No evidence of xylem embolism refilling during recovery from drought stress in intact sunflowers

Abstract

Refilling of xylem embolism during recovery from drought has been studied for decades, although the legitimacy of supporting evidence has been debated in recent years due to potentially widespread methodological artifacts associated with destructive sampling. Evidence from multiple destructive/indirect measurements of embolism suggests sunflower can refill, but more recent evidence suggests this species is particularly susceptible to artifacts that cause the appearance of refilling. Here, we investigated drought-induced xylem embolism and its potential reversibility in intact sunflowers using low-radiation micro-computed tomography. While plants were drying, turgor loss and stomatal closure occurred at ca. −1.0 MPa, whereas comparable losses of photosystem II efficiency as well as leaf and stem xylem embolism occurred at ca. −1.5 MPa. After re-watering, xylem embolism that accumulated in stems and leaves during the drought event did not reverse despite significant recovery of transpiration and photosynthesis. We found no evidence of xylem embolism refilling in sunflowers, highlighting the potential widespread nature of methodological artifacts and the need to revisit conclusions of ‘refilling’ in other species as well as physiological and developmental consequences of irreversible embolism.

Failure of the water-transporting system during drought stress was not reversed during recovery in sunflower, which challenges previous evidence and highlights the need to revisit similar assumptions in other species.

Introduction

Plants become stressed during drought conditions when the rate of water loss from the plant exceeds the rate at which they can acquire water from the soil. One consequence during extreme drought events is that water-transporting xylem conduits can develop embolism (i.e. become air-filled), and the accumulation of embolism leads to proportional reductions in the water-transporting capacity of the plant. Hydraulic capacity is closely aligned with photosynthesis across species (Brodribb et al., 2005), so the occurrence of drought-induced embolism and its reversibility (or lack thereof) have implications for plant and ecosystem function, terrestrial carbon fluxes, and agricultural productivity (Torres-Ruiz et al., 2024). Yet, our understanding of this foundational process has been obscured by methodological issues and a biological system that has been markedly difficult to study non-destructively until recently (Brodersen et al., 2010; Brodribb et al., 2016a, b; Stewart et al., 2025).

Functional reversal of embolism, known as ‘refilling’, is an adaptive process in which water-transporting functionality is restored to conduits after the onset of embolism (Stewart et al., 2025). Numerous studies over the past several decades have reported results consistent with daily embolism–refilling cycles (e.g. Tyree et al., 1986), although very few have included any direct observation of embolism refilling in intact plants (Stewart et al., 2025). Sunflower has been reported to reverse embolism in the xylem of roots (Buchard et al., 1999), stems (Hacke et al., 2001; Stiller and Sperry, 2002), and leaves (Trifilò et al., 2003; Nardini et al., 2008). However, the methods employed by these studies, in which embolism of intact plants was estimated based on measurements of excised tissues, have been increasingly scrutinized over the past decade (Sperry, 2013; Rockwell et al., 2014). This shift occurred after Wheeler et al. (2013) demonstrated that the common practice of excising tissue from plants under negative tension can cause embolism and lead to the appearance of ‘refilling’ (see also Sperry, 2013; Trifilò et al., 2014). Torres-Ruiz et al. (2015) subsequently demonstrated that this artifact can be especially pronounced in sunflower, even at relatively mild water potentials (−0.4 MPa).

To date, studies observing and quantifying xylem embolism in intact plants with non-invasive methods are rare. Methods that can be employed non-destructively include the optical (refraction) method applied to leaves (Brodribb et al., 2016a, b), micro-computed tomography (µCT; Brodersen et al., 2010; Stewart et al., 2025), and magnetic resonance imaging (MRI; Kaufmann et al., 2009; Meixner et al., 2021). Unfortunately, these methods are technically challenging and/or require prohibitively expensive equipment for most researchers. These methods also impose logistical challenges (e.g. equipment portability, tissue size constraints, radiation hazards) for repeated and entirely non-destructive study of embolism during drought and recovery. The resulting lack of in vivo evidence and persistent knowledge gaps with crop species are particularly concerning. The consequences of embolism on plant function and agriculturally relevant traits (e.g. reproductive success) should be addressed, especially if embolism is not rapidly reversible and/or embolism reversibility varies across species. Elucidating variation in embolism onset and/or reversibility in crop species should inform efforts towards maintaining/increasing food production in the face of warming temperatures and increasingly frequent and severe droughts (Gleason, 2015; Torres-Ruiz et al., 2024).

