Secreting salt glands constrain cuticle fracture to enhance desalination efficiency

Plants responding to excessive soil salinity by discharging brine onto their leaf surface risk10 dehydration through the osmotic continuity between the living tissue and the surface brine, which further enriches11 with evaporation. Cuticle cracks have long been identified as essential for salt to reach the leaf surface but provide12 the potentially desiccating continuity between the brine and the gland interior. Using the secreting salt gland of13 Nolanamollis asamodelsystem,weintegratemathematicalmodeling,imaging,andphysiologicalmeasurementsto14 examine the mechanical and biochemical processes required for efficient desalination. We find that the subcuticular15 space between the concentrated surface brine and the more dilute secreting cell eases the energetic limits of active16 desalination by reducing the concentration gradient of salt across the cell membrane. We show that crack size plays17 a critical role in balancing the osmotic and pressure gradients required for salt removal without runaway foliar18 desiccation.19 1 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. Soil salinity poses both a biochemical and a physical challenge for terrestrial plants. Elevated exposure to salt20 interferes with proper osmoregulation and induces protein malfunction and aggregation, leading to stunted growth,21 impaired development, and increased mortality [1–6]. As ions are transported through the xylem, higher rates of22 transpiration exacerbate salt accumulation within the plant [7]. Yet, preventing this accumulation by exclusion23 at the roots produces large osmotic gradients that can impair hydraulic function [8]. To address the full range of24 issues presented by salt stress, strategies for salt tolerance are expected to encompass not only biochemical but also25 physical solutions.26 Halophytic plants have evolved specialized mechanisms for salt tolerance and exist within several plant lineages27 [9, 10]. Many halophytes address belowground salt stress by excluding salt from their roots or by accumulating salt28 in the roots to reduce accumulation in the leaves [11–14]. However, for most plants, there is often not enough water29 flow around the roots to wash away the buildup of excluded salts. As salts accumulate in the root zone, replacement30 of fouled roots incurs significant energetic costs for the plant [6]. Notable exceptions are mangroves, whose roots31 grow in waterlogged soils where mobile seawater can advectively or diffusively dilute the locally hypersaline32 water [12, 15]. Nevertheless, salt absorption cannot be fully avoided, and accumulation of excess salts leads to the33 induction of leaf succulence until the salts are physically cleared from the plant during leaf senescence [15].34 Recretohalophytesrelylessonexclusionandinsteaduseenergytodriveeliminationofsaltfromtheirleavesvia35 specialized cellular structures known as salt glands. Present in many plant genera, salt glands are morphologically36 and functionally diverse, with two main strategies for salt removal [16]. One strategy is sequestering salt from other37 tissues in the vacuoles of bladder-like cells that are eventually shed [17]. The other strategy is secretion of excess38 saltontotheleafsurface(exo-recretohalophytes)[ 18].Secretorycellsattheglandapexactivelydepositsaltbeneath39 the cuticle, either through membrane transporters or exocytosis of microvacuoles [19,20]. The resulting osmotic40 gradient across the cell membrane pulls water from the gland into the subcuticular space, generating pressure [21].41 Cracks in the cuticle allow the salt solution to reach the leaf surface, where it becomes further concentrated by42 evaporation into a brine [22–24]. These cracks also form an aqueous continuum between the resulting surface brine43 and the living tissue. How these plants manage to prevent self-desiccation despite this strong osmotic gradient is44 not understood.45 Here we examine how the secreting salt gland’s multicompartment morphology contributes to the system’s46 success. We develop a steady-state mathematical model of the secreting salt gland to determine the role of the47 intermediate subcuticular space in mediating salt removal without catastrophic water loss. We find that the success48 of secreting salt glands relies not solely on biochemical processes operating at the cell membrane but also on their49 2 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency Fig. 1.Salt gland structure ofNolana mollis. a,N. mollisis a shrub native to Pan de Azúcar, Atacama Desert, Chile. b, The leaves ofN. mollis secrete brine onto their surface (circled). c-d, Cryo-SEM imaging of the adaxial leaf surface (c) reveals multicellular brine-secreting salt glands (d) in depressions of the epidermis (c, white arrows). e, Confocal imaging of the salt gland highlights a chamber formed between the cuticle (yellow) and the cell wall (white) at the gland apex. f, Conceptual diagram of the chambered secreting salt gland ultrastructure. Cracks in the cuticle allow for aqueous salt to be secreted into the surface brine. Scale bars: 1 mm (b), 400µm (c), 10µm (d,e). structural organization and the fracture mechanics of the cuticle.50 RESULTS51 Ultrastructure of the secreting salt gland52 We useNolana mollis(Phil) Johnston as the model species for our analysis of secreting salt glands.N. mollis53 is a succulent-leaved shrub dominating the Pan de Azúcar coastal valley of the Atacama Desert (Fig. 1a) and54 is distinctively covered with a persistent and concentrated brine (Fig. 1b) while neighboring species remain dry55 [25, 26]. N. mollis’s deep taproot provides access to a consistent, but saline, source of groundwater [27]. Secretion56 of the brine maintains a stable whole-leaf osmolality (500 mmol/kg) despite high soil salinity (Fig. S1).57 The salt glands, with a diameter of 20-30µm, lie in depressions of the leaf epidermis and cover approximately58 1% of the leaf surface (Fig. 1c,d). Confocal imaging of the salt gland (Fig. 1e) reveals the cuticle (yellow) visibly59 3 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. Fig. 2.Conceptual framework for the secreting salt gland model. a, Salt (orange) is actively loaded into the chamber while water (blue) moves passively across the cell membrane. The aqueous salt solution in the chamber reaches the surface through cracks in the cuticle. Evaporation from the leaf surface concentrates the surface solution into a brine. b, Through cracks in the cuticle, pressure-driven flow of water can move salt advectively out toward the brine (straight