Stem functional traits vary among co-occurring tree species and forest vulnerability to drought

Keywords: bark, die-off, eucalyptus, heartwood, heatwave, jarrah, marri, sapwood.

Historical climate is a key factor in driving the evolution of functional traits in trees, and climate change in some regions is expected to progress at rates that exceed the ability of some species to adapt (Hoffmann and Sgrò 2011). Regions projected to experience the combination of increasing temperatures and decreasing precipitation may be especially at risk for changes in forest species composition (Allen et al. 2010), since excessive temperatures compound the effects of drought and shorten the time to tree die-off and mortality (Adams et al. 2009; Williams et al. 2013). A growing body of research suggests recent increases in temperature and changing precipitation patterns are pushing species past critical hydraulic tipping points, leading to widespread tree die-off and mortality (Allen et al. 2010, 2015; Mitchell et al. 2014; Goulden and Bales 2019). Given these changes, understanding which functional traits are important for survival and the plasticity of those traits is critical for determining which species and ecosystems are more vulnerable to projected climatic changes (Lopez-Iglesias et al. 2014; Skelton et al. 2015).

Climate-induced forest die-off events commonly cause structural state shifts in ecosystems (Fensham and Holman 1999; Matusick et al. 2016) and may also result in species compositional shifts when co-occurring species show differential effects and tolerances (Mueller et al. 2005; Chen et al. 2011; Fauset et al. 2012). Varying responses to extreme climatic events are explained by different physiological tolerances (Breshears et al. 2009), functional traits (Condit et al. 1996), and ecological strategies (Zeppel et al. 2015). For example, in the semi-arid Pinyon pine–juniper ecosystem of south-western United States, repeated drought events have preferentially killed the less drought-tolerant Pinus edulis, shifting composition towards juniper (Juniperus spp.) dominance in certain circumstances (Mueller et al. 2005). These differences in drought-tolerance are partially explained by functional traits, including higher resistance to xylem cavitation, lower tracheid diameter, and xylem conductance in Juniperus than in P. edulis, highlighting the trade-off between resistance to drought and water use (Linton et al. 1998).

Stem sapwood is a critical component of the hydraulic infrastructure in trees and can represent an important constriction in the soil–plant–atmosphere hydraulic pathway (Tyree and Ewers 1991; Ryan and Yoder 1997). The production and maintenance of sapwood are critical for facilitating water movement, and can be limited by drought (Scholz et al. 2011). For example, less sapwood was produced when Pinus laricio Poiret was treated with reduced precipitation, and this was attributed to reduced water transport capacity resulting in significantly less stand transpiration (Cinnirella et al. 2002). Additionally, the proportion of sapwood and heartwood can be a critical component of stem water storage, which buffers trees during daily and seasonal dry periods (Scholz et al. 2011), especially in water-limiting environments (Hu et al. 2018). For example, Pfautsch and Adams (2013) identified the importance of stem water storage in resisting the effects of extreme drought and heat events in Eucalyptus. A recent study has found that drought resistance in Eucalyptus is enhanced after being exposed to repeated drought, in part, by altering stem morphology (Pritzkow et al. 2021). These studies have collectively illustrated the importance of stem functional traits and trait plasticity in drought resistance. At larger scales, understanding stem sapwood × tree diameter relationships is critical for determining site-stand-level transpiration dynamics, which is necessary for deciphering forest-level adaptation to climate change. In water-limited ecosystems, such as those in semi-arid and Mediterranean climate regions, where seasonal drought naturally limits tree growth, size and survival, adjustment to sapwood and heartwood areas may be an important strategy for regulating water use and preventing drought-induced tree die-off and mortality (Martínez-Vilalta et al. 2009; Markesteijn et al. 2011). In these dry regions, where climate change is extending seasonal drought patterns and contributing to extreme drought and heat events (Hoerling et al. 2012), it is important to determine species adaptation potential and which strategies are being utilised among co-occurring species to estimate ecosystem compositional changes.

