EuphyticaPlant shape is a major component of the visual quality of ornamental plants. It is the result of their architectural construction. It can be analyzed by breaking down the plant into entities (axis, metamer) that can be characterized morphologically, topologically and geometrically. Eight bush rose cultivars were selected for their contrasting shapes (from upright to spreading) and their architecture was digitized at two scales, the plant and the axis, differentiating between short and long axes. Thirty-five variables were measured. Measurement acquisition is nevertheless tedious and time-consuming and not really compatible with an analysis involving a large number of individuals. To diminish these constraints, our approach aimed at reducing the number of variables measured, limiting ourselves to the ones most relevant for describing the architecture. A selection of variables was made using the following criteria: to represent the different categories of variables describing the plant architecture; to explain the variability observed; to present the weakest correlation between them. Seven variables were selected: at the plant scale, the number of determined axes, the number of long axes of order 3 and the branching order number; at the long axis scale, the number of metamers and the length of the axis; and at the short axis scale, the basal diameter of the axis and the branching angle of the cord in relation to the vertical axis. Four architectural profiles were differentiated based on these seven variables. Moreover, a high correlation was revealed between some of these architectural variables and a shape descriptor

The shape and, therefore, the architecture of the plant are dependent on genetic and environmental factors such as water supply. The architecture determines the visual quality, a key criterion underlying the decision to purchase an ornamental potted plant. The aim of this study was to analyze genotypic responses of eight rose bush cultivars to alternation of water restriction and re-watering periods, with soil water potential of -20 and -10 kPa respectively. Responses were evaluated at the architectural level through 3D digitalization using six architectural variables and at the physiological level by measuring stomatal conductance, water content, hormones [abscisic acid (ABA), auxin, cytokinins, jasmonic acid, and salicylic acid (SA)], sugars (sucrose, fructose, and glucose), and proline. Highly significant genotype and watering effects were revealed for all the architectural variables measured, as well as genotype × watering interaction, with three distinct genotypic architectural responses to water restriction – weak, moderate and strong – represented by Hw336, ‘Baipome’ and ‘The Fairy,’ respectively. The physiological analysis explained, at least in part, the more moderate architectural response of ‘Baipome’ compared to ‘The Fairy,’ but not that of Hw336 which is an interspecific hybrid. Such physiological responses in ‘Baipome’ could be related to: (i) the maintenance of the stimulation of budbreak and photosynthetic activity during water restriction periods due to a higher concentration in conjugated cytokinins (cCK) and to a lower concentration in SA; (ii) a better resumption of budbreak during the re-watering periods due to a lower concentration in ABA during this period. When associated with the six architectural descriptors, cCK, SA and ABA, which explained the genotypic differences in this study, could be used as selection criteria for breeding programs aimed at improving plant shape and tolerance to water restriction.

The effect of genotype factors, year and their interaction was assessed on six architectural variables of eight cultivars of rose bush. Plants were grown in pots in a greenhouse in the spring of 2011 and 2012, two highly contrasted years in terms of the quantity of cumulative radiation, with a relative deviation (for 2012 compared to 2011) ranging from À24.6% (April) to +13.7% (March). Their architecture was digitized at two observation scales, the plant and the axis. Highly significant genotype (G) and year (Y) effects were revealed for all of the variables measured, as well as a G 9 Y interaction. Concerning the year effect, it was significantly higher in 2012 and for all of the variables measured. The G 9 Y interaction was due to (i) different genotype groupings according to year, (ii) difference response amplitudes between years according to genotype. Broad-sense heritability was calculated for each of these variables. It was moderate to high, ranging from 48% for the length of long axes to 98% for the number of metamers on long axes.