Introduction

Southwest Australia is a global biodiversity hotspot (Myers et al. 2000), one of only 36 in the world (Habel et al. 2019), with some of the most phosphorus (P)-impoverished soils globally (Kooyman et al. 2017). Within the Southwest Australian Biodiversity Hotspot, tall eucalypt forests are characterised by a low diversity of vascular plants at the community level and limited diversity of growth forms (Hopper and Gioia 2004). For example, we collected 21 species at a Eucalyptus patens site for our previous paper (Zhou et al. 2022); this comprised most of the species at this site, except for some understorey orchids, sundews and sedges. The low diversity of this site is in strong contrast with numbers of 40–50 species per 100 m2 in kwongan shrublands in the Southwest Biodiversity Hotspot (Zemunik et al. 2016). We therefore referred to the E. patens habitat as a ‘cool spot in a biodiversity hotspot’ (Zhou et al. 2022).

In southwest Australia, the greatest plant diversity is associated with some of the most P-impoverished soils in the world (Lambers et al. 2010). All Eucalyptus species in this region occur on P-impoverished soils, but their habitats differ in terms of rainfall (https://florabase.dpaw.wa.gov.au/browse/profile/21824). Compared with extremely low soil P availability, nitrogen (N) is relatively abundant in southwest Australia (Laliberté et al. 2012; Lambers et al. 2010; Turner and Laliberté 2014). Soil P availability declines during pedogenesis (Walker and Syers 1976), leading to a steadily increasing soil N:P ratio during long-term soil development which makes P the major macronutrient limiting plant productivity in southwest Australia (Guilherme Pereira et al. 2019; Lambers et al. 2010; Turner and Laliberté 2014). Many native plant species that evolved in this P-limiting environment are highly P-efficient, with specialised P-acquisition and P-utilisation strategies (Lambers 2022).

Proteaceae is one of the dominant plant families in southwest Australia; most of them are non-mycorrhizal (Brundrett 2017; Lambers et al. 2010). They efficiently acquire P from the soil through their specialised cluster roots that exude carboxylates into the soil and mobilise P from poorly-soluble P complexes (Shane and Lambers 2005). Proteaceae are prominent, with many species showing relatively fast rates of photosynthesis per unit leaf area (Aarea) and a very high photosynthetic P-use efficiency (PPUE) in severely P-impoverished ecosystems in southwest Australia (Guilherme Pereira et al. 2019; Sulpice et al. 2014; Denton et al. 2007). Eucalyptus belongs to the family Myrtaceae, which plays an important role in southwest Australia; Eucalyptus patens (yarri or Swan River blackbutt) occurs in eucalypt forests in high-rainfall zones, where it dominates the forests (Zhou et al. 2022; French and Nicolle 2019.). Eucalypts do not have cluster roots, but some do exude carboxylates into the soil and mobilise P (Lambers et al. 2021; Zhou et al. 2022; de Andrade et al. 2022). Cluster roots in Proteaceae are ephemeral structures, and in Hakea prostrata, they live for about 20 days, from rootlet emergence to senescence (Shane et al. 2004). In H. prostrata (Shane et al. 2004) and Lupinus albus (Fabaceae) (Watt and Evans 1999), citrate and malate are released in an exudative burst at their mature stage. Also, in L. albus, acidification of the rhizosphere is associated with the exudation of protons just before the mature stage of cluster roots (Zhu et al. 2005). Cluster-root formation and carboxylate release in L. albus (Watt and Evans 1999) and H. prostrata (Shane et al. 2003) are suppressed at a high P availability, but how variation in P supply affects Eucalyptus root activity, especially the release of carboxylates, is unknown. Also, it is quite common in Proteaceae for excess P supply to cause P-toxicity symptoms (Shane et al. 2003). As a rare example among Myrtaceae, E. marginata also exhibits P-toxicity symptoms at elevated soil P availability when grown without mycorrhizal symbionts, but it does not when colonised by ectomycorrhizal and arbuscular mycorrhizal fungi (Kariman et al. 2014). We do not know whether E. patens, which co-occurs with E. marginata in high-rainfall areas (French and Nicolle 2019), is P sensitive, like many Proteaceae, or tolerant like most Myrtaceae (Lambers 2022; Liu et al. 2022).

Eucalyptus is a C3 genus of woody species distributed across the Australian continent, with species occupying both mesic and xeric environments (Schulze et al. 2006; Cernusak 2020; Lamoureux et al. 2018; Nolan et al. 2017). Foliage δ13C is considered a good indicator of a time-integrated water-use efficiency in C3 plants, with more negative leaf δ13C values indicating lower water-use efficiency (Farquhar and Richards 1984; Adams et al. 2020). Eucalyptus species native to mesic environments tend to have a lower water-use efficiency than those native to xeric environments (Schulze et al. 2006). The photosynthetic activity of leaves is closely correlated with their N concentration ([N]) (Jia et al. 2021; Evans 1989; Field and Mooney 1986) because N is a component of photosynthetic enzymes such as Rubisco and proteins of the electron transport chain (Evans 1989). There is an inverse relationship between photosynthetic N-efficiency and intrinsic water-use efficiency (iWUE) (Hu et al. 2023; Querejeta et al. 2022; Fredeen et al. 1991; van den Boogaard et al. 1995). Under limited availability of N, stomatal conductance may be higher, increasing PNUE at the expense of iWUE (Renninger et al. 2015). Our study site of E. patens occurs in a region with an annual rainfall of 1224.5 mm (1934–2022; http://www.bom.gov.au/climate/data/); water is available throughout the year, even in the dry summer that is common in southwest Australia. To date, there is no information on the iWUE of E. patens and co-occurring species in E. patens forests.