Bark thickness is another stem functional trait that could explain current and future tree species distributions with climate change. Bark is the organ used by trees to resist the effects of fires (Nolan et al. 2020) and bark attributes help explain species distributions in fire-prone regions (Lawes et al. 2021). Climate change is either currently altering or expected to alter the frequency and intensity of fire in many regions (Bowman et al. 2020). Therefore, understanding differences in bark attributes among co-occurring species is critical for predicting future species compositions under altered climate and fire regimes.

In the Mediterranean climate region of south-western Australia, a severe heatwave-compounded drought event corresponded with tree die-off and mortality in the Northern Jarrah Forest (NJF) ecosystem in 2011 (Matusick et al. 2013). Although there was little evidence of consistent compositional shifts 49 months following the event (Matusick et al. 2016), the two dominant overstorey species exhibited different responses (Matusick et al. 2013; Ruthrof et al. 2015). For example, the dominant tree species Eucalyptus marginata Donn ex Sm. was more likely to experience rapid crown die-off in response to the event than was the co-dominant species Corymbia calophylla (Lindl.) K.D.Hill & L.A.S.Johnson, whereas both species exhibited significant resprouting responses in stems experiencing die-off (Matusick et al. 2013). In effect, C. calophylla exhibited higher resistance to a drought and heatwave, whereas the pronounced resprouting response in E. marginata illustrated its resilience to extreme hot and dry conditions (Matusick et al. 2013; Ruthrof et al. 2015). The shift observed from single large old stems to multiple small young stems in E. marginata has implications for forest sapwood, heartwood area, and water use (Macfarlane et al. 2010). For example, areas of regrowth E. marginata used double the amount of water compared with old-growth forest. This was attributed to differences in the relative amounts of sapwood and heartwood area (and not sapwood velocity or density; Macfarlane et al. 2010). The shift from large stems to small stems also has implications on the effects of future fires, because small stems tend to have thinner bark and are less resistance to fire as a result (Nolan et al. 2020).

To determine whether three key stem functional traits are associated with the contrasting observed responses between E. marginata and C. calophylla, bark area, sapwood area, and heartwood area were assessed in forest patches that had experienced severe forest canopy die-off (defined as >70% canopy die-off) and minimal forest canopy die-off (exhibiting <10% canopy die-off) during drought in 2011 (Matusick et al. 2013). From here on, we differentiate these forest patches on their vulnerability to drought, and categorise them as having either high (experiencing severe canopy die-off) or low (experiencing minimal canopy die-off) drought vulnerability. Our objective was to examine whether these three functional traits varied among the two co-occurring species and two levels of drought vulnerability (high and low).

The NJF represents the northern portion of the Jarrah Forest ecosystem, spanning from latitude −30.761570, longitude 115.98676 to latitude −33.504507, longitude 117.082340, and is classified as open forest in the north and tall, more closed forest in the south (Specht et al. 1974). The upland forest overstorey is dominated by E. marginata (jarrah) and C. calophylla (marri), with E. wandoo (wandoo, Blakely) occurring predominately in the eastern region. The midstorey is composed of a mixture of small trees and the understorey is highly diverse (Myers et al. 2000; Gioia and Hopper 2017).

The climate of the NJF is considered a Mediterranean-type, characterised by cool, wet winters, and an extended summer drought period, lasting from 6 to 8 months. Historically, annual precipitation has ranged from 635 to ∼1300 mm (Gentilli 1989). A strong precipitation gradient occurs from the north-east (drier) to the south-west (wetter) across the NJF. A well-documented shift in precipitation has occurred in the south-west of Western Australia during the past 40+ years (Cai et al. 2011; Matusick et al. 2013). The mean maximum temperature in the town of Dwellingup, Western Australia (within the NJF), is 21.9°C, and high temperatures and the frequency of extreme heat events have been increasing in the study region (Matusick et al. 2016; Breshears et al. 2021).