arrows). A concentration gradient across the cuticle drives diffusion of ions through the cracks (curved arrow). If the brine is more concentrated than the chamber, diffusion will run counter to the advective flux. c, Theoretical response of ion transporter activity in the cell membrane as a response to a concentration gradient∆c across the membrane between the gland and the chamber. Transporter activity vanishes above a threshold value∆c∗. detached from the cell wall (white), forming an inflated subcuticular chamber at the gland apex. Details of the60 ultrastructure of theN. mollissalt gland are summarized in Fig. 1f.61 The conceptual framework of the secreting salt gland is presented in Fig. 2. Salt moves from the secretory cells62 into the chamber via active processes. The details of this remain unclear, as specific ion transporters involved in63 foliar salt glands have yet to be identified [28–30]. Microvesicular transport has also been hypothesized but not yet64 confirmed[ 31].Regardlessofthespecificmechanism,movementagainstaniongradientrequiresaninputofenergy65 [32], whether to induce protein conformational changes [33], operate proton pumps [34–36], or drive vesicular66 formation and movement [31]. Histological studies have consistently found mitochondrial enrichment in secretory67 cells, revealing the substantial energetic cost of salt removal [19, 20, 23, 24, 37, 38]. When the concentration68 gradient across the cell membrane becomes steep enough, the energetic cost of moving an ion against its gradient69 becomes too large, and export activity vanishes. A theoretical response of transporter activity to the concentration70 gradient across the membrane is sketched in Fig. 2c. The exact shape of this curve is unknown for real systems,71 and we find that, as long as the response vanishes at a large concentration gradient, our results are qualitatively72 unaffected (Fig. S2).73 The electroosmotic gradient established across the cell membrane by ion transporters drives an osmotic flow of74 water into the chamber via membrane-bound aquaporins. The resulting buildup of pressure inside the chamber75 drives advective flow of the enclosed salt solution out toward the exterior brine via cracks in the cuticle. Diffusion76 of salt also occurs through these channels. If the concentration of the brine exceeds that of the chamber, the net77 4 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency Fig. 3.Model results as a function of fracture pore radius (rf, nm). a, Chamber pressure. b, Total per-area water loss from the gland, normalized by total gland area(πr2). c, Chamber salt concentration (left) and osmotic pressure (right). Ion transporters stop functioning when the concentration gradient between the chamber and the gland is above an activation threshold (horizontal orange dashed line). d, Advective (black) and diffusive (red) salt flux normalized by crack area(πr2 f), referred to as the salt flux density through the cuticle cracks. Positive flux densities indicate movement from the chamber out toward the brine. e, The net molar salt flux from the gland, normalized by total gland area, is a function of the concentration gradient across the cell membrane between the cell and chamber. f, Export efficiency (µmol Na+ per mL water) from the gland. Vertical orange dashed lines indicate the points at which the chamber concentration crosses the transporter activation threshold. The vertical gray dotted line corresponds to the crack size with maximized advective salt flux density and the first inflection point of the water flux profile. diffusive flux will drive ions back into the chamber, counter to the advective current (Fig. 2b).78 The volume and concentration of the surface brine is modulated by evaporation, crystallization of salts, and79 dripping of the brine off the leaf surface [25]. Pre-dawn brine osmolality was measured to be around 2000 mmol/kg80 (Fig.S1).Weexpecttheconcentrationgradientbetweenthebrineandthegland,ifnotseparatedbytheintermediate81 chamber, to be too steep to be overcome by active transport processes. In our model, we consider the steady state,82 with the brine concentration set to its empirical value, to determine the contribution of the intervening subcuticular83 space to the efficiency of salt expulsion as a function of cuticle crack size.84 Model results85 Model results are illustrated in Fig. 3. When cracks are small, pressure builds up in the chamber, and when cracks86 are too large to contribute meaningful resistance to flow, the pressure relaxes to atmospheric levels (0.1 MPa) (Fig.87 3a).Thetotalwaterflux,correspondingtothewaterlostfromthegland,increaseswithlargercracksizes,eventually88 plateauing as the increasing crack size corresponds to the complete destruction of the cuticle (Fig. 3b). Considering89 5 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. salt gland coverage at approximately 1% of the leaf surface, the maximal water loss from the salt gland is about90 0.3 mL/cm2/day, or 3 mm/day, comparable to the potential evapotranspiration of arid lands (4-6 mm/day) [39].91 Whereas transpiration is limited to times the stomata are open, gland-mediated water loss is continuous, potentially92 exceeding net daily transpiration and risking desiccation if the cracks are too large. Moreover, the brine itself is93 composed of gland secretions and nighttime deliquescence. Lower water loss rates at smaller cracks suggest that94 deliquescent water accounts for a larger volume fraction of the brine towards this limit. This decreases the water95 demandfordesalinationbysubsidizingthetotallandevapotranspirationwithwatercondensedfromtheatmosphere.96 Atboththesmallandlargefracturelimits,saltconcentrationinthechamberexceedsthethresholdfortransporter97 activity, suppressing salt export. With large cracks, the chamber and brine become indistinguishable compartments,98 causing the chamber concentration to approach that of the brine. High resistance across the cuticle through small99 cracks impairs both water and salt export, leading to a retentive regime where salt builds up within the chamber. At100 intermediate crack sizes, the concentration within the chamber drops by around 40% (Fig. 3c). Within this range,101 advective fluxes move salt to the leaf surface faster than the diffusion of salt backward into the chamber.102 Fig. 3d illustrates the advective and diffusive components of the salt flux, normalized by the crack area (πr2 f),103 which we refer to as the flux density. While the gross advective and diffusive fluxes monotonically increase in104 magnitude with crack area (Fig. S3), their densities achieve local optima (Fig. 3d). Large resistance across small105 cracksrestrictsflowacrossthecuticleandmaintainsahighconcentrationofsaltwithinthechamber.Withincreasing106 crack size, resistance drops, leading to larger advective fluxes, with a maximum occuring at a crack size just under107 100 nm (Fig. 3d), corresponding to the first inflection point of the water flux profile (Fig. 3b). Larger fluxes more108 effectively clear salt from the chamber. The drop in concentration below the activity threshold (dashed line) allows109 ion transporter activity to resume, permitting active salt export from the secreting cell in this range (Fig. 3e).110 Despite increasing conductance with even larger cracks, greater crack area ultimately reduces the advective flux111 density by both increasing the flux area and by lowering the pressure gradient across the cuticle (Fig. 3a).112 The inward diffusive flux reaches a maximum with a crack size near the lower limit of transporter activation.113 This corresponds to the increasing concentration gradient across the cuticle resulting from a more dilute chamber114 solution. With larger cracks, a smaller gradient across the cuticle, coupled with a larger crack area, reduces the115 diffusive flux density through the cracks. We note that the advective and diffusive flux densities scale differently116 with rf and therefore reach their optima at different crack sizes. Within the range of crack sizes that allow for117 positive transporter activity, a large, outward advective flux of salt overcomes a smaller, inward diffusive flux,118 leading to positive salt export. Immediately outside this range, a small amount of salt export occurs, corresponding119 6 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency Fig. 4.Fracture of the salt gland cuticle. a, Model predictions for functional regimes of cuticle crack sizes. Exceedingly narrow cracks ( 400 nm) lead to impairment of ion transporters, arrest salt export while enabling catastrophic water loss. b-c, Cryo-SEM imaging reveals cracks in the cuticle surface through which brine escapes. cracks were observed on the order of 20 nm (b) or 100 nm (c) in width. d, Conceptual graphic of fracture in the cuticle. (i) The cuticle of the naive salt gland is crack-free, and the chamber has not yet fully inflated. (ii) As salt (orange) and water (blue) are exported into the subcuticular space, building pressure inflates the chamber. (iii) Fracture initiates when the pressure exceeds the cuticle strength. (iv) cracks grow until the chamber pressure drops below the propagation stress. (v) If cracks reach a stable size within the operable range of the salt gland, the leaf will be able to export salt without catastrophic water loss. (vi) If cracks grow too large, the salt gland exits the operable range, leading to desiccation. Scale bars: 500 nm (b,c) to the smooth shoulders of the chosen activity profile. Other profile shapes do not qualitatively affect the results and120 are explored in the SI. Beyond this range, while these flux densities are low at this limit, the large crack area allows121 for nonzero advective and diffusive fluxes that balance each other, resulting in negligible salt export (Fig. S3).122 The export efficiency (Fig. 3f) measures the amount of salt removed (Fig. 3e) relative to the water lost from the123 gland (Fig. 3b). While water loss is monotonically minimized with smaller cracks, export efficiency vanishes at this124 limit as transporter activity halts against the steep concentration gradient. This establishes a global optimum in the125 export efficiency atrf≈ 20 nm, corresponding to a state in which transporters have been fully activated while the126 water flux remains low, one or two orders of magnitude below the potential evapotranspiration [39].127 Fracture mechanics128 Cryo-SEM imaging of the salt gland cuticle revealed cracks in plants that were secreting brine (Fig. 4b-c). Glands129 from plants watered with dH2O did not secrete brine, and cracks were not observed on their cuticle surfaces. The130 7 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. observed crack width from active glands ranged between 10 and 200 nm, less than 1% of the gland’s diameter. We131 recognize that flexure of the cuticle surface may result in variable crack width along its depth, restricting the true132 effective crack width [40, 41].133 Fig. 4d illustrates the development of cuticle fractures and the mechanics necessary to prevent excessive crack134 propagation and, ultimately, salt gland dysfunction. A naïve salt gland begins with an intact, crack-free cuticle135 surface. Upon activation of salt export, the flux of salt and water out of the cell inflates the chamber by pushing the136 flexible cuticle away from the rigid cell wall. When the chamber pressure exceeds the strength of the cuticle, a137 crack forms. The growing crack releases pressure in the chamber and eventually arrests. If the crack arrests within138 the operable range of the salt gland, salt can be exported without significant water loss. However, if the crack grows139 beyond the range of operable sizes, export efficiency will vanish as transporters become overwhelmed and osmotic140 movement of water from the gland leads to dysfunction and desiccation. This process is likely irreversible, as there141 are no known mechanisms for cuticle repair after its detachment from the cell wall. Moreover, we observed that142 leaves previously treated with salt continued to secrete fluid even after the watering regimen returned to exclusively143 deionized water. This implies a point of no return for salt gland function, which may contribute to the observed144 high leaf turnover rate of wildN. mollis.145 How does the cuticle achieve a crack of intermediate size to stay within the desalinating range (Fig. 4a)?146 Exceedingly small cracks restrict water flow and block salt export, leading to retention of salt. Strength estimates147 for fruit cuticles suggest that the salt gland’s cuticle is not strong enough to resist the high pressures at the small148 crack limit [42,43]. This apparent weakness is, in fact, essential for the crack to grow large enough to escape the149 retentive regime and activate salt efflux processes. To avoid runaway propagation into the desiccating regime, the150 crack’s growth must arrest within the intermediate range. The fractocohesive length is a length scale describing the151 dependenceofamaterial’svulnerabilitytocrackpropagationonthesizeofexistingcracks.Foracracksmallerthan152 the fractocohesive length, the crack tip does not concentrate stress, and the strength of the material is independent153 of the