Plants and water: the search for a comprehensive understanding

Abstract

From the individual plant to the global scale, plant–water relations is focused on pools and fluxes of water. Soil water content, atmospheric absolute humidity, plant water content and the thermodynamic potentials driving the transfer of water molecules are all expressed on a volumetric basis, justifying the description of the soil–plant–air continuum as a simple series of buckets connected by streams of water (Sperry et al., 1998). Yet the persistence of plants in the face of environmental stress often involves adaptations that manipulate the shape of water, as in the response of water droplet contact angles to leaf hairs and the sculpting and composition of cuticular surfaces (Barthlott and Neinuis, 1997). Adaptations may also transduce changes in water content into changes in plant shape that directly affect water flux, such as the wilting or rolling of leaves to minimize energy loading. It is through such explorations of how shape contributes to function that the adaptive significance of the extraordinary morphological diversity of plants can be understood, and the spandrels sorted from the evolutionary innovations (Gould and Lewontin, 1979). The beauty (and for students the consternation) of plant–water relations is the ability to describe in rigorous terms how plants have been able to manipulate physicochemical gradients to ensure water flux into plants, and to hold onto the water once acquired. Unlike animals, plants cannot transport themselves to water, and thus must create favourable chemical potential gradients of water (termed water potential) such that water spontaneously enters the plant. The ability of plants to acquire and move water can be astounding, as demonstrated by mangrove trees that extract freshwater from seawater, Sequoia trees that move water >100 m straight up and desert xerophytes that pull water from soils at tensions 50-fold larger than anything experienced in animals. The ability to do this involves, in part, manipulation of osmotic gradients, but, more fundamentally, plants use structure – notably within the design of cell walls – to generate favourable water potential gradients. The ability of plants to use rigid structure to establish favourable water potential gradients for water uptake allows them to directly couple water flux to solar energy-driven transpiration (Zimmermann, 1983). This avoids the need to spend large amounts of metabolic energy acquiring and transporting water, particularly where significant height is involved. Indeed, one of the characteristic differences between plants and animals is the ability of plants to use rigid tissues to generate tensions capable of extracting water from undersaturated soils. The consequence is a world of sessile plants – silent, immutable, steadfast – surrounded by the busy activity of the often noisy and mobile animal world.

As anthropogenic forcing continues to warm the climate, the simple and robust thermodynamic prediction is that wet areas will on average become wetter and dry areas drier, even as extreme events become more frequent and intense (Held and Soden, 2006; Durack et al., 2012; Liu et al., 2021). Drought kills plants through the effects of dehydration on metabolism and cytosolic chemistry, which involve changes in effective water concentration and so volume. Dehydration also kills through irreversible effects on the mechanical integrity of the plant, for example by collapsing conduits or living cells (Zimmermann, 1983). Intense rainfall and flooding can be just as dangerous for plants, for the simple reason that gases diffuse about 10 000 times more slowly through liquid than through air, such that anoxia in flooded roots becomes a threat. Paradoxically, flooding can lead to shoot drought stress because hydraulic flux through the roots is impaired by flooding, which may mean that leaves are water deficient (Tournaire-Roux et al., 2003). While survival of extremes of water availability is important for most plants and influences many of their fitness attributes, the use of water during non-extreme periods influences their ability to efficiently capture and transport resources such as carbon and soil nutrients, and to efficiently do so. Patterns of water use during times of hydraulic sufficiency have significance for the ability of plants to compete against neighbours and thus to persist in their ecological communities over multiple generations. Water relations is not simply a matter for dry environments, because poor regulation and transport in an otherwise moist environment can create hydraulic imbalances within the plant that could desiccate parts of the plant. For farmers, crop yields are often dependent upon optimal water use of plants in non-stressful conditions, with the efficiency of plant water use influencing irrigation timing and costs. Photosynthetic carbon gain comes at the direct expenditure of water, simply because stomata need to open to let CO₂ in to feed photosynthesis in water-filled mesophyll cells, with the inevitable loss of water from those cells. Stomata are the main regulators of the balance between CO₂ acquisition and water loss, but in this task plants have often evolved complementary structures to help control water loss, such as leaf rolling in grasses, hairs and waxes to influence boundary layers and leaf energy balance, and stomatal plugs and crypts.

