2.1 Study site and species

Sun-exposed terminal branches <1 m were collected from mature trees of A. corniculatum and R. stylosa from similar canopy positions <3 m in height from the ground and growing naturally along the banks of the Daintree River, Daintree National Park, Far North Queensland, Australia (16.1700°S, 145.4185°E). PV curves were constructed for the youngest, healthiest, fully expanded leaf from each branch. Three estuarine sites (designated Upper, Middle and Lower, Figure 1a) were selected along the river banks where the geographical position in the estuary and the mangrove vegetation were indicative of different ranges of salinity, driven largely by differences in tidal influence combined with seasonal variation in river discharge. The Upper estuarine site (typically low salinity, 2−9 ppt) was located near the upstream limits of mangrove distribution (16.257421°S, 145.343296°E), the Middle estuarine site (typically mid salinity, 10−24 ppt) was located near the Daintree ferry crossing (16.260275°S, 145.393811°E) and the Lower estuarine site (typically high salinity, 25−26 ppt) was located near the river mouth (16.283785°S, 145.451308°E).

The study was limited to trees on the river bank margins of a tidal system (tidal range approximately 3 m) for three reasons. First, safety restrictions for working in a saltwater crocodile (Crocodylus porosus) habitat necessitated rapid sampling of mangroves that could be accessed by boat at high tide. Second, the conditions along the river bank are more similar to each other than to sites even a few metres inland. Finally, along river margins where sediment is loosely consolidated (accreting banks) or where the river banks are well drained and reflooded with successive tides (eroding banks, Figure 1b), the estuarine surface water salinities are likely to be representative of those used by root systems which were partially exposed to river water, consistent with split root studies of mangroves and other halophytes (Bazihizina et al., 2009, 2012; Reef et al., 2015).

Measurements of mid-dry season leaf water relations in A. corniculatum and R. stylosa were made in August 2019 after a severe dry-season heatwave and low rainfall event (late 2018, please see ‘Climate’ for further details) and its relief in the following wet season. To take advantage of this event, we compared the data collected in 2019 with prior mid-dry season leaf water relations data collected during 2016 and 2017 for R. stylosa, and 2018 for A. corniculatum. These prior measurements captured multiple years of typical estuarine saline conditions and were thus considered baseline measurements for before/after analysis of the effects of the severe drought.

Leaf samples of R. stylosa were collected from three trees (n = 3) from each of the Lower and Middle estuarine sites in October 2016 and the Upper estuarine site in late July of 2017, and samples of A. corniculatum (n = 5 trees) were collected from all three estuarine sites in July of 2018 early in the drought (Supporting Information: Table S1), hereafter called the pre-drought condition. A. corniculatum (n = 5) and R. stylosa (n = 5 except Upper estuarine site where n = 2) were then resampled at all sites in August of 2019 hereafter called the post-drought condition (Supporting Information: Table S1). Different individuals from the same cohorts of trees at the same three estuarine sites were sampled for each collection period.

Estuarine surface water samples were collected at each sampled tree and salinity was measured with a hand-held refractometer (A. S. T. Co. Ltd). ψ_salinity (Table 2) was calculated as a fraction of seawater, where standard seawater has a salinity of 35 ppt and a ψ of −2.4 MPa. Values for ψ_salinity in the 2016−2017 and 2018 collection periods were reflective of the average range of conditions in this ecosystem. Values for ψ_salinity collected in 2018 were more negative than those collected in 2016−2017 as these measurements were made at the onset of the 2018 dry-season drought.

2.2 Climate

The 2018 drought was characterised by higher minimum and maximum temperatures (Supporting Information: Figure S1), greater solar exposure (Supporting Information: Figure S2) and reduced rainfall (Supporting Information: Figure S3) relative to the long-term mean (data from: http://www.bom.gov.au/climate accessed 16/02/2021). As the drought developed, August through October experienced higher average maximum daily temperatures than the previous 30-year average (p < 0.01); similarly, July, August and November experienced significantly higher average minimum daily temperatures than the 30-year average (p < 0.001) (Supporting Information: Figure S1). The Australian Bureau of Meteorology reported a record air temperature of 42°C in Cairns, 105 km from the field site (data from: http://www.bom.gov.au/climate accessed 17/08/2021), while Daintree River Crossing recorded 45.6°C on 27 November 2018 (Figure 1c).