We aimed to quantify the PPUE, PNUE, WUE and iWUE of common plant species from a low-diversity spot in a biodiversity hotspot. We hypothesised, first, that with a low leaf [P] and releasing carboxylates, the dominant species at the E. patens site would exhibit a very high PPUE, while leaf [N] would be low in the dominant species (except B. aquifolium; Fabaceae) and these species would exhibit a very high PNUE. Conversely, we hypothesised WUE would be low with a low leaf δ13C value, and iWUE would be low, with a high stomatal conductance compensating for low leaf [N]. Second, we also tested the hypothesis that E. patens would not show P toxicity at high [P] in the root environment, in contrast with E. marginata where P toxicity is acute, because E. patens would down-regulate its P-uptake capacity at supraoptimal P supply. Third, we hypothesised that carboxylate release in E. patens would not decline by increasing [P] in the hydroponic solution. To test this hypothesis, we compared root exudation by measuring the carboxylate-release rate of the root system from plants that received increasing [P] in the hydroponic solution. Understanding how Eucalyptus species acquire and use P, N and water at physiological and ecological levels will increase our understanding of ecosystem functioning in a low biodiversity spot in a biodiversity hotspot.

Materials and methods

In situ measurements in a Eucalyptus patens forest

Site description

The study site was located at Cypress Form, Waroona (32°47′35”S, 116°0′46″E), ~100 km south of Perth in southwest Australia (Fig. 1a), a region characterised by a Mediterranean climate, with hot dry summers (December to February) and cool wet winters (June to August) with a mean annual rainfall of 1224.5 mm. The mean annual temperature is 22.0 °C and the mean monthly temperature maxima range from 16.7 °C in July to 32.5 °C in January (http://www.bom.gov.au/climate/data/; measured at Dwellingup, 1934–2022). Soils were acidic; pH (CaCl2) was 4.8 ± 0.1 at the E. patens sampling site, and soil total [P] were very low (34 ± 2 mg P kg−1). Readily-available soil [P], measured as resin-P concentrations, were very low (1.1 ± 0.1 mg kg−1) across the site, and soil organic-P was the major fraction of soil P at all sites with a proportion of 88% (Zhou et al. 2022).

Fig. 1
figure 1

Study site of the low-diversity ‘cool’ spot and studied species. a Study site of the Eucalyptus patens (yarri) forest at Cypress Form, Waroona (32°47′35”S, 116°0′46″E) in southwest Australia, located ~100 km south of Perth. Shades with different intensities of green represent the density of forest cover. The inset shows the map of Australia. Scale bars: on main map: 50 km; on inset: 1000 km. Source: https://www.google.com/maps/place/Waroona+WA. b Banksia grandis, c Bossiaea aquifolium, and (d) Eucalyptus patens at the study site; e E. marginata and (f) Corymbia calophylla photographed elsewhere. Insets of b-f are the flowers of each species. Photos: b, c, d Xue Meng Zhou; e, f Sophie Xiang

Full size image

Selection of species

We used Banksia grandis Willd (Proteaceae), Bossiaea aquifolium Benth (Fabaceae), and three Myrtaceae, Eucalyptus patens Benth, E. marginata Sm. and Corymbia calophylla Hill & Johnson (Fig. 1b-e), including saplings (1–3 years old) and regrowth (around one year after coppicing) of E. patens, for gas-exchange measurements and leaf chemical analyses. These eucalypt species are the dominant species of the overstorey at the field site, and all have a high leaf [Mn] in their natural habitat and release carboxylates when grown in a nutrient solution with a low [P] (Lambers et al. 2021; Zhou et al. 2022). We used B. prionotes to show the sunken stomata of its thick leaves, but it was not sampled together with the other species as it does not occur in the same habitat.

Analyses of leaf samples

Leaf P concentrations

The mature leaves of selected species were sampled to determine leaf [P], just after measuring the leaf gas exchange rates on 15 and 16 January 2021, and five to 10 plants were sampled per species. After drying the leaves at 65 °C for three days, samples were ground in a ball-mill grinder (2010 Geno/Grinder, Metuchen, USA), using plastic vials and yttrium-stabilised zirconium ceramic beads. Samples were digested with hot concentrated HNO3:HClO4 (3:1) and analysed for [P] using ICP-OES (Model 5300DV, Perkin Elmer, Shelton, CT, USA) (Zarcinas et al. 1987).

Leaf N concentrations and carbon-isotopic composition

The oven-dried finely-ground leaf samples of selected species were analysed for leaf [N] and carbon-isotopic composition (δ13C). Total [N] was measured by the combustion method using a Leco analyser (FP628, St. Joseph, MI, USA). Stable isotopes (δ13C) were measured using a Flash 1112 Elemental Analyser coupled via a continuous flow interface to Delta V Plus (Thermo-Finnigan, Bremen, Germany) isotope ratio mass spectrometer (Schulze et al. 2006). The natural stable isotope ratios were expressed as parts per thousand (‰), according to the following equation:

$$\delta X\left(\%_0\right)=\left({\mathrm R}_{\mathrm{sam}}/{\mathrm R}_{\mathrm{sta}}-1\right)\times1000$$

where X is 13C and Rsam and Rsta are the corresponding ratios of 13C /12C in sample and standard, respectively. We referenced our samples against Vienna Pee Dee Belemnite (V-PDB) for δ13C (Skrzypek 2013). Standards were tested repeatedly after the completion of each batch of 10 measurements. Within a run, precision was ±0.10‰ for δ13C (Coplen et al. 2006). Water-use efficiency (WUE) was calculated from δ13C values (Du et al. 2021), as described by Farquhar and Richards (1984).

$${\Delta}^{13}\textrm{C}=\left({\updelta}^{13}{\textrm{C}}_{\textrm{a}}-{\updelta}^{13}{\textrm{C}}_{\textrm{p}}\right)/\left(1+{\updelta}^{13}{\textrm{C}}_{\textrm{p}}/1000\right)$$

in which δ13Ca and δ13Cp represent the δ13C value of air and plant samples, respectively. The δ13Ca was assumed to be −8‰, following Farquhar et al. (1989).

$$\textrm{WUE}=\left(\textrm{b}-{\Delta}^{13}\textrm{C}\right)/1.6\left(\textrm{b}-\textrm{a}\right)$$

where a is the diffusive fractionation of 13C in air through the stomata (4.4‰), and b represents the net fractionation caused by carboxylation (27‰) (Du et al. 2021; Farquhar et al. 1989).

