Genomic consequences of apple improvement

Abstract

The apple (Malus domestica) is one of the world's most commercially important perennial crops and its improvement has been the focus of human effort for thousands of years. Here, we genetically characterise over 1000 apple accessions from the United States Department of Agriculture (USDA) germplasm collection using over 30,000 single-nucleotide polymorphisms (SNPs). We confirm the close genetic relationship between modern apple cultivars and their primary progenitor species, Malus sieversii from Central Asia, and find that cider apples derive more of their ancestry from the European crabapple, Malus sylvestris, than do dessert apples. We determine that most of the USDA collection is a large complex pedigree: over half of the collection is interconnected by a series of first-degree relationships. In addition, 15% of the accessions have a first-degree relationship with one of the top 8 cultivars produced in the USA. With the exception of 'Honeycrisp', the top 8 cultivars are interconnected to each other via pedigree relationships. The cultivars 'Golden Delicious' and 'Red Delicious' were found to have over 60 first-degree relatives, consistent with their repeated use by apple breeders. We detected a signature of intense selection for red skin and provide evidence that breeders also selected for increased firmness. Our results suggest that Americans are eating apples largely from a single family tree and that the apple's future improvement will benefit from increased exploitation of its tremendous natural genetic diversity.

Fig. 1. Principal components analysis (PCA) showing population structure for M. domestica (N = 749), M. sieversii (N = 115), and M. sylvestris (N = 11).

A PC axes were established using equal sample sizes (N = 11) of the wild species and the remaining accessions were projected onto the axes. The percent variance explained by each PC is indicated in parentheses. B Comparison of PC1 values for cider and dessert apples demonstrates that dessert apples derive more ancestry from M. sieversii, while cider apples derive more ancestry from M. sylvestris

Fig. 2. Genome-wide scans for selection during apple domestication.

A Fst and B XP-EHH selection scans comparing M. domestica and M. sieversii. The horizontal lines indicate the top 5% of values for each test across the genome

Fig. 3. Clonal and first-degree relationships within the USDA apple collection.

A Network of clonal relationships among accessions labelled as M. domestica. Only accessions with at least one clonal relationship (N = 279) are included. Each accession is represented by a dot and each line represents a clonal relationship. B Number of clonal relationships across entire collection (N = 1005). C Network of first-degree relationships among apple accessions. Only accessions with at least one first-degree relationship (N = 535) are included. Each accession is represented by a dot and each line represents a first-degree relationship. D Number of first-degree relationships for each of the unique accessions within the collection (N = 826)

Fig. 4. Network of first-degree relationships with the top 8 apple cultivars sold in the United States.

In addition to the top apple cultivars, only accessions with at least one first-degree relationship (N = 129) are included. Each accession is represented by a dot and each line represents a first-degree relationship

Fig. 5. Genome-wide scan for selection during apple improvement.

A Fst and B XP-EHH selection scan profiles comparing dessert and cider accessions. The horizontal lines indicate the top or bottom 5% of values for each test across the entire genome. For XP-EHH, high values indicate selection in dessert apples while low values indicate selection in cider apples

Fig. 6. GWAS and XP-EHH selection scan for apple skin colour. Dashed vertical lines indicate the location of the redTE retrotransposon just upstream of the MYB1 gene, a known key regulator of apple skin colour.

A Manhattan plot of GWAS results for fruit colour. p-values are log-transformed and the horizontal line indicates a Bonferroni-corrected p-value threshold for significance. B XP-EHH selection scan profiles comparing red and green accessions. Positive values indicate selection for red colour. The horizontal lines indicate the top 5% of values across the genome

Fig. 7. GWAS and XP-EHH selection scan for fruit firmness.

The position of a previously identified SNP in NAC18.1 (D5Y) is indicated with a green dot. A Manhattan plot of GWAS results for fruit firmness. p-values are log-transformed and the horizontal line indicates a Bonferroni-corrected p-value threshold for significance. B XP-EHH selection scan comparing soft and firm accessions. Positive values indicate selection for firmer apples. The horizontal lines indicate the top 5% of values across the entire genome