In addition to the higher temperatures, and record-breaking heatwave, little rain was received from June through October 2018. Both during this period (p < 0.0001), and in November (p < 0.001) rainfall was lower than the previous 30-year average (Supporting Information: Figure S3). Consequently, water discharge (mL/day) measured at Bairds Crossing, a water monitoring site on the Daintree River, was smaller from August to November in 2018 than the 30-year average since 1988 (p < 0.0001) and reached a low of 105 mL/day in November. Mean river height was lower (~7 cm) in 2018 from August through November than the 30-year average river height since 1990 (p < 0.0001) (data from: https://water-monitoring.information.qld.gov.au accessed 24/02/21, comparisons made using t-tests).

Extremely high temperatures and drier atmospheres imposed greater evaporative demand, which would have increased water use from trees, reduced the water level of the Daintree River and concentrated estuarine salinity. Reduced water discharge due to low rainfall would have allowed tidal seawater intrusion farther upstream further increasing estuarine salinity at all three sites. Thus, drought here refers to the combined effects of atmospheric drought and low rainfall that increased estuarine salinity, inducing severe limitations on water available for hydration at all field sites as the dry-season drought progressed.

A wet season relieved the 2018 drought. In January 2019, total rainfall was 1007 mm (data from: http://www.bom.gov.au/climate accessed 16/02/2021). This wet season had a prolonged effect on the river system as the volume of freshwater discharged from the watershed delayed the upstream penetration of a tidal salt wedge. This was shown by the persistence of fresh water at the Upper and Middle estuarine sites in the mid-dry season of 2019, even though the average salinity had recovered at the river mouth (Lower estuarine site, Table 2).

2.3 Leaf water relations

One sun-exposed terminal branch <1 m from each of 2−5 trees was collected at midday from the Daintree River from Upper, Middle and Lower estuarine sites and transported to the field lab in plastic bags to prevent moisture loss. The youngest, fully expanded, healthy leaf from each branch was cut at the petiole under perfusion solution (1% seawater for R. stylosa and 5% seawater for A. corniculatum) and rehydrated overnight with the petiole immersed in perfusion solution in a beaker. Each beaker was covered by a wet paper towel and plastic wrap to increase humidity and prevent transpiration. The composition of the perfusion solution was based on analyses of the ionic composition of xylem sap in A. corniculatum and R. stylosa grown under lab (Ball, 1988b) and field conditions (Scholander et al., 1966; Stuart et al., 2007). A PV curve was constructed and analysed for each of the leaves as described by (Nguyen, Meir, Wolfe, et al., 2017). Upon completion of the measurements, leaf area was measured using the ‘LeafScan' app (Anderson & Rosas-Anderson, 2017) and the leaves were oven dried at 70°C for 24 h and to constant mass before determining the leaf dry mass (DM).

The π_FT and ψ_TLP were determined by standard methods (Tyree & Hammel, 1972) as shown in Supporting Information: Figure S4. Data for leaf physical traits (Supporting Information: Table S2) were determined as described by Nguyen, Meir, Wolfe, et al. (2017) and combined with PV parameters determined from the PV relationships to calculate three additional PV parameters, ε, C and water storage (WS). Bulk modulus of elasticity was calculated as per Nguyen, Meir, Sack, et al. (2017). Capacitance was calculated according to Tyree and Hammel (1972), estimated as the linear slope of an appropriate region in the PV curve above the TLP (Supporting Information: Figure S5), normalised by leaf WC_A (Brodribb & Holbrook, 2003; Koide et al., 2000). WS was here estimated as the total amount of water released with drying from a fully hydrated state to the TLP (Nguyen, Meir, Sack, et al., 2017). Average PV curves for each species, pre- and post-drought from each estuarine site can be found in Supporting Information: Figure S6. Converting salinities to ψ (Table 2) enabled positioning of environmental conditions on respective PV curves to estimate, for example, the extent of WS that could be achieved through root water uptake alone, calculated as per Nguyen, Meir, Sack, et al. (2017); Nguyen, Meir, Wolfe, et al. (2017).