Even the acquisition of other resources such as soil nutrients is dependent on the water supply, as well as flux patterns and resistance mechanisms designed to maintain safe and efficient use of water. To ensure sufficient nutrient acquisition, plants will often manipulate water fluxes and pools to enhance nutrient acquisition and storage, as demonstrated by hydraulic lift where deep roots transport water up from nutrient-deficient ground water to re-moisten nutrient-rich layers near the soil surface (Dawson, 1993). The role of stomata in restricting water loss is well known, but plants also construct valves termed hydathodes to help retain nutrients coming into leaves via the transpiration stream. Hydathodes may be particularly important in maintaining a favourable nutrient balance; by filtering nutrient ions from the water stream, they enable water flux through leaves at night, which allows for sustained uptake of nutrient-laden water when transpiration is nil (Mehltreter et al., 2022).

Our understanding of plant–water relations has grown over the decades, in no small part because of the biophysical contributions underpinning the theory of plant water potential, now some 60 years old (Slatyer and Taylor, 1960). However, this theory, and other contributions by the pioneers of plant–water relations, often required simplifying assumptions, for example that water vapour in intercellular air spaces is saturated, that stomata are the principal resistance points controlling water flux, that solutes in the water of the soil–plant–atmosphere continuum are present in low concentration, that the xylem is incompressible and that the water column from deep in the soil to the leaf is isothermal. However, many plants live in situations where simplifying assumptions may not hold and, accordingly, they have evolved structures and functions that reflect this. For the plant biologist, it is important to keep in mind these assumptions, and be aware of when they may not hold. Where the assumptions fail, there are often fascinating insights to be discovered into how plants manipulate water use in their natural environments.

Thus, an important theme in plant growth and environmental response at a variety of scales is the work of shaping of water into a mix of liquid and gaseous phases capable of supporting photosynthetic life. In this special issue of Annals of Botany, we present a series of papers that investigate the consequences of water use for plants. The papers investigate the mechanisms plants have evolved to manipulate the available volumes of water imposed by the environment into the most resilient shapes for the maintenance of growth and survival. The papers extend beyond the classic approaches of hydraulic limitations and stomatal control to address unique and often-overlooked aspects of plant–water relations. Long-standing assumptions are re-examined, notably the assumption that water vapour in the intercellular air spaces is saturated in most cases. Through the papers, a broader perspective on plant–water relations emerges, allowing the reader to better appreciate the many ways in which plants are adapted to stay hydrated when they lack the means to simply get up and pour themselves a drink.

The special issue begins with a perspective on the costs and benefits of mesophyll conductance (Mizokami et al., 2022). Mesophyll conductance (g_m) refers to the series of internal conductances to CO₂ diffusion from the intercellular air space just inside the stomata to the site of carboxylation in the chloroplast stroma. These limitations associated with the individual points of conductance along the diffusion pathway from the intercellular air spaces to the stroma have a marked impact on carbon gain because they reduce the stromal CO₂ concentration about 25 % in addition to the already significant drop across the boundary layer and stomata (also near 25 %; Von Caemerer and Evans, 2015). If the component processes controlling mesophyll conductance are inexpensive, then it follows that the plants should enhance investment in these processes to minimize the limitations; for example, as Mizokami et al. (2022) indicate, more cooporins and aqua-cooporins in the mesophyll plasma membrane could accelerate CO₂ influx without much additional cost in terms of protein. They identify conductance across the cell wall as a major limitation to overall conductance, which they argue explains, in part, a correlation between construction costs of leaves and the value of g_m. For example, as growth light intensity increases, both g_m and leaf construction costs rise, driven in part by greater numbers of mesophyll cells, chloroplasts and the associated cell walls. This correlation allows leaf construction costs to serve as a proxy for costs of greater g_m, but not always. Drought, for example, can also create an inverse relationship between construction costs and g_m, in part due to greater wall content in droughted leaves. In addition to higher construction costs, droughted leaves have lower g_m, possibly due to regulated deactivation of porins by abscisic acid (ABA) and dephosphorylation. Mizokami et al. (2022) hypothesize that drought creates a case where thicker cell walls lead to long-term fitness benefits such as improved drought survival. If so, this would represent a case where optimized investments into g_m are superseded by essential investments such as thicker cell walls that have a high benefit during drought. Within cells, construction and maintenance costs of specific g_m components such as porins or carbonic anhydrase are not onerous by the analysis of Mizokami et al.; however, they note that there is much uncertainty and probably numerous costs that cannot yet be accounted for. It is probable that the multiple steps within the CO₂ flux pathway could collectively represent a significant cost that would favour selection to optimize the costs vs. the benefits of g_m.