Leaf anatomy

We collected intact mature leaves of the selected species for anatomical study. We used carrots as the supporting material to hold the leaves and cross-sections were prepared using a microtome (Leica RM2125 RTS; Leica Biosystems Nussloch GmbH, Heidelberg, Germany). Unstained cross-sections were viewed under an epifluorescence microscope in the darkfield with a UV filter, and photos were taken using an Axiocam digital camera (Carl Zeiss Microscopy GmbH, Oberkochen, Germany).

Leaf gas-exchange measurements

We measured the leaf gas exchange of selected species with an infrared gas analyser (LI-COR 6800XT; LI-COR Inc., Lincoln NE, USA) between 9 am and 12 am on 15 and 16 January 2021. Light-saturated photosynthesis (Aarea) rates were measured in situ at a CO2 concentration of 400 μmol CO2 mol−1, a chamber temperature of 25 ± 0.2 °C, and a vapour pressure deficit of 0.5–1.0 kPa. The photosynthetic photon flux density was 1500 μmol m−2 s−1, with a relative humidity of 50–70%. Measurements were taken using undamaged, youngest fully-expanded and sun-exposed mature leaves, For the measurement, we cut a large branch from the tree and then measured the gas exchange rate immediately to minimise any errors on the reading as previously described (Grant 1992; Samuelson 1998). Stomatal conductance did not change after cutting the branch and during the gas-exchange measurements. We sampled five to 10 individuals per species.

After finishing each gas-exchange measurement, we collected the leaf material enclosed within the chamber of the LI-COR 6800XT. The area of those leaves was determined by scanning them using an Epson 1680 scanner (Epson America, Long Beach, CA, USA), and the images were analysed using the WINRHIZO v.4.1c scanner program (Regent Instructions, Quebec, Canada) to obtain leaf area data. Those leaves were then dried in an oven at 70 °C for three days to measure the dry mass. We calculated leaf mass per unit area (LMA) from the area and dry mass.

Experiments in a controlled-environment glasshouse

In vivo determination of changes in rhizosphere pH

We determined pH changes in the rhizosphere as described previously (Marschner et al. 1982). Agar (0.75% w/v; Sigma Chemical Company, St Louis, MO, USA) was heated to a clear liquid at ca. 80 °C, and 0.06% (w/v) bromocresol purple (pH indicator in the range of pH 5–8; Aldrich Chemical Company, Milwaukee, WI, USA) was added to the melted agar. The pH of the solution was adjusted to 6.5. The agar solution with pH indicator was divided into four equal parts and the pH of each part was adjusted to 5, 6, 7 and 8, respectively, by adding either 0.1 M HCl or 0.1 M NaOH. These solutions were poured into small Petri dishes and cooled to slightly below room temperature to observe the pH-dependent colour change of the agar solutions.

To observe the colour change of the agar solution with bromocresol purple as affected by root exudates, the pH of the agar medium was adjusted to 6.5 by adding 0.1 M NaOH (purple). The agar solution was poured into large Petri dishes and cooled to slightly below room temperature. To determine which parts of the root system changed the rhizosphere pH the most, we cut branched first-order lateral roots from the main root of E. patens seedling in hydroponics and separated mature, intermediate and younger zones, which contained second-order fine lateral roots. We also harvested younger tip regions of unbranched first-order lateral roots. We washed excised roots using double-distilled water and gently blotted them dry with paper towels and placed them in the agar medium separately. Root-induced changes in the rhizosphere pH by exudates, which were indicated by the colour change of the agar medium, were observed approximately 10–20 min after placing roots on the agar medium (Marschner et al. 1982; Shane et al. 2006; Tomasi et al. 2013).

Determination of root carboxylate-exudation rate in nutrient solution

We grew 10 seedlings of E. patens in a glasshouse at the University of Western Australia, under the controlled environment with an average daytime temperature of 25 °C from March to May 2022. Seedlings (around nine-months old) were purchased from Natural Area Holdings Pty Ltd. (Whiteman, WA 6068, Australia). All seedlings were transferred to aerated hydroponics in cooling tanks that maintained roots at 18 °C to 20 °C to start the pre-treatment period (Shane et al. 2004). Roots of the seedlings were washed and disinfected with 1% (v/v) sodium hypochlorite for 20 s before transferring them to 5-L plastic pots containing 4 L nutrient solution, which is commonly used for growing southwest Australian native plants (the nutrient solution contained, μM, 2.5 PO43−, 500 NO3−, 200 Ca2+, 500 K+, 36 SO42+, 36 Mg2+, 10 Fe-EDTA, 0.24 Mn2+, 0.10 Zn2+, 0.018 Cu2+, 2.4 H3BO3, 50 SiO32−, and 0.3 Mo4+. pH = 5.8 (Liu et al. 2022). Nutrient solutions were changed twice a week. Six weeks later, we excised one branched first-order lateral root from the main root and separated it into different parts, i.e. (1) younger zone of the branched lateral root with fine roots (YZBL), (2) mature zone of the branched lateral root with fine roots (MZBL), and (3) younger zone of unbranched lateral roots (YZUL) for the collection of root exudates and carboxylate analysis (Fig. S1). Root exudates were collected from these different parts of the roots in 3–10 mL of the same nutrient solution that was used to grow plants in hydroponics, but without P and Ca, for 1.5 h (Shane et al. 2006). Root exudates were then filtered using a 0.2 μm filter and stored at −20 °C until the samples were analysed. Root carboxylates were analysed using reverse phase liquid chromatography on an Alltima C-18 reverse phase column (250 × 4.6 mm, Alltech, Deerfield, IL, USA). The standards of malic, malonic, lactic, acetic, maleic, citric, succinic, cis-aconitic and trans-aconitic acid (ICN Biomedicals Inc., Aurora, OH) were used to identify carboxylates. The detection limit of the 10 acids ranged from 0.05 to 24 μM, as previously described by Cawthray (2003).