Furthermore, the sequential response and alignment of physiological systems during embolism-inducing drought events and upon recovery are not well understood. Identifying these sequences will improve our ability to predict the impact of drought on individuals and ecosystems and guide future crop improvement studies to increase food security. Three specific examples that are especially pressing are: (i) the order in which hydraulic (e.g. stem and leaf xylem conductance, stomatal conductance) and photosynthetic (e.g. CO₂ assimilation, photosystem II efficiency) systems fail during drought and recover after drought in intact plants; (ii) the extent and rate of recovery of these systems, or even if recovery is possible after exposure to embolism-causing water potentials (Cardoso et al., 2018; Johnson et al., 2018; Brodribb et al., 2021); and (iii) specific to food security, most efforts to improve crop species have focused on cell- and leaf-level responses (e.g. stomatal responses, photosynthetic capacity, osmotic regulation, plasma membrane stability) or else gene and gene regulatory networks, but the whole-plant response and recovery from low water potential is still poorly understood.

In the present study, we aimed to observe the extent of embolism in both leaves and stems of intact sunflowers during a drought event as well as during recovery after re-watering. Whole-plant and leaf-level gas exchange were characterized in the same plants during drought and recovery to assess the consequences of xylem functionality on downstream physiological processes of afflicted shoots as well as individual leaves. Sunflower was chosen as our study system for two reasons: (i) this species had been reported to refill previously from studies using indirect estimates of xylem embolism from excised tissue (Hacke et al., 2001; Stiller and Sperry, 2002; Trifilò et al., 2003; Nardini et al., 2008), and (ii) this species was reported as being particularly vulnerable to embolism induced by tissue excision (Torres-Ruiz et al., 2015).

Materials and methods

Plant material and growth conditions

Sunflower (Helianthus annuus L. cv. Mammoth) plants were grown from seed (USDA Germplasm Resources Information Network, Ames, IA, USA) in a greenhouse (USDA Crops Research Laboratory, Fort Collins, CO, USA). Plants were grown in D40H Deepot cells (Stuewe and Sons., Inc.; Tangent, OR, USA) that had a diameter of ca. 6 cm and were cut down to a height of ca. 10 cm, which was necessary for the detection of xylem embolisms (described below; see also Stewart et al., 2025), and contained ca. 250 ml of Profile Porous Ceramic Greens Grade (Profile Products LLC; Buffalo Grove, IL, USA). Plants were watered every 6 h starting at 05.00 h for 2 min with 1.9 liters h⁻¹ emitters (Rainbird, Corp.; Tucson, AZ, USA). Nutrients were supplied by Osmocote Smart-Release Plant Food Plus 15-9-12 granular (The Scotts Miracle-Gro Company; Marysville, OH, USA) and Jack's Professional 20-20-20 General Purpose liquid (JR Peters Inc.; Allentown, PA, USA) fertilizers. Supplemental lighting was provided by Philips GreenPower LED toplighting (Signify N.V.; Eindhoven, Netherlands) from 05.30 h to 20.30 h—the approximate sunrise and sunset times, respectively, on the summer solstice.

Drought–recovery experiments

For drought–recovery experiments, six seeds were sown per day on 10 separate days, which resulted in 55 characterized plants. The first set of plants (n=25) was grown between 8 August 2021 and 23 September 2021, and another set of plants (n=30) was grown between 15 March 2022 and 15 April 2022. When plants were 3 weeks old, drought stress treatments were initiated with a cohort of plants (n=5 or 6 plants per day) by removing the emitters, which typically occurred after the 05.00 h watering event on the day before measurements were scheduled. Functional characterization of droughted plants (see below) occurred between ca. 10.00 h and 14.00 h the following day (i.e. ca. 30 h after the last watering). Recovery was initiated by saturating the substrate with water, and then plants were returned to the pre-stress watering schedule provided by the same emitter.