2.4 Statistical analysis

In this design there were two species by three estuarine sites by two conditions (i.e., pre-drought and post-drought) with varying numbers of replicate trees. This design contrasts data collected during the mid-dry season in the succession of years when a gradual salinity gradient was present (Table 2) with data collected after a severe drought had been relieved by the subsequent wet season.

Statistical analyses were performed using the R statistical software package (R Core Team, 2020; version 4.0.2). The basic model for all dependent variables was a linear mixed model using the lmer() function in the lmerTest package (Kuznetsova et al., 2017) with condition, species, estuarine site and all two-way interactions as fixed effects, and a variable indicating the unique combinations of estuarine site, condition and species as a random intercept. For some measurements, within-unit variation was so low that the linear mixed model failed to converge; to improve model convergence, we added some random noise to these measurements. In comparison to a fixed-effects linear model, this approach is a more conservative analysis that better identifies estuarine site effects by accounting for the repeated measures made at estuarine sites, leading to larger standard errors (SE).

The anova() function was used to assess the significance of main effects and interactions at the p < 0.05 level. Model fit was assessed through a number of diagnostic tests including the Shapiro−Wilk test for normality of residuals. Where necessary dependent variables were log-transformed to improve model fit. Model estimated marginal means, trendlines and SE for figures were produced post hoc using the emmeans R package (Lenth, 2020), with the Tukey method used for p-value adjustment. Figures were produced using the package ggplot2 (Wickham, 2016). Observed means ± SE and ANOVA tables are given in supporting information. All data are publicly accessible [(data set) Beckett et al., 2021].

3.1 Physical properties of leaves

R. stylosa leaves were larger in area, mass, leaf mass per area and water content than A. corniculatum leaves (p < 0.04, Figure 2). Yet, with the exception of leaf area which for both species was smaller post-drought (p = 0.016), neither species' physical properties differed in response to drought. Leaf area in both species differed with estuarine site (p < 0.0001), with the smallest leaf area occurring at the Lower estuarine site where salinity was highest both pre- and post-drought. Saturated leaf WC_DM in both A. corniculatum and R. stylosa responded differently to drought depending on the estuarine site of leaf growth (p = 0.004, Figure 2). At the Upper estuarine site, WC_DM was greater in both species post-drought than pre-drought, while the opposite trend occurred at the Lower estuarine site.

3.2 Turgor loss point and its components

There were no significant differences in PV parameters between A. corniculatum and R. stylosa within either the pre-drought or post-drought samples (Figures 3 and 5). Both A. corniculatum and R. stylosa showed strong shifts in all PV parameters between pre- and post-drought (p < 0.05), except for RWC_TLP where there were no significant drought effects (Figure 3). Post-drought values for ψ_TLP and π_FT were more negative than pre-drought (p < 0.0001), while ε increased between pre- and post-drought (p = 0.047). At the Upper estuarine site, ψ_TLP pre- and post-drought was less negative than in Middle or Lower estuarine sites (p = 0.0002); however, the effect of estuarine site on ψ_TLP differed pre- and post-drought (p = 0.01, Figure 3a). Pre-drought ψ_TLP became increasingly negative across Upper, Middle and Lower estuarine sites, while post-drought ψ_TLP was most negative at the Middle estuarine site. Similar trends occurred in π_FT, which was less negative at the Upper estuarine site relative to the Lower estuarine site pre- and post-drought (p = 0.04). Values for ε correlated significantly with ψ_TLP (Figure 4) such that RWC_TLP averaged 90.1% and 89.9% for A. corniculatum and R. stylosa, respectively (Figure 3d).