A long-standing assumption in plant biology is that water vapour in the intercellular air spaces is saturated at leaf surface temperature, which in turn reflects an assumption that the critical resistance controlling transpiration through plants (i.e. from the soil to the air) occurs in the stomatal pores. As Rockwell et al. (2022) and Santrucek (2022) separately discuss, the assumption of internal saturation may not hold, such that the interior air spaces of leaves are drier than previously thought. Undersaturation in leaves has numerous consequences, but two stand out. First, since gas exchange estimates of intercellular CO₂ concentration depend upon assumptions of saturation, recent isotopic inferences of undersaturation call into question decades of study on the effects of water stress on carbon gain by plants (Cernusak et al., 2018). Second, even a ‘modest’ reduction in relative humidity inside the leaf of 10 % implies liquid phase water potentials in the cell walls equivalent to –15 MPa. Because such water potentials are incompatible with a living symplast in most plants, either the plasma membrane or the cell wall itself must be capable of being a limiting resistance to water loss that could rival the stomatal resistance. Moreover, such a barrier would be a critical target for improving crop water use efficiency and would also be a prime site for natural selection to act. However, a material that conserves water and yet is invisible to CO₂ is perhaps ‘too good to be true’: why bother with the stomata and cuticle then? One source of alternative explanations for apparent undersaturation is the failure of the Ohmic circuit analogy that rests on averages of local quantities such as stomatal aperture, incident radiation, temperature and humidity. As long as the distributions of the local values of these parameters remain tightly unimodal, ‘leaf-level’ quantities calculated based on the local averages remain well behaved. However, it has long been known that as the distribution of stomatal aperture transitions between unimodal and bimodal (as, for example, can happen at high vapour pressure deficit), quantities calculated based on the averages are problematic, as they can indicate changes in mesophyll conductance that are artefactual. Rockwell et al. (2022) make the case that such prosaic explanations must be excluded before the improbable result of a material blocking water but not CO₂ needs to be accepted.

Santrucek (2022) addresses the functional significance of the various forms of stomatal morphology collectively termed sunken stomata. In the classic case, sunken stomata are substantially recessed below the leaf margin, often in crypts lined with hairs. The degree of recession varies and can include stomata whose apertures are partially covered by overhanging epidermal lips or waxy plugs inserted into the stomatal pore. Most people first learn about sunken stomata in xerophytes, where it is presumed they are important for preventing excess water loss in arid environments. However, as Santrucek (2022) discusses, this presumption reflects association more than mechanism, such that the functional role of stomatal crypts and plugs remains to be described. On the one hand, they may effectively enhance the boundary resistance to water loss and hence buffer guard cells from external extremes. They may also act to deter entry of pathogens into leaves and prevent nutrient leaching. Using a novel gas exchange technique to quantify the relative contribution of sunken stomata to the water flux pathway, Santrucek (2022) shows that in plants with sunken stomata, the relative contribution of the guard cells to leaf conductance is reduced, indicating that the control over gas exchange has been partially offloaded from the stomata (gas-phase pathway) to the leaf structure (solid-phase cuticle). Where environmental conditions exceed the functional capabilities of stomata, such offloading of control can be valuable. The relative cost of conductance reductions due to sunken stomata may be less than imagined however, because stomata have greater relative control over the water flux pathway than they do over the CO₂ flux pathway (due to the added flux control of g_m over the CO₂ pathway). Hence, increasing resistance around stomata due to crypts or plugs reduces water loss proportionally more than CO₂ gain, improving plant water use efficiency.