Gas exchange and chlorophyll fluorescence

During the drought–recovery experiments, pots were covered with plastic around the base of the plant to minimize substrate evaporation, and their mass was measured on a balance with 0.001 g resolution (GX-600; A&D Store, Wood Dale, IL, USA) every ca. 15 min. Whole-plant transpiration rate was calculated based on changes in mass during the time elapsed. Net CO₂ assimilation rate per leaf area and photosystem II efficiency, which was calculated as (F_m′−F_s)/F_m′, were measured with LI-6400/XT Portable Photosynthesis Systems equipped with 6400-40 Leaf Chamber Fluorometers (LI-COR Biosciences; Lincoln, NE, USA). For measurements with the LI-6400/XT, several leaves per plant were measured to provide a plant-level average because we were more interested in effects of stem xylem embolism on the function of apical leaves generally. Light intensity was set to 500 µmol photons m⁻² s⁻¹ to match the maximal light intensity in the greenhouse, and reference CO₂ was set to 400 ppm. Each plant was measured several times over several hours at midday (typically between 10.00 h and 14.00 h).

Water potential and xylem embolism

Leaf water potential (Ψ_Leaf) was measured with a Scholander pressure chamber (Model 3005, SoilMoisture Equipment Corp.; Goleta, CA, USA). For the drought–recovery experiments, minimum Ψ_Leaf was measured from at least one leaf of each of the four of five plants selected for destructive measurements. At the same time, the one or two plants per cohort selected for non-destructive measurements were enclosed in plastic (to minimize additional water loss) and then imaged with a vivaCT 80 in vivo microCT scanner (SCANCO Medical AG, Bruttisellen, Switzerland). The field of view was 31.9 mm, resolution was 4.8 μm, integration time was 356 ms, power was 55 kVp, and scan time was 45 min. The cross-sectional area (µm²) of the stem as well as xylem embolism in stems, petioles, and midribs in each image was measured with Fiji (ImageJ; Schindelin et al., 2012; Schneider et al., 2012) and the ‘BoneJ’ plugin (Doube et al., 2010). Minimum stem water potential (Ψ_Stem) of these plants was estimated based on the extent to which each stem had shrunk (Bourbia et al., 2021). For this estimation, a calibration curve was generated with 20 plants that were terminally dried under laboratory conditions (Supplementary Fig. S1). As plants dried, stem girth was measured continuously with custom-built contact dendrometers (Gleason et al., 2024), and Ψ_Stem was measured periodically with the pressure chamber from bagged leaves on the same plants. The calibration curve to predict Ψ_Stem from stem shrinkage (i.e. change in stem girth from its maximum) was then generated with a loess model (Supplementary Fig. S1).

Leaf xylem vulnerability curves

To produce a complete vulnerability curve for leaves, in which cumulative xylem embolism reaches 100%, leaf xylem embolism was measured optically (via changes in light transmission) based on methods described by Brodribb et al. (2016a, b) using custom-built camera systems (Gleason et al., 2024; Allen et al., 2025). Fifteen plants were terminally dried under laboratory conditions. Two leaves from five intact plants were imaged for embolism, and Ψ_Leaf was measured from excised leaves of 10 additional ‘companion’ plants that were dried at the same time. Ψ_Leaf for the intact plants was then predicted from elapsed time with a loess model (Supplementary Fig. S2). Images were captured every 5 min, and image stacks were processed with Fiji (ImageJ) and the ‘OSOV’ plugin.

Pressure–volume curves

The fresh mass and Ψ_Leaf of excised leaves were measured with a 0.0001 g resolution balance (OHAUS Explorer; Parsippany, NJ, USA) and a Scholander pressure chamber, respectively. Prior to excision of selected leaves, six plants were brought to the laboratory, fully watered, and allowed to equilibrate for several hours. Following excision, fresh mass and Ψ_Leaf of each leaf were measured several times before and after stomatal closure, and dry mass of each leaf was measured after leaves had dried at 70 °C for 1 week. Ψ_Leaf at turgor loss was then extracted from the relationship between the reciprocal pressure (i.e. 1/Ψ_Leaf) and the inverse of relative water content (i.e. 1−RWC; Supplementary Fig. S3).

Sap exudate flow rate

After the 1 week recovery period, the shoots of seven plants that were characterized via µCT (i.e. observed levels of xylem embolism) were excised, and a cotton pad was placed on top of the cut surface (ca. 5 cm above the soil surface), secured by a rubber band within a plastic bag, and allowed to absorb fluid exudated. Sap exudate flow rate, which can be used as a proxy for root pressure (Comas et al., 2020), was calculated as the change in mass of the cotton pad, which was measured with a 0.0001 g resolution balance (Explorer series, OHAUS), divided by time elapsed.

