Two presentations of an F1 rootstock population (V. rupestris Scheele ‘B38’ (USDA PI#588160) X V. riparia Michx. ‘HP1’ (USDA PI#588271)) were utilized for the objectives of this research (Bhattarai et al., 2021) 1) the F1 grapevine progeny on their own roots and 2) replicate F1 progeny grafted with the common scion 'Marquette'. The grapevines were grown in greenhouse under natural light and 25-30/20-25°C day/night temperature at South Dakota State University Brookings, SD (44.31°C N, 96.80°W) in 2021 and 2022 (vines were 3 and 4 years old, respectively). Ecodormant spur-pruned vines were removed from the cooler the first week of June, root pruned, and repotted in a growing medium consisting of soil, perlite, and peat (1:2:2 by volume). Vines were watered daily and fertilized bimonthly using a custom trickle irrigation system with one dripper in each pot. The vines were maintained with 5 spurs, and a single shoot was trained vertically from each spur. The study used 135 own-rooted genotypes for both years (2021 and 2022) and a total of 139 (2021) and 71 grafted vines (2022, note several grafted vines were moved to a field planting and not available for this study). Leaves were collected from each vine during the growing season at natural seasonal daylengths of 14 h and 13 h (mid-August and first week of September, Supplementary Table 1 ). All vines were actively growing and with no shoot tip senescence or periderm development present during this period. At each timepoint an apical, middle, and basal leaf was collected from the same shoot. A total of 2,880 leaves were collected and landmarked in the two years. To capture the full range of morphological and developmental variation along the grapevine shoot, we sampled three leaves per vine representing young, middle, and old positions. Grapevine leaves vary in shape systematically during development, for example young expanding leaves at the shoot apex have prominent veins and reduced blade area, while mature leaves at the base of the shoot exhibit distinct heteroblastic shape compared to mid-shoot leaves (Chitwood et al., 2016b). Additionally, grapevine leaves have been shown to follow conserved allometric scaling, with vein-to-blade ratios decreasing as leaves expand, such that young, middle, and old leaves capture the physiological extremes and intermediate states of this shape change (Chitwood et al., 2021). The leaf from the apical region was identified as the first unfolded expanding leaf where the leaf blade oriented perpendicular to the direction of shoot growth. The middle leaves were fully expanded and midway between tip and base of the shoot. The basal leaves originated from the first node closest to the point where the shoot emerges from spur on the vine. The collected leaves were placed in Ziplock plastic bags and stored at 4°C until scanned (24 h). A genotype label, calibration card and leaf set comprising apical, middle, and basal leaves from each vine, were scanned using a Plustek OpticPro A320E (Lakewood NJ, US) Flatbed Scanner.

A total of 21 landmarks were manually labeled on the abaxial side of the images using Image J software (Abràmoff et al., 2004). As shown in Figure 1 , these 21 landmarks consisted of: (1) left side of the proximal vein base, (2) right side of the proximal vein base/left side of the distal vein base, (3) right side of the distal vein base/left side of the midvein base, (4) right side of the midvein base, (5) distal base of petiolar vein, (6) proximal base of petiolar vein, (7) width of proximal vein at petiolar vein branch point, (8) distal base of distal vein branch, (9) proximal base of distal vein branch, (10) width of distal vein at branch point, (11) distal base of midvein branch, (12) proximal base of midvein branch, (13) width of midvein at branch point, (14) tip of petiolar vein, (15) tip of proximal lobe, (16) proximal sinus, (17) tip of distal vein branch, (18) tip of distal lobe, (19) distal sinus, (20) tip of midvein branch, and (21) tip of the leaf (Migicovsky et al., 2022; Chitwood et al., 2021; Bryson et al., 2020). To evaluate for position errors, ggplot in R (Wickham and Sievert, 2009) was used to replot the landmark coordinates into images, and images with errors were re-landmarked.

The ampelographic finite traits were derived using the 21 landmark coordinates. A total of 16 different phenotypes were obtained for each leaf sample, as outlined in Table 1 . Distances between landmarks were also derived using the Pythagorean theorem (Maor, 2019). Ten phenotypes related to leaf size were computed, using the overall area of the leaf and the vein and blade area, utilizing the shoelace algorithm (Pure and Durrani, 2015). In addition, six different ratios were derived using area and length metrics as indirect measures describing proportion or pattern related to leaf shape.