After the first harvest of roots to measure carboxylate-exudation rates, we left plants in hydroponics for two weeks to recover before harvesting roots again for the next carboxylate-exudation measurements. During this time, we changed the nutrient solution twice a week. After that, we gradually increased the [P] in the hydroponic medium from 1 μM to 20 μM, then to 40 μM, and finally to 80 μM. At each concentration, the seedlings were grown for two weeks before increasing [P] to the next level. The nutrient solution was changed twice a week. At each stage with different [P] in the medium, we collected exudates from one healthy branched lateral root from each plant, and analysed carboxylate exudation as described above for plants grown at 1 μM P. We also collected one or two mature fully-expanded mature leaves per plant grown at different P concentrations to analyse leaf [P]. We quantified P-toxicity symptoms (including chlorotic and necrotic areas on leaves) by ranking plants into six classes from 0 to 5, where the absence of any toxicity symptoms corresponded to 0, traces to 20% of symptomatic leaf tissue area (SLTA) as 1, 20 to 40% SLTA as 2, 40 to 60% of SLTA as 3, 60 to 80% of SLTA as 4 and more than 80% of SLTA as 5 (Kariman et al. 2014).

Statistical analyses

Individual trees were considered as the unit of replication. Differences among species, seedlings and regrowth for E. patens were analysed using one-way ANOVA. For each model, the normality of the residuals was assessed using the Shapiro-Wilk test, and heteroskedasticity was assessed by preparing a residual plot and visual inspection. Where appropriate (leaf [P], N:P ratio, PPUE and root exudation rates), data were log10 transformed before ANOVA. We used repeated-measures ANOVA with the R package rstatix to examine the difference among P treatments for the same individuals (Fig. 7b, c). We performed Tukey-Kramer HSD tests for a post-hoc mean comparison and differences were considered to be significant at P ≤ 0.05. All analyses were conducted in R 4.2.1 (R_Development_Core_Team 2022) using base R packages.

Results

Leaf P and N concentrations

The dry mass (DM)-based leaf [P] and [N] across all species in this study ranged from 0.23–0.60 mg g−1 and 8.8–26.5 mg g−1, respectively (Fig. 2a, b). Bossiaea aquifolium (Fabaceae) had the greatest leaf [N], 26.5 mg g−1, and N:P ratio, 71, among all species studied (Fig. 2c). In contrast, B. grandis (Proteaceae) had the lowest leaf [P] (0.23 mg g−1), and leaf [N] (8.8 mg g−1). However, its leaf N:P ratio was intermediate (38). Mature leaves of E. patens saplings had significantly greater leaf [P] (0.60 mg g−1) than those of adult trees (0.42 mg g−1), whereas, it was intermediate for E. patens regrowth, following coppicing (0.55 mg g−1 for P). The mature leaf [P] of E. marginata and C. calophylla (0.47 and 0.44 mg g−1, respectively) were the same as those of adult E. patens, and they were indistinguishable from each other. Leaf [N] of all Myrtaceae were similarly low, approximately 12 mg g−1. The mature leaf N:P ratio of E. patens regrowth (23) was similar to that of adult E. patens, E. marginata and C. calophylla, while the N:P ratio of E. patens saplings (18) was significantly lower than that of adult E. patens.

Fig. 2
figure 2

a Leaf phosphorus concentrations ([P]), b leaf nitrogen concentration ([N]) and (c) leaf N:P ratio, on a dry mass (DM) basis of Eucalyptus patens, yarri (Myrtaceae) and co-occurring species at Cypress Form, Waroona. Data are means ± SE (n = 5). Different lowercase letters above the bars indicate significant differences among means, determined by analysis of variance (ANOVA) with Tukey’s HSD test at P ≤ 0.05 level

Full size image

Leaf gas exchange and carbon-isotope composition

Across all studied adult species, the mean values of area-based light-saturated photosynthetic rates of mature leaves (Aarea) ranged from 7.2 (B. aquifolium) to 12.6 μmol CO2 m−2 s−1 (E. patens) (Fig. 3a). Eucalyptus patens regrowth had the fastest Aarea (22.2 μmol CO2 m−2 s−1) followed by E. patens saplings (17.3 μmol CO2 m−2 s−1). The former had significantly faster Aarea than the adult leaves of all co-occurring species. The leaf stomatal conductance (gsw) was similar for all studied species (approximately 0.2 mol H2O m−2 s−1), except for B. aquifolium, which had the lowest gsw (0.08 mol H2O m−2 s−1) (Fig. 3b). The LMA varied among the co-occurring species, ranging from 56.5 g m−2 (B. aquifolium) to 177.8 g m−2 (B. grandis) (Fig. 3c). Adult plants of E. patens, E. patens saplings, E. patens regrowth, E. marginata and C. calophylla had a similar LMA, approximately 100 g m−2. Rates of photosynthesis per unit P (PPUE) (Fig. 3d) and N (PNUE) (Fig. 3e) of Myrtaceae were significantly greater than that of B. aquifolium, whereas it was intermediate for B. grandis (Proteaceae).