Data processing and statistical analysis

All statistical analyses were conducted with R (4.3.2) and RStudio (2023.12.1+402). Differences between ‘stress’ versus ‘recovery’ time points were compared with Student’s paired t-test. Relationships between water potential and the loss of transpiration, net CO₂ assimilation, photosystem II (PSII) efficiency, and onset of leaf and stem xylem embolism were analysed with Weibull models using the ‘fitplc’ package (Duursma and Choat, 2017). Prior to this analysis, the relationship between cross-sectional area of stem xylem embolism and Ψ_Stem was fit with a three-parameter logistic model using the ‘drc’ package, and stem embolism data were normalized to a 0–100% range based on the lower and upper asymptotes, respectively. Transpiration and net CO₂ assimilation values for each plant were normalized to a 0–1 range using the ‘datawizard’ package (Patil et al., 2022), and mean values of PSII efficiency were normalized to a maximum value of 1 (for raw data, see Supplementary Dataset S1). The sums of these latter three normalized values were used to generate a color gradient with the ‘colourvalues’ package, and the color assigned to each plant is used consistently in all figures.

Results

Physiological failures and embolism formation during dehydration

As sunflower plants dried, stomatal closure in leaves occurred at Ψ_Leaf of ca. −1 MPa (Fig. 1; Supplementary Fig. S3). Turgor loss point occurred at Ψ_Leaf of −0.91±0.05 MPa (mean ±SD; Supplementary Fig. S3). Whole-plant transpiration rate (Fig. 1A) and net CO₂ assimilation rate per leaf area (Fig. 1B) exhibited similar sigmoidal declines as plants dried. These rates exhibited 50% loss at Ψ_Leaf of −0.95 MPa and −1.06 MPa, respectively (Table 1). Initial declines in PSII efficiency (Fig. 1C; Table 1) corresponded to initial declines in transpiration and CO₂ assimilation rates, with 12% losses at Ψ_Leaf of −0.65, −0.62, and −0.80 MPa, respectively. However, further declines of transpiration and net CO₂ assimilation rates occurred more rapidly (i.e. at higher Ψ_Leaf) than declines of PSII efficiency, with 88% losses at Ψ_Leaf of −1.21, −1.27, and −2.45 MPa, respectively (Table 1). The onset of leaf (Fig. 1D) and stem (Fig. 1E) xylem embolism coincided with 50% loss of transpiration and CO₂ assimilation rates (Table 1). The point at which leaf xylem embolism reached 50% (P₅₀; Fig. 1D) was closely aligned with the 50% loss of PSII efficiency, which occurred at Ψ_Leaf of −1.46 MPa and −1.45 MPa, respectively (Fig. 1C, D), and P₅₀ for stems occurred at Ψ_Stem of −1.56 MPa (Fig. 1E; Table 1). Leaf xylem embolism accumulated more rapidly than stem xylem embolism, and 88% loss of leaf xylem capacity occurred at Ψ_Leaf of −1.67 MPa (Fig. 1D), whereas the 88% loss of stem xylem capacity occurred at Ψ_Stem of −2.08 MPa (Fig. 1E).

Fig. 1.

Relationships between water potential and (A) whole-plant transpiration rate, (B) net CO₂ assimilation rate per leaf area, (C) photosystem II efficiency, as well as xylem embolism in leaves (D) and stems (E) of sunflower plants during dehydration stress. The solid black lines represent the fitted Weibull models, and the dashed vertical lines indicate the water potentials at which each trait is reduced by 12%, 50%, and 88% (Table 1). For (A–C), symbols represent mean values (n=55 plants); for raw data, see Supplementary Dataset S1. Each symbol in (B, C) represents a plant average from measurements of multiple leaves per plant. For (D), n=5 plants; for (E), n=10 plants. Colors for each plant were generated based on the sum of their values in (A–C).

Table 1.

Water potentials at which 12%, 50%, and 88% losses occurred for whole-plant transpiration rate, net CO₂ assimilation rate per leaf area, photosystem II efficiency, as well as leaf and stem xylem functional capacity (i.e. inverse of xylem embolism; Fig. 1)

	12% loss (MPa)	50% loss (MPa)	88% loss (MPa)
Transpiration rate	−0.62	−0.92	−1.21
Net CO2 assimilation rate	−0.80	−1.06	−1.27
Photosystem II efficiency	−0.65	−1.45	−2.45
Leaf xylem capacity	−1.19	−1.46	−1.67
Stem xylem capacity	−1.02	−1.56	−2.08