The Kruskal-Wallis test (McKight and Najab, 2010), a non-parametric approach was used to determine whether there were significant differences between treatment groups while comparing trait measures across time points or presentation types. The exploratory data analysis and visualization was performed in R utilizing ggplot and ggpubr (Kassambara, 2018). Pearson correlation analysis of traits was conducted using Pearson method and visualized using ‘ggcorrplot’ package in R. PCA was implemented using ‘prcomp’ builtin function in R to reduce dimension separating leaf morphological traits into size and shape categories. Relationship between distinct factors and phenotypic traits were explored generalized additive model (GAM) using ‘mgcv’ package in R. For descriptive analysis, the data were z-transformed to account for the scale differences associated with leaf positions and allow for standardized comparison in a unified figure. However, the original data were retained for QTL analysis preserving the accuracy of effect estimation.

A pathway enrichment analysis was performed for loci hotspots (> 3 QTL) where individual size-related traits were identified with overlapping confidence intervals and same peak marker position. Genes located within 700 Kb either side of the peak position were retrieved for enrichment analysis. This range was used to meet the criterion that candidate genes be within 3–4 cM of the peak position as determined by the size of the grape genome; therefore, a total of 1.4 Mb centered on the peak position was used (Cipriani et al., 2011; Hugalde et al., 2021). The VitisNet functional annotation of the genes within this 1.4 Mb region a Fisher’s test (p-value < 0.05 for gene network enrichment analysis was conducted (Grimplet et al., 2012; Alahakoon and Fennell, 2023; Osier, 2016).

The sixteen different ampelographic phenotypes revealed different relationships according to leaf position and daylength as shown by Pearson correlation analysis ( Supplementary Figures 4 , 5 ). Across all leaf positions, size related traits (area and length), had strong positive correlation with each other suggesting coordinated growth patterns in both presentation types. Veins traits correlated well with blade size, but their relationship with shape traits was low and highly variable across leaf positions. ( Figures 2A, B ) shows how leaf shape and size varied by position on the plant (apical, middle, and basal) under different daylengths in 2021 and 2022. In both own-rooted ( Figure 2Ai ) and grafted plants ( Figure 2Bi ), leaf position had a strong effect on leaf shape and size changes related to daylength. For both own-rooted F1_Vruprip and grafted vines, the apical leaf z transformed scores presented in Figure 2 are consistently clustered in the lower range of values for 14 h and 13 h daylength. The 14 h was more similar for both presentations, while the 13 h violin plots showed a broader spread in the grafted presentation ( Figures 2Ai, Bi ). In contrast, the visualizations using shape z transformed scores showed a broad distribution and less influence of environmental factors (year, photoperiod) ( Figures 2Aii, Bii ).

The Kruskal-Wallis test indicated that daylength influenced the leaf size in own-rooted and grafted presentations ( Supplementary Table 2 ). Shape related traits were not similarly affected in the own-rooted presentation but were noted in the middle and apical leaf positions in the grafted presentation. To further understand daylength influence on leaf morphology, individual traits were examined for each leaf position. ( Figures 3Ai, Aii ) showed that the size traits of own-rooted vines had similar patterns of change for the 13 h relative to the 14 h daylength. For the own-rooted presentation, the highest median leaf trait values were observed at 13 h for apical and basal leaves, in contrast to 14 h for middle leaf. In the grafted presentation, the scenario was different for the basal leaves, where 13 h and 14 h daylength measurements were similar for most traits, except for distal vein length. In the case of shape related traits, the shape changes in own-rooted and grafted presentation in 13 h relative to the 14 h were similar; however, the magnitude of change between the 14 h and 13 h daylength appeared to differ between the own-rooted and grafted presentations for some traits ( Figures 3Bi, Bii ). The median trait differences between daylengths were greater for the grafted presentation compared to the own-rooted for traits like proximal to distal vein length ratio and distal to mid vein length ratio in middle leaves.

Principal component analysis performed on leaf morphological features, including size and shape attributes indicated that the first five PC components accounted for more than 95 percent of the overall variation observed in both own-rooted ( Figure 4Ai ) and grafted ( Figure 4Aii ) presentations. Year, genotype, position, daylength, and their interaction were determined to be the key factors in explaining the variance in leaf morphological phenotype (as represented by each PC score) using generalized additive models (GAM) for each PC score ( Figures 4Ai, Aii ). The findings showed that these factors had varying effects on distinct features of leaf morphology, with some aspects being more strongly influenced by one given factor than by another. According to the color-coded illustrations, the interaction effect between position and root genotype accounted for the highest percentage of variance contributing to PC1 for both own-rooted (4Ai) and grafted (4Aii) presentations.

