Cultivated sunflower (Helianthus annuus L.) exhibits numerous phenotypic and transcriptomic responses to drought. However, the ways in which these responses vary with differences in drought timing and severity are insufficiently understood. We used phenotypic and transcriptomic data to evaluate the response of sunflower to drought scenarios of different timing and severity in a common garden experiment. Using a semi-automated outdoor high-throughput phenotyping platform, we grew six oilseed sunflower lines under control and drought conditions. Our results reveal that similar transcriptomic responses can have disparate phenotypic effects when triggered at different developmental time points. Leaf transcriptomic responses, however, share similarities despite timing and severity differences (e.g., 523 differentially expressed genes (DEGs) were shared across all treatments), though increased severity elicits greater differences in expression, particularly during vegetative growth. Across treatments, DEGs were highly enriched for genes related to photosynthesis and plastid maintenance. A co-expression analysis identified a single module (M8) enriched in all drought stress treatments. Genes related to drought, temperature, proline biosynthesis, and other stress responses were overrepresented in this module. In contrast to transcriptomic responses, phenotypic responses were largely divergent between early and late drought. Early-stressed sunflowers responded to drought with reduced overall growth, but became highly water-acquisitive during recovery irrigation, resulting in overcompensation (higher aboveground biomass and leaf area) and a greater overall shift in phenotypic correlations, whereas late-stressed sunflowers were smaller and more water use-efficient. Taken together, these results suggest that drought stress at an earlier growth stage elicits a change in development that enables greater uptake and transpiration of water during recovery, resulting in higher growth rates despite similar initial transcriptomic responses.

Six sunflower maintainer (B) lines bred for oil-rich seeds from the SAM population (HA124, HA370, HA412HO, HA850, HAR4, and SF193, also known as XRQ) were grown at the Heliaphen outdoor high-throughput phenotyping platform at INRAE Toulouse (France) in 2018. Seeds were planted on 17 April 2018 in 10 L pots of Terreau Proveen PAM 2 substrate. Plants were fertilized with 300 mL of Peter’s Professional 17-07-27 and 200 mL of Hortrilon per pot on 5 May, 25 May, and 13 June, and were provided Ortiva Top fungicide on 28 May and 15 June. Pots were maintained at specified soil moisture levels by measuring the pot weights daily and refilling them with water to return the pot weights to target levels. Soil around the plants was covered with a silicone cover to limit soil evaporation and was protected from rain using a cone-shaped polystyrene skirt. Drought treatment regimes were imposed by permitting pot weights to drop to specified levels for set periods of time.

Plants were grown under one of five irrigation treatment groups. Control plants were provided daily water to maintain a pot weight equal to the recorded weight of the pot left to drain for two hours after full water saturation, defining a fraction of transpirable water (FTSW) of 1. The early moderate (EM) and early severe (ES) drought stress groups were subjected to drought stress during the vegetative growth phase from 7 May until target stress levels (EM = 0.4 FTSW; ES = 0.2 FTSW) were reached. Early drought stress target soil moisture levels were reached between 1 June and 11 June. Following these early drought stress treatments, these plants were returned to and maintained at control levels of soil moisture (FTSW = 1) until the end of the experiment. The late moderate (LM) and late severe (LS) drought stress groups were subjected to drought stress during the reproductive growth phase starting on 12 June (plants initiated flowering between 18 June and 2 July). Late drought stress was maintained at target drought levels (LM = 0.4 FTSW; LS = 0.2 FTSW) until 3 August, after which all plants were denied irrigation and left to dry until harvest on 31 August. Thus, treatment groups for this study were control (C), early moderate (EM), early severe (ES), late moderate (LM), and late severe (LS). Six replicates of each combination of treatment group (n = 5) and genotype (n = 6) were grown, resulting in 180 plants, of which 174 survived.

The volume of water provided daily to each plant was measured automatically, permitting us to track the time series of water provision to each plant in each treatment group. One potential source of error introduced by this irrigation regime is that not all water provided to the plant was transpired. Some small but unmeasured portion of the water inevitably evaporated from the soil surface despite the use of silicone pot covers. More notably, error was introduced by water being taken up and stored in the plant. Water stored in the plant increases the pot weight, reducing the amount of water the automated system will provide in subsequent rewatering events, resulting in a soil moisture lower than targeted levels. The scale of this effect is, however, likely to be minimal due to the large difference in weight between the pot/soil and plant biomass.

Several phenotypic traits (plant height, leaf number and area, collar diameter) were measured multiple times over the course of the experiment, producing the time-series data. Because the phenotyping platform (RapidoScan RS-C-025-1600-MOD) records a maximal plant height of 1.5 m, plant height data were truncated to those collected before 28 June, the date when some plants began to reach this height. Total leaf area was calculated for each plant by finding individual leaf areas for every other leaf and doubling the sum of those values.

Other traits were collected at the end of the experiment after harvest and after 48 h dry-down at 80°C for yield components (seed number and weight, plant weight). Aboveground plant weight was divided into total seed weight, capitulum weight, and vegetative (leaf and stem) plant weight. Total water added was defined as the total water provided to a single plant over the course of the experiment. We defined water use efficiency (WUE) as the total dry plant weight (vegetative plant weight plus capitulum weight plus total seed weight) per total water added per plant.

At the end of each stress treatment (12 June and 3 August, respectively), dental putty imprints of both sides of leaves one-third of the way down the plant were taken for stomatal data collection. Clear nail polish was applied to the dental putty imprints and removed with clear tape. Images were taken from three or four 100× magnified areas per leaf print using a light microscope. Stomata were counted manually for each image and averaged per leaf side. Ten individual stomata per leaf side were imaged at 400× and measured for pore length and guard cell width (as a proxy for pore depth). Using these stomatal measurements (density, pore size, pore depth), we calculated the maximum anatomical stomatal conductance (gsmax hereafter). It should be noted that, given the duration and implementation of our drought treatment, the first sampled leaves began development before the early drought stress and finished during the early drought stress period, though the second sampled leaves fully developed during the drought stress period.