Xylem embolism is not reversed in sunflower stems or leaves

Sunflower plants with observed stem xylem embolism during drought stress were re-watered and then characterized via µCT at two time points: (i) after an overnight recovery period under laboratory conditions, and/or (ii) 1 week after being returned to the pre-stress (i.e. well-watered) growth conditions in the greenhouse (Figs 2, 3; Supplementary Figs S4–S13). The extent and spatial distribution of stem xylem embolism were similar at peak stress (Fig. 2A, D), after the overnight recovery (Fig. 2B, E), and 1 week after being returned to the well-watered conditions in the greenhouse (Fig. 2C, F), which suggests no refilling occurred during the 1 week recovery period. Embolism remained relatively unchanged in one plant even after 3 weeks (Supplementary Fig. S8), during which time the stem had grown considerably (ca. 240% increase in stem cross-sectional area). This result was not influenced by the severity of the stress event, as stem xylem embolism did not reverse upon re-watering in sunflower plants that experienced relatively mild stress (−1.36 MPa) with relatively low levels of stem embolism (ca. 10%; Fig. 2A–C) or in plants that experienced relatively severe stress (−2.10 MPa) with relatively high levels of embolism (ca. 90%; Fig. 2D–F). In contrast, the extent to which stems had shrunk (Fig. 2A, D) and/or wilted (Fig. 3A) at peak stress was mostly reversible overnight under laboratory conditions (Fig. 2B, E). The extent of this recovery was maintained when plants were returned to the greenhouse for 1 week (Figs 2C, F, 3B), and the stems of all plants were upright and erect at the end of this recovery period (Supplementary Figs S4–S13).

Fig. 2.

Micro-computed tomographic images of xylem embolism (highlighted within the outlined stems) in sunflower stems from (A–C) a plant that was mildly stressed (Ψ_Stem of −1.36 MPa) and (D–F) a plant that was severely stressed (Ψ_Stem of −2.10 MPa). Plants were scanned while stressed (i.e. at the end of the dry down) (A, D) as well as the morning after re-watering (B, E), and after 1 week of recovery under well-watered conditions (C, F). Images without annotated features as well as photographic images of the plants corresponding to panels (A–C) and (D–F) are shown in Supplementary Figs S9 and S13, respectively. Scale bars are 1 mm.

Fig. 3.

Xylem embolism within a sunflower plant during drought stress and well-watered recovery. (A, B) Photographic images of a sunflower plant while stressed (i.e. at the end of the dry down) (A) and after 1 week of recovery under well-watered conditions (B). (C–E) Micro-computed tomographic images of xylem embolism (highlighted within the outlined tissues) in the stem, petioles (Pet.), and leaf midribs when the plant was subjected to drought stress (Ψ_Stem of −1.23 MPa) (C), re-watered and allowed to recover overnight (D), and allowed to recover for 1 week (E). The arrows in (A, B) indicate the approximate locations of scanned tissue. For images without annotated features, see Supplementary Fig. S7. For (C–E), scale bar is 1 mm.

Similarly, the extent and spatial distribution of xylem embolism in the petioles of lower leaves and the midribs of upper leaves at peak stress (Fig. 3C) were similar (i.e. not reversed) after re-watering and an overnight recovery period (Fig. 3D) or after being returned to well-watered conditions in the greenhouse for 1 week (Fig. 3E). In some plants, the lowest leaves (i.e. first true leaves) desiccated completely when plants were returned to the greenhouse (e.g. Supplementary Fig. S4). Some additional leaves (i.e. second/third pair of true leaves) experienced partial or complete desiccation (e.g. Supplementary Fig. S13) in plants that were subjected to the most severe stress (Ψ_Stem of ca. −2.0 MPa). The most apical leaves and shoot apical meristem survived in all characterized plants (Supplementary Figs S4–S13). The extent to which petioles and midribs had shrunk during drought (Fig. 3C) was reversible after the overnight recovery period (Fig. 3D) in some plants, and these leaves remained relatively turgid at midday in the greenhouse the following week (Fig. 3B) despite relatively high levels of xylem embolism (Fig. 3E).