crack’s size [44,45]. We use the measured properties of fruit cuticles to estimate the fractocohesive length of154 the salt gland cuticle to be on the order of 1µm to 1 mm [44,46,47]. As this fractocohesive length is larger than155 the upper limit of the predicted operable range, and potentially larger than the length scale of the gland itself, we156 expect that the strength of the cuticle does not drop as the crack grows, allowing it to resist runaway propagation.157 Moreover, the pressure gradient across the cuticle vanishes with a crack size just above 100 nm, removing the158 driving force for propagation. Consequently, the crack arrests well within the operable range.159 8 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency DISCUSSION160 Its persistent surface brine is a distinctive characteristic ofNolana mollis, but the risk of desiccation is not a unique161 problem. Even if the surface brine fully evaporates, the remaining surface salt content is not likely to be fully162 decoupledfromthesubcuticularspace,duetoitshygroscopicordeliquescentproperties[ 48].Hereweuse N.mollis163 as a model system because the concentration of its persistent brine can be directly measured, whereas the brine in164 contact with the glands cannot be directly sampled for plants like black mangroves (Avicennia germinans) whose165 leaf surfaces during the day are covered in salt but appear dry.166 The limitations and design considerations explored here likely apply even to secreting glands without visibly167 detached cuticles [20]. Without a subcuticular chamber, the processes described in Fig. 2a would be confined to the168 hydrated cell wall. The inflated chamber can be considered simply as an extension of this space and provides a169 more convenient calculation of the water potential in which matric contributions from cell wall polymers can be170 ignored [21]. In non-chambered salt glands, the cuticle is often thickest around basal cells and can isolate secreting171 cells from adjacent epidermal cells, suggesting that hydraulic separation of the subcuticular space is important even172 in these systems [16].173 Therehasbeensubstantialinterestindesigningbiomimeticdesalinationdevicesandengineeringsalttolerancein174 crops,withsignificanteffortdedicatedtounderstandingthebiologicalproblemsofregulatinggenesandengineering175 transporters for efficient salt transport [30, 49–52]. This work highlights that the physical problem of desalination176 must also be considered. Our model finds that structural solutions can circumvent energetic limitations on moving177 salt against a steep concentration gradient. Dilution of the intervening subcuticular chamber reduces the energetic178 cost of efflux and relies on the mechanical properties of the cuticle itself. Efforts to introduce similar strategies in179 other plants may benefit from a focus on structural properties instead of solely on transporter optimization.180 Similar subcuticular chambers are found in oil-secreting glands, called elaiophores, in some floral structures181 [53–56] and as glandular trichomes, such as in tomato, among others [50, 57, 58]. Development of the cavities182 in these glands begins with digestion of the pectin strata to initially detach the cuticle and cell wall, followed by183 distention of the cuticle chamber with the accumulation of secreted oils [55]. It is likely that a similar process184 applies for the development of chambered salt glands, such as those recently described in some mangroves [21].185 However, we note that elaiophores are designed for storage and mechanical disruption by either a pollinator or other186 visitor, and do not include unprovoked cuticle fracture as part of their development [59], as we propose is relevant187 for secreting salt glands. However, these structures may be potential candidates for endeavors seeking to establish188 9 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. desalination pathways in other plants.189 We developed a model of the secreting salt gland to understand the interplay between the cuticle’s fracture190 mechanics and the biochemical processes involved in foliar desalination. Thermodynamic limitations to transporter191 energy balance suggest that ion transporters cannot be infinitely active. This constrains the functional parameter192 space of the salt gland, as both direct contact with the brine and highly restrictive flow across the cuticle lead193 to steep concentration gradients across the cell membrane and subsequent deactivation of export processes. Our194 model finds that desalination is achieved when cuticle crack size remains on the order of 10–400 nm, which we195 have confirmed through cryo-SEM imaging of functional glands. We estimate the cuticle’s fractocohesive length to196 be on the order of 1µm to 1 mm, well beyond the required range to avoid runaway crack propagation and remain197 within the operable region. Our work highlights the importance of the elasticity, strength, and fractocohesive length198 of the cuticle for secreting salt gland function and raises important questions regarding how plants achieve the199 appropriate values for these material properties to create a functional salt-exporting phenotype.200

AND METHODS201 Plant material202 We useNolana mollis, a shrub native to the Pan de Azúcar region of the Atacama Desert, Chile, which has been203 previously identified as a salt-secreting plant [25, 26]. Seeds were germinated according to the protocol presented204 by Ref. [60]. To germinate, seeds were first surface sterilized using a bleach (3% NaOCl), followed by 70% ethanol,205 then incubated in 500ppm gibberellic acid at room temperature for 5 to 7 days. After incubation, seeds were plated206 in groups of 15–30 onto Petri dishes lined with dampened circular filter paper. The plates were then covered with207 their clear lids and tilted at a small angle to prevent pooling of water around the seeds. A cotton ball soaked in208 water was also placed at the lower edge of the plate to maintain humidity. Plates were then loosely enclosed in clear209 plastic greenhouse trays and kept in a growth chamber with a 12h light/12h dark light cycle, 85% RH, and 21◦C.210 Plates were monitored for three months. When a seed’s radicle reached 1–5cm long, the seedling was planted in a211 seedling cell with a mix of 60% potting mix (PRO-MIX BK, Premier Tech) and 40% coarse sand. Over the course212 of 12 months, seedlings were transferred to 4in and then 6in pots when necessary, and the relative humidity was213 reduced to 65% RH. All plants were watered exclusively with dH2O for at least one month prior to any treatments.214 