Hydathodes provide an interesting twist on plant–water relations not commonly discussed in the mainstream of the discipline (Mehltreter et al., 2022). As with the traits that enable long-distance transport in the transpiration stream, the function of hydathodes is enabled by structural modifications in the apoplast that establish osmotic and positive pressure potentials in the xylem that are akin to root pressure. The function of these gradients is multi-fold, leading to some uncertainty on the role of hydathodes in specific organisms, as Mehltreter et al. (2022) discuss. Positive xylem pressure could drive flux at night when transpiration is nil, supporting continued movement of mineral nutrients from roots to shoots. Positive xylem pressure could also provide sufficient hydraulic flux to enable excess or toxic ions to be excreted from leaves, thereby maintaining a proper stoichiometry of mineral nutrients in leaves. Positive xylem pressure could also refill conduits embolized over the previous day, thus restoring hydraulic continuity in the xylem, or, alternatively, there could be no specific role for positive xylem pressure, as it may be an inevitable consequence of continued nutrient flux into the vascular cylinder at night when transpiration is nil. Whatever the function of positive pressure, the hydathodes serve as a regulatory filter that operates in tandem with the positive pressure to ensure effective screening of nutrients in the guttation stream and, possibly, to help regulate the efflux rate and thus the size of the pressure gradient. Thus, hydathodes recover essential nutrients from the guttation stream to prevent their loss, while allowing non-essential ions and toxins to pass. Hydathodes may also regulate pressure in the xylem stream by controlling flux into guttation drops, potentially relieving pressure which could otherwise push water into the intercellular air spaces, filling them in what would be a botanical version of drowning in their own internal fluids (Field et al., 2005). Despite their apparent value, hydathodes are not ubiquitous in the plant kingdom, raising questions of whether their distribution reflects phylogenetic ancestry or environmental adaptation, or perhaps both, in that hydathodes might serve as a technology that allows certain phylogenetic lineages to proliferate within specific environments. To assess these issues, Mehltreter et al. (2022) examined the distribution of hydathodes across the fern phylogeny and compared the structure and functions of hydathodes in divergent clades of ferns. A clear phylogenetic pattern was apparent, in that hydathodes were absent from longer branched, and hypothetically older, clades more often than from diverging clades with shorter branch lengths. In these hypothetically younger clades, estimated to have arisen in the past 50 million years, there were multiple examples of hydathode acquisition and loss, indicating that the hydathode trait is dependent upon environmental selection factors. High relative humidity is important for hydathode function, as positive xylem pressure and guttation occur when relative humidity is >90 %. Evaluation of salts in the guttation water revealed that calcium is consistently among the highest excreted minerals, with silica also being concentrated in the sap of certain fern species. High calcium excretion makes sense given that this mineral creates solubility problems in the cellular solution, and its effectiveness as a critical secondary messenger is greatest at low cytosolic concentrations. However, some essential nutrients, notably phosphorus, appeared in the guttation drop, indicating imperfect filtering of essential vs. non-essential nutrients. It may also be that phosphates serve as effective counter-ions to help mobilize excess calcium and toxic aluminium for efflux into guttation drops, as part of the effort to maintain healthy stoichiometries of essential minerals in leaves.

Wimmler et al. (2022) explore the functional implications of root grafting in mangroves. While the benefits of mechanical linkages between trees for stabilizing mangrove forests against storm surges and tidal forces is intuitive, root grafts in mangrove are functional, with exchange of phloem- and xylem-transported resources occurring between potentially unrelated neighbours, which is harder to explain. Wimmler et al. (2022) start by reviewing the literature on root grafting, moving on to a discussion of a variety of hypotheses driving the evolution of grafting behaviours. One hypothesis is that, should damage to an above-ground shoot occur, grafts connecting the surviving root stock to undamaged shoots would hasten the regrowth of the damaged individual, thereby increasing the overall resilience of the stand to disturbance. A second hypothesis for functional grafting, more specific to mangroves, is the idea that given significant spatial heterogeneity in pore water salinity in a stand of trees, grafting would allow individuals with roots in salty pores to access less saline water. Wimmler et al. (2022) then explore these ideas by modelling the effects of grafting based on field observations of actual linkages to the ideal case of all possible pairwise linkages between individuals in a stand. The observation-based mapping of grafts between individuals showed weaker flows between trees than the ideal case, as would be expected from circuit theory (Horowitz and Hill, 2015). By construction in the model, small trees had greater hydraulic root and shoot resistances than large trees, driving flow through grafts from large trees to small. Interestingly, including heterogeneity in pore salinity in the model, from 35 to 40 p.p.t. salinity, had larger effects than the relative difference in tree size between pairs, such that flows could occur from small trees accessing less saline water to larger trees with access to more saline water. This result opens the door to experimental and observational studies to test whether functional grafts in mangroves contribute more to fitness by speeding the recovery of damaged trees (via flows from large/undamaged to small/damaged), thereby improving the resilience of the entire stand to disturbance, or, given heterogeneous and shifting pore water salinities, by simply maximizing stand-level transpiration and carbon capture.

