Image data set of sun and shade-grown leaves of Cabernet Sauvignon (CS) and Blaufränkisch (BF) grapevine (Vitis vinifera L.) cultivars. When using this dataset, please cite: Théroux-Rancourt G, Herrera C, Voggeneder K, Luijken N, Nocker L, Savi T, Scheffknecht S, Schneck M, Tholen D. (accepted) Analyzing anatomy over three dimensions unpacks the differences in mesophyll diffusive area between sun and shade Vitis vinifera leaves. AoB Plants Data acquisition methodology Plants were brought to the TOMCAT tomographic beamline of the Swiss Light Source at the Paul Scherrer Institute (Villigen, Switzerland) in pots (Cabernet Sauvignon (CS)) or as cut shoots with the cut end placed in water (Blaufränkisch (BF)). Before scanning, a leaf was cut from the stem and a thin strip of ~1.5 mm width and 1.5 cm length was cut in between apparent higher-order veins, immediately wrapped in polyimide tape and inserted into a styrofoam block glued with wax onto a holder. Three (CS) or two (BF) strips were cut at different locations on the leaf surface to ensure within-leaf replications and to get better leaf-level averages. The strip was immediately scanned by imaging 1801 projections of 100 ms under a beam energy of 21 keV and magnified using a 40x (CS) or 20x (BF) objective, yielding respective final voxel sizes of 0.1625 µm (field of view: ~416x416x312 µm) and 0.325 µm (field of view: ~832x832x624 µm). Scanned projections were reconstructed to cross-sectional view using both absorption (gridrec; Marone et al. 2012) and phase contrast enhancement (Paganin et al. 2002) reconstructions. Dataset description After aligning the stacks to be parallel to the image edges using ImageJ (Schneider et al. 2012), at least nine slices were hand labeled using a graphics pen display tablet to precisely segment the background, the epidermis, and the vasculature. The mesophyll cells and the intercellular airspace were segmented by thresholding each absorption and phase contrast scan to maximize airspace volume (background) and taking care to avoid false segmentation within the cells (i.e. false segmentation of airspace). The hand-labeled slices were then used to automatically segment the whole stack using a Python-based random-forest machine learning approach (Théroux-Rancourt, Jenkins, et al. 2020). File naming convention For all stacks, files start with: Cultivar_Treatment_Plantn_leaf_N_ Cultivar: CS for Cabernet Sauvignon, BF for Blaufränkisch Treatment: Sun for plants grown under high light, Shade for plants grown under low light Plantn: Plant number; 1-6 for CS, 1-5 for BF leaf: Leaf number; 1-4 N specifies the type of stack: GRID (gridrec reconstruction) PAGANIN (phase contrast enhancement reconstruction) labelled-stack (hand labeled slices / ground truth; stacks have been hand labeled in cross-sectional view.) SEGMENTED (automatically segmented stack using random-forest machine learning approach) STOMATAL_REGIONS_BBOX_CROPPED (Stack with individually segmented stomatal vaporsheds, i.e. airspace closest to a stoma. The stack has been cropped in paradermal view around the stomata closest to the stack's edges, i.e. a bounding box (BBOX) with stomatal vaporsheds fully enclosed). All stacks are provided as 8-bit grayscale TIF files. Stacks have been hand labeled in cross-sectional view. Plant material and growth conditions The experiment was carried out over two consecutive years, in 2018 and 2019, at the facilities of BOKU UFT (Tulln, Austria). In the first year, rooted grafts of Vitis vinifera ‘Cabernet Sauvignon’ (clone 191E) on 101-14 rootstock (hereafter named CS) were acquired from a local nursery (Reben Iby, Neckenmarkt, Austria). In the second year, rooted grafts of Vitis vinifera ‘Blaufränkisch’ (clone 13-3 GM) on 5 BB rootstock (hereafter named BF) were acquired from the same nursery. Blaufränkisch is known to have originated from Lower Styria (present day Styria, Slovenia, Maul et al. 2016) and to be genetically different from Cabernet Sauvignon (Magris et al. 2021), which originated in the Bordeaux region in France. The rooted grafts were planted in 7-L pots and allowed to grow in a glasshouse without any environmental control. For CS, nutrient-rich, sieved vineyard soil mixed with perlite (3:1 ratio) was used, while for BF pots were filled with commercial pot substrate containing slow release fertilizer (10 g pot-1, 15-5-20 NPK “Entec vino”). Pots were watered to pot capacity automatically every day. Clones were planted June 1 and April 1 in the first and second year, respectively. When all the plants had at least three mature leaves on one shoot, plants were pruned so that only one dominant shoot remained. To ensure that only leaves fully developed under different light conditions were used for further analyses, the last developing leaf below the tip was marked before moving half of the plants to the shaded environment (on June 29, 2018 for CS and on April 18, 2019 for BF). For the shade treatment, a tent of about 2 m (height) x 1.5 m (width) x 1.5 m (depth) was made from black polypropylene cloth (HaGa-Welt GmbH & Co. KG, Elze, Germany), resulting in a 60% reduction in photosynthetic photon flux density (PPFD). A spectrometer (FLAME-S-VIS- ES, Ocean Optics Inc. Largo, USA) was used to confirm that under both light conditions, relative differences in the contribution of red, green blue and far-red light to the total PPFD were below 5% (i.e. spectrally neutral shade). During the week before synchrotron microCT scanning (last week of August in 2018 and first week of September in 2019), one mature leaf per plant was selected at least three leaves above the previously mentioned mark indicating the last developing leaf at the start of the shade treatment. The measured leaves were estimated to be about one month old, resulting in an average daily light integral (DLI) of 30 (CS sun), 12 (CS shade), 24 (BF sun), and 10 (BF shade) mol m-2 day-1. These estimates were computed using solar radiation measured at a weather station a few meters from the glasshouse, and using PPFD values measured inside the glasshouse. Average daily PPFD was below 700 μmol m-2 s-1 under full light, with maximum recorded values at leaf level of ~1200 μmol m-2 s-1 under full light and ~500 μmol m-2 s-1 under shade, i.e. ~60% reduction. References Marone F, Stampanoni M. 2012. Regridding reconstruction algorithm for real-time tomo- graphic imaging. J. Synchrotron Radiat. 19: 1029–1037. Paganin D, Mayo SC, Gureyev TE, Miller PR, Wilkins SW. 2002. Simultaneous phase and amplitude extraction from a single defocused image of a homogeneous object. J. Microsc. 206: 33–40. Schneider CA, Rasband WS, Eliceiri KW. 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9: 671–675. Théroux-Rancourt G, Jenkins MR, Brodersen CR, McElrone A, Forrestel EJ, Earles JM. 2020. Digitally deconstructing leaves in 3D using X-ray microcomputed tomography and machine learning. Appl. Plant Sci. 8: e11380.

Project supported by the Austrian Science Fund (FWF) projects M2245 and P30275, and by the Vienna Science and Technology Fund (WWTF), project LS19-013.