Leaf Shrinkagewith Dehydration: Coordination with Hydraulic Vulnerability and Drought

C.Scoffoni et al. SupplementalMaterial1

Leaf shrinkagewith dehydration: coordination with hydraulic vulnerability and drought tolerance

Christine Scoffoni, Christine Vuong, Steven Diep, HervéCochard and Lawren Sack

Includes:

Supplemental Results

Supplemental Discussion

Supplemental Materials and Methods

Supplemental Tables4 and 5

SUPPLEMENTAL RESULTS

Leaf shrinkage with dehydration: variation across diverse species of other shrinkage parameters

All additional shrinkage parameters (Table S4) strongly correlated with their related shrinkage traits from Table III. Thus, ΨT50, RWCT50and PLT’leaf,TLPwere strongly correlated with PLTleaf,TLPandPLA’leaf,TLPwith PLAleaf,TLP (P < 0.01; Table S3).

The Ψleafat 50% loss of thickness (ΨT50) varied 8-fold among species from -0.7 MPa for Helianthus annuus to -5.5 MPa for Q. agrifolia. Relative water content at 50% loss of thickness (RWCT50) varied 2-fold among species from 0.42 in Q. agrifolia to 0.90 for Magnolia grandiflora.

Percentage shrinkage in volume was primarily driven by shrinkage in thickness rather than area. Loss of volume at turgor loss point (PLVleaf,TLP) correlated more strongly with PLTleaf,TLP (rp and rs= 0.96, P < 0.001) than PLAleaf,TLP (rp and rs= 0.69-0.64, P < 0.05). The PLVleaf,TLP varied 9-fold across species from 4.6% for R. indica to 43% for L.camara. The Ψleafat 50% loss of volume (ΨV50) was strongly correlated with ΨT50(Table S3) and varied 4-fold among species from -1.2 MPa for Salvia canariensis to -4.8 MPa for C. diversifolia. Relative water content at 50% loss of volume (RWCV50) was strongly correlated with RWCT50(Table S3) and varied 2-fold among species from 0.50 for H. canariensis to 0.86 for P. racemosa. The maximum shrinkage in volume (PLVdry), i.e., that for a dry leaf, was strongly correlated with both PLTdry and PLAdry (Table S3) and ranged 3.5-fold across species from 25% for Q. agrifolia to up to 89% for L. camara (Table S2).

Leaf shrinkage with dehydration: variation between species of wet and dry habitats

Species from dry and wet habitats differed significantly in their shrinkage. Species from dry habitats showed less thickness shrinkage at turgor loss point (PLTleaf,TLP = 14.6 ± 1.6 vs. 22.1 ± 2.0 %and dTleaf/dΨ = 41.3 ± 2.4 vs. 51.7 ± 2.0 %·MPa-1, P = 0.013 and 0.001 respectively; one-way ANOVAs on log-transformed data; Table S2), reflecting 1.4% less shrinkage of cells and 8.2% less shrinkage of intercellular airspaces (dTC/dΨ = 3.9 ± 0.26 vs. 6.0 ± 0.4 %·MPa-1, P < 0.001 and dTA/dΨ = 1.8 ± 0.64 vs. 7.5 ± 1.7 %·MPa-1, P = 0.004; Table S2). Dry habitat species also showed less maximum shrinkage in leaf thickness, i.e., by 47.2 ± 2.6% on average compared with 57.4 ± 3.2% for species from moist habitat (P = 0.021; one-way ANOVA on log-transformed data; Table S2). No significant differences were found between species of dry and moist habitats in their leaf shrinkage in area at turgor loss point, or their maximum shrinkage in area (PLAleaf,TLP, P = 0.81 and PLAdry, P = 0.07; one-way ANOVA on log-transformed data).

Dry habitat species lost on average 50% of their initial thickness at -3.2 ± 0.27 MPa compared to -1.7 ± 0.13 MPa for leaves of moist habitat species (P < 0.001, one-way ANOVA on log-transformed data; Table S2).

Correlation of other leaf shrinkage parameters with pressure-volume parameters and cuticular conductance

A strong trend was observed for pressure volume curve parameters with ΨT50 (rp and rs= 0.91-0.93, P < 0.001; Table S3). However, no correlations were found between parameters of leaf shrinkage and the RWCTLP (|rp| and |rs| = 0.04-0.53, P > 0.05; Table S3), except the directly-related percentage of cell shrinkage at turgor loss point (PLTC,TLP: rp = -0.79, P < 0.01).Cuticular conductance (gmin) was positively correlated with ΨT50, RWCT50 (rp and rs = 0.65-0.79; P < 0.05; Table S3).