3.3 WS and capacitance

Leaf C was significantly lower post-drought than pre-drought (p = 0.03, Figure 5). However, there were no significant species or site effects for leaf C. Similarly, despite mean values of leaf WS decreasing between pre- and post-drought, there were no significant species, site or drought effects for leaf WS (Figure 5). Both leaf C and WS tended to increase across Upper, Middle and Lower estuarine sites in R. stylosa pre- and post-drought, with no strong site trend present in A. corniculatum.

3.4 The TSM

Figure 6 shows the maximum range of leaf ψ over which turgor could be maintained from saturation (0 MPa) to ψ_TLP (indicated in red). However, due to the requirement to maintain favourable ψ gradients for uptake of water through roots, the ψ of the estuarine water in which the roots reside (ψ_salinity) sets the upper limit to leaf hydration achievable through roots. The difference between ψ_TLP and the ψ_salinity (indicated in yellow) gives a measure of the TSM, the range of ψ over which turgor can be maintained with access to root water alone. The TSM (shown in the dark green) pre- and post-drought was smallest at the Lower estuarine sites compared to Upper estuarine sites (p = 0.04, Figure 6), despite more negative values for ψ_TLP at Lower than Upper estuarine sites. In other words, TSM were smaller at high than low salinity both pre- and post-drought because TLPs varied less than salinity. Therefore, the hydration deficit—the range of ψ that could not be filled by root water in the absence of a lower salinity water source—increased between low and high salinity. However, the lower TLPs post-drought than pre-drought enabled greater TSM at all sites following relief from the drought (p = 0.01, Figure 6).

3.5 The salinity-induced WS deficit

Just as ψ_salinity sets the upper limit of turgor pressure in the absence of alternative sources of low salinity water, it also sets the upper limit on WS that can be filled through the roots. Accordingly, the WS attainable through the roots as set by ψ_salinity (dark green) was considered in relation to the total WS (i.e., WS at saturation, see Figure 7). Although there was no significant difference in total WS between the species, the attainable WS was larger in R. stylosa than in A. corniculatum (p = 0.03), with a trend in both species towards attainable WS decreasing between Upper and Lower estuarine sites (p = 0.05). Pre-drought, there was a WS deficit at all sites as attainable WS was less than total WS, with a greater WS deficit in leaves grown at the Lower compared to the Upper estuarine site both pre- and post-drought (Figure 7). In contrast, there was no deficit between attainable WS and total WS in leaves from Upper and Middle estuarine sites post-drought where saline conditions had not yet re-established after the wet season. Hence, plants could achieve total WS through the uptake of estuarine water alone. The results obtained pre- and post-drought imply a greater requirement for alternative low-salinity water sources to achieve total WS in high than low-salinity environments (p = 0.04) as evidenced by the difference between attainable and total WS.

4.1 Osmotic adjustment in response to drought was similar in species distributed over a broad salinity gradient

Physical properties revealed substantial physical differences between the species, where R. stylosa had larger leaf DM, area, LMA and WC than A. corniculatum leaves (Figure 2). However, while A. corniculatum and R. stylosa may have differed in the scale of their responses pre- and post-drought, the species did not differ in their adjustment of key water relations characteristics in response to drought, including in ψ_TLP, π_FT, ε, RWC_TLP, WS and C (Figures 3 and 5), as hypothesised. Furthermore, despite different collection periods, there were no significant differences in ψ_TLP, π_FT, ε and RWC_TLP between the species pre-drought.