In adapting to arid environments, leaf traits may follow divergent strategies. Meng et al. (2022) investigated the leaf economic and drought resistance traits of 41 cycad species (20 from the Cycadaceae and 21 from the Zamiaceae) to understand the role of co-ordinated leaf trait evolution in the radiation of extant cycads into their current environmental ranges. They report that arid native ranges were associated with thicker but mechanically weaker leaves, and, perhaps most surprisingly, less negative turgor loss points. Structurally, these trait changes could plausibly be explained as consequences of an observed increase in the amount of accessory transfusion tissue, water-conducting cells that exist outside the bundle sheath of the vascular system and conduct water toward the leaf margins (Brodribb and Holbrook, 2005). As the walls of these extraxylery conduits are not completely lignified, they can elastically (i.e. reversibly) collapse, providing a source of hydraulic capacitance (water storage and release) and a reduction of hydraulic conductance that may protect axial xylem from cavitation (Brodribb and Holbrook, 2005). Meng et al. (2022) interpret the shift in traits with aridity as the evolution of a particularly gymnospermic form of succulence, where the role of water storage is played by apoplastic transfusion tissue, rather than thin-walled parenchyma. As discussed in a Commentary by Bartlett (2022), the negative relationship found among cycads between drought tolerance, as represented by the turgor loss point, and leaf economic traits such as mechanical toughness, thickness and construction costs runs counter to both the behaviour of angiosperms co-occurring in the native ranges of the sampled cycads and leaf economic theory, which posits that aridity should select for both resource conservation (tougher leaves) and greater drought tolerance. Diverging strategies of tolerance (drought resistance by more negative turgor loss points) vs. avoidance (increased succulence) may be an important driver of these different relationships.

For leaves to efficiently function, plant stems must time the initiation and maturation of new xylem in a manner that supports seasonal fluctuations in foliar mass. This challenge is of course most acute for deciduous species. Valdovinos-Ayala et al. (2022) report on the phenology of xylem development in boles and branches of three winter-deciduous trees. Of the three, two (red maple and black cottonwood) are diffuse porous, meaning that when looking at a cross-section of a stem, there is little variation in vessel diameter within a growth ring, from early- to late-formed wood; the third species, red oak, was ring porous, meaning that the vessels in the early wood tend to be much larger than those formed in the late wood. Freeze–thaw embolism in the native ranges of diffuse and ring-porous trees renders large, early-wood vessels of ring-porous trees non-functional, and the assumption has long been that ring-porous trees need to replace their lost hydraulic conductivity with new vessels prior to the flushing of new leaves (Zimmermann, 1983). Surprisingly, this is not what these authors found. Rather, as further discussed in a Commentary by Groover (2022), a delay of weeks between xylem formation and maturation of functional vessels meant that at the onset of leaf flushing both the ring-porous and diffuse-porous species relied on xylem from the previous year. However, shortly after the onset of leaf flushing, the ring-porous red oak appears to rapidly transition from depending on previous-year to current-year wood for its functionally active xylem (see fig. 5 in Valdovinos-Ayala et al., 2022). In the diffuse-porous cottonwood, they find a more gradual replacement of the previous-year conduits with current-year xylem over the growing season. A possible ecophysiological interpretation of this difference in timing between ring porous and diffuse porous is that a strategy of trying to meet full transpirational demand with only early wood requires large diameters. However, as these conduits are unlikely to survive winter freeze–thaw embolism (which is more probable in larger conduits), late-wood formation then must focus on small conduits that do not add to the support of current-year transpiration, but are essential for supporting bud break in the following spring. This is not necessarily a better strategy than the diffuse approach of more uniform xylem that survives the winter intact (or can be easily refilled by root/stem pressure in spring) and so can be replaced more slowly over the growing season; however, once a species enters a regime where some conduit diameters are large enough to be permanently lost to freeze–thaw embolism, then the results of Valdovinos-Ayala et al. (2022) do seem to indicate that selection ought to rapidly push their early- and late-wood vessels toward divergent mean diameters.