In much of the NJF, an extensive lateritic duricrust overlies a predominately granitic basement. The laterites weather to yield gravel and sand, whereas the weathered granite below the duricrust is represented by a deep clay (Churchward and Dimmock 1989). Over time, the deep-rooted overstorey trees have tunnelled through the duricrust, forming root channels, allowing access to water stored in the underlying clay (Churchward and Dimmock 1989; Dell and Havel 1989). This globally unique hydrogeology, with a deeply weathered lateritic profile and large soil water storage capacity, supports tall trees, which are reliant on stored water during the annual summer drought as well as intermittently dry winters (Dell and Havel 1989; Schofield et al. 1989; Farrington et al. 1996).

Corresponding with the ninth heatwave of the 2010/2011 summer period, significant changes were observed in a broad range of terrestrial and marine flora and fauna in south-western Australia (Ruthrof et al. 2018, 2021). This included patches of the NJF, which began experiencing die-off in February 2011 (Matusick et al. 2013). Forest patches with shallow soils, and limited access to stored water were most vulnerable to die-off, including those on upper slope positions surrounding granite rock outcrops (Brouwers et al. 2013). Patches experiencing die-off were larger and more densely clustered in xeric areas (Andrew et al. 2016). Structural characteristics, combined with the lack of the drought-susceptible midstorey species, B. grandis (see also Steel et al. 2019), strongly suggest that these patches have experienced previous periods of die-off and may be conditioned to living through large oscillations in moisture availability.

Each study site contained patches of high and low drought-vulnerable forest. A subset of eight study sites was selected randomly from a larger sample of sites (20) for sapwood measurements (see Matusick et al. 2013 for how study sites were identified). Briefly, forest patches with high vulnerability to drought were delineated using a differential GPS (Pathfinder Pro XRS receiver, Trimble Navigation Ltd, Sunnyvale, CA, USA). These drought-vulnerable patches were defined as having >70% of the forest canopy showing severe effects, including yellow, orange, and red leaves. A band of forest outside the high drought patches was considered to have a low vulnerability to drought (Matusick et al. 2013).

Twenty-six months following the die-off event, between 3 and 25 (average: 10) E. marginata and C. calophylla trees ranging in size between 1.3 and 92 cm diameter at breast height (DBH, 1.3 m height) were sampled within the high and low drought-vulnerable patches (total n = 143 for each species, Table 1). All samples were taken within a 21-day period. For each tree, the DBH was measured and stems were cut transversely at 1.3 m height and a slice of stem cross-section was collected for measurements. The width of the bark, sapwood and heartwood was taken on each collected cross-section at three random locations on the sample by using a digital calliper (mm). Sapwood was discriminated from heartwood primarily by using the distinct natural colouration differences. For samples difficult to discriminate, 1% methyl orange was used to enhance the differences between tissue types, as described in Macfarlane et al. (2010). The total amount of cross-sectional bark, sapwood, and heartwood area was calculated for each sample tree from these measurements. The ratio of sapwood to heartwood area was also calculated for each tree.

Table 1.  The number of sampled trees in five DBH classes within forest sites with high and low vulnerability to drought, and for canopy tree species Eucalyptus marginata and Corymbia calophylla in the Northern Jarrah Forest, south-western Australia.

To determine whether bark, sapwood and heartwood areas, as well as the ratio of sapwood to heartwood area, vary by tree species and drought vulnerability (high and low) independently or interactively, variables were used in a linear mixed effects model in the MIXED procedure in SAS (SAS ver. 9.3, Cary, NC, USA). Bark, sapwood and heartwood area were log-transformed, whereas the ratio of sapwood to heartwood area was square-root-transformed to correct for normality and variance. Tree species, drought vulnerability, and the interaction between tree species and drought vulnerability were included as fixed effects, and site was included as the random factor in the mixed effects model. Diameter at breast height (DBH, log-transformed) was included as the covariate in each model. Tukey’s multiple-comparison tests were used to discriminate differences within and across factors.