The PCA biplot aids to further elucidate this relationship ( Figures 4Bi, Bii ) where the plots are categorized by daylength (14 h and 13 h), with distinct plots representing two distinct years (2021 in the top row and 2022 in the bottom row). Separation of the leaf positions across PC1 showed that each leaf position had its own unique morphological profile. There is a greater spread for the genotypes within each leaf position in both PC1 and PC2 for the own-rooted vines. Consistent patterns across the two years implied that the size and shape traits responded similarly to daylength differences. In both own-rooted (Bi) and grafted (Bii) presentations, the apical leaf group exhibited a distinct set of morphological characteristics that were detected by PC1 and PC2. The middle and basal leaf clusters overlapped in both 14 h and 13 h daylengths. When comparing presentation types, the points representing basal, middle, and apical leaf are more tightly clustered in grafted plants indicating lower genetic variability for all three leaf types. In contrast, the own-rooted plants show a broader range of variability representing the genetic differences within own-rooted leaves.

There was no significant difference in size related traits when compared across the own-rooted and grafted presentation types as shown by the Kruskal-Wallis test ( Supplementary Table 3 ). However, for the shape related measures, two traits were significantly different between the presentations, the proximal to distal vein length ratio and petiolar sinus length to leaf area ratio. The difference is shown in the violin plots for 2021 and 2022 ( Figures 5A, B ). The median values for the proximal to distal vein length ratio the response to daylength was consistent between the own-rooted and grafted presentations in 2021 and 2022. The petiolar sinus length to leaf area ratio ( Figure 5B ) presented a notable difference in the pattern of variation, with the distribution of genotype samples being more spread out in the own-rooted presentation than in the grafted vines in both years.

During map curation, 133 singleton and 901 monomorphic markers were found and no markers were identified as not grouping with a linkage group (LG). A total of 962 (48.1%) of the genotyped markers were anchored in 19 chromosomes spanning a total genetic distance of 1741 cM ( Supplementary Table 4 ). The number of markers varied from 29 to 90 markers per LG, with an average of 50 markers anchored on each LG. The length of LG ranged from 71.1 cM (LG 15) to 154.4 cM (LG 14), with an average distance of 1.8 cM between markers ( Supplementary Table 3 ). Genome-wide recombination rate of the integrated map was 0.24 cM/Mbp. The largest gap occurred on LG14 (33.3 cM) with the maximum gap on the other chromosomes between 6.7 to 17.5 cM. The genetic map covered 85% to 99% across the LG and there was strong collinearity between the genetic map when compared with the reference V. vinifera PN40024 12X.v2 genome providing a high genetic map quality ( Supplementary Figure 1 ). No marker ordering errors were identified in this final map ( Supplementary Figure 2 ) The F1_Vruprip rhAmpSeq was validated using sex type for a portion of the population grown in field using the binary QTL mapping method and subsequently the map was used for QTL analysis of leaf traits (Alahakoon et al., 2022).

Five QTL hotspots, two in the own-rooted (rh14_5687927, rh19_472320) and three in grafted (rh2_17347640, rh5_2193315, and rh16_5479328) containing three or greater collocated QTL with identical peak markers were identified and pathway enrichment analyzed ( Table 5 ). The QTL regions for the own-rooted QTL on chromosomes 14 and 19 identified 11 and 12 enriched pathways for DNA replication, amino acid metabolism, and transcription factors ( Supplementary Table 8 ). In the grafted presentation there were eight, five, and 11 enriched pathways for hotspots on chromosomes eight, five, and 11 respectively ( Supplementary Table 8 ). Transporters, transcription factors, and carbohydrate metabolism were found in these enriched pathways. The total lists of the genes analyzed for pathway enrichment can be found in Supplementary Table 9 .

Enriched pathways, including galactose metabolism, protein export, accessory factors involved in transport, BES1, and FHA, occurred repeatedly in multiple QTL regions. The cumulative enriched pathway results from all five QTL hotspot regions are presented in Supplementary Table 8 . A total of five distinct enriched pathways associated with amino acid metabolism were identified. Various processes, including DNA replication, thylakoid targeting pathway, cell cycle, and RNA polymerase, were identified regarding protein modifications and cellular processes and growth. A collective of five distinct enriched pathways were identified regarding transport mechanisms. In addition, twelve distinct transcription factors were identified, including BES1, GNAT, FHA, and CAMTA ( Supplementary Table 8 ). Signal transduction-related enriched pathways for brassinosteriod, auxin, and ABA signaling were found. Likewise, a total of six distinct pathways for carbohydrate and lipid metabolism were identified.

