping appears to be a dominant trait (Sampson and Cambron 1965). The gene responsible for the weeping trait, despite being unknown, has been named as W (weeping) in the literature (Alston et al. 2000). Lately, we genetically mapped the weeping trait to four genomic regions on chromosomes 13, 10, 16, and 5 (Figure 6), named W, W2, W3 and W4, respectively (Dougherty et al. 2018). Of the four genomic loci, the one on chromosome 13 showed the largest genetic effect on the trait, presumably representing the W gene named previously (Alston et al. 2000) (Figure 6). Figure 5. Over-expression analysis of the columnar gene CO. In this study, we used a non-traditional approach for map- The picture shows young transgenic apple lines ping the weeping trait, which is called pooled genome sequencing over-expressing the Co gene (left) and a control (Dardick et al. 2013). This approach takes advantage of the power gene (right) (Okada et al. 2016). of Next Generation Sequencing (NGS) technology. In addition, we also exploited the hidden segregation types of DNA variants in the massive sequence data, making the approach much faster and more efficient than the traditional method (Dougherty et al. ropism. Gene MdLAZY1 and other candidate genes have been 2018). This study also demonstrated an essential application of targeted for further functional analysis. The finding of the weep- NGS technologies in apple gene discovery by eliminating a major ing genes further deepens our understanding of the directional bottleneck: i.e., genetic mapping of traits. growth of apple tree branches, and is expected to contribute to So far, we have characterized the W locus and identified efforts in designing apple cultivars with ideal tree architecture. several candidate genes, including MdLAZY1, a new member in One of the major modern apple orchard systems is Solaxe a gene family known to be involved in plant response to gravit- (Figure 7), which uses extensive manual limb bending to achieve FRUIT QUARTERLY . VOLUME 26 . NUMBER 3 . FALL 2018 31 a balance between vegetative growth and cropping (Robinson et al. 2013). Since weeping trees naturally grow downward branches, another possible application of the weeping genes is to use them to develop weeping cultivars for areas where the Solaxe system is preferred. In summary, at least two different prototypes of fully automated robotic fruit harvesters have been tested for commercial applications by high-end Figure 6. Identification of genomic regions W,( W2-W4) relevant to the weeping users, but apple tree canopy poses trait. The plot shows the distribution of a set of informative DNA variants a major challenge to these robotic identified using pooled genome sequencing. The peaks W and W2-W4 are systems. Important genes for apple significantly higher than the genome average (Dougherty et al. 2018). columnar and weeping tree forms have been or are being identified. Such knowledge will help the effort to develop apple cultivars of ideal tree architecture for both fruit production and orchard mechanization.

Acknowledgments This work is financially supported by a grant (IOS-1339211) from the Plant Genome Research Program of the Na- tional Science Foundation (NSF).

References Alston, F. H., Phillips, K. L., and Evans, K. M. 2000. A Malus gene list. Acta Hort. 538: 561–570. Bai, T., Zhu, Y., Fernández-Fernández, F., Keulemans, J., Brown, S., and Xu, K. 2012. Fine genetic mapping of the Co locus controlling columnar growth habit in apple. Mol. Genet. Genom. 287: 437–450. Conner, P. J., Brown, S. K., Weeden, N. F. 1997. Randomly amplified Figure 7. An orchard after adopting the Solaxe training system. It suggests a polymorphic DNA-based genetic possible application of weeping tree form in apple production. (https:// linkage maps of three apple cultivars. www.pressreader.com/new-zealand/the-orchardist/20170801). J. Am. Soc. Hort. Sci. 122: 350–359. Dardick, C., Callahan, A., Horn, R., Ruiz, K. B., Zhebentyayeva, T., Hollender, C., Whitaker, M., J. Am. Soc. Hort. Sci. 101: 133–135. Abbott, A., and Scorza, R. 2013. PpeTAC1 promotes the Lespinasse, J. M. and Delort, J.F. 1986. Apple tree management horizontal growth of branches in peach trees and is a member in vertical axis: appraisal after ten years of experiments. Acta of a functionally conserved gene family found in diverse plants Hort. 160: 139–156. species. Plant J. 75: 618–630. Moriya, S., Iwanami, H., Kotoda, N., Takahashi, S., Yamamoto, T., De Wit, I., Cook, N., and Keulemans, J. 2004. Characterization and Abe, K. 2009. Development of a marker-assisted selection of tree architecture in two-year-old apple seedling populations system for columnar growth habit in apple breeding. J. Jap. of different progenies with a common columnar gene parent. Soc. Hort. Sci. 78: 279–287. Acta Hort. 663(62): 363–368. Okada, K., Wada, M., Moriya, S., Katayose, Y., Fujisawa, H., Wu, Dougherty, L., Singh, R., Brown, S., Dardick, C., and Xu, K. 2018. J., Kanamori, H., Kurita, K., Sasaki, H., Fujii, H., Terakami, S., Exploring DNA variant segregation types in pooled genome Iwanami, H., Yamamoto, T., and Abe, K. 2016. Expression of sequencing enables effective mapping of weeping trait in a putative dioxygenase gene adjacent to an insertion mutation Malus. J. Exp. Bot. 69: 1499–1516. is involved in the short internodes of columnar apples (Malus Lapins, K. O. 1969. Segregation of compact growth types in × domestica). J. Plant Res. 129: 1109–1126. certain apple seedling progenies. Can. J. Plant Sci. 49: 765–768. Otto, D., Petersen, R., Brauksiepe, B., Braun, P., and Schmidt, E. Lapins, K. O. 1976. Inheritance of compact growth type in apple. 2014. The columnar mutation (“Co gene”) of apple Malus( × 32 NEW YORK STATE HORTICULTURAL SOCIETY domestica) is associated with an integration of a Gypsy-like retrotransposon. Mol.