Fig. 3
figure 3

Leaf gas exchange, leaf mass per area, and photosynthetic phosphorus (P)- and nitrogen (N)-use efficiencies of Eucalyptus patens, yarri (Myrtaceae) and co-occurring species at Cypress Form, Waroona measured in January 2021. a Area-based maximum photosynthetic rate (Aarea); b leaf stomatal conductance (gsw); c leaf mass per area (LMA); d photosynthetic P-use efficiency (PPUE); and (e) photosynthetic N-use efficiency (PNUE). Data are means ± SE (n = 5–10). Different lowercase letters above the bars indicate significant differences among means, determined by analysis of variance (ANOVA) with Tukey’s HSD test at P ≤ 0.05 level

Full size image

Banksia grandis had the most negative δ13C value of −33.3‰ (Fig. 4a), and hence showed the lowest WUE calculated from δ13C values (11 μmol mol−1) (Fig. 4b); it also showed the lowest iWUE, calculated as Aarea /gsw (57 μmol mol−1) (Fig. 4c). The δ13C value of B. aquifolium was slightly less negative than that of B. grandis (−32.3‰), and hence B. aquifolium had a slightly higher WUE and iWUE (26 and 81 μmol mol−1, respectively). Eucalyptus patens saplings and regrowth plants had the least negative δ13C values of −29.9 and − 30.5‰, respectively, like that of C. calophylla (−30.6‰). The Myrtaceae had the highest WUE and iWUE (62 and 120 for E. patens saplings, 54 and 88 for E. patens regrowth and 52 and 82 μmol mol−1 for C. calophylla, respectively. The δ13C values of adult plants of E. patens (−31.1‰) and E. marginata (−31.3‰) were intermediate, and both had an intermediate WUE and iWUE (45 and 62; 41 and 64 μmol mol−1, respectively), among the species. For all species, WUE calculated from δ13C values was lower than iWUE calculated from Aarea/gsw. Importantly, all δ13C values, WUE and iWUE values were low when compared with global values (O’Leary 1993) or values for other eucalypts (Schulze et al. 2006), indicating a very low efficiency of water used of all studied plants in the present habitat.

Fig. 4
figure 4

Leaf carbon-isotope composition and water-use efficiency of Eucalyptus patens, yarri (Myrtaceae) and co-occurring species at Cypress Form, Waroona measured in January 2021. a Leaf carbon-isotope composition (δ13C); b water-use efficiency (WUE) calculated from δ13C values.; c intrinsic water-use efficiency (iWUE) calculated as Aarea/gsw. Boxplots show medians, and 25th and 75th percentiles. Whiskers extend to 1.5 times the interquartile range. Data presented beyond whiskers represent outliers. (n = 5–10). Different letters indicate statistically-significant differences among means, analysed with one-way ANOVA and Tukey’s HSD test at P ≤ 0.05 level

Full size image

Leaf anatomy

Eucalyptus patens, E. marginata, C. calophylla, and B. aquifolium showed dorsiventral leaf anatomy with a single layer of palisade mesophyll and abundant spongy mesophyll cells (Fig. 5a-d). They also had a thick cuticle on both the upper and lower epidermis. Like other studied species, Banksia grandis also showed dorsiventral leaf anatomy (Fig. 5e). Its lignified and thick-walled fibrous bundles of sclerenchyma cells (blue autofluorescence) divided the mesophyll tissue into small compartments. Despite their thickness, leaves of B. grandis did not exhibit deep stomatal crypts with sunken stomata and abundant long hairs originating from its lower epidermis, which are common in other thick-leaved Banksia species such as B. prionotes (Fig. 5f). Instead, B. grandis only had shallow stomatal crypts without sunken stomata (Fig. 5e).

Fig. 5
figure 5

Cross-sections of mature scleromorphic leaves of a range of species co-occurring with Eucalyptus patens (except Banksia prionotes, which was collected elsewhere and used to demonstrate the deep stomatal crypts and sunken stomata that are typical for thick-leaved Banksia species), viewed in darkfield with UV light using a UV filter. a Eucalyptus patens, b E. marginata, c Corymbia calophylla, d Bossiaea aquifolium, e Banksia grandis, f B. prionotes. UV-induced autofluorescence is visible in all micrographs, and different molecules and tissues exhibit different colours of autofluorescence. a-f Chlorophyll in the palisade and mesophyll parenchyma autofluoresced red, whereas lignified xylem vessels autofluoresced blue. a-d, f Blue/green and (e) purple/blue autofluorescence in the outer tangential walls of the upper and lower epidermis indicate the presence of thick cuticles. e, f Blue autofluorescence in the upper and lower epidermis of both Banksia species indicates lignified epidermal cell walls. Yellow arrow shows a shallow stomatal crypt in B. grandis. White arrows show deep stomal crypts (SC), and sunken stomata (SS) in B. prionotes. Scale bars: 200 μm (a-d) and 100 μm (e-f)

Full size image

Changes in rhizosphere pH and carboxylate release of plants grown at 1 μM P in nutrient solution

The addition of 0.1 M NaOH to the agar medium that contained a pH indicator, bromocresol purple, resulted in a notable colour change from yellow/orange (pH 5) to dark purple (pH 8) (Fig. 6a, b). To test the proton release from roots to the medium, the pH of the agar was adjusted to 6.5 (Fig. 6c, d). The strong colour changes from dark purple to bright yellow around the cluster roots (rhizosphere) of B. grandis indicated a strong acidification of the medium by cluster roots (Fig. 6c), which release carboxylates (Lambers et al. 2002). This colour change was mild for very young cluster roots (1–4 days old), whereas it was strong for mature cluster roots (13–15 days old). Eucalyptus patens roots also acidified the agar around them as indicated by the colour change (Fig. 6d). The mature zones of the branched lateral roots (MZBL) showed the strongest acidification of the medium, evidenced by the strongest colour change, followed by those around the young zone of the unbranched lateral roots (YZUL), and intermediate and younger zones of branched lateral roots (YZBL) with fine roots.

