Plant material.

All 28 UMN cultivars, their recorded parents, and ancestral selections and cultivars were genotyped for pedigree reconstruction (Supplemental Table 1). The individuals used in the study were cataloged by MUNQ codes (for Malus UNiQue genotype codes; Denancé et al., 2020), which were defined to facilitate international comparison of apple genetic resources as a development from the FBUNQ code described by Urrestarazu et al. (2016) based on SSR marker data. Individuals with the same MUNQ code are genotypic duplicates. These accessions are part of a large MUNQ code dataset (Denancé et al., 2020). Individuals that lacked SSR data and thus typical MUNQ attribution were given provisional MUNQ codes, typically derived from accession id numbers. Leaves from UMN cultivars were collected from orchards at the UMN Horticultural Research Center near Chaska, MN, and several other locations. Because many cultivar releases were reported as resulting from OP or unknown parentage, an effort was made to sample cultivars known to be historically grown in Minnesota (Green, 1903) and extant cultivars listed in early UMN pollination records extending back to 1916 (Horticultural Research Center, 1922, 1927, 1931, 1935, 1941, 1949, 1951, 1955). Additionally, SNP array data for more than 5000 cultivars and germplasm accessions included in a concurrent large-scale collaborative apple pedigree reconstruction project (Howard et al., 2018a) and for more than 1400 cultivars included in a previous pedigree reconstruction study (Muranty et al., 2020) were also included for the pedigree reconstruction of the UMN cultivars.

Genetic data.

Cultivars and breeding selections were genotyped either on the Illumina Infinium 20K apple SNP array (Bianco et al., 2014) with DNA extracted as described by Clark et al. (2014), or on the Axiom Affymetrix apple 480K SNP array (Bianco et al., 2016) as described by Muranty et al. (2020). SNP call data from both arrays were integrated in Howard et al. (2021b). A combined dataset was processed and curated as described in Vanderzande et al. (2019) to address Mendelian errors, creating a highly accurate dataset suitable for detailed pedigree reconstruction. The genetic map used was an edited form of the iGLmap (Di Pierro et al., 2016) described in Howard et al. (2021b) and included 10,295 SNPs.

Pedigree reconstruction.

Parent–offspring and parent–parent–offspring relationships were identified following the methods of Vanderzande et al. (2019), which relied on the identification of Mendelian inconsistent errors. In short, SNP calls between individuals and their prospective parent(s) were checked to be consistent with Mendelian inheritance. These relationships were considered true if they lacked Mendelian inconsistent errors across >99.9% of SNPs. The possibility of <0.1% errors was accepted because there are occasionally rare, undiagnosed technical issues or rare biological peculiarities (small deletions, duplications, new mutations, etc.) that can cause Mendelian errors that are not easily resolved or explainable in cluster plot data. Typically, there were no unexplainable Mendelian errors between individuals and their parent(s) following the SNP data curation methods used. Additionally, the possibilities of Mendelian errors arising from eu/aneuploidy were checked using methods described in Vanderzande et al. (2019). Other relationship types, particularly grandparent–grandchild relationships, were identified using methods described in Howard et al. (2021a). These methods made use of the interpretation of summed potential lengths of shared haplotype information generated using HapShared, a custom Python script. Phased SNP genotypic data used in these methods were generated using FlexQTL (Bink et al., 2014; Howard et al., 2021a). Individual grandparent–grandchild relationships were considered confirmed when they followed the type of logic described in the grandparent–grandchild relationship case study included in Howard et al. (2021a). In short, the following evidence was required for a grandparent–grandchild relationship to be considered confirmed: 1) extended haplotypes in a likely grandparent needed to be shared with about the expected 50% of the homologs of the likely grandchild; 2) if phased haplotype data for the prospective grandparent and its prospective grandchild were available, some evidence of extended haplotypes in the grandchild being composed of recombinant haplotypes from the grandparent was needed; 3) haplotypes from the prospective grandparent had to cover roughly 50% of the ends of the homologs of the prospective grandchild; and 4) a grandparent needed to be older than its grandchild (when reputable provenance information was available) such that they could reasonably have been in the lineage of a UMN cultivar based on introduction date (Table 1). Pairs of grandparents constituting the pedigree of an ungenotyped parent of an individual were considered confirmed when there were no Mendelian inconsistent errors present in such a matchup as also described in Howard et al. (2021a).

Pedigree reconstruction of UMN cultivars.

Partial or complete pedigrees could be reconstructed for each UMN cultivar based on SNP markers. The parentages as recorded by staff and as either confirmed (plain typeface) or identified in this study (bold typeface) are presented in Table 1. Available markers could not distinguish whether a parent served as female or male parent. For some cultivars, grandparents through an unknown or ungenotyped parent could be identified but not the actual parent itself.