10 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency Imaging of foliar surface features215 Plants were bottom watered with 200 mL 100 mM NaCl 9 and 2 days prior to imaging to induce brine secretion.216 Branchlets were detached pre-dawn, wrapped loosely in damp paper towels, and kept in a plastic bag until they217 were processed for imaging at the Center for Nanoscale Systems (CNS) at Harvard University. Branchlets were218 directly imaged using a digital light microscope (VHX-7100, Keyence) with serial recording of 3D image stacks.219 For cryo-SEM imaging, a leaf from the third to fifth whorl of the branch was extracted and rinsed with dH2Oto220 remove surface salts and gently blotted dry with lint-free tissue paper (Kimwipe). The leaf was then submerged221 and frozen in degassed liquid nitrogen and mounted in a clamped sample holder with the adaxial surface exposed.222 Using a cryo-transfer shuttle (VCT 500, Leica), the sample was mounted in a Leica EM ACE600 to etch off surface223 ice for 10 min at -95◦C. The sample was then cooled to -150◦C and sputter coated with a 9 nm layer of platinum224 and palladium. The sample was transferred in the cryo-shuttle to be imaged at 3.0 kV in a FESEM (Zeiss Gemini225 360) at -150◦C.226 Confocal imaging of salt gland ultrastructure227 Individual leaves ofN. molliswere collected and fixed in 4% w/v acrolein (Polysciences, Warrington, PA, USA) in228 a modified piperazine-N, N’-bis (2-ethanesulfonic acid) (PIPES) buffer adjusted to pH 6.8 (50 mM PIPES and229 1 mM MgSO4 from BDH, London, UK; and 5 mM EGTA) for 24 h. After rinsing them thrice in the buffer, 15230 min each, the cellulose of the cell walls was stained with 0.01% w/v calcofluor white in 10 mM CHES buffer231 with 100 mM KCl (pH = 10) [61] for 1h, washed with water 10 min, and then counterstained with auramine O in232 0.05 M Tris/HCl buffer, (pH = 7.2), that labels cutinized lipids [62]. Then, the individual leaves were cleared with233 a solution containing ethanol:benzoyl benzoate 3:1 (v/v) for 6 h, ethanol:benzoyl benzoate 1:3 (v/v) and finally234 benzoyl benzoate:dibutyl phthalate 1:1 (v/v) for several days [63]. Whole mounts of leaves were carefully mounted235 onto slides with a small cavity, and imaged with a Zeiss LSM700 Confocal Microscope connected to an AxioCam236 512 camera and Zen Blue 2.3 software, using a 63x/1.40 Oil DIC M27 Plan-Apochromat objective. 3D images237 were reconstructed from Z-stacks and assembled with the Image J1.51d software (National Institutes of Health,238 Bethesda, MD, USA).239 11 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. Leaf osmolality and water potential240 Plants were divided into six groups: one control (dH2O) and five salt treatment groups (NaCl, KCl, LiCl, CaCl2,241 MgCl2,withelectricalconductivity16.45mS).Eachgroupcontainedfiveplantsandwassplitintotwosubgroupson242 opposite sides of the growth chamber to account for potential spatial variations in the growth chamber environment.243 Each plant was placed in a plastic saucer and bottom watered at dawn of day 0, with 200mL of either dH2O244 (Control) or its assigned salt solution. The saucers were covered with aluminum foil to limit evaporation before245 absorption. After day 2, all pots were flushed top-down with dH2O until the conductivity of the flowthrough was246 similar to that of the control.247 Leaves were collected pre-dawn and placed into 0.5mL Eppendorf tubes pre-loaded with a filter paper disc248 to absorb the surface brine. These brine discs were then processed using a Wescor Vapro-5600 vapor pressure249 osmometer to determine osmolality of the surface brine.250 To determine whole-leaf osmolality, leaves were washed with dH2Oto remove surface salts, then gently blotted251 drywithlint-freetissuepaper(Kimwipe).Leavesweresnapfrozeninliquidnitrogenandlefttothawfor15minutes.252 Thawed leaves were again blotted dry to remove surface condensation, then loaded into an Eppendorf tube fitted253 with a 0.22-µm pore cellulose acetate membrane spin filter (Costar 8161, Corning). Leaves were gently crushed254 with a spatula, careful not to disturb the filter. Samples were spun at 12,000 RPM for 10 minutes to extract liquid255 (Microfuge 18 and F241.5P Rotor, Beckman Coulter). 10µL of the extracted liquid was pipetted on a filter paper256 disc to determine osmolality with a vapor pressure osmometer (Wescor Vapro-5600).257 On day 1 (24 h after treatment), additional leaves from three plants each from the control and NaCl groups258 were collected pre-dawn for whole-leaf water potential measurements. Leaves were rinsed with dH2O to remove259 surface salts and debris and blotted dry before loading 5 to 6 leaves into the chamber of a psychrometer. Samples260 were allowed to equilibrate for 16 to 24 hours in a water bath held at 25◦C. Psychrometer measurements were261 obtainedevery15minuteswith10secondsofcoolingat-6000 µampsusingaCR6datalogger(CampbellScientific).262 The psychrometer voltage measurement was determined as the mean of the output plateau during evaporation.263 An exponential fit was calculated for each voltage time series to determine the equilibrated value, which was264 then converted to the corresponding water potential using the empirical calibration curves for each individual265 psychrometer.266 12 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency Mathematical model267 Wedevelopedasteady-statemathematicalmodeltodescribethemovementofwaterandsaltthroughthechambered268 secreting salt gland. Model parameters are summarized in Table 1. The primary compartment of the model269 corresponds to the subcuticular chamber, with unknown concentrationcc, pressurePc, and water potentialψc. The270 water potentialψ = P− cRT describes the potential energy density of water. The chamber is connected to the271 gland’s secreting cell, with known water potentialψg and osmotic potentialΠg = cg RT as determined above, and272 the surface brine, held at atmospheric pressurePb = 0.1MPa with a concentrationcb, as determined above. We did273 not find significant differences in the leaf water potential or brine concentration across the various salt treatments,274 so we set these quantities at their mean observed values (Table 1).275 Aquaporin-mediated movement of water between the secreting gland cell (g) and the chamber (c) is driven by276 the water potential gradient:277 Qgc =− πr2Lgc ¯v (ψc− ψg ) (1) where r istheradiusofthecell, Lgc istheaquaporin-mediatedpermeabilitytowaterofthecellwallandmembrane,278 and ¯v is the molar volume of water.279 As the cuticle is not a selective membrane and is traversed by aqueous salt through cracks in