In a global meta-analysis, Bartlett et al. (2012) found that ψ_TLP was strongly associated with water availability and drought tolerance. Further, Nguyen, Meir, Sack, et al. (2017) identified soil water salinity as the key variable driving ψ_TLP adjustment in mangroves, such that leaf ψ_TLP varies with salinity (Bryant, Fuenzalida, Brothers, et al., 2021; Paliyavuth et al., 2004; Suárez & Sobrado, 2000; Suárez et al., 1998) to maintain a water balance favourable for maintenance of turgor and water uptake. Here, we suggest that species made similar adjustments in ψ_TLP and its components due to the requirement to maintain turgor and favourable gradients for water uptake under increasing salinity, both along a salinity gradient and in response to severe water stress. Indeed, although drought responses differed from those in the present study, Fajardo and Piper (2021) found that all angiosperm tree species studied responded similarly to drought such that there was functional coordination in drought acclimation. Thus, although A. corniculatum and R. stylosa differ in salinity tolerance and salt management, their water relations characteristics revealed no fundamental differences in the nature of acclimation to salinity variation and related drought drivers. As such, species are discussed collectively below.

4.2 Coordinated adjustment in ψ TLP and ε maintained turgor and water content with increasing water stress

The maintenance of turgor and cell volume is essential for growth, cellular and intracellular communication, maintenance of cellular function and sugar transport (Cabon et al., 2020; Fricke, 2017; Peters et al., 2021; Potkay et al., 2022; Steppe et al., 2006). Additionally, loss of turgor and cell volume can impact structural integrity, cellular function and carbon gain, and eventually lead to cell death (Guadagno et al., 2017; Lamacque et al., 2020; Lambers & Oliveira, 2019; McDowell et al., 2022; Nguyen, Meir, Wolfe, et al., 2017; Sapes & Sala, 2021). Consequently, the TLP is recognised as an important indicator of both soil water availability and drought tolerance (Bartlett et al., 2012). Given this, the plasticity to adjust ψ_TLP to more negative values with dynamic variation in salinity over time and space would contribute to defining the salinity range over which a plant can maintain hydraulic functions, gas exchange and growth. Both mangrove species adjusted ψ_TLP sufficiently to enable the maintenance of water uptake by the roots (Figure 3). Indeed, adjustments in TLP in the present study were consistent with the prediction of ψ_TLP with adjustment in π_FT (Supporting Information: Figure S6) as shown in the Bartlett et al. (2012) meta-analysis of a wide range of species across biomes and environments.

Through adjustments in ψ_TLP, π_FT and ε (Figure 3), both A. corniculatum and R. stylosa maintained RWC_TLP at ~90% across the range of salinities and environmental conditions (Figure 3). This highlights the eco-physiological importance of maintaining both volume and a high hydration state. It has been noted in previous studies that RWC below 75% inhibits protein, ATP and RuBP production (Lawlor & Cornic, 2002). Indeed, the Bartlett et al. (2012) meta-analysis revealed no species with a RWC_TLP below 75%. Furthermore, drought-induced mortality increased as RWC declined below TLP (Sapes & Sala, 2021). Thus, maintenance of RWC_TLP at 90% during drought could provide a buffer from cellular damage incurred at lower RWC (Martinez-Vilalta et al., 2019).

The constant RWC_TLP across sites pre- and post-drought (Figure 3), was largely a consequence of proportional adjustments in ψ_TLP and ε (Figure 4) consistent with the cell water conservation hypothesis (Bartlett et al., 2012; Cheung et al., 1975). Under this hypothesis, more rigid cells will show a greater change in ψ for a given change in RWC, allowing for the maintenance of cellular water content at more negative values of ψ (Cheung et al., 1975). If leaves had decreased only π_FT and ψ_TLP then the RWC_TLP would have lowered with potential negative impacts on cellular function, geometry (Bartlett et al., 2012) and survival (Sapes & Sala, 2021). Thus, leaf plasticity enabling adjustments in water relations minimised changes in cellular hydration in an environment of strongly varying soil water availability.