Fig. 6
figure 6

Root-induced changes in rhizosphere pH after ca. 10–20 min on agar with a pH indicator (bromocresol purple). a The reference pH values and associated colours in the agar medium. b Roots of Eucalyptus patens when the initial pH of the agar medium with bromocresol purple was 5.0, but it was changed to 6.5 by adding 0.1 M NaOH (c and d). c Roots of Banksia grandis, from left to right: a senesced cluster root (16–19 days old), a mature cluster root (13–15 days old), and a young cluster root (1–4 days old). d Roots of E. patens, from left to right: younger zone of an unbranched lateral root, mature basal zone of a branched lateral root, intermediate zone of a branched lateral root and the younger zone of a branched lateral root with numerous second-order laterals (fine roots). The colour changes from purple to yellow around the roots on the agar medium reflects acidification of the medium from the initial pH of 6.5 to less than 5.0. The intensity of the yellow colour around the roots reflects the level of acidification of the medium (strong acidification results in intense bright yellow)

Full size image

Different from the agar experiment, the younger zone of the branched lateral (YZBL) with fine roots of E. patens had the fastest rate of carboxylate release (0.13 nmol g−1 FM s−1), followed by the young zone of the unbranched lateral roots (YZUL; 0.08 nmol g−1 FM s−1) and mature zones of the branched lateral roots (MZBL; 0.04 nmol g−1 FM s−1) (Fig. 7a). Remarkably, the root carboxylate-exudation rate of E. patens was independent of the [P] in the nutrient solution, in which even 80 μM P treatment failed to reduce the exudation rate significantly from that at 1 μM P (F(3.12) = 0.96, p = 0.44), even though there was a slight reduction of carboxylate exudation with 80 μM Pi treatment (Fig. 7b).

Fig. 7
figure 7

Carboxylate-exudation rates of roots, and leaf phosphorus (P) concentrations ([P]) of Eucalyptus patens seedlings subjected to a gradual increase in [P] in the hydroponic medium. At each concentration, the seedlings were grown for two weeks, before measuring carboxylate release and leaf [P], followed by increasing [P] to the next level. a Carboxylate-exudation rates of different parts of the root system grown in nutrient solution with 1 μM P. YZBL: younger zone of branched lateral roots; MZBL: mature zone of branched lateral roots; YZUL: younger zone of unbranched lateral roots. b Carboxylate-exudation rates of branched lateral roots with fine roots, grown in different P treatments. c Phosphorus concentration in the mature leaves of E. patens seedlings subjected to a gradual increase in [P] in the nutrient solution. Data are means ± SE (n = 5). Different letters indicate significant differences among means, analysed with one-way ANOVA (a), repeated measures of ANOVA (b) and (c), Tukey’s HSD test (a, b and c) at P ≤ 0.05

Full size image

Changes in total leaf P concentration with different P treatments

There was no P left in the hydroponic medium three days after replacing the nutrient solution of the 1 μM P treatment, whereas more than 0.5 μM P was left in the medium three days after replacing the nutrient solution with 20 μM P. In the 40 and 80 μM P treatments, even more P was left in the hydroponic medium. Total leaf [P] increased with increasing [P] in the nutrient solution (F(3.12) = 9.27, p = 0.002) (Fig. 7c). In the 1 μM P treatment, total leaf [P] was approximately 0.9 mg g−1 DM, increasing to 2.8 mg g−1 DM in the 80 μM P treatment, a 3.1-fold change in leaf [P] and an 80-fold change for the medium [P]. We did not find any P-toxicity symptoms (such as chlorotic and necrotic areas on leaves) of E. patens; this absence of any toxicity symptoms corresponded to toxicity class 0 (Kariman et al. 2014). All the plants had green and healthy leaves, without typical P-toxicity symptoms as found in previous studies (Kariman et al. 2014; de Campos et al. 2013; Shane et al. 2003), and maintained normal growth, even with higher P levels (Supplementary Fig. S1).

Discussion

The present study builds on recent findings that dominant mycorrhizal overstorey and understorey species in a tall eucalypt forest in southwest Australia depend on carboxylate release to acquire P (Lambers et al. 2021; Zhou et al. 2022). However, unlike cluster-root-producing Fabaceae (Keerthisinghe et al. 1998) and Proteaceae (Shane et al. 2003), E. patens did not show a decrease in carboxylate release from its roots with increasing P supply and leaf [P] (Fig. 7b, c), indicating that carboxylate release was not dependent on the [P] in the medium or leaves. Being mycorrhizal, E. patens would be protected against pathogens (Albornoz et al. 2017; Cordier et al. 1998; Marx 1972; Pozo and Azcón-Aguilar 2007; Pozo et al. 2002), and it combines this with an effective carboxylate-releasing P-mobilising P-acquisition strategy. This implies that its P acquisition does not need to be facilitated by carboxylate-releasing P-mobilising neighbours (Lambers et al. 2018). Even if plants grow next to a carboxylate-releasing neighbour, this does not guarantee that their P uptake is facilitated by that neighbour (Yu et al. 2021; Yu et al. 2023). It takes two to tango (Yu et al. 2021). It requires the facilitated species to sense where P is mobilised and proliferate its roots accordingly (de Britto et al. 2021). With their effective P-acquisition strategy in a severely P-impoverished habitat, carboxylate release must give the P-mobilising species a competitive advantage, as has been shown for Proteaceae (Lambers 2022). The key difference, however, is that mycorrhizal Myrtaceae are protected against oomycete pathogens, whereas non-mycorrhizal Proteaceae are highly susceptible to these native pathogens (Albornoz et al. 2017). Therefore, these P-mobilising Myrtaceae combine the best of both worlds and are expected to be strong competitors, also against N2-fixing species that lack a carboxylate-releasing P-acquisition strategy, such as Acacia drummondii and Mirbelia dilatata (Zhou et al. 2022). Below, we explore if this trait might make E. patens and other carboxylate-releasing eucalypts ‘Darwinian demons’ (sensu Law 1979).