Partial restoration of plant-level physiology during recovery

Across all plants, the extent of xylem embolism in stems when plants were stressed did not change during recovery (Fig. 4A). The extent of xylem embolism in petioles of lower leaves (Fig. 4B) and midribs of upper leaves (Fig. 4C) did not change after the overnight recovery period, although the extent of embolism worsened (i.e. significantly increased) in both petioles and midribs after being returned to well-watered conditions for 1 week. In some cases, relatively large increases in petiole (Fig. 4B) and midrib (Fig. 4C) embolism corresponded to complete desiccation of the leaf after the plant was returned to the greenhouse (e.g. Supplementary Fig. S9). In contrast, whole-plant transpiration (Fig. 5A) as well as net CO₂ assimilation rate per leaf area (Fig. 5B) and PSII efficiency (Fig. 5C) of non-desiccated leaves improved (i.e. significantly increased) during the 1 week recovery period. The latter two metrics are likely overestimated for some plants since no measurements were taken on completely desiccated leaves, the number of which varied across plants (see Supplementary Figs S4–S13). However, transpiration (Fig. 5A), CO₂ assimilation (Fig. 5B), and PSII efficiency (Fig. 5C) of most plants during the 1 week recovery period remained depressed relative to well-watered control plants (see also Supplementary Fig. S14). Some plants with relatively high levels of embolism in stems had relatively low rates of whole-plant transpiration and PSII efficiency in leaves, although these relationships were not significant across all plants (Supplementary Fig. S14). Sap exudation was observed at the conclusion of the 1 week recovery period. Sap exudate flow rates were 0.0749±0.0447 g h⁻¹ (mean value ±SD; n=7 plants) with a range of 0.0231–0.1572 g h⁻¹ (for individual values, see legends of Supplementary Figs S4, S7, S9–S13).

Fig. 4.

Xylem embolism in stems (A), petioles of lower leaves (B), and midribs of upper leaves (C) of sunflower plants while stressed (i.e. at the end of the dry down) and during recovery under well-watered conditions. Connected symbols represent individual plants (n=10 plants) (A) or the same petiole (n=8 plants) (B) or midrib (n=7 plants) (C). Asterisks denote significant differences between stressed and recovery values: *P<0.05, ***P<0.001 (n.s., not significantly different).

Fig. 5.

Whole-plant transpiration rate (normalized for leaf area) (A), net CO₂ assimilation rate per leaf area (B), and photosystem II efficiency (C) of sunflower plants while stressed (i.e. at the end of the dry down) and during recovery under well-watered conditions. Connected symbols represent individual plants [n=7 plants for (A), n=8 plants for (B, C)]. Each symbol in (B, C) represent a plant average from measurements of multiple leaves per plant. Asterisks denote significant differences between stressed and recovery values: ***P<0.001. Dashed lines and shaded regions denote mean values and mean values ±SD, respectively, of a well-watered control group (n=6 plants).

Discussion

We found no evidence of short-term (i.e. overnight to 1 week) xylem refilling after re-watering in sunflower. Our findings are consistent with those from similar studies of gymnosperm (Choat et al., 2015; Rehschuh et al., 2020) and angiosperm trees (Knipfer et al., 2018; Choat et al., 2019; Avila et al., 2022; Gauthey et al., 2022; Tonet et al., 2024), as well as the annual grass wheat (Triticum aestivum; Johnson et al., 2018; Harrison Day et al., 2024). Recently, we reported complete and functional overnight refilling in the stems of another annual grass, barnyard grass (Echinochloa crus-galli), using the same methodology as the present study (Stewart et al., 2025). This total reversal was observed in all barnyard grass plants, which included embolism levels up to 88%, and all conduits remained refilled in all barnyard grass plants after they were returned to the growing conditions. Some evidence of embolism reversal has been reported in desiccation-tolerant ferns (Holmlund et al., 2019; Prats and Brodersen, 2021) and, to a lesser extent, a desiccation-tolerant monocot (Nadal et al., 2023). In contrast to barnyard grass, in which xylem refilling preceded tissue rehydration, these desiccation-tolerant species exhibited tissue rehydration and physiological recovery that occurred more rapidly than embolism reversal. Some evidence of refilling has also been reported in grapevine (Brodersen et al., 2010; Knipfer et al., 2015; Charrier et al., 2016; Vuerich et al., 2023) and poplar (Secchi et al., 2021; Tricerri et al., 2025), although the extent of refilling, the conditions under which refilling occurs, and its functionality when the plant is returned to growing conditions remain important research questions. When combined, the evidence from direct observation of embolism in intact plants suggests the prevalence of complete, rapid, and functional refilling is rare.

