Summary
A bioRxiv preprint analysing 16,645 bench-grafted grapevines found that scion genetics explained a limited share of success variation. The study identified one modest, reproducible signal on chromosome 7, while most variation was linked to events affecting entire grafting bundles.
A bioRxiv preprint analysing 16,645 bench-grafted grapevines finds that the genetic identity of the scion—the upper part joined to a rooted plant—explained only a limited share of variation in grafting success. The researchers identified one modest, reproducible genomic region on chromosome 7, while most variation within the experiment was attributed to non-genetic events affecting entire grafting bundles.
The study, posted on September 16, 2026, examined 138 progeny from a Riesling × Cabernet Sauvignon cross, along with both parent plants. Each was grafted onto the 1103P rootstock, with 40 grafts per genotype in each of three separate field blocks and grafting performed on separate days. Overall grafting success averaged 91.1%.
Contents
What the experiment measured
Bench grafting is a propagation process in which scion material is joined to a compatible rootstock so the resulting vine combines the desired upper-plant traits with an established root system. The experiment focused on whether differences among scion genotypes could explain why some grafts formed successfully and others failed.
Scion genotype explained 19% of the variance among the field blocks. The researchers also reported an entry-mean heritability of 0.42, an estimate of how much variation among the tested genotype means was attributable to genetic differences under this experimental design. That result indicates a measurable inherited component, but not genetic control of most of the observed variation.
The interaction between genotype and block was larger than the variation attributed to genotype alone. A few genotypes also failed almost completely in one block. This means that the same genetic material could perform very differently under different block or batch conditions. Grafting success was not correlated with the yield of the mother vines—the plants supplying the scion material.
A small but repeatable genetic signal
The researchers used quantitative-trait-locus, or QTL, mapping to search for genomic regions associated with grafting success. They tested six definitions of the trait and used four haplotype-resolved assemblies of the parental genomes, which represent the two inherited chromosome copies separately.
One modest QTL on chromosome 7 was reproducible across the analyses. The preprint associates the effect with Cabernet Sauvignon alleles and estimates an increase of 3.9 percentage points in grafting success. This region accounted for 11% to 15% of the variance, depending on the analysis. The researchers also reported suggestive regions on chromosomes 5 and 9.
The chromosome 7 result is a genomic region associated with the trait rather than a reported causal gene or mechanism. The study therefore places an approximate scale on the genetic contribution in this cross: useful genetic differences exist, but the experiment did not find a large-effect locus that dominates grafting outcomes.
Why process control matters
Grapevine nurseries produce tens of millions of bench-grafted vines each year, making consistency important for propagation. The study’s main practical implication is that improving grafting success may require careful control of shared process conditions, not only selection of scion genotypes.
Because the largest share of variation was attributed to events affecting entire bundles, failures may be correlated among grafts handled together. That pattern can obscure genetic effects and makes replication across blocks and grafting days important when studying the trait.
The findings are specific to the tested elite Vitis vinifera cross, the 1103P rootstock and the study’s field-block design. They provide a baseline for the size and reproducibility of genetic effects in that setting; the preprint also describes the chromosome 5 and 9 findings as suggestive rather than as reproducible QTLs. The work is presented as a bioRxiv preprint.