The relationship between DBH and the measured stem functional traits were also of interest because of differences in forest structure mentioned previously. Simple linear regression equations were fitted for each drought vulnerability, tree species, and stem trait separately by using PROC REG, including log-transformed DBH and log-transformed stem traits to determine the strength of the relationship between DBH and stem traits.

Bark area varied between species, with E. marginata having slightly more bark than C. calophylla (12 303 mm² vs 12 230 mm²); however, the difference was not statistically significant (F = 3.46, P = 0.0640). Drought vulnerability had no statistical effect on bark area (F = 1.12, P = 0.2912), nor was there a statistically significant interaction (F = 0.02, P = 0.8780). A very significant linear relationship between bark area and DBH was observed for both species (Fig. 1). Bark area was greater in smaller E. marginata trees and in larger C. calophylla trees.

Fig. 1.  Linear regression equations for bark area and diameter at breast height (DBH) for Corymbia calophylla and Eucalyptus marginata in the Northern Jarrah Forest of south-western Australia. Dashed line indicates E. marginata.

Sapwood area varied by tree species (F = 22.88, P < 0.0001) and drought vulnerability (F = 7.41, P = 0.0069) independently. Corymbia calophylla had a larger sapwood area than did E. marginata (Fig. 2a). When controlling for DBH, trees located in low drought-vulnerability patches had a significantly greater sapwood area than did those in high drought-vulnerability patches (Fig. 2b). From the regression analysis, the relationship between sapwood area and DBH was strong for both species and across drought vulnerabilities. Corymbia calophylla had more sapwood than did E. marginata in smaller trees, but less in larger trees (Fig. 3a). Similar results were found between high and low drought-vulnerability patches. Smaller trees had more sapwood area in high-drought patches and larger trees had more sapwood area in low-drought patches (Fig. 3b).

Fig. 2.  Average sapwood area (mm²) (a) in the two co-dominant canopy species, Corymbia calophylla and Eucalyptus marginata, and (b) in high and low drought-vulnerability forest, in the Northern Jarrah Forest, south-western Australia, following drought-induced die-off in 2011. Means are statistically significant between species and drought-vulnerability plots independently from mixed-effects models controlling for diameter at breast height (DBH) at alpha = 0.05. Error bars represent the standard error of the mean.

Fig. 3.  Linear regression equations for sapwood area and diameter at breast height (DBH) (a) for trees in low and high drought-vulnerability forest, and (b) for the co-dominant canopy species Corymbia calophylla and Eucalyptus marginata in the Northern Jarrah Forest of south-western Australia, following drought-induced die-off in 2011. Dashed line indicates E. marginata in a; dashed line indicates low drought-vulnerability class in b.

There was a significant statistical interaction between tree species and drought vulnerability (F = 4.54, P = 0.0340). Corymbia calophylla had a significantly smaller heartwood area than did E. marginata (F = 7.21, P = 0.0077) in high drought-vulnerability patches, but this relationship was not statistically significant in low drought-vulnerability patches (Fig. 4). The relationship between the heartwood area and DBH was strong for both species (Fig. 5). Eucalyptus marginata maintained more heartwood in smaller trees, but the relationship was similar between species in larger trees.

Fig. 4.  Average heartwood area (mm²) for Corymbia calophylla and Eucalyptus marginata in high and low drought-vulnerability forest, Northern Jarrah Forest, south-western Australia. Bars with the same letter are not significantly different from Tukey’s pairwise comparisons following mixed-effects modelling controlling for diameter at breast height (DBH). Error bars represent the standard error of the mean.