Recovery of leaf shrinkage in thickness

Leaf shrinkage was only partially recoverable after one-hour rehydration, to a degree that depended on the species. No significant differences were found in thickness recoverybetween leaf discs previously cut in air versus cut in water after one-hour rehydration, whether discs were dehydrated between full turgor and turgor loss point, or past turgor loss point (paired t-test; P = 0.08-0.59), and thus the data for these treatments were pooled.

Bauhinia galpinii and Lantana camara showed less recovery when dehydrated past turgor loss point (18 and 23% less recovery respectively), although still recovering 50-60% of their initial thickness (Table S5).

SUPPLEMENTAL DISCUSSION

Impact of leaf shrinkage on leaf hydraulic vulnerability

A recent study using the rehydration kinetics method (RKM) for leaf discs showed no effect of dehydration on Kleafat high Ψleaf, and concluded that xylem embolism is the main driver of Kleafdeclines (Johnson et al., 2012). However, several methodological issues challenged the ability of that approach to test the specific role of shrinkage. First, the RKM does not mimic the natural transpiration pathways for water movement, and may partly or fully remove the extra-xylem component of the hydraulic resistance (Scoffoni et al., 2008). Second, in that experiment, the leaf discs rehydrated during the water uptake measurement, alleviating tissue shrinkage, and thus this factor might not influence Kleaf as in dehydrating whole leaves (as discussed previously in relation to the RKM by Brodribb and Holbrook, 2003). Third, the RKM requires precise knowledge of rehydration time to calculate Kleaf, and although the authors used leaf disc submersion time, rehydration of cells is likely to continue even after the leaf is withdrawn from water, enhancing the leaf discs’ apparent hydraulic conductance.

Mechanisms of leaf shrinkage

Our experiments on leaf shrinkage measured at macro-level (i.e., for whole leaves) provided further insight into processes occurring within the leaf during shrinkage. Previous studies used light and cryo-scanning electron microscopy (cryo-SEM) to examine shrinkage of cells and tissues within the leaf for a few species (Fig 1). Those studies found that round spongy mesophyll cells are more vulnerable to shrinkage than cylindrical palisade cells (Fellows and Boyer, 1978; McBurney, 1992; Colpitts and Coleman, 1997;Canny et al., 2012) and that species with greater ratios of palisade: spongy mesophyll experienced less shrinkage with dehydration (Burquez, 1987). Spongy mesophyll cells have fewer chloroplasts than palisade mesophyll cells, and less negative osmotic pressures (Nonami and Schulze, 1989; Koroleva et al., 1997), and have thinner cell walls relative to their lumen volume (John et al., in press); these effects would result in greater loss of turgor and shrinkage for a given drop in water potential.

Epidermal cell shrinkage can be an important contributor to leaf shrinkage. Because of their higher water storage properties, epidermal cells are thought to shrink more rapidly, and potentially to reflect leaf thickness changes (Meidner, 1952; Edwards and Meidner, 1978). Having epidermal cells resistant to shrinkage might be an important drought tolerance trait. Species with resistant epidermis can maintain leaf integrity, even as intercellular airspaces expand during dehydration due to shrinkage of the mesophyll cells, and thus drought tolerant species tend to show an increase in their intercellular airspaces (Kennedy and Booth, 1958; Burquez, 1987; Sancho-Knapik et al., 2010; Sancho-Knapik et al., 2011), while drought sensitive species tend to show a decrease in intercellular airspaces with shrinkage (Kennedy and Booth, 1958. Meidner, 1955; Fellows and Boyer, 1978; Burquez, 1987; Colpitts and Coleman, 1997). Our data supported and extended those findings, indicating that species with the most shrinkage at the leaf level were most drought sensitive, and experienced most shrinkage in their intercellular airspaces as well as their cells.