4.3 Adjustments in turgor loss points with increasing salinity enabled maintenance of water uptake but were not sufficient to prevent a decline in TSM

The TSM gives the range of ψ over which leaves can function with turgor if the only source of water was that absorbed by roots. A greater leaf TSM would enable longer periods of carbon gain under greater salinities. Under extreme conditions when stomata may be closed, uptake of soil water would be sufficient to maintain a minimal level of turgor unless there was a substantial decrease in ψ_salinity, as turgor loss points were always more negative than ψ_salinity (Figure 3). However, the decrease in ψ_TLP was not proportionate to the decrease in ψ_salinity between Upper and Lower estuarine sites. As a result, the TSM tended to be smallest at the Lower estuarine site where salinity was highest. This was consistent with findings for A. marina grown along gradients in salinity and aridity (Nguyen, Meir, Sack, et al., 2017).

Similarly, total leaf WS capacity tended to be greatest in leaves grown at high salinity at the Lower estuarine site (Figure 5). However, due to the more negative ψ at highly saline sites, leaves grown under these conditions would realise the smallest fraction of total leaf WS capacity based on root water uptake alone. If the water supply from the roots becomes restricted due to increasing salinity, then the availability of stored water may become increasingly important for carbon gain and maintenance of cellular functions (Scholz et al., 2011). Indeed, total WS alone was sufficient to support substantial transpiration rates in A. marina for up to 2 h before turgor loss (Nguyen, Meir, Wolfe, et al., 2017) and contributes to the maintenance of stem hydraulic function during the late dry season in A. marina (Coopman et al., 2021) and Sonneratia alba (Bryant, Fuenzalida, Brothers, et al., 2021).

Limitations of decreasing TSM and increasing WS deficits might be offset by foliar water uptake as suggested by Nguyen, Meir, Sack, et al. (2017). Indeed, widespread capacity for foliar water uptake occurs in mangroves (Bryant, Fuenzalida, Zavafer, et al., 2021; Hayes et al., 2020) including A. corniculatum (Schaepdryver et al., 2022), enabling recovery of leaf hydraulic conductance (Fuenzalida et al., 2019), embolism repair (Fuenzalida et al., 2022), protection of stem hydraulic function (Coopman et al., 2021) and long-term benefits to turgor-dependent growth (Schreel et al., 2019; Steppe et al., 2018). However, greater reliance on atmospheric water may result in greater vulnerability to drought conditions that lead to increased estuarine salinity, while drier atmospheric conditions would also reduce the occurrence of foliar uptake of atmospheric water. Future studies are needed to address the relative contribution of foliar water uptake to plant hydration.

There may be costs associated with maintaining a large TSM driven by a low ψ_TLP year-round. These include osmotic adjustment of cytoplasmic compartments (Flowers & Colmer, 2008) and wall reinforcement to increase rigidity, both of which reduce carbon that might otherwise be available for growth. These carbon costs are small at an individual leaf level but could be significant at whole canopy scales. However, a greater cost linked to increasing ε may be reductions in mesophyll conductance, incurring a long-term cost to assimilation rates. For example, studies have shown a trade-off between cell wall properties and mesophyll conductance, whereby decreasing diffusivity across thicker or less permeable walls could impede carbon gain (Carriquí et al., 2020; Niinemets et al., 2009; Onoda et al., 2017). This suggests that a high ε is associated with increased diffusive resistance to CO₂ (Nadal et al., 2018). However, the benefits of increasing ε together with decreasing ψ_TLP to enable maintenance of both turgor and the RWC_TLP at greater salinities may outweigh the potential cost of reduced maximum assimilation rates.