Sixteen of the 21 cultivars introduced in the 20th century before ‘Honeycrisp’ had incorrect or incomplete recorded parentages that we could at least partially elucidate. In the early decades of the program, breeders selected prolifically among seedlings having OP origins to quickly identify adapted types (Dorsey, 1919). More than 1000 selections from OP seedlings were made from the 1910s through 1930s. Of the 14 cultivars introduced from these early selections, records indicated that five had unknown parentages and nine were from OP of specified maternal parents. We were able to identify at least one of the parents for each of the five cultivars with recorded unknown parentage. Of the nine cultivars with OP origin, the recorded seed parents of only three were supported by marker data. Misidentification of trees serving as seed parents, either by UMN staff or the owners of orchards where seeds were collected (Dorsey, 1919), or mislabeling of seeds or germinated seedlings may account for incorrect seed parents. Grandparents were identified for several early cultivars even though parents remain unknown. ‘Wealthy’, which was widely grown in the Midwest United States in the late 19th and early 20th centuries (Green, 1903), was a parent of 10 of the 14 UMN cultivars introduced from 1920 through 1940 (Table 1). These include ‘Haralson’ and ‘Fireside’, which were the two most widely grown cultivars in Minnesota in the mid to late 20th century. The parentage of ‘Haralson’ was recorded as ‘Malinda’ OP. ‘Malinda’ was confirmed as a parent and the other parent was identified as ‘Wealthy’. The parentage of ‘Fireside’ was recorded as unknown, but the actual parents were confirmed as ‘Wealthy’ and ‘Northwest Greening’.

Seed from controlled pollinations provided the basis for selection as the 20th century progressed and selection was more tempered. From the late 1930s through the 2000s, only ≈600 seedlings were selected for advanced testing, resulting in the introduction of 14 cultivars (Table 1). Our results confirmed recorded parentages for 10 of the 14 cultivars. Incorrect recorded parentages for four cultivars could be elucidated. Both recorded parents of ‘Honeycrisp’ were incorrect, as previously reported (Cabe et al., 2005; Howard et al., 2017). Likewise, the recorded parents of ‘Regent’ were refuted, and ‘Haralson’ × ‘McIntosh’ was confirmed as the correct parentage. For ‘State Fair’, ‘Mantet’ was confirmed as a parent and ‘Haralson’ was identified as the other parent, rather than ‘Oriole’. For ‘Centennial’, ‘Dolgo’ was confirmed as a parent and ‘Chestnut’, rather than ‘Wealthy’, was identified as the other parent. To avoid issues with incorrect pedigrees as we have identified with past cultivars, each new selection in the UMN breeding program is genotyped on the Illumina Infinium 20K apple SNP array to confirm its identity and parentage.

Pedigree connections to important ancestors.

Extended pedigrees of UMN cultivars connect to multiple important ancestors of European and North American origin (Fig. 1). Several important founders of European apple germplasm (Muranty et al., 2020) that we identified as ancestors of the UMN breeding program include, most prominently, ‘Duchess of Oldenburg’ (synonyms ‘Borowitsky’, ‘Borovinka’, and others listed in Bussey, 2016), but also ‘Alexander’ (synonym ‘Kaiser Alexander’) and ‘Reinette Franche’.

View Figure

• Download as Powerpoint
• Download Figure

‘Duchess of Oldenburg’ was an ancestor of overwhelming importance in the UMN breeding program (Table 1, Fig. 1). It was an ancestor of 27 of the 28 UMN cultivars, including as a parent of two cultivars and a grandparent of 15 cultivars, including ‘Honeycrisp’. ‘Duchess of Oldenburg’ was a parent of ‘Wealthy’ (Muranty et al., 2020), which was a parent of 10 of the 14 UMN cultivars introduced from 1920 through 1940. Green (1903) noted that in the late 19th century, ‘Duchess of Oldenburg’ was “the standard of hardiness in Minnesota and more generally grown than any other variety.” A sport, referred to in the UMN breeding records as ‘Red Duchess’ or ‘Daniels Red Duchess’, was used extensively as a seed parent of OP seedlings and in early controlled crosses through the 1930s. According to Bussey (2016), ‘Daniels Red Duchess’ was selected in 1902 as a sport of ‘Duchess of Oldenburg’ at Excelsior, MN which is near the UMN Fruit Breeding Farm.