Clément et al. (2022) investigated the impact on water uptake from different soil layers of variations in root structure and xylem vessel diameter, as functions of root length and branching order. As noted in a Commentary by Pierret (2022), the paradox of stressed crops failing to exploit deep soil water suggests that selection for crop lineages that demonstrate greater rooting depth may not suffice to improve the drought resilience of agricultural systems. This is because roots are ‘leaky pipes’ that can draw water axially from distant sources or pull water radially from the local soil (Landsberg and Fowkes, 1978). The balance at any point in the root between radial and axial inflow depends on the axial conductance and xylem pressure to some deeper part of the root, and the radial conductance and local soil water availability. Clément et al. (2022) studied intermediate wheat grass, with its typical monocot-type fibrous root system, and alfalfa, with a tap root system typical of dicots. For field-grown plants, the tap root system had higher hydraulic conductivities at each depth and made more use of deeper soil water (1.5 and 2.5 m) than the fibrous system. Yet, both species showed similar patterns of declining calculated axial root hydraulic conductivity with increasing soil depth as well as with distance along an individual root from stem to tip. However, the underlying drivers of this decline varied both between species and the choice of single root vs. soil depth as the explanatory variable. Interestingly, while vessel diameter did decline with soil depth in wheatgrass, there was no such pattern in alfalfa, or in either species along individual roots. This is very different from the patterns of vessel tapering that have been reported both in stem and leaf xylem of the shoot (Olson et al., 2014; Carvalho et al., 2017).

Trait-based approaches to ecological questions such as community assembly or species composition along environmental gradients, and increasingly in response to climate change, can be powerful where traits can be cheaply and easily measured across a large number of species. A further consideration, often beyond the scope of any individual study, is how well the suite of traits measured captures the relevant dimensions of plant performance. For example, it would be very convenient if root traits co-varied with leaf traits across environmental gradients in a predictable manner. This might occur along a gradient from high to low resource acquisition, for example as leaves become thicker and smaller, and roots become thicker and shorter. Yet, as explained by Weemstra et al. (2022), root and leaf responses may become uncoupled if edaphic gradients influencing roots are themselves uncoupled from climatological gradients affecting leaves. Furthermore, they point out that if soils vary at smaller spatial scales than climate, then the scales at which root and shoot traits vary, in addition to the directions of variations, may also be uncoupled. To test these ideas, Weemstra et al. (2022) investigated 11 species across a 1000 m elevational gradient in the French alps, measuring pairs of leaf and root traits that would, from a resource economics perspective, be expected to co-vary. The results confirmed their hypothesis that root traits and leaf traits are uncoupled, with only two of the 11 species showing coupling across the elevational gradient. Surprisingly, root traits showed stronger responses to climate variables (e.g. mean annual temperature and mean annual precipitation) than to soil variables (e.g. nitrogen, pH and cation exchange capacity). Also, while leaves generally responded in the expected manner to climate (becoming smaller and thicker as the climate became cooler and drier with elevation), roots responded inconsistently and non-linearly. The authors conclude that root traits might respond at even finer scales than they measured, and that mycorrhizal status may complicate the relationship between root traits and resource acquisition. Alternatively, that leaves responded more strongly to the measured soil variables than roots might indicate that leaf economic traits were more sensitive to variations in successful root nutrient capture than the analogous traits in roots. Some analogies between roots and shoots may not work.

Pritzkow et al. (2022) report on the pattern of embolism spread in three tree species experiencing natural drought. As the authors discuss, a long-standing concept in plant hydraulics has been the idea that larger diameter conduits are more vulnerable to cavitation, and so embolize first under drought. For the two angiosperm and one conifer species studied in a nearby forest experiencing natural drought, these workers found no support for a role for conduit size in the distribution of embolism. Rather, connectivity between conduits appears to control the pattern, which is consistent with the idea that embolism spreads by air-seeding from conduit to conduit across pit membranes (Zimmermann, 1983). In this study, the distribution of embolism showed a strong circumferential pattern of spread within individual growth rings. At first glance, this result seems at odds with previous studies showing primarily radial spread along a file of vessels from pith to cambium, with circumferential spread between adjacent files only occurring where short linking vessels occur (e.g. Brodersen et al., 2013). However, such radial spread patterns have been observed in studies focused on a single growth ring (current-year wood). Reports of circumferential vs. radial spread may not be in conflict, but rather reflect the patterns that hold at different levels of organization (e.g. current vs. multi-year wood). Interestingly, Pritzkow et al. (2022) also report that they saw discreet regions of embolism that extended across a number of growth rings where branch insertions intersected with the axial stem xylem, yet these clusters quickly dissipate and spread apart as one moves up or down the stem, due to the fact that significant numbers of vessels do not appear to remain localized in a single file, but can change position with neighbours many times along their meandering lengths (Zimmermann, 1983). Observations such as these are challenging to make, due to the extreme aspect ratio (metre scale length vs. micron scale diameter) of xylem vessels. However, such observations may provide the key to understanding how plant vulnerability to cavitation emerges from the interplay of pit membranes, conduit-level properties and xylem architecture.