4.4 Severe but nonlethal dry season conditions lead to greater leaf salinity tolerance in the subsequent dry season, consistent with ESM

When mid-dry season samples were collected post-drought, freshwater conditions occurred at the Upper and Middle estuarine sites (Table 2) likely due to increased freshwater discharge from the watershed following the wet season. Despite leaf growth mainly occurring over the much lower ranges of salinities during the wet season (Saenger & Moverley, 1985), post-drought measurements of both π_FT and ψ_TLP were lower while ε was greater at all three sites in both A. corniculatum and R. stylosa relative to the pre-drought measurements (Figure 3). We know of no evidence in the literature of a lowering of ψ_TLP together with an increase in ε in response to decreasing salinity. Rather, the leaves collected post-drought possessed water relations characteristics consistent with osmotic adjustment to increase drought and salinity tolerance (e.g., Bartlett et al., 2012; Bryant, Fuenzalida, Brothers, et al., 2021; Naidoo et al., 2002; Nguyen, Meir, Sack, et al., 2017; Paliyavuth et al., 2004; Suarez & Medina, 2008) and thus we hypothesise that these shifts in ψ_TLP and ε occurred in response to the severe 2018 dry season drought.

Post-drought results imply acclimation to higher salinity may have been initiated either in meristematic tissue or during leaf primordium development in response to the 2018 drought. There is precedent in the physiological literature for priming of salinity tolerance through ESM of prior exposure to heat, drought and salinity stresses (Caparrotta et al., 2018; Gong et al., 2001), where stress may induce epigenetic responses involving modification of DNA (Bruce et al., 2007; Hilker & Schmülling, 2019; Sharma et al., 2022). For example, exposure to environmental stress during early development can induce lasting enhancement of stress tolerance in the tissue or plant, as demonstrated by epigenetic determination of traits characterising provenances of Picea abies (Johnsen et al., 1996, 2005; Kvaalen & Johnsen, 2008). Indeed, mangroves exhibit a correlation between epigenetic methylation and morphological variation along salinity gradients (Lira-Medeiros et al., 2010; Miryeganeh et al., 2021). Thus, the influence of the preceding drought on the development of leaf traits, which resulted in enhanced salinity tolerance in the subsequent dry season as hypothesised, was consistent with ESM.

In the long-lived leaves (15−21 months) of A. corniculatum and R. stylosa (Saenger & West, 2016), ESM of the higher salinities of the previous dry season would enable greater leaf function with minimal osmotic and structural adjustments as conditions progressed from wet through dry seasons. Indeed, Galiano et al. (2017) suggested that long-term acclimation to the legacy of drought reflected the prioritisation of long-term survival over improved performance in the short-term following drought relief. For example, a decrease in leaf area and adjustment of vessel anatomical traits reduced water use during drought, reduced plant daily water loss and increased resistance to subsequent drought (Tomasella et al., 2019), and may explain the significant reduction in individual leaf area post-drought in both species measured here. Furthermore, ESM may explain observations of little difference between wet and dry season water relations parameters in coexisting mangroves Avicennia germinans and Laguncularia racemosa (Sobrado & Ewe, 2006). Notably, if acclimation occurs primarily in response to the previous dry season, then plants experiencing a prolonged period of cool, moist conditions (e.g., La Niña) would be more vulnerable to drought with a rapid switch to hot dry conditions (e.g., El Niño).

4.5 Limitations of the study

Our results do not distinguish between water stress from atmospheric drought, extreme temperature and associated stress from increased salinity, all of which would produce similar effects on leaf water relations. Furthermore, due to the remote nature of the field site, detailed environmental data are limited. We thus cannot distinguish a particular driver and so have discussed the results in terms of salinity tolerance, recognising that changes in salinity do not occur in isolation from other factors that affect it and the response of plants to salinity.

Despite different years of collection, the lack of significant differences in PV parameters of leaves among samples collected pre-drought indicates the suitability of our pre-drought samples as a baseline for comparison with samples collected over time, which included a drought superimposed upon year-to-year variability. Our field-based study thus captured responses to complex conditions as indicated by statistically significant differences in water relations parameters measured pre- and post-drought, which did not differ significantly between species. However, this study is limited in both the time span and scale of measurements. Consequently, ESM of drought that influenced salinity tolerance of future leaf growth remains a hypothesis, strongly supported by our data. Due to the implications for mangrove acclimation and survival under future extreme weather and the design of future work, the results presented here urgently warrant further investigation into this phenomenon in mangroves.