Similarly, epidermal cell shrinkage could play an important role in the shrinkage of leaf area, as the epidermal cells are all directly connected with no intercellular airspaces. Across our species, we found a strong correlation between PLAdry and ε, such that species with lower ε shrank more strongly in area.Maximum area shrinkage could be related to the elasticity of the cells leading to stronger collapse as the dehydration of flaccid protoplasts pull the cell walls inward below turgor loss point.The relationship of the rigidity of the epidermis and cuticle to leaf shrinkage can explain the correlation of area shrinkage with minimal epidermal conductance (gmin). A high gmincan arise from a highly permeable cuticle or from leaky stomata (Sinclair and Ludlow, 1986; Cavender-Bares et al., 2007). Epidermal shrinkage may have opposing effects ongmin, as it may tighten the structure of the waxes in the cuticle, reducing its permeability (Fellows and Boyer, 1978; Boyer et al., 1997), whereas area shrinkage in sunflower enhanced the sealing of stomata as the leaf lost turgor, improving water conservation (Tang and Boyer, 2007). However, epidermal shrinkage may also reduce epidermal pressure against guard cells, and thus promote leakiness from partially closed stomata (Buckley et al., 2011). Our data suggests that overall, area shrinkage may contribute to higher gmin across species. Indeed, results from our partial correlations showed that when partialing out the effect of εor osmotic pressure, the correlation between PLAdry and gmin remained, whereas when partialing out the effect of PLAdry the correlation between gminand ε found in our study, and in a previous study of eight other species (Sack et al., 2003) was lost. This finding suggests that ε influences PLAdry, which is most directly linked withgmin.

Resistance to leaf shrinkage: an important trait contributing to drought tolerance?

Species of dry habitats experienced less shrinkage, on average, than species from moist habitats. This pattern has been observed in previous studies of a total of ten species (Kennedy and Booth, 1958; Burquez, 1987) and our study has extended this finding for more than double the species. This association can be explained by the fact that those species resistant to shrinkage were those with more negative πo and high ε. In a recent meta-analysis of data for more than 300 species, the πTLP, which is driven by the species’ πo, was a strong predictor of species drought tolerance (Bartlett et al., 2012). That study showed that a more negative πTLP enables maintained function at lower Ψleaf, and a high ε played only an indirect role in drought tolerance, allowing such leaves to avoid cell shrinkage and maintain relative water content above lethal levels at πTLP. Here we showed an additional potential role of high ε in drought tolerance, because resistance to shrinkage allows a leaf to better maintain its structural and hydraulic integrity outside the xylem, and thus to maintain a high Kleaf especially at relatively high Ψleaf during incipient drought.

We note that in succulents, an opposite mechanism for drought tolerance may operate. Outside of specialist plants such as succulents, not included in our study, most species adapted to drought do so via a more negative πTLP, rather than a high water content (Bartlett et al., 2012). Succulence involves water storage typically based on cells that can shrink without sustaining damage (Ogburn and Edwards, 2012). In such species, tissue shrinkage would benefit the mechanism providing the ability to survive in dry soil. We found that species with high CFT, CTLP and SWC showed stronger shrinkage in thickness and area. These species also lost hydraulic function rapidly during leaf dehydration. However, for species with specialized water storage cells isolated from those in the hydraulic pathway cell shrinkage would likely be tolerable, and shrinkage would be an important drought tolerance mechanism for succulent species.

Species of dry habitat tend to have high LMA, though a high LMA does not predict drought adaptation, as many moist habitat species also have high LMA, and succulent leafed or malacophyll species of dry habitats can have low LMA (Bartlett et al., 2012). Our results showed that species with high LMA and leaf density experienced less shrinkage. This linkage was probably due to their very tight correlation with p-v parameters (Sack et al., 2003); partial correlation analysis could not separate LMA or leaf density from the πo or ε in our species set.

SUPPLEMENTAL MATERIALS AND METHODS

Leaf shrinkage experiments: testing leaf responses to dehydration

The precision of the digital calipers was sufficient to resolve leaf shrinkage with dehydration. The accuracy of the calipers for measuring leaf thickness was confirmed by testing for correlation with leaf thickness measured from images of cross sections measured using ImageJ (Rasband, 1997-2012) for fully hydrated leaves of the species in our study (using data from John et al., in press), We found a strong relationship between thickness measured using calipers (Tcalipers) and that measured from cross sections (Tcs); the least squares regression fitted through the origin was Tcalipers = 1.05 × Tcs; r2 = 0.82; P < 0.001).Thickness was measured each time at a slightly different spot on the lamina to avoid any crushing of the leafby the calipers. To confirm that thickness measurements were not influenced by crushing, two control leaves per species were kept fully hydrated with petioles in water, and repeatedly measured for thickness in the same way during the course of the experiment. The thickness of control leaves did not vary with repeated measurements, indicating that crushing was not a problem.