O R K Breed. 33: 863–880. Y S T A W T E E Robinson, T., Hoying, S., Sazo, M. M., DeMarree, N

H A., and Dominguez, L. 2013. A vision for O Y T R E T I IC C apple orchard systems of the future. NY U S O LT U R A L Fruit Q. 21: 11–16. New York State Horticultural Society Sampson, D. R. and Cambron, D. F. 1965. NYSHS Inheritance of bronze foliage, extra petals and pendulous habit in ornamental crabapples. Proc. Am. Soc. Hort. Sci. 86: 717–722. Silwal, A., Davidson, J.R., Karkee, M., Mo, C., Zhang, Q., and Lewis, K. 2017. Design, integration, and field evaluation of a robotic apple harvester. J. Field Robot. 34: 1140– Check out 1159. Tobutt, K.R. 1984. Breeding Columnar apple our web page varieties at East Malling. Sci. Hort. 35: 72–77. NYSHS.org Wolters, P. J., Schouten, H. J., Velasco, R., Si- Ammour, A., and Baldi, P. 2013. Evidence for regulation of columnar habit in apple by a putative 2OG-Fe(II) oxygenase. New Phytol. 200: 993–999.

Kenong Xu is a research and extension professor located at the Geneva Campus of Cornell University who leads Cornell’s program in applied fruit genomics.

Chain Top Bin Trailer

Building specialized orchard equipment to meet your unique requirements! 5 Old Route 31 Lyons, NY 14489 LaGasse Works, Inc. (888) 946-9202 [email protected] Orchard Hedgers

FRUIT QUARTERLY . VOLUME 26 . NUMBER 3 . FALL 2018 33 APPRAISALS • PAYROLL SERVICES • BENCHMARKS TAX PLANNING • TAX PREPARATION • ESTATE PLANNING FARM BUSINESS CONSULTING RECORD-KEEPING

When it comes to maintaining long-term profitability, there are few tasks more important than keeping reliable and timely records. Our financial specialists will tailor a program to fit your accounting and record-keeping needs — either from our office or on your farm. Learn how we can help keep your business strong at the roots with record-keeping services from Farm Credit East.

800.562.2235 FARMCREDITEAST.COM/RECORDKEEPING

The 3 row sprayer is the first piece of Platforms reduced labour time spent by 39% in 20% savings when you hedge sides, top, and equipment I have purchased to pay for dormant pruning, 50% in hand thinning, 29% bottom. It takes the guess work out for those itself in the first year! in summer pruning and 18% in harvesting. factors. You set the box, then they do big cuts - Leslie Huffman, past OMAFRA Apple Specialist and uprights. Also, by hedging you can fill the - David Knight, Scarlett Acres, Ontario John Molenhuis, OMAFRA Business Analysis and space better and have a consistent canopy. Cost of Production Specialist - Management, Green Tree Orchards, Michigan GREEFA PERFECT ORSI GROUP FAMA FRUIT TEC PULSE INSTRUMENTS BARTLETT SINCLAIR H.S.S. SHUR FARM

Contact us today for more information Visit our new Facebook page for on any of our product lines. news on demos, shows and product updates.

