Introduction

Plants need a constant supply of energy and a suite of at least 17 chemical elements in adequate proportions to carry out their metabolic functions (Sterner and Elser 2002; Schlesinger and others 2011; Kaspari and Powers 2016). These elements, excluding carbon (C), are obtained mainly from the soil. These soil elements include essential macro- and micronutrients such as nitrogen (N), potassium (K), sulfur (S), calcium (Ca), magnesium (Mg), iron (Fe), copper (Cu), zinc (Zn), and manganese (Mn) (Chapin 1980; Kabata-Pendias 2001