Daylength varies by latitude and season and is a well-known regulator of plant development and physiology (Jackson, 2009). In Arabidopsis thaliana, daylength reductions have been reported to reduce the absolute rate of leaf expansion, suggesting that photoperiod has a direct influence on the dynamics of leaf development (Cookson et al., 2007). The progression or temporal shift related to leaf size and shape in response to a decrease in photoperiod in this study varied with the i) type of leaf trait (size or shape) and ii) leaf position. Visual examination of the leaves suggested that the shape related traits changes were less variable than size-related traits across daylength. While leaf shape was more conserved, it still exhibited variability within certain bounds to changing daylength conditions as evidenced by the results of Kruskal-Wallis test. Specifically, the middle leaf differed significantly when compared across daylengths for both years in the grafted presentation and for 2022 in the own-rooted presentation. Similarly, in Vitis hybrid ‘Marquette’ grafted to five commercial rootstocks: 1103 Paulsen (1103P), 3309 Couderc (3309C), Teleki 5C (5C), Freedom (FREE), Selection Oppenheim 4 (SO4), as well as homografted controls, physiological assessments using a portable photosynthesis system (LI-6800; LI-COR Biosciences, Lincoln, NE, USA) showed a decrease in the net assimilation rate (A), the maximum rate of carboxylation (Vcmax), and the maximum electron transport rate (Jmax) in middle leaf, when daylength decreased from 15 to 14 hours (Sharma et al., 2024a, Sharma et al., 2024b). These three studies suggest that the middle leaf is more developmentally plastic in terms of structural features and responds to environmental influences to a greater extent. In contrast, the basal leaf remained consistently non-significant for both own-rooted and grafted presentations, these leaves were formed before the leaf measurement began; therefore, would not be expected to show morphological shifts during the study, although slight variations could occur as basal leaves can be shed as they age. This finding aligns with a previous V. riparia study where oldest leaf was found to be invariant likely because they focused on maximizing early season growth or are controlled by genetics that limit changes in their shape and size (Chitwood et al., 2016b; Baumgartner et al., 2020). In contrast, intermediate leaves showed greater phenotypic plasticity and were interpreted to have strongest climate signals (Chitwood et al., 2016b). The major takeaway is that there was significant difference in leaf size and shape traits across different daylengths. However, the underlying drivers of this phenotypic plasticity are complex and could be related to combination of environmental cue like daylength, allometry (varying growth rates across an organ) or heteroblasty (trait differences based on successive nodes) or combination of these (Baumgartner et al., 2020; Evans, 1972; Spriggs et al., 2018; Chitwood et al., 2014a; Coleman et al., 1994).

To develop a greater understanding of the major factors contributing to variation in leaf morphology (shape and size) several factors were considered i.e., i) genotype effect ii) environmental effect (year, daylength) and iii) developmental effect (leaf position). These studies suggested that interaction between genotype and position were major contributors to the variation in leaf morphological traits. PC1 separated the root genotypes based on leaf positions, where apical leaf was separated from middle and basal leaf in all cases (multiple daylength and year) ( Figures 4Aii, Bii ). Moreover, the variation between the data points along y axis showed the variation within specific leaf positions for genotypes. This y axis or PC2 dimension primarily showed how genotypes varied at particular position for given presentation type. Klein et al. (2017), leaf morphometrics study illustrates differences between V. rupestris Sheele and V. riparia Michx. with linear discriminant analysis and was able to predict their leaf shape with accuracy of >98%. Our study utilized two F1 presentations derived from these two species, and the observed differences observed in morphology could be attributed to the genetic contribution of either parent. It is noted in the PC biplot that own-rooted had a greater distribution than that conferred to the common scion in the grafted presentation. This implied that the common scion ‘Marquette’ may have minimized variations within the grafted presentations. When comparing presentation types, only two traits i.e., proximal to distal vein length ratios and petiolar sinus to leaf area ratios were significantly different between own-rooted and grafted types ( Supplementary Table 3 ). This indicated that the genetically different root systems conferred an influence on the common scion and affected the narrowness/width of leaves when measured perpendicular to midrib as well as the length of petiolar sinus. Pearson correlation of the 2023 pruning weights (year following leaf study) with the leaf total area and blade area (2022) of this study showed significant correlation and provide further support for the rootstock influence on leaf morphology subsequently impacting the scion shoots ( Supplementary Table 10 ). Rootstock-scion interaction is a complex and dynamic phenomenon in grapevines, and the extent to which rootstock influences scion remains an area of active research. Previous studies in grapevines have demonstrated that rootstock has a direct influence on elemental compositions of leaf, its shape, as well as physiological responses (Migicovsky et al., 2019; Harris et al., 2021; Sharma et al., 2024a, Sharma et al., 2024b). For example, research on V. vinifera 'Italia' grapes comparing ungrafted vines and grafted vines (two distinct rootstocks) under two irrigation settings revealed that the leaf area was significantly influenced by rootstock-by-irrigation interaction (Sabir, 2016). Like these findings, our study identified certain morphological features associated with rootstock effect. This suggests broader implications of rootstock selection in viticulture aiming for vine performance improvement, optimizing canopy structure and tailoring growth responses based on production goals.