the cuticle, we280 model the flux of water between the chamber and the brine (b) as a pressure-driven (advective) flux, where the281 Table 1.Model parameters Parameter Symbol Value Ideal gas constant R 8.314 J K−1 mol−1 Temperature T 298 K Molar volume of water ¯v 1.807× 10−5 m3 mol−1 Viscosity of water η 10−3 Pa s Free diffusion of sodium Dfree 1.6× 10−9 m2 s−1 [64] Atmospheric pressure Patm 0.1 MPa Gland water potential ψg −1.5MPa Gland radius r 15 µm Cuticle thickness lc 1 µm [23] Membrane permeability Lgc 10−12 m2 s kg−1 [65] Brine concentration cb 2000 mmol kg−1 Transport gradient threshold ∆c∗ 1400 mmol kg−1 Gland Na+ concentration cg 100 mmol kg−1 [3] Maximum salt efflux Jt 400 nmol m−2 s−1 [66] Number of pores n 1 Activity steepness h 100 13 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. brine is held at atmospheric pressure,Patm:282 Qcb =− 1 Rcb ¯v(Patm− Pc) (2) where Rcb is the resistance across the cuticle. We model the fractures as circular pores, so we describe this283 resistance through a perforated plate, following Ref. [67]:284 Rcb = 1 n ( 8ηlc πr4 f + 3η r3 f ) (3) with nthe number of pores, which for our primary analysis we take as 1,η the viscosity of water,lc the thickness of285 the cuticle, andrf the radius of the fracture.286 Through these cracks, salt flux is coupled to water transport and described by an advection-diffusion equation:287 Jcb = Qcb cc− Deff(cb− cc) (4) with the effective diffusionDeff described as the diffusion through a perforated plate [68]:288 Deff = 4rf ( 1 + nπrf 4lc ) Dfree (5) Active transporters mediate the secretion of salt from the living tissue into the chamber. While the precise289 nature of these processes is so far undetermined, we model this behavior as a function of the concentration gradient290 ∆cgc = cc− cg between the chamber (cc) and gland (cg) and a threshold gradient value,∆c∗:291 Jgc = Jt 1 + exp[−h(∆c∗− ∆cgc)] (6) with Jt the maximum transport rate andh a coefficient to describe the steepness of the activation function. Other functions for the activity profile are explored in Fig. S2. We can estimate the activity threshold for ion transport through, as an example, a Na+/H+ antiporter [36] using the Nernst potential. The energy required to move 14 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency one Na+ cation across the cell membrane is − e(∆V− VNernst Na+ ) + e(∆V− VNernst H+ ) (7) = ( ln cc cg + ∆pH ln 10 ) kBT (8) where ∆V is the membrane potential andVNernst is the Nernst potential of each ion,e is the magnitude of the292 ion charge, andkB is the Boltzmann constant. ATP provides around 19kBT of energy to operate the transporter.293 Inducingtheconformationalchangesfortransporterfunctionrequires8–16 kBT [33],limitingtheremainingenergy294 provided by ATP after the loss of some energy as heat. If the energy from Eq. 8 exceeds the remaining allotted295 energy, the transporter will not have enough energy to perform the transfer across the membrane. This required296 energy can be reduced by decreasing the pH of the chamber using separate proton pumps, which also require297 energetic input, or by reducing the concentration gradient. For our model, we estimate a concentration gradient of298 1400 mM as the activity threshold, though we find that, as long as this threshold is above the lowest achievable299 gradient across the membrane, the gland is able to function (Fig. S2).300 Sweeping acrossrf, which alters bothRcb and Deff, we use Newton’s method to solve for the steady state solution of the chamber’s concentrationcc, pressurePc, and water potentialψc, for which Qgc = Qcb (9) Jgc = Jcb (10) REFERENCES301 1. A. R. Yeo and T. J. Flowers. Varietal differences in the toxicity of sodium ions in rice leaves.Physiologia Plantarum, 59:189–195, 1983. doi:302 10.1111/j.1399-3054.1983.tb00756.x.303 2. Marilyn C. Ball and Rana Munns. Plant responses to salinity under elevated atmospheric concentrations of CO2. Australian Journal of Botany, 40(5):304 515–525, 1992. doi: 10.1071/BT9920515.305 3. Timothy J. Flowers, Rana Munns, and Timothy D. Colmer. Sodium chloride toxicity and the cellular basis of salt tolerance in halophytes.Annals of306 Botany, 115(3):419–431, 2015. doi: 10.1093/aob/mcu217.307 4. Rana Munns, Richard A James, Matthew Gilliham, Timothy J Flowers, and Timothy D Colmer. Tissue tolerance: an essential but elusive trait for308 salt-tolerant crops.Functional Plant Biology, 43:1103–1113, 2016.309 5. Zohreh Heydarian, Min Yu, Margaret Gruber, Cathy Coutu, Stephen J. Robinson, and Dwayne D. Hegedus. Changes in gene expression in Camelina310 sativa roots and vegetative tissues in response to salinity stress.Scientific Reports, 8(1):1–22, 2018. doi: 10.1038/s41598-018-28204-4.311 6. Eva Van Zelm, Yanxia Zhang, and Christa Testerink. Salt tolerance mechanisms of plants.Annual Review of Plant Biology, 71(1):403–433, 2020. doi:312 10.1146/annurev-arplant-050718.313 15 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. 7. Paul M Hasegawa, Ray A Bressan, Jian-Kang Zhu, and Hans J Bohnert. Plant Cellular and Molecular Responses to High Salinity.Annual Review of314 Plant Physiology and Plant Molecular Biology, 51:463–499, 2000.315 8. Saverio Perri, Gabriel G. Katul, and Annalisa Molini. Xylem–phloem hydraulic coupling explains multiple osmoregulatory responses to salt stress.316 New Phytologist, 224(2):644–662, 2019. doi: 10.1111/nph.16072.317 9. TimothyJ.Flowers, HanaaK.Galal,andLindell Bromham. Evolutionofhalophytes: multipleoriginsofsalttolerance inlandplants. FunctionalPlant318 Biology, 37(2):604–612, 2010. doi: 10.1071/FP09269.319 10. John M. Cheeseman. The evolution of halophytes, glycophytes and crops, and its implications for food security under saline conditions.New320 Phytologist, 206(2):557–570, 2015. doi: 10.1111/nph.13217.321 11. J. Gorham. Photosynthesis, transpiration and salt fluxes through leaves of Leptochloa fusca L. Kunth.Plant, Cell & Environment, 10(2):191–196,322 1987. doi: 10.1111/1365-3040.ep11602151.323 12. Marilyn C. Ball. Salinity tolerance in the mangroves Aegiceras corniculatum and Avicennia marina. I. Water use in relation to growth, carbon324 partitioning, and salt balance.Australian Journal of Plant Physiology, 15(3):447–464, 1988. doi: 10.1071/PP9880447.325 13. Timothy J. Flowers and Timothy D. Colmer. Salinity tolerance in halophytes.New Phytologist, 179(4):945–963, 2008. doi:326 10.1111/j.1469-8137.2008.02531.x.327 14. Mingjing Zhang, Yanlu Liu, Guoliang Han, Yi Zhang, Baoshan Wang, and Min Chen. Salt tolerance mechanisms in trees: research progress.Trees,328 35(3):717–730, 2021. doi: 10.1007/s00468-020-02060-0.329 15. P. B. Tomlinson. Water