Carboxylate and proton exudation

To assess the effect of external P supply and carboxylate release of E. patens, we gradually increased the P supply in the nutrient solution and measured the root carboxylate-exudation rate after each increase. Interestingly, E. patens did not respond as is known for Fabaceae (Keerthisinghe et al. 1998) and Proteaceae (Shane et al. 2003), because it did not show a significant decrease in carboxylate release with increasing P supply and increasing leaf [P] (from 0.9–2.8 mg g−1), similar to some other Eucalyptus species studied by de Andrade et al. (2022). The absence of an effect of leaf [P] on the rate of carboxylate release is inconsistent with that in species producing cluster roots (Shane and Lambers 2005) or dauciform roots (Shane et al. 2006). However, this absence of a response to leaf [P] has also been noted for other species without specialised root structures (He et al. 2021; Huang et al. 2017; Wouterlood et al. 2004). This suggests that carboxylate release in E. patens may function to dispose of surplus carbon (Prescott 2022; Prescott et al. 2020), rather than releasing carboxylates in response to P limitation (Fig. 7c). However, independent of their control, the root exudates will mobilise sorbed P. The proton exudation of E. patens in agar suggests that the mature zones of branched lateral roots (MZBL) released more protons than the younger zone of branched laterals (YZBL) with fine roots (Fig. 6d), even though, the mature roots did not exhibit the fastest carboxylate-exudation rate in nutrient solution; that rate was faster in the younger roots (Fig. 7a). Further research is needed to test if exudate release from mature main roots of E. patens facilitates the P uptake of branched laterals with fine roots for the entire root system to acquire P efficiently, as shown for Oryza sativa (rice) (Kuppe et al. 2022).

Leaf P concentrations, photosynthesis, and photosynthetic P-use efficiency

The leaf [P] of B. grandis (Fig. 2a) was very low in this low-P habitat (Zhou et al. 2022), as recorded before (Barrick 2003), and other populations of this species (Denton et al. 2007; Sulpice et al. 2014; Guilherme Pereira et al. 2019). Leaf [N] of B. grandis (Fig. 2b) was also similar to that recorded for other B. grandis populations (Barrick 2003), and other Proteaceae in southwest Australia (Sulpice et al. 2014; Guilherme Pereira et al. 2019). In accordance, the leaf N:P ratio of B. grandis (45.5) was similar to that of three other Banksia species (Sulpice et al. 2014). Banksia grandis in the ‘cool spot’ showed relatively slow photosynthetic rates (Aarea) compared with other Banksia species that have stomatal crypts in similar low-P habitats (Guilherme Pereira et al. 2019; Sulpice et al. 2014; Denton et al. 2007). Even though B. grandis has thick leaves, it did not have deep stomatal crypts and sunken stomata as found for B. prionotes (Fig. 5e vs. f) and other Banksia species with thick leaves (Hassiotou et al. 2009; Sulpice et al. 2014). The absence of deep stomatal crypts likely accounts for its relatively slow rates of photosynthesis, as observed in similarly thick-leaved Hakea species without stomatal crypts (Lambers et al. 2012). Thick leaves without deep stomatal crypts imply a longer path for CO2 diffusion from the air to the chloroplasts, thus restricting photosynthesis (Roth-Nebelsick et al. 2009).

The leaf PPUE of B. grandis was similar to that reported before for three other Banksia species (Denton et al. 2007). The leaf [P] and [N] of E. patens, C. calophylla and E. marginata were also similar to or somewhat lower than published data for these species (Grigg et al. 2009; Qiu et al. 2013). In accordance, the present N:P ratios for C. calophylla and E. marginata were similar to published data (Grigg et al. 2009); they were all >20, which indicates that their productivity was limited by P, rather than N (Güsewell 2004). There is no published record of the Aarea of E. patens, but the present results of Aarea of C. calophylla and E. marginata (Fig. 3a) were similar to or somewhat higher than published values (Grigg et al. 2009). The Aarea of saplings and regrowth of E. patens were greater than that of adult E. patens (Fig. 3a). We did not find data on leaf [N] and [P] in the literature for B. aquifolium, but the present values were similar to published values for B. laidlawiana (renamed to B. aquifolium subsp. laidlawiana) in karri forest of southwest Australia (Grove 1990). The Aarea of B. aquifolium was lower than that of the other four species at our study site, associated with a lower PPUE (Fig. 3d). The generally low leaf [N] and N:P values for the present species that do not fix N2 agree with the low N availability at the E. patens site. This is possibly due to strong competition for P by P-mobilising eucalypts with N2-fixing species in the study system, that are far less common and do not release carboxylates, unlike B. aquifolium, which is far more common and releases carboxylates (Zhou et al. 2022). In addition, there may be substantial leaching and/or denitrification associated with the high rainfall at this site which is typical for E. patens habitats, whereas C. calophylla and E. marginata also occur in low-rainfall habitats.

Leaf N concentrations and the trade-off between photosynthetic water-use efficiency and photosynthetic nitrogen-use efficiency

A low leaf [N] implies a low photosynthetic activity (Evans 1983; Field et al. 1983); this low activity can be partially compensated by a higher stomatal conductance (van den Boogaard et al. 1995; Wright et al. 2003). Plants tend to reduce stomatal conductance under water stress so that iWUE is maximised at the expense of PNUE (Limousin et al. 2015). Conversely, under limited availability of N, stomatal conductance may increase, increasing PNUE at the expense of iWUE (Renninger et al. 2015). This implies a trade-off between efficient use of either water or N (Fredeen et al. 1991; Limousin et al. 2015; van den Boogaard et al. 1995; Wright and Westoby 2002).

In the ‘cool spot’ ecosystem where P is severely limiting (Lambers et al. 2021; Zhou et al. 2022), we found that species belonging to three families exhibited different strategies for resource-use efficiency. Banksia grandis (Proteaceae), E. patens, C. calophylla and E. marginata (Myrtaceae) had similarly low leaf [N], restricting their photosynthetic activity, but the relatively high availability of water allowed them to partially compensate for their low photosynthetic activity by a high stomatal conductance. This suggests resource substitution in these species with limited access to N, maximising PNUE at the expense of iWUE, in agreement with previous studies (Fredeen et al. 1991; Nolan et al. 2017; Renninger et al. 2015; van den Boogaard et al. 1995). Conversely, B. aquifolium (Fabaceae) strongly depends on symbiotic N2 fixation (Zhou et al. 2022) and functioned at a much higher leaf [N] (Fig. 2b). Despite this, it exhibited a low photosynthetic activity (Fig. 3a) and surprisingly low iWUE (Fig. 4c). This may account for the restriction of this species to high-rainfall regions (Hansen et al. 1991).