The QTL for the own-rooted and grafted presentations provided different genetic loci associations. For the own-rooted presentation, QTL were frequently identified on chromosomes 8,14, and 19 ( Table 2 ). These QTL explained from 12-17% variation for a given phenotype. In contrast, the QTL for the grafted presentation of the F1 population were mostly identified in chromosome 2, 5 and 16 and explained 11- 27% total phenotypic variation ( Table 3 ). For shape related trait measures, there were no hotspots from co-localized traits, but there was a QTL with exceptionally higher LOD score 19.63 ( Table 4 ). This QTL was responsible for around 48% of phenotypic variation in the apical leaf of grafted presentation and was associated with chromosome 19. This is a significant foundation for further investigation as this suggests a strong potential leaf morphometric phenotype.

The QTL hotspots regions selected from size traits identified pathways known to be associated with leaf growth and development ( Table 5 ). The enriched pathways identified under QTL hotpot regions were associated with development related processes like cell cycle, thylakoid targeting pathway, DNA replication as well as protein export. Similarly, enriched carbohydrate-related pathways (starch, sucrose, galactose) and amino acid (glutamate, cysteine, lysine, tyrosine) related metabolic pathways were identified. Leaf size increase or other leaf growth characters requires substantial demand of carbon skeleton molecules to build new structures like cellulose and hemicellulose in cell walls. It also involves the use of photosynthetic carbon to initiate various energy consuming mechanisms like cell cycle and protein synthesis. The leaf growth rate is correlated with activity of carbon metabolism enzymes or level of carbon metabolites (Sulpice et al., 2009; Fichtner et al., 1993; Cross et al., 2006). Similarly, amino acids serve as precursor for the synthesis of variety of compounds necessary for plant developments like chlorophyll, nucleotides, and secondary metabolites (Tegeder, 2012). The rh19_472320 QTL hotspot had one of the most interesting characteristics with three significant interrelated pathways Brassinosteroid_biosynthesis vv10905, Brassinosteroids_signaling vv30005, and Brassinosteroid-activated BRI1-EMS-Supressor transcription factors vv60010BES1. Brassinosteroids (BR) are growth-promoting hormones known to have role in plant cell expansion, elongation, and proliferation impacting overall leaf size (Zhang et al., 2021; Clément et al., 2018; Zhiponova et al., 2013; Gudesblat and Russinova 2011). One other hormone signaling process identified was ABA signaling. ABA is found to impact both expansive and structural growth inhibiting cell and plastid division (Pantin et al., 2012; Wang et al., 1998; Galpaz et al., 2008). ABA not only regulates stomata and aquaporins controlling incoming carbon dioxide, but it is also found to regulate carbon metabolism at enzymatic level at both transcriptional and post transcriptional scenarios (Zhu et al., 2011). BES1 transcription factor, which was identified in two hotspot peak regions, is known to activate BRinduced gene expression in plants (Yin et al., 2005). The other repeatedly occurring transcription factor was fork-head transcription factor (FH). FHA or fork-head associated proteins are known to be associated with multiple functions in regulating plant organ development, signal transduction, hormone response, and DNA damage repair in Arabidopsis (Wang, 2023). In summary, findings of leaf growth associated pathways reveal that the identified QTL are important and need to be further validated in natural environment settings.