Our results contrast with those of several studies (Buchard et al., 1999; Hacke et al., 2001; Stiller and Sperry, 2002; Trifilò et al., 2003; Nardini et al., 2008) that concluded embolism in sunflower could be reversed. The results presented in these studies represent estimations of embolism in intact plants based on measurements (e.g. loss of hydraulic conductivity) of excised tissues. In contrast, the results in the current study represent direct and repeated observation of embolism in intact plants. Our results (i.e. no evidence of refilling) are unlikely to be the consequence of an experimental artifact, especially when considering we observed complete and functional refilling in barnyard grass grown in the same small pots and using the same methodology (Stewart et al., 2025). It is also unlikely that our methodology would have prevented refilling in sunflower specifically, as one plant from our study that was not scanned via µCT during stress nevertheless exhibited extensive stem embolism after 1 week of well-watered recovery (Supplementary Fig. S15). Likewise, in the present study, we did not observe any adverse effects that have been reported with other µCT scanners (Savi et al., 2017; Petruzzellis et al., 2018; Mekarni et al., 2024). Compared with the radiation produced by these µCT scanners (ca. 8500–30 000 mGy per scan and entrance rates of, for example, 47 mGy s⁻¹; Petruzzellis et al., 2018), the radiation produced by the scanner used in the present study is extremely low (462 mGy per scan based on an entrance rate of 0.171 mGy s⁻¹; Oliviero et al., 2019). This difference in radiation may explain the discrepancy between the P₅₀ for stem xylem embolism reported in the present study (−1.56 MPa) and the P₅₀ (−0.87 MPa) reported in a similar study (Savi et al., 2017), which produced radiation dosages from a single scan that were subsequently demonstrated to cause significant tissue damage in sunflower (Petruzzellis et al., 2018). However, these interpretations are limited since we did not measure µCT-induced cellular damage in the present study.

Methodological differences remain a concern for reconciling studies of xylem embolism and refilling (Cochard et al., 2013; Rockwell et al., 2014). The wide range of methods for measuring embolism may be responsible for the wide range of reported P₅₀ values for sunflower, ranging from greater than −1.0 MPa (Savi et al., 2017) to −3.0 MPa or lower (Stiller and Sperry, 2002; Rico et al., 2013; see also Ahmad et al., 2018). The relatively low P₅₀ values reported by the latter studies were all generated by inducing embolism in fully hydrated stem segments via centrifugal force, although this same method has resulted in a P₅₀ similar to the one in the current study [approximately −1.8 MPa (Hacke et al., 2001) versus −1.56 MPa, respectively]. This method should be somewhat exempt from cutting artifacts since these artifacts are expected only in plants under tension (Torres-Ruiz et al., 2015) and are most problematic in the context of embolism refilling (Wheeler et al., 2013; see also Sperry, 2013). The slightly higher P₅₀ reported in our study was generated from younger and less developed plants, and stem P₅₀ of juvenile plants can be approximately 1 MPa higher than that of adult plants in some species (Lucani et al., 2019). Cardoso et al. (2018) observed a similar onset of leaf embolism (approximately −1.0 MPa) but in much older plants compared with those in our study (8 weeks old versus 3 weeks old, respectively). One possible unifying explanation could be that stem xylem development and maturation confers increasing resistance to embolism spread, resulting in lower stem P₅₀ but similar leaf P₅₀ as plants age. This explanation would be consistent with the evidence of greater hydraulic segmentation in older versus younger sunflower plants (Lo Gullo et al., 2004). Other factors, such as genotype (Ahmad et al., 2018) and growth conditions (Cardoso et al., 2018), can influence P₅₀ in sunflower and should be considered when interpreting results across studies.

Additionally, opposing conclusions of the impact of repeated drought–recovery events on this species—from decreased embolism resistance (i.e. ‘cavitation fatigue’; Hacke et al., 2001; Stiller and Sperry, 2002) to increased embolism resistance (Cardoso et al., 2018)—need to be revisited. Given our results and the likelihood of cutting artifacts in this species (Torres-Ruiz et al., 2015), the reported ‘cavitation fatigue’ in sunflower (Hacke et al., 2001; Stiller and Sperry, 2002) is likely an artifact from tissue excision and/or pre-existing embolism rather than an exhaustion of ‘refilling’. This discrepancy between the findings of the present study (i.e. no evidence of refilling) and those of previous studies of sunflower (i.e. evidence in support of complete and rapid refilling) should be cause for concern and highlight the importance of at least some level of direct observational evidence of embolism in intact plants.