Can we consider E. patens to be a ‘Darwinian demon’?

Law (1979) used the term ‘Darwinian demons’ for species that can maximise all aspects of fitness simultaneously but acknowledged there are costs associated with specific strategies. These are generally referred to as trade-offs, preventing the evolution of Darwinian demons (Pfautsch et al. 2018; Rees 1993). In biodiverse kwongan vegetation in the Southwest Biodiversity Hotspot, efficient acquisition of P in non-mycorrhizal Proteaceae is traded off against susceptibility to oomycete pathogens (Albornoz et al. 2017; Lambers et al. 2018). Some of the tall eucalypts we studied appear to combine the best of both worlds, because they exhibit a carboxylate-releasing P-mining strategy (Lambers et al. 2021; Zhou et al. 2022), whilst their association with mycorrhizal fungi would provide protection against native oomycete pathogens (Albornoz et al. 2017; Cordier et al. 1998; Marx 1972; Pozo and Azcón-Aguilar 2007; Pozo et al. 2002). We surmise that this accounts for the low plant diversity in ‘cool spots’, but does it make these tall eucalypts Darwinian demons?

So far, we have only provided evidence for a combination of efficient P acquisition and pathogen defence in eucalypts from high-rainfall areas and found that this combination does likely not exist in E. todtiana, which grows in low-P habitats with less rainfall and exhibits a low leaf [Mn] (K. Ranathunge & H. Lambers, unpubl.). We use leaf [Mn] as a proxy for carboxylate-releasing P-acquisition strategies, based on the correlation between leaf [Mn] in the field and root carboxylate release in glasshouse studies (Lambers et al. 2021). We surmise that Eucalyptus species with high leaf [Mn] that release carboxylates are very strong competitors, leading to low-diversity habitats (Zhou et al. 2022), unlike the habitats where Myrtaceae depend on facilitation by carboxylate-releasing neighbours (Muler et al. 2014; Lambers et al. 2019). The low leaf [N] of the non-N2-fixing species in the studied low-diversity environment was associated with a high stomatal conductance, as evidenced by low leaf δ13C values (Fig. 4a). For B. grandis, the δ13C values were slightly more negative (−33.2‰) than those for expanding leaves of B. prionotes during the wet growing season (−29‰ to −31‰), and considerably more negative than those of the same leaves later in the season, when they gradually increase to −26‰ while rainfall declines (Pate et al. 1998). This may account for B. grandis being restricted to habitats with greater availability of water than that of B. prionotes (https://florabase.dpaw.wa.gov.au/browse/profile/21316). For the eucalypts (C. calophylla, E. marginata, E. patens) the δ13C values ranged from −30.7‰ to −31.3‰ (Fig. 4a). For 65 Eucalyptus species growing along rainfall gradients in southwest and central Western Australia, species-specific leaf δ13C values decrease with increasing annual rainfall from −27.0‰ at 300 mm to −28.2‰ at 1000 mm (Schulze et al. 2006). For E. marginata at Wungong Catchment, in the same region as Waroona, leaf δ13C values vary between −29‰ and − 31‰ (Qiu et al. 2013). These comparisons show similar or more negative δ13C values, suggesting greater stomatal conductance in the present tree species. The high stomatal conductance of the study species (Szota et al. 2011), allowed the leaves to function at a high internal CO2 concentration (around 280 μmol mol−1) to compensate for a low [N] (van den Boogaard et al. 1997; Warren and Adams 2006; Wright et al. 2001). In accordance, for E. marginata, a positive correlation between leaf [N] and leaf δ13C was found at Wungong Catchment (Qiu et al. 2013). This shows a trade-off between efficient use of either water or N in photosynthesis.

In summary, the combination of traits involving pathogen defence, efficient P acquisition and efficient N use in plants was likely traded off against the water-use efficiency of the studied species. This nutrient-efficient strategy comes at the expense of profligate water use, restricting the nutrient-use efficient but water-use inefficient species to habitats with an abundant water availability but low nutrient availability throughout the year. Therefore, none of the eucalypts we have studied can be considered Darwinian demons, and they may be disadvantaged in a drying climate.

Concluding remarks

The present results provide further insights into a low-diversity ‘cool spot’ in a biodiversity hotspot. They indicate that E. patens in southwest Australia, and possibly other eucalypts elsewhere in Australia where they occur on P-impoverished sites and exhibit high leaf [Mn] (Lambers et al. 2021), function differently from what we might believe, based on mycorrhizal surveys (Albornoz et al. 2021; Brundrett and Abbott 1991). Surprisingly, carboxylate exudation in E. patens was not suppressed with increasing [P], and hence might function to dispose of surplus carbon (Prescott 2022; Prescott et al. 2020). Yet, the exuded carboxylates would mobilise sorbed P.

The combination of being mycorrhizal and releasing carboxylates would make E. patens a very strong competitor, also against N2-fixing neighbours that do not release carboxylates. This likely accounts for the low diversity at the study site (Zhou et al. 2022), and also for the relatively low leaf [N] of E. patens. This low leaf [N] implies a low photosynthetic activity, but this was compensated for by a high stomatal conductance, and thus low iWUE, but high PNUE (Wright et al. 2003; Wright and Westoby 2002). This trade-off would only be feasible in a high-rainfall region or in discharge zones in the landscape, and not outside these areas. Hence, E. patens is not a Darwinian demon that combines every single trait to be competitive everywhere.