The lack of compelling evidence that vessel diameters are constrained by trade-offs between safety and efficiency raises the question of what are the trade-offs that do shape wood structure. Doria et al. (2022) investigate the allocation of stem volume to different xylem cell types in 75 (42 families) from the Brazilian Cerrado. As described by the authors, wood density has long been relied upon as a broad indicator of where plants lie on an ‘economic’ spectrum from slow growth and resource acquisition to fast growth and resource use. High wood density, they also note, is associated with higher embolism resistance, as vessel walls and the matrix of fibres must be strong enough to support the tensions that develop in the vessel lumens. Doria et al. (2022) sought to shed light on these relationships by investigating the patterns of allocation to cell types that drive variations in density. Across their studied species, fibres (largely dead xylem cells that provide capillary storage of water, filling and draining at water potentials proportional to their diameter) occupied the largest fraction of stems, on average about 50 %. Furthermore, most of the 3-fold variation in wood densities across species was explained by variations in the fibre wall and lumen fractions, with more lumen and less wall driving lower density. The authors also point to a trade-off between fibres and living cell (ray and parenchyma) fractions in wood, which suggests a trade-off between capillary water storage in fibres and starch storage along a high-density to a low-density axis. This result raises an interesting question about constraints and capillary storage. For a tree pursuing a slow growth and persistence strategy, allocating space to energy storage parenchyma would reduce space for fibres, and so fitting in enough fibre cell wall would then require shrinking fibre cell lumens. Small lumens not only hold less water, but discharge at more negative water potentials. This matters as the advantage of stem storage is that it can fill overnight when the tension in water in the soil is most relaxed, and discharge during the day to protect the highly vulnerable root–soil interface, a site that can develop very large resistances under high transpiration rates (Cowan, 1965). Stem capillaries that hold water too tightly therefore risk greater cavitation as well as greater loss of root–soil contact before they drain. Whatever the merits of this particular argument, these kinds of functional considerations could lead to a richer understanding of the trade-offs embedded in such integrative measures of xylem structure as wood density.

Once soil drought intensifies to the level at which stomata close, incident levels of light energy could become excessive, potentially inducing photoinhibition within leaves. For many leaves, desiccation results in changes in shape and/or display angles that reduce light interception. These effects may occur across the entire leaf, for example when the whole leaf wilts, or can be localized to specific cell types whose changes in volume drive leaf rolling, folding and curling. Hura et al. (2022) investigate structural and biochemical changes associated with the presence and absence of leaf rolling in different lines of triticale, a hybrid of rye and wheat. As non-rolling leaves experience no reduction in solar energy loading, they must contend with excess light energy which endangers the integrity of electron transport and can result in the production of reactive oxygen species. Hura et al. (2022) hypothesized that non-rolling leaves would meet this challenge by upregulating proteins crucial for the structural integrity of the photosynthetic apparatus, proteins binding carotenoids in the antenna complex and those supporting efficient electron transport between photosystems II and I. However, photoprotection is not the whole story: rolling and wilting reduce the radiative energy load upon leaves, and so help suppress leaf heating and associated increases in transpirational water losses. Hura et al. (2022) found that non-rolling leaves under drought developed a 5-fold increase in cell wall-bound phenolics, even as the rolling variety that had similar levels of cell wall phenolics in the unstressed state saw no increase under drought. In the cell wall, these phenolic compounds form cross-links with carbohydrates, increasing the rigidity and hydrophobicity of the wall, as well as blocking UV radiation. Whether increased cell wall stiffness contributed to a lack of rolling, or there were changes in the bulliform cells, remains unknown, but Hura et al. (2022) did observe for non-rolling leaves a one-third decrease in water loss rates between stressed and non-stressed leaves, a difference associated with a >3-fold increase in cell wall phenolics. Such a stress-induced, and apparently irreversible, change in wall properties could conceivably contribute to undersaturation of water vapour in leaf airspaces. A one-third decrease in water loss, assuming no change in minimum stomatal and epidermal conductance, implies a one-third reduction in the transpirational driving force. If this apparent change in hydraulic resistance due to increased cell wall phenolics is localized to mesophyll cell walls, it would be large enough to explain reports of 80 % relatve humidity in the airspaces of leaves (Cernusak et al., 2018).