Fig. 5.  Linear regression equations for heartwood area and diameter at breast height (DBH) for Corymbia calophylla and Eucalyptus marginata in the Northern Jarrah Forest of south-western Australia, following drought-induced die-off in 2011. Dashed line indicates E. marginata.

A significant tree species effect was detected (F = 9.38, P = 0.0024). Corymbia calophylla had a much higher ratio than did E. marginata (2.4 vs 0.8). On average, C. calophylla had more sapwood than heartwood, whereas E. marginata had more heartwood than sapwood. Drought vulnerability was not statistically significant in the linear mixed model (F = 0.22, P = 0.6373).

This study has shown differences in certain stem functional traits between co-occurring tree species in a forest considered to have high and low vulnerability to drought. The observed differences in sapwood are perhaps most applicable in terms of drought responses in trees, because sapwood is an essential component of hydraulic infrastructure. Corymbia calophylla had a greater sapwood area than did E. marginata and was observed to be more resistant to die-off than was E. marginata (Matusick et al. 2013; Ruthrof et al. 2015). Sapwood can function as a significant water-storage organ (i.e. hydraulic capacitance; Goldstein et al. 1998; Köcher et al. 2013), including in Eucalyptus (Pfautsch and Adams 2013), and represents a critical water source during dry periods (Stout and Sala 2003; Hartzell et al. 2017; Salomón et al. 2017). These findings, combined with a greater stomatal control in C. calophylla (Szota et al. 2011), suggest sapwood storage and stomatal control combine to enhance C. calophylla resistance to drought, compared with E. marginata (Ruthrof et al. 2015).

The two species in the current study have some similar, and some disparate, ecological characteristics. For example, C. calophylla flowers on a 3- to 5-year cycle (Robinson 1960) and E. marginata flowers intermittently every 4–6 years (Abbott and Loneragan 1986; Abbott et al. 1989). Corymbia calophylla has much larger fruit (29.6 mm width, cf. 12.8 mm for E. marginata; Gill et al. 1992), a higher number of seeds per fruit (3.1 cf. 0.8 seeds per fruit respectively; Gill et al. 1992), and unusually large seeds (94 mg cf. 12 mg respectively; McChesney et al. 1995), compared with E. marginata because. The size of fruit and seed may be driven by predation pressure, C. calophylla is a favoured food source by three cockatoo species (Biggs et al. 2011; Lee et al. 2013). Both of these serotinous species also invest in a bank of suppressed, lignotuberous seedlings (Schuster 1980), which grow rapidly (∼1 m/year) once a disturbance occurs, and canopy competition is decreased (Ward and Koch 1995). The survival rate of these seedlings is considerably higher in C. calophylla than in E. marginata (Abbott 1984). Wood density for the study species is similar, at 0.67 and 0.65 t/m³ respectively (Forest Products Commission 2020), which is known to explain drought response (Hoffmann et al. 2011). The physiological response to water deficit is among the biggest differences between the species. Szota et al. (2011) showed that although E. marginata showed a high tolerance to dehydration and was able to survive drought via resprouting C. calophylla had higher drought tolerance (via lower stomatal conductance, higher leaf osmotic potential and leaf water content). Observations made following the response of the species to the 2011 die-off event support the findings by Szota et al. (2011). They collectively have implications for the species under a drier future. Pfautsch and Adams (2013) highlighted the importance of stem water storage in the survival of E. regnans during hot and dry conditions, including the importance of refilling depleted stem water stores during the night. Under the expected drier and hotter future in south-western Australia (Andrys et al. 2017), the capacity of Eucalyptus species to refill stem water and access stored water will be critical for determining competitive relationships in natural ecosystems. If the increased sapwood area in C. calophylla proves to represent a greater opportunity to store and access water, we would expect C. calophylla to have a competitive advantage over E. marginata under both short, high-intensity heat and drought events and protracted periods of water deficits. Specifically, increased survival and a high probability of maintaining a dominant stem would advantage C. calophylla in re-populating forest patches experiencing die-off, particularly given the reproductive characteristics mentioned previously. Future work should conclusively determine the role of sapwood and stem water storage in the increased drought resistance of C. calophylla and search for evidence of compositional shifts driven by repeated die-off events to assist predictions of future climate changes.