relations and salt balance. InThe botany of mangroves, pages 116–130. Cambridge University Press, 1986. ISBN330 9780521255677.331 16. Maheshi Dassanayake and John C. Larkin. Making plants break a sweat: The structure, function, and evolution of plant salt glands.Frontiers in Plant332 Science, 8(406):1–20, 2017. doi: 10.3389/fpls.2017.00406.333 17. Siegmar-W. Breckle. Salinity tolerance of different halophyte types. In N. El Bassam, M. Dambroth, and B.C. Loughman, editors,Genetic Aspects of334 Plant Mineral Nutrition, pages 167–175. Springer, Dordrecht, 1990. doi: 10.1007/978-94-009-2053-8_26.335 18. Feng Ding, Jian-Chao Yang, Fang Yuan, and Bao-Shan Wang. Progress in mechanism of salt excretion in recretohalopytes.Frontiers in Biology, 5(2):336 164–170, 2010. doi: 10.1007/s11515-010-0032-7.337 19. N Campbell and W W Thomson. The Ultrastructure of Frankenia Salt Glands.Annals of Botany, 40(4):681–686, 1976. doi:338 10.1093/oxfordjournals.aob.a085181.339 20. Fang Yuan, Bingying Leng, and Baoshan Wang. Progress in Studying Salt Secretion from the Salt Glands in Recretohalophytes: How Do Plants340 Secrete Salt?Frontiers in Plant Science, 7(977):1–12, 2016. doi: 10.3389/fpls.2016.00977.341 21. Bing-Jie Chi, Ze-Jun Guo, Ming-Yue Wei, Shi-Wei Song, You-Hui Zhong, Jing-Wen Liu, Yu-Chen Zhang, Jing Li, Chao-Qun Xu, Xue-Yi Zhu, and342 Hai-Lei Zheng. Structural, developmental and functional analyses of leaf salt glands of mangrove recretohalophyte Aegiceras corniculatum.Tree343 Physiology, 44(1):tpad123, 2023. doi: 10.1093/treephys/tpad123.344 22. Yunquan Deng, Zhongtao Feng, Fang Yuan, Jianrong Guo, Shanshan Suo, and Baoshan Wang. Identification and functional analysis of the345 autofluorescent substance in Limonium bicolor salt glands.Plant Physiology and Biochemistry, 97:20–27, 2015. doi: 10.1016/j.plaphy.2015.09.007.346 23. C. Alice Levering and William W. Thomson. The ultrastructure of the salt gland of Spartina foliosa.Planta, 97(3):183–196, 1971. doi:347 10.1007/BF00389200.348 24. Zouhaier Barhoumi, Wahbi Djebali, Chedly Abdelly, Wided Chaïbi, and Abderrazak Smaoui. Ultrastructure of Aeluropus littoralis leaf salt glands349 under NaCl stress.Protoplasma, 233(3-4):195–202, 2008. doi: 10.1007/s00709-008-0003-x.350 25. H. A. Mooney, S. L. Gulmon, J. Ehleringer, and P. W. Rundel. Atmospheric water uptake by an Atacama Desert shrub.Science, 209(4457):693–694,351 1980. doi: 10.1126/science.209.4457.693.352 26. P. W. Rundel, James R. Ehleringer, H. A. Mooney, and S. L. Gulmon. Patterns of Drought Response in Leaf-Succulent Shrubs of the Coastal Atacama353 16 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency Desert in Northern Chile.Oecologia, 46:196–200, 1980.354 27. Jess Gersony, Anju Manandhar, Uri Hochberg, Nora Abdellaoui, Paula Llanos, Jacques Dumais, N Michele Holbrook, and Fulton E Rockwell.355 Making dew in the Atacama Desert: a paradigmatic case of plant water uptake water from an unsaturated atmosphere fails a test.Annals of Botany,356 2024. doi: 10.1093/aob/mcae221.357 28. Rainer Hedrich. Ion channels in plants.Physiological Reviews, 92(4):1777–1811, 2012. doi: 10.1152/physrev.00038.2011.358 29. Bernhard Huchzermeyer and Tim Flowers. Putting halophytes to work-genetics, biochemistry and physiology.Functional Plant Biology, 40(9), 2013.359 doi: 10.1071/FPv40n9-FO.360 30. Jesper T. Pedersen and Michael Palmgren. Why do plants lack sodium pumps and would they benefit from having one?Functional Plant Biology, 44361 (5):473–479, 2017. doi: 10.1071/FP16422.362 31. Timothy J. Flowers, Edward P. Glenn, and Vadim Volkov. Could vesicular transport of Na+ and Cl− be a feature of salt tolerance in halophytes?363 Annals of Botany, 123(1):1–18, 2019. doi: 10.1093/aob/mcy164.364 32. Michael R. Blatt. A charged existence: A century of transmembrane ion transport in plants.Plant Physiology, 195(1):79–110, 2024. doi:365 10.1093/plphys/kiad630.366 33. Fabrizio Marinelli and José D. Faraldo-Gómez. Conformational free-energy landscapes of a Na+/Ca2+ exchanger explain its alternating-access367 mechanism and functional specificity.Proceedings of the National Academy of Sciences, 121(16):e2318009121, 2024. doi: 10.1073/pnas.2318009121.368 34. Alessio Accardi, Michael Walden, Wang Nguitragool, Hariharan Jayaram, Carole Williams, and Christopher Miller. Separate ion pathways in a369 Cl−/H+ exchanger. Journal of General Physiology, 126(6):563–570, 2005. doi: 10.1085/jgp.200509417.370 35. Qing Xie, Yang Zhou, and Xingyu Jiang. Structure, Function, and Regulation of the Plasma Membrane Na+/H+ Antiporter Salt Overly Sensitive 1 in371 Plants. Frontiers in Plant Science, 13(March), 2022. doi: 10.3389/fpls.2022.866265.372 36. Xiang-yun Zhang, Ling-hui Tang, Jia-wei Nie, Chun-rui Zhang, Xiaonan Han, Qi-yu Li, Li Qin, Mei-hua Wang, Xiahe Huang, Feifei Yu, Min Su,373 Yingchun Wang, Rui-ming Xu, Yan Guo, Qi Xie, and Yu-hang Chen. Structure and activation mechanism of the rice Salt Overly Sensitive 1 (SOS1)374 Na+/H+ antiporter. Nature Plants, 9(11):1924–1936, 2023. doi: 10.1038/s41477-023-01551-5.375 37. W. W. Thomson. The Structure and Function of Salt Glands. InPlants in Saline Environments, pages 118–146. Springer-Verlag, Berlin, 1975. doi:376 10.1007/978-3-642-80929-3_9.377 38. Masahiro Koyama and Takao Oi. Morphology and excreting-function of microhairs in salt-tolerant Zoysia japonica, comparing adaxial and abaxial378 leaf surfaces.Flora, 312:152472, 2024. doi: 10.1016/j.flora.2024.152472.379 39. Richard G Allen, Luis S Pereira, Dirk Raes, Martin Smith, and Others. Crop evapotranspiration-Guidelines for computing crop water380 requirements-FAO Irrigation and drainage paper 56.FAO, 300(9):D05109, 1998.381 40. Anneline H. Christensen and Kaare H. Jensen. Viscous flow in a slit between two elastic plates.Physical Review Fluids, 5(4):1–10, 2020. doi:382 10.1103/PhysRevFluids.5.044101.383 41. Jean François Louf, Jan Knoblauch, and Kaare H. Jensen. Bending and stretching of soft pores enable passive control of fluid flows.Physical Review384 Letters, 125(9):98101, 2020. doi: 10.1103/PhysRevLett.125.098101.385 42. Hendrik Bargel and Christoph Neinhuis. Tomato (Lycopersicon esculentum Mill.) fruit growth and ripening as related to the biomechanical properties386 of fruit skin and isolated cuticle.Journal of Experimental Botany, 56(413):1049–1060, 2005. doi: 10.1093/jxb/eri098.387 43. Eva Dominguez, José Alejandro Heredia-Guerrero, and Antonio Heredia. The biophysical design of plant cuticles: An overview.New Phytologist,388 189(4):938–949, 2011. doi: 10.1111/j.1469-8137.2010.03553.x.389 44. Chao Chen, Zhengjin Wang, and Zhigang Suo. Flaw sensitivity