Our poor understanding of embolism and its reversibility across species is not a trivial shortcoming, especially when considering that embolism is apparently irreversible in many species (Choat et al., 2015, 2019; Knipfer et al., 2018). On the one hand, permanent xylem embolism should affect hydraulic capacity and, therefore, photosynthetic productivity (Brodribb et al., 2005), which is consistent with the partial but incomplete recovery of transpiration and photosynthetic activity observed in most plants in the present study. On the other hand, the survival and continued growth of all plants with embolism in this study suggests some species can mitigate detrimental effects of irreversible embolism via hydraulic segmentation favoring apical leaves (Lo Gullo et al., 2004), shedding of older/afflicted leaves, and/or the development of new leaves and xylem in the stem (Gauthey et al., 2022). While xylem embolism in leaves may be more directly linked to tissue death in some cases (Brodribb et al., 2021), several sunflower leaves with substantial embolism in the present study survived after being returned to well-watered growing conditions (see also Allen et al., 2025). Furthermore, all sunflowers that were re-watered in the present study survived despite stem embolism levels in excess of 50%. Our findings also cast doubt on the role of positive pressure (i.e. ‘root pressure’; see Schenk et al., 2021) as the sole mechanism responsible for xylem refilling. The sap exudate flow rates of sunflower plants in the current study, which did not refill, were relatively similar to those (mean value ±SD of 0.2013±0.1478 g h⁻¹; n=8 plants) of barnyard grass plants in our previous study (Stewart et al., 2025), which refilled completely. However, both studies are limited in that only proxies of root pressure were measured (for details, see Comas et al., 2020). Identifying refilling mechanisms as well as the long-term impacts of non-lethal but irreversible xylem embolism on subsequent stress tolerance/resilience, reproductive fitness, and seed yield should be priorities for ecological and agricultural research efforts.

We found no evidence that xylem embolism is rapidly or functionally reversible in stems, petioles, and midribs of the sunflower genotype chosen for our study. These findings contrast with those of previous studies on sunflower, although additional investigations using similar methods on a broader range of genotypes, conditions, and tissues are needed to draw more general conclusions about refilling in this species. The lack of a clear relationship between stem xylem embolism and physiological recovery of leaves illustrates the importance of direct observational evidence (e.g. low-radiation µCT) and highlights the need for future studies to investigate the link between irreversible embolism and physiological recovery more directly. Interpretation of the results in the present study is limited to plant-level because we did not track individual leaves before and after embolism had occurred. However, all measured leaves for physiological recovery were apical to the stem xylem embolism measurements, so their recovery occurred despite impaired hydraulic transport through the stem to the leaves. Likewise, our results should be considered when evaluating conclusions from studies of xylem embolism refilling that have relied on measurements from excised material (see Wheeler et al., 2013) and/or inference from physiological measurements (e.g. hydraulic conductance) per se. Specifically, future research should address the consequences of embolism on reproductive success, the results of which would have ecological, evolutionary, and agricultural relevance.

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Fitted loess model used to predict stem water potential.

Fig. S2. Fitted loess model used to predict leaf water potential.

Fig. S3. Pressure–volume curves for six sunflower plants.

Fig. S4. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −0.88 MPa).

Fig. S5. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.01 MPa).

Fig. S6. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.02 MPa).

Fig. S7. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.23 MPa).

Fig. S8. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.24 MPa).

Fig. S9. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.36 MPa).

Fig. S10. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.38 MPa).

Fig. S11. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.59 MPa).

Fig. S12. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −1.98 MPa).

Fig. S13. Photographic and μCT images of a scanned plant (min. Ψ_Stem of −2.10 MPa).

Fig. S14. Relationship between stem xylem embolism and leaf functional traits during recovery.

Fig. S15. Photographic and μCT images of a scanned plant (min. Ψ_Leaf of −2.24 MPa).

Dataset S1. Raw data for whole-plant transpiration (mmol s⁻¹), CO₂ assimilation (µmol m⁻² s⁻¹), and photosystem II efficiency used to generate mean values shown in Fig. 1.

Author contributions

JJS and SMG: conceptualization, resources; JJS, BSA, and SMG: methodology, visualization; JJS, BSA, SKP, CH, and SMG: investigation; JJS, TWO, and SMG: formal analysis; JJS: funding acquisition, writing—original draft; BSA, SKP, CH, TWO, and SMG: writing—review & editing.

Funding

This work was supported by the National Science Foundation (grant number IOS-1907338 to JJS).

Data availability

All data are available in the manuscript and Supplementary data.