Sapwood areas were greater, regardless of tree species, in forest with a low vulnerability to drought-induced die-off. In the absence of other measurements, it remains unclear what this finding means. Measurement of leaf area would enable calculation of the Huber value, which describes the relationship between leaf area and sapwood area (Tyree and Ewers 1991) and is an indicator of functional adaptation to environmental conditions in trees (Gotsch et al. 2010). Although we cannot directly measure the Huber value here, we can infer that those forests highly vulnerable to drought had maintained less leaf area in advance of the drought than had forests less vulnerable from indirect evidence, including the fact that this high drought-vulnerable forest had experienced die-off in the recent past, had higher levels of recent tree mortality and partial crown die-off (Matusick et al. 2013). Indeed, Eucalyptus adjusts their Huber value to account for differences with water supply to maintain homeostasis (Carter and White 2009; Zolfaghar et al. 2015). Lower levels of leaf area in a drought-prone forest may explain the observed differences in sapwood area, because less sapwood is required to meet transpirational demands in trees with smaller tree crowns and fewer leaves (Paine et al. 1990). However, the exact relationship between Huber value and forest vulnerability to die-off in the NJF has yet to be determined.

The findings presented here show that the relationship between DBH and sapwood area changed with tree size. Small trees contained greater sapwood areas in high-drought patches, whereas larger trees contained greater sapwood areas in low-drought patches. These results may reflect leaf area adjustments to water availability (Pook 1985), in relation to competitive position (Crous et al. 2021), or simply the effects of age differences (Roberts et al. 2001). In high drought-vulnerable forest, repeated die-off events have caused retraction in overstorey trees, tree mortality, and vigorous re-sprouting and growth of small stems (Matusick et al. 2013). This exchange of growth between large trees and small, young stems is also likely to be reflected in relative leaf area changes, where proportional leaf area has shifted from large trees to small trees. With a greater leaf area per unit DBH in smaller trees, than in larger trees, more sapwood area would be required to satisfy transpirational demands (assuming similar sap velocities; Köstner et al. 2002). Because low drought-vulnerability sites did not experience overstorey die-off, large trees remained dominant and small trees have remained suppressed (with potentially a lower leaf area:DBH ratio), explaining why proportional sapwood areas are greater in larger trees in low-drought patches. Alternatively, our findings are the result of an age-related phenomenon. Small stems in high drought-vulnerable patches are likely to be younger than are small stems in low drought-vulnerable patches, owing to recruitment opportunities following recent die-off events. If this is true, we might expect higher sapwood areas per unit DBH in younger stems, because proportional sapwood naturally decreases with age as heartwood increases (Santos et al. 2021).

How functional traits, such as sapwood and heartwood areas, interact with drought and heatwave events has important consequences for stand function, particularly for water use and vulnerability to future drought events (Greenwood and Weisberg 2008). In the water-limited NJF, stand structural characteristics and the associated amounts of sapwood and heartwood are important regulators of stand water use (Macfarlane et al. 2010). Specifically, large trees contain greater quantities of non-water-conducting heartwood than do smaller trees, and site water use in the NJF can be attributed to these structural attributes (compared with sap velocity) (Macfarlane et al. 2010). Therefore, the replacement of large trees with higher densities of smaller trees (containing much less heartwood), as is seen following bauxite mining (Grigg and Grant 2009) and drought-induced die-off (Matusick et al. 2016), is likely to alter site water use by the forest canopy. In a drying climate, increasing site water use from disturbance-induced structural shifts may interact to result in repeated tree die-off and mortality events.