of highly stretchable materials.Extreme Mechanics Letters, 10:50–57, 2017. doi:390 10.1016/j.eml.2016.10.002.391 45. Jie Ma, Xizhe Zhang, Daochen Yin, Yijie Cai, Zihang Shen, Zhi Sheng, Jiabao Bai, Shaoxing Qu, Shuze Zhu, and Zheng Jia. Designing Ultratough392 Single-Network Hydrogels with Centimeter-Scale Fractocohesive Lengths via Inelastic Crack Blunting.Advanced Materials, 36(23):1–11, 2024. doi:393 17 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Mai et al. 10.1002/adma.202311795.394 46. Antonio J. Matas, Eward D. Cobb, Dominick J. Paolillo, and Karl J. Niklas. Crack resistance in cherry tomato fruit correlates with cuticular395 membrane thickness.HortScience, 39(6):1354–1358, 2004.396 47. Laura Espańa, José A. Heredia-Guerrero, Patricia Segado, José J. Benítez, Antonio Heredia, and Eva Domínguez. Biomechanical properties of the397 tomato (Solanum lycopersicum) fruit cuticle during development are modulated by changes in the relative amounts of its components.New398 Phytologist, 202(3):790–802, 2014. doi: 10.1111/nph.12727.399 48. HuanhuanZhang,WenjunGu,YongJieLi,andMingjinTang. Hygroscopicpropertiesofsodiumandpotassiumsaltsasrelatedtosalinemineraldusts400 and sea salt aerosols.Journal of Environmental Sciences, 95:65–72, 2020. doi: 10.1016/j.jes.2020.03.046.401 49. Timothy John Flowers, Aurora Garcia, Mikiko Koyama, and Anthony Richard Yeo. Breeding for salt tolerance in crop plants - the role of molecular402 biology. Acta Physiologiae Plantarum, 19(4):427–433, 1997.403 50. Joris J. Glas, Bernardus C.J. Schimmel, Juan M. Alba, Rocío Escobar-Bravo, Robert C. Schuurink, and Merijn R. Kant. Plant glandular trichomes as404 targets for breeding or engineering of resistance to herbivores.International Journal of Molecular Sciences, 13(12):17077–17103, 2012. doi:405 10.3390/ijms131217077.406 51. Lukasz Kotula, Pedro Garcia Caparros, Christian Zörb, Timothy D. Colmer, and Timothy J. Flowers. Improving crop salt tolerance using transgenic407 approaches: An update and physiological analysis.Plant Cell and Environment, 43(12):2932–2956, 2020. doi: 10.1111/pce.13865.408 52. Maria Di Vincenzo, Alberto Tiraferri, Valentina Elena Musteata, Stefan Chisca, Rachid Sougrat, Li Bo Huang, Suzana P. Nunes, and Mihail Barboiu.409 Biomimetic artificial water channel membranes for enhanced desalination.Nature Nanotechnology, 16(2):190–196, 2021. doi:410 10.1038/s41565-020-00796-x.411 53. Beryl B Simpson, John L Neff, and David S Seigler. Floral Biology and Floral Rewards of Lysimachia (Primulaceae).The American Midland412 Naturalist, 110(2):249–256, 1983.413 54. Malgorzata Stpiczyńska and Kevin L. Davies. Elaiophore structure and oil secretion in flowers of Oncidium trulliferum Lindl. and Ornithophora414 radicans (Rchb.f.) Garay Pabst (Oncidiinae: Orchidaceae).Annals of Botany, 101(3):375–384, 2008. doi: 10.1093/aob/mcm297.415 55. Clivia Carolina Fiorilo Possobom, Elza Guimarães, and Silvia Rodrigues Machado. Structure and secretion mechanisms of floral glands in416 Diplopterys pubipetala (Malpighiaceae), a neotropical species.Flora, 211:26–39, 2015.417 56. Clivia Carolina Fiorilo Possobom and Silvia Rodrigues Machado. Elaiophores: Their taxonomic distribution, morphology and functions.Acta418 Botanica Brasilica, 31(3):503–524, 2017. doi: 10.1590/0102-33062017abb0088.419 57. Ryan W. Shepherd and George J. Wagner. Phylloplane proteins: emerging defenses at the aerial frontline?Trends in Plant Science, 12(2):51–56, 2007.420 doi: 10.1016/j.tplants.2006.12.003.421 58. Robert Schuurink and Alain Tissier. Glandular trichomes: micro-organs with model status?New Phytologist, 225(6):2251–2266, 2020. doi:422 10.1111/nph.16283.423 59. A.N Sérsic. Pollination Biology in the Genus Calceolaria L.(Calceolariaceae).Stapfia, 82:1–121, 2004.424 60. Liesbeth van den Brink, Melissa H Mai, Lorenz Henneberg, Margret Ecke, and Rafaella Canessa. Gibberellin enhances germination of nolana mollis425 and heliotropium pycnophyllum seeds.bioRxiv, 2025. doi: 10.1101/2025.01.10.632421.426 61. J. Hughes and Margaret E. McCully. The Use of an Optical Brightener in the Study of Plant Structure.Stain Technology, 50(5):319–329, 1975. doi:427 10.3109/10520297509117082.428 62. Yolande Heslop-Harrison and K. R. Shivanna. The receptive surface of the angiosperm stigma.Annals of Botany, 41(6):1233–1258, 11 1977. ISSN429 0305-7364. doi: 10.1093/oxfordjournals.aob.a085414.430 63. Charles F. Crane and John G. Carman. Mechanisms of Apomixis in Elymus rectisetus from Eastern Australia and New Zealand.American Journal of431 Botany, 74(4):477–496, 1987. doi: 10.2307/2443827.432 64. E. A. Guggenheim. The diffusion coefficient of sodium chloride.Transactions of the Faraday Society, 50:1048–1051, 1954. doi:433 18 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint Secreting salt glands leverage cuticle structural properties to regulate desalination efficiency 10.1039/tf9545001048.434 65. Paul J. Kramer and John S. Boyer.Water Relations of Plants and Soils. Academic Press, Inc., New York, 1 edition, 1995. ISBN 9780124250604. doi:435 10.1097/00010694-199604000-00007.436 66. John W. Oross, Robert T. Leonard, and William W. Thomson. Flux rate and a secretion model for salt glands of grasses.Israel Journal of Botany, 34437 (2-4):69–77, 1985. doi: 10.1080/0021213X.1985.10677012.438 67. Kaare Hartvig Jensen, Daniel Leroy Mullendore, Noel Michele Holbrook, Tomas Bohr, Michael Knoblauch, and Henrik Bruus. Modeling the439 hydrodynamics of phloem sieve plates.Frontiers in Plant Science, 3(3):1–11, 2012. doi: 10.3389/fpls.2012.00151.440 68. William F Pickard. Lumped Circuit Approximations for Flux in Systems Governed by the Laplace Equation.Chemical Engineering Science, 36:441 191–197, 1980.442 ACKNOWLEDGMENTS443 This work was supported by the National Science Foundation through the Harvard University Materials Research444 Science and Engineering Center (DMR-2011754). We acknowledge Adam Graham from the Harvard University445 CenterforNanoscaleSystems(CNS)forhissupportonthecryo-SEMandlightmicrosopy.MHMrecognizessupport446 from the Fannie and John Hertz Foundation Fellowship and the National Science Foundation Graduate Research447 Fellowship (Grant no. DGE1745303) and thanks Sophie Everbach, Liesbeth van den Brink, Tomás Fuenzalida,448 and Jacques Dumais for helpful discussions, and Ayman Fayad and Cory Hahn for greenhouse support. JML was449 funded by a grant from the Agencia Estatal de Investigación (PID2021-129074OB-100) and a Fulbright-CSIC450 fellowship (FULBR23036).451 19 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 28, 2025. ; https://doi.org/10.1101/2025.02.27.640653doi: bioRxiv preprint