Q. Chen, L. chen, J. Fan, X. Yang, J. Zhang, W. Du, W. Liu, Z. Liu, H. Hu
Germplasm collections are expanding worldwide, yet pedigree errors, duplicate accessions, and undocumented kinship remain bottlenecks in genebank curation. Existing tools such as CERVUS and COLONY rely on likelihood-based frameworks that require allele-frequency estimates and operate in pairwise mode, providing no collection-wide view of Mendelian consistency. We applied the Mendelian Mismatch Matrix (MMM) framework to publicly available apple (Malus domestica; 1,085 accessions, 15 SSRs) and sweet cherry (Prunus avium; 383 accessions, 13 SSRs) datasets from the German Fruit Genebank. MMM converts the pairwise Mendelian mismatch number (Mmn) into an N x N symmetric matrix, from which four collection-scale indices are derived: Mmin (Mendelian minimum mismatch number), Mavg (Mendelian average mismatch number), Mzmp (Mendelian zero-mismatch partner number), and Mmmd (Mendelian mismatch-mode duplication index), together with the Mendelian Exhaustive Stratification (MES) grading system (grades D0-Dn). Apple displayed a gradient-type structure (Mendelian Grade-gap Index, MGI = 0.125), whereas sweet cherry exhibited a broken-type structure (MGI = 0.545). The Mmin-Mavg strategy identified 20 and 10 highly distinct accessions in apple and sweet cherry, respectively, including the scab-resistance donor Malus x floribunda 821 (Rvi6) and the fire-blight donor M. x robusta 5. Of CERVUS-confirmed parent-offspring pairs, 97.31% fell within Mmn <= 1 in apple and 100% of high-confidence pairs in sweet cherry. Mzmp rankings matched documented breeding history, with 'Cox Orange' (Mzmp = 57) and 'Lubecker Bunte' (Mzmp = 49) ranking first in each species. All 588,070 apple pairwise comparisons were computed in 7.24 s on a standard workstation. As implemented in the open-source MMMAS (Mendelian Mismatch Matrix Analysis System), this deterministic, parameter-free pre-screening layer supports routine quality assessment in germplasm repositories.