Several species required special treatments before the experiments. For P.racemosa, the dense trichomes on the abaxialsurface were gently shaved off with a razor blade in regions in the center of the bottom, middle and top thirds of the leaf to ensure accurate thickness measurements. Because leaves ofC. betuloides and Q. agrifolia are convex and did not lay flat on the scanner, leaf area could not be determined accurately during the dehydration. For these species, five additional leaves were cut at the margins such that the lamina could lie flat, and subjected to the same measurements during progressive dehydration. For those species, the parameters related to shrinkage in area and in volume were obtained from the cut leaves whereas parameters related to shrinkage in thickness were obtained from the uncut leaves. M. grandiflora leaves curled during the dehydration such that the measurements of area were inaccurate, and parameters relating to shrinkage in area or volume were not calculated for this species. For M. grandiflora, dried leaf area was measured for separate leaves that had been weighed at full turgor, and then cut into pieces to avoid leaf curling before oven drying for three days at 70°C.

Two lines of evidence suggested that the parameters of shrinkage and hydraulics measured for excised leaves should be representative of those for leaves dehydrating on the plant. First, studies that focused on daily and seasonal patterns of leaf shrinkage found a similar degree of shrinkage as observed for detached leaves (Kadoya et al., 1975; Tyree and Cameron, 1977; Syvertsen and Levy, 1982; Rozema et al., 1987; Ogaya and Peñuelas, 2006; Seelig et al., 2012) and one study that specifically compared leaf shrinkage in the laboratory and field from five woody species found comparable leaf shrinkage (Meidner, 1952). Second, previous studies of leaf hydraulic vulnerability showed statistically similar behavior for dehydrating detached leaves asfor attached leaves on plants during progressive drought for two species, indicating similar responses of the water pathways(Brodribb and Holbrook, 2004; Pasquet-Kok et al., 2010).

Leaf shrinkage experiments: determination of the other parameters of leaf shrinkage

We calculated the percent loss of volume at turgor loss point (PLVleaf, TLP), and we determined the percent loss of volume for the dry leaf. We calculated a second relative measure of the percent loss of thickness, volume and area at turgor loss point, indicated by a prime symbol, to indicate the percentage of the size of the shrinkable component (i.e., the difference between the fully hydrated leaf and the fully desiccated leaf) that was lost at a given level of dehydration (see Table S4 for derivations and definitions).

We calculated the water potential and relative water content at 50% loss of relative thickness and volume (ΨT50, ΨV50, RWCT50, RWCV50)(see Table S4 for derivations and definitions).

Leaf structural and compositional traits

The fraction of air, water and solid in the leaf were measured for 4 to 10 leaves per species by water infiltration into the airspaces(after Roderick et al., 1999; Sack et al., 2003). Briefly, shoots were collected the night previous to measurements, and re-cut by two nodes under pure water and rehydrated overnight. The next day, leaf area and thickness were measured. Initial mass was measured after leaves were dipped in water, and toweled dry. Leaves were then placed under water in a vacuum flask with a few drops of detergent as surfactant, and vacuum infiltrated for > 6h. Once leaves were completely infiltrated, they were toweled dry, and their mass was determined. Leaf volume was determined by water displacement in a 10 ml graduated cylinder (± 0.1 ml). Leaf dry mass was determined after at least 72 h in an oven at 70°C. Percentage air fraction was calculated as (infiltrated leaf mass – turgid leaf mass) divided by the density of water and by the leaf volume, water fraction as(turgid leaf mass - leaf dry mass)divided by the density of water and by the leaf volume, and solid fraction as 1- the percentages or air and water.

C.Scoffoni et al. SupplementalMaterial1

Table S4. Symbols, terms, unit, derivation and biological significance of 9additional leaf thickness, area and volumeshrinkage traits this study. *We assumed a linear function between leaf dimensions and Ψleaf (or RWC) between the two measurements surrounding 50% loss of thickness/area/volume or turgor loss point.