Since heartwood formation is an active process, trees may regulate the amount of sapwood by altering the rate of heartwood formation (Spicer 2005). Heartwood area, sapwood area, and transpirational demands are inherently linked, and some have postulated that drought and associated cavitation of the inner sapwood precedes and facilitates heartwood formation (Beauchamp et al. 2013; Chan et al. 2013). Therefore, heartwood formation can be a function of age (Santos et al. 2021), climate (Almeida et al. 2020), growth resources (Miranda et al. 2006, 2009), site quality (Climent et al. 1993), and stress (Paine et al. 1990). However, in this study, we found no differences in heartwood area between high and low drought-vulnerable forest. Instead, we found differences in heartwood area between competing species, with E. marginata having greater quantities of heartwood in smaller trees. Inherent genetic differences in rates of heartwood formation are common among Eucalyptus (Miranda et al. 2014). However, since trees were not aged, it remains unclear whether smaller E. marginata trees were of ages similar to those of similar-sized C. calophylla trees. Eucalyptus marginata can arrest its growth and development and persist as saplings for many years under intense competition (Abbott et al. 1989). Because age is an important determinant of heartwood quantity (Climent et al. 2002), this potential confounding variable needs to be understood prior to making conclusions regarding tissue allocation.

Bark thickness in eucalypt forests and woodlands is a function of stem diameter and confers greater resistance to fire damage (Wesolowski et al. 2014; Schubert et al. 2016), which is frequent in many eucalypt-dominated ecosystems (Murphy et al. 2013). It is therefore not surprising that co-occurring E. marginata and C. calophylla have similar bark areas when controlling for DBH, because they experience identical frequent fire regimes comingling in the upland NJF. Our findings suggest that smaller E. marginata trees have more bark than do C. calophylla trees, which may confer greater fire protection during the early stages of stand development.

This study has provided further information about three stem functional traits in C. calophylla and E. marginata in a forest with high and low vulnerability to drought; however, making strong conclusions regarding the role of these traits requires additional measurements to add appropriate physiological context. Examining individual functional traits in isolation is unlikely to explain the overall response of a species to drought. A suite of additional measurements on functional traits are needed to explain differences in species responses to drought. These include tree age, sapwood:leaf area ratio, and vulnerability to cavitation because it relates to sapwood:leaf area ratio, specifically at drought-vulnerable sites. Sapwood capacitance is not only a function of sapwood area, but also, critically, wood anatomy and density, which is commonly found to be associated with drought responses in trees (Hoffmann et al. 2011). This study does not infer how much sapwood is active within the trees, nor sap flow rates, which are critical for examining water use among forest patches (Doody et al. 2015). Additional information regarding natural embolism rates or sap flow rates could better estimate water use on drought-vulnerable sites. Further research encompassing wood density, stomatal conductance, rooting depth, water age and differences between species and how they relate to drought vulnerability is required. Importantly, these measurements should be undertaken on drought-vulnerable sites with limited connection to groundwater, which can buffer interseasonal and interannual surface soil dryness.

In conclusion, our study has shown that the two co-occurring canopy species in the NJF have different sapwood and heartwood areas. Specifically, C. calophylla had a larger sapwood area and a smaller heartwood area than did E. marginata (in high-drought patches), which may help explain observations of differential species responses to drought. Regression equations developed in this study may be helpful for projecting quantities of sapwood and heartwood across forest patches, which may be required for future modelling exercises and understanding of the spatial and temporal vulnerability of the NJF. This study has highlighted important and fruitful research pathways, investigating the differences among the co-occurring species to inform predictions regarding future forest composition. Greater knowledge about the responses to drought by co-occurring forest tree species will increase our predictive ability regarding forest responses in a drying and warming climate.

The data will be made available upon request to the main author.

The authors declare no conflicts of interest.

This research was conducted as part of the Western Australian State Centre of Excellence for Climate Change Woodland and Forest Health, which is a partnership between private industry, community groups, Universities, and the Government of Western Australia.