905.563.8261 | 1.800.263.1287 | [email protected] | www.provideag.ca | Beamsville, ON, Canada

34 NEW YORK STATE HORTICULTURAL SOCIETY 3 New York State Horticultural Society

NYSHSEDUCATING, PROMOTING and PROTECTING New York’s Commercial Fruit Industry

O R K Y S T A W T Founded in 1855, the mission of the New York E E N State Horticultural Society is to foster the H O Y T R 2 E T I growth, development and profitability of the IC C U S O LT U R A L fruit industry in New York State.

NYSHS accomplishes this by: • Supporting educational opportunities for members • Promoting the industry Figure 3. Modified sprayer and sensors in the trial orchard

• Representing the industry in matters of public policy

OBJECTIVES Education - providing education programs for members that include: • obtaining and disseminating information to the fruit industry

• sponsor and/or cooperate with other groups to provide and Figure 2. RR test site, Wayne Co., NY using Gala trees on M.9 rootstock planted May 3, 2014.

support tours • sponsor trade shows

• cooperate with and encourage others to provide educational opportunities Promoting the Fruit Industry by: • promoting ideas which will benefit the economic health of the fruit industry • educate the general public about the New York fruit industry NYSHS ISSUES Representing the New York fruit industry by: • Pesticide Registrations • Have cemented our role as legislative voice in both Albany and Washington, DC for the fruit industry • Food Quality Protection Act • representing the industry’s interest as well as other agencies and • Integrated Pest Management institutions • Agricultural Labor and Immigration • Fruit Industry Economic Development • Cornell Research and Extension • Education Public Officials • Educating Industry • Food Safety

FRUIT QUARTERLY . VOLUME 26 . NUMBER 3 . FALL 2018 35 NYSHS MEMBERSHIP INVOICE Yes! I will support the NYSHS and its mission to Educate, Promote and Protect the New York Fruit Industry. Yearly membership includes Hort Flash Newsletter, and the New York Fruit Quarterly. Dues are paid once a year, good from July 1st until June 30th. Please enclose check or cash. What the NYSHS Does for You: Growers Membship ...... $225 FRUIT INDUSTRY EDUCATION Per Each Multiple Membership after 1st ...... $100 Growers w/50 Acres or Less...... $100 Sponsor/Cosponsor of: Industry Professional ...... $225 • Annual Empire State Fruit & Vegetable Direct Academic Professional ...... $100 Marketing Expo • Industry Leadership Forums Sponsors: • Cornell Extension Events Bronze level ...... $300 • Sponsor of Labor Conferences Silver leve ...... $500 Gold level ...... $1000 Publisher of: Platinum ...... $2500 and up • The New York Fruit Quarterly • Hort Flash Newsletter ADDITIONAL SUPPORT: Presentations at: AREAS YOU’D LIKE NYSHS TO SPEND MORE EFFORT ON: • Growers’ Meetings H2A Reform ...... $100 • Discussion in Albany on All Labor Topics Speaker Programs ...... $100 Achievements: Your Thought______• Twice awarded $500,000 for Applied Apple ______$______Research TOTAL AMOUNT...... $______• Cornell Apple Labor Survey Articles in: Name______• Fruit Industry Publications Company______FRUIT INDUSTRY PROMOTION Mailing address______• Empire State Council of Agricultural Organizations City______State_____ Zip______• NYS Berry Growers Association and NYS Ag Affiliates County ______Cooperate with: Phone______Fax______• The U.S. Apple Association E-mail ______• The United Fresh Fruit and Vegetable Association • The New York Apple Association Thank You for Your Support!!! Membership dues are not deduct- FRUIT INDUSTRY PROTECTION ible for Federal Income Tax Purposes. They may be tax-deductible Seats on Advisory Boards: under other provisions of the IRS Code. Please check with your tax advisor. • Agricultural Affiliated; make certain all understand our industry’s often unique concerns and needs. Please remit to: NYSHS www.NYSHS.org 630 W. North Street To educate, promote and protect the Hedrick Hall New York State Fruit Industry. Geneva, NY 14456

Contact Us: NYSHS 630 W. North Street Hedrick Hall Geneva, NY 14456 www.NYSHS.org Ph 315-787-2404 Fx 315-787-2216 [email protected] www.NYSHS.org

36 NEW YORK STATE HORTICULTURAL SOCIETY