Only stomata can exert a controlling influence over transpiration as the rest of the leaf surface is covered by cuticular waxes that suppress water loss. In wet climates, the cuticle serves both to suppress non-stomatal water loss when humidities are low and to promote the roll-off of water droplets during rainfall events that otherwise develop into water films on the leaf surface that can impede gas exchange. In dry environments such as the Patagonian steppe, however, the situation is somewhat different; Cavallaro et al. (2022) present evidence that leaf surface properties are tuned to capture dew and small rainfall events that do not sufficiently wet the soil. In their system, a trade-off is observed between rates of foliar water uptake and the hydrophobicity of the leaf surface, as measured by water droplet contact angles. What was surprising was that in absolute terms, the leaves appeared to be highly wettable, with many leaves showing contact angles <90°. Cavallaro et al. (2022) show that this high wettability aids interception of rainfall and the adhesion of water droplets, increasing the amount of water ‘stored’ on the leaf surface after such events; such water is then available to be absorbed across the cuticle, a slow process occurring over a time scale of hours. Interestingly, the authors report that total foliar absorption over 3 h, close to the typical duration of natural wetting events, was about five times less than the amount of water present on the leaf, suggesting that adhesion and storage are critical for maximizing uptake from wetting events. While the gas exchange impact of water films on the leaf surface was not studied, an interesting avenue for future work would be to compare the total ‘extra’ transpiration (and thus assimilation) permitted by the volume of water taken up from the leaf surface with the ‘opportunity cost’ of total carbon gain lost due to a period of gas exchange suppression by leaf surface water films.

Leaf economic traits may not respond to all environmental stress gradients. Pan et al. (2022) performed a meta-analysis to inquire into whether plant flooding response traits (shoot elongation, root porosity and the root–shoot ratio) were independent of leaf economic traits, as well as plant size. They find that flooding response traits are largely independent not only of leaf economic traits and the size of a plant, but also of each other. Plants should then be free to respond to specific flooding-related stressors independently, unburdened by trade-offs. For example, plants can address anoxia though increases in root porosity, and/or reduce below-ground oxygen requirements by reducing the root–shoot ratio, while elongating the shoot above flood water or initiating adventitious roots to help prop up stems in moving water, without compromise. According to Pan et al. (2022), an important implication is that we should expect the same ability to respond to flooding from oligotrophic to eutrophic systems. One exception to this pattern of independence was that, between species, leaf nitrogen showed a partial trade-off with root porosity. That this pattern was evident across species, rather than within, suggests that the anti-correlation is not due to an inescapable mechanistic linkage between foliar nitrogen and porosity. Indeed, from a mechanistic perspective, given that anoxia suppresses nitrate uptake (Liu et al., 2015), one might expect that higher porosity would in general positively correlate with higher foliar nitrogen. That a trade-off at the interspecific level exists suggests that foliar nitrogen levels and the amount of porosity induced in roots by flooding are sensitive to genetically hard-wired life history or resource acquisition strategies. Inducing relatively low porosity makes sense if you are adapted to fertile flood plain conditions, as the typically moving waters will be better oxygenated than stagnant (and typically nutrient-poorer) bogs.

In considering the papers in this special issue, we note that many are focused on adaptations, or suites of adaptations, to particular hydraulic and climactic environments. These studies contain important information on the trait variations, within and between species, that govern resilience at the individual and community level. However, in thinking about plant resilience in the face of future climatic shifts, it will probably not be sufficient to simply look at how trait values or physiological performances shift along gradients in annual means of climactic variables, in the classic observational approach of substituting space for time. Due to the sessile nature of plants, persistence in the landscape requires that they build their bodies to withstand the greatest local stress that typically occurs in the interval between growth and senescence. Often, as in the case of the xylem, adaptation and acclimation to a particular level of stress represents a gamble about future conditions that then becomes genetically hard-wired. In this view, with respect to the intensity and frequency of climate extremes, the faster the past becomes a poor predictor of the future, the less existing variation within populations or migration will be able to mitigate future shocks. Predicting plant responses to extreme events that may occur only once in a human life time thus involves many non-linearities that are poorly represented by extrapolation from more easily observed incremental changes. For this reason, clear mechanistic explanations of plant adaptations, as pursued in these pages, will be required to understand the fate of vegetation in our changing world.

LITERATURE CITED