The use of hybrid breeding systems to increase crop yields has been the cornerstone of modern agriculture and is exemplified in the breeding and improvement of cultivated sunflower (Helianthus annuus). However, it is poorly understood what effect supporting separate breeding pools in such systems, combined with continued selection for yield, may have on leaf ecophysiology and specialized metabolite variation. We analyzed 288 cultivated H. annuus lines to examine the genomic basis of several specialized metabolites and agronomically important traits across major heterotic groups. Heterotic group identity supports phenotypic divergences between fertility restoring and cytoplasmic male-sterility maintainer lines in leaf ecophysiology and specialized metabolism. However, the divergence is not associated with physical linkage to nuclear genes that support current hybrid breeding systems in cultivated H. annuus. Further, we identified four genomic regions associated with variation in leaf ecophysiology and specialized metabolism that co-localize with previously identified QTLs in cultivated H. annuus for quantitative self-compatibility traits and with SPH-proteins, a recently discovered family of proteins associated with self-incompatibility and self/nonself recognition in Papaver rhoeas (common poppy) with suggested conserved downstream mechanisms among eudicots. Self-compatibility is a derived trait in cultivated H. annuus with quantitative variation in selfing success, suggesting that trait linkage to divergent phenotypic traits may have partially arisen as a potential unintended consequence of historical breeding practices. Further work is necessary to confirm the self-incompatibility mechanisms in cultivated H. annuus and their relationship to the integrative and polygenic architecture of leaf ecophysiology and specialized metabolism in cultivated sunflower. 

An association mapping population of sunflower (Helianthus annuus) was grown in growth chambers in 3-4 replicates. For each replicate of every genotype, we identified the pair of most recently fully expanded leaves. For this leaf pair, one of the two leaves was used to assess ecophysiological traits: leaf fresh mass, dry mass, water content, leaf area, perimeter, aspect ratio, circularity, solidity, leaf mass per area (LMA), lamina thickness, lamina toughness, lamina density, and chlorophyll content on an area and mass basis. From the selected ecophysiology leaf, we first took chlorophyll content measurements in SPAD units and then clipped the leaf at the base of the petiole and weighed it for fresh mass. We scanned the leaf with a digital scanner to later derive leaf area, perimeter, aspect ratio, circularity, and solidity via ImageJ. Following scanning, we measured lamina thickness with digital calipers at the midpoint of the leaf's length, avoiding major veins. Lamina toughness was assessed using a digital penetrometer, averaging three measurements of force required to penetrate the leaf lamina with a millimeter-wide flat-tipped needle. Leaves were then dried at 60°C in a forced-air oven for 96 hours and weighed for dry mass. SPAD scores were converted into an area-basis (μmol m^-2). Dry mass was used to calculate other traits, including water content, LMA, and mass-based chlorophyll content. In an unfortunate turn of events, dried leaves were accidentally destroyed for a subset of lines (N=125) while in storage awaiting processing, resulting in missing data for these lines for dry mass and the three traits derived using it. The missing data is listed as n/a in each respective cell.

We took the opposing leaf to phenotype coarse defense traits and sample tissue for analytical chemistry analysis of secondary metabolism. We split this leaf down the midrib, and one half was rolled, placed into a microcentrifuge tube, and snap-frozen in liquid nitrogen. This leaf sample was stored at -80°C for later analysis. The other half of the leaf was dried at 60°C for at least 36 h. This dried leaf material was used to assess four coarse leaf defense metrics: trichome density, ash content, total phenolics, and total flavonoids. Trichome density was calculated using a dissecting scope to count the trichomes present in a 0.25 cm² region. The dried leaf tissue was then ground into a fine, homogenous powder. This material was used to estimate total flavonoids via the aluminum complexation assay and estimate total phenolics via the Folin-Ciocalteu assay. To estimate total ash content, leaf powder was combusted in crucibles in a muffle furnace at 600°C for 12 hours, and the pre-and post-combustion mass was compared.

To characterize broad metabolomic variation across cultivars, we used an untargeted approach. For each genotype, replicate frozen leaf samples were pooled in even proportion by mass, and these samples were ground to homogeneity under liquid nitrogen. A total of 10 mg of homogenized frozen material was extracted in 0.4 ml of 1:1 methanol:chloroform (v/v. in a sonicated ice-water bath for 30 min. Post-extraction, 0.2 ml of HPLC-grade water was added to each sample, and the mixture was vortexed and centrifuged. The aqueous layer was collected and further vortexed and centrifuged. The resulting aqueous layer extract was used for analysis by high-performance liquid chromatography (HPLC).

three µl of extract from each sample was injected into an Agilent 1200 HPLC with a ultraviolet diode array detector (Agilent Technologies, Santa Clara, California, USA). Separation was achieved using an Agilent ZORBAX Rapid Resolution Eclipse XDB-C18 column (4.6 ⋅ 50 mm 1.8 μm), with mobile phase solvents of water:acetonitrile:formic acid = 97:3:0.1 (A) and 3:97:0.1 (B). The flow rate was set to 1 ml min^-1, and the elution gradient was as follows: 3% B from 0-1 min, linear gradient to 17% B over 2 min, isocratic at 17% B for 2 min, linear gradient to 60% B over 4 min, then to 98% B over 2 min. The ultraviolet diode array detector was set at 260, 270, 280, 310, and 350 nm. We chose only the 350nm wavelength for analyses based on the high number of peaks detected via the diode array detector at this wavelength versus the others. We linearly calibrated retention time and peak area based on the averages of retention time (RT) and peak areas for two common peaks (RT 4.11 and 7.05) found among all samples. We further used manual binning of peaks based on retention time similarity, producing a list of metabolites present among at least 200 of 288 lines.