‘Duchess of Oldenburg’, along with ‘Alexander’, ‘Tetofsky’, and ‘Red Astrachan’ (the latter of which was not represented in the UMN pedigree), are noted by Bussey (2016) as four important Russian cultivars first imported to Massachusetts from England in 1835. ‘Alexander’, which Green (1903) described as “lacking in hardiness and productiveness”, contributed only to the pedigrees of two relatively obscure early 20th-century cultivars, Folwell and Redwell.

‘Tetofsky’ is an important ancestor of the early ripening UMN cultivars Beacon, State Fair, Minnewashta, Minneiska, and MN55. Compared with ‘Duchess of Oldenburg’, which was among the cultivars Green (1903) considered to be of “first degree of hardiness,” he wrote that “Tetofski [sic] is of second degree of hardiness.” However, he described it as “one of the earliest if not the earliest apple to ripen in Minnesota and should have a place in every home orchard,” although premature fruit drop prevented it from being “profitable for market.” ‘Tetofsky’ is a grandparent of ‘Beacon’, an early-ripening cultivar that was regionally popular in the mid-20th century. ‘Tetofsky’ features in the ancestry of several early-ripening Canadian-bred cultivars: Petrel, Melba, Mantet, and Goodland. ‘Petrel’ and ‘Melba’ are grandparents of ‘July Red’ (Bussey, 2016), which is likely a great-grandparent of ‘MN55’ (discussed subsequently). ‘Mantet’ and ‘Goodland’ contributed to the early-ripening UMN cultivars State Fair, Minnewashta, and Minneiska.

‘Reinette Franche’, a 16th-century cultivar from Normandy, France, and an important ancestor of European germplasm (Muranty et al., 2020), was also identified as an important ancestor of UMN cultivars through ‘Wealthy’, via its parent, ‘Jonathan’ (Fig. 1). ‘Reinette Franche’ is also an ancestor to some UMN cultivars via two other historically important U.S. cultivars, Golden Delicious and Northern Spy. Their descent from ‘Reinette Franche’ was elucidated through a concurrent large-scale apple pedigree reconstruction project (Howard et al., 2018a) and is first reported here.

In addition to European ancestors, several cultivars that were widely grown in the United States in the first half of the 20th century (Dolan, 2009) entered the UMN extended pedigree, including ‘Jonathan’, ‘Northern Spy’, ‘McIntosh’, ‘Rome Beauty’, and ‘Grimes Golden’ (Fig. 1). ‘Northern Spy’, discovered circa 1800 in western New York, was estimated by the USDA to be the third most important cultivar in the United States during the period 1909–13 (Magness, 1941). ‘Northern Spy’ was recognized for its very good eating and culinary quality as well as storage ability but required an excessively long time to begin bearing fruit (Davis, 1925). Early UMN breeders likely sought to combine its fruit quality with the local adaptation of ‘Frostbite’ (then known as MN447). This cross gave rise to many selections, including two that were released, ‘Keepsake’, which is one parent of ‘Honeycrisp’, and ‘Sweet Sixteen’. ‘McIntosh’ appears in the pedigrees of UMN cultivars as a parent of ‘Victory’, ‘Northland’, and ‘Regent’. It is a more distant ancestor of ‘State Fair’, ‘Minnewashta’, and ‘Minneiska’ through UMN breeders’ use of Canadian cultivars, Mantet and Goodland. ‘McIntosh’ also enters the pedigree of UMN cultivar MN80 through its parent, ‘Liberty’, which contributed the allele for resistance to apple scab [Venturia inaequalis (Cooke) G. Winter] at the Rvi6 locus. ‘MN80’ is the only UMN cultivar derived from the introgressive backcrossing by the cooperative Purdue–Rutgers–Illinois breeding effort (Crosby et al., 1992) that introduced resistance to apple scab into M. ×domestica from M. floribunda Siebold ex Van Houtte.

‘Delicious’, a dominant North American cultivar of the mid to late 20th century (Volk et al., 2015), was used in crossing at UMN as early as 1918 (Dorsey, 1921), yet was notably rare in the pedigrees of UMN cultivars. Although discovered in the neighboring state of Iowa (Bussey, 2016), ‘Delicious’ has not exhibited consistent fruiting and tree survival for commercial production in Minnesota except in the far southeastern corner of the state. ‘Delicious’ occurred only as a distant relative in the pedigree of ‘MN55’ through its parent, AA44, a breeding selection from the University of Arkansas that was not formally released (C. Rom, University of Arkansas, personal communication) but has been in the public domain as ‘MonArk’ since at least 1993 (Norton and Way, 1999). The parentage of AA44 is unclear from breeding records and its pedigree could not be fully elucidated in this study. Rom et al. (1998) reported that AA44 was derived from a cross between two selections from the Rutgers University breeding program, 674016 and NJ40. Breeding records at Rutgers University (J. Goffreda, Rutgers University, personal communication) indicate that the parentage of selection 674016 was PCF4-56 (‘Mollie’s Delicious’ × ‘July Red’) × NJ40 {314049 [‘Blackjon’ (sport of ‘Jonathan’) × 55737 (‘Yellow Newtown’ × ‘Edgewood’)]} × 81248 {‘Jonathan’ × 11387 [‘Melba’ × (‘Williams’ × ‘Starr’)]}. None of the breeding selections, nor ‘Edgewood’, was available for genotyping, so the recorded pedigree of AA44 could not be confirmed nor denied. However, one parent of AA44 was confirmed to be a cross between ‘Mollie’s Delicious’ × ‘July Red’, suggesting that PCF4-56 was this parent. ‘Delicious’ is a great-grandparent of ‘Mollie’s Delicious’ {‘Mollie’s Delicious = (‘Wealthy’ × ‘Golden Delicious’) × [‘Orleans’ (‘Delicious’ × ‘Deacon Jones’) × (‘William’s Favorite’ × ‘Starr’)]} and is, thus, an ancestor of ‘MN55’, six generations removed. The haplotypes from the second parent of AA44 were found to be completely composed of haplotypes present in ‘Jonathan’, ‘Melba’, ‘Newtown Pippin’, ‘Starr’, F2 26829-2-2, and ‘Duchess of Oldenburg’.

‘Malinda’, although not widely grown in the United States, was frequently listed as a parent of early UMN cultivars in breeding records (Table 1). It was prized in the 19th century for its fruit appearance and long storage ability (Dorsey, 1919; Green, 1903). Dorsey (1919) recounted how seeds were collected in 1907 from trees of ‘Malinda’ that were topworked on ‘Duchess of Oldenburg’ in a private orchard in Waterville, MN, to establish seedling plantings in the new UMN apple breeding program. Dorsey (1919) noted that trees of “[Duchess of] Oldenburg, Wealthy, Scott Winter, Hibernal, English Russet, Patten Greening, Northwestern Greening, and a number of other varieties” were present in the orchard where the seeds resulting from open-pollination were collected. Cabe et al. (2005) found SSR markers did not support ‘Malinda’ as parent of ‘Frostbite’ or ‘Chestnut’. We confirmed ‘Malinda’ was not in the pedigree of either cultivar, nor in the pedigrees of ‘Folwell’ and ‘Minnehaha’. ‘Malinda’ also was not a parent of ‘Beacon’, although we could identify it as both a maternal and paternal grandparent. Nevertheless, ‘Malinda’ was an important program ancestor as a parent of three cultivars: Haralson, Lakeland, and Minjon. Through ‘Haralson’, ‘Malinda’ was an ancestor of several late 20th century cultivars (Fig. 1).

‘Honeycrisp’ pedigree.

The extended pedigree of ‘Honeycrisp’ described in this study (Fig. 1) is especially relevant as it is a major U.S. cultivar, and its descendants are being introduced from breeding programs globally (Kostick and Evans, 2018, 2020). Its pedigree was previously elucidated in Howard et al. (2017). We have extended this pedigree to the ancestors ‘Reinette Franche’ and ‘Utter’s Large Red’. We also confirmed that ‘Duchess of Oldenburg’ is a great-great-grandparent through grandparent ‘Frostbite’ (Fig. 1). Although ‘Honeycrisp’ is unconnected to most elite apple germplasm via first-degree relationships (Migicovsky et al., 2021), the ‘Reinette Franche’ ancestry connects ‘Honeycrisp’ to a large group of European and North American cultivar descendants of ‘Reinette Franche’ (Muranty et al., 2020). Haplotype contributions of ‘Reinette Franche’ to ‘Honeycrisp’ were responsible for the runs of homozygosity on chromosomes 7 and 15 previously reported in Howard et al. (2017), as ‘Reinette Franche’ was an ancestor of both the maternal and paternal parents of ‘Honeycrisp’. ‘Duchess of Oldenburg’ was previously speculated as a grandparent of ‘Frostbite’ (Howard et al., 2017). We confirmed this relationship, and thus, ‘Duchess of Oldenburg’ was responsible for runs of homozygosity on chromosomes 1 and 10 in ‘Honeycrisp’ previously reported in Howard et al. (2017). The pedigree of ‘Honeycrisp’s other great-great grandparent, ‘Utter’s Large Red’ (synonym ‘Utter Red’, ‘Utter’), a cultivar discovered in the Midwest United States in the early 19th century (Bussey, 2016), remains unknown. It was considered by Green (1903), despite its large size, to be “entirely unworthy of planting” but was nevertheless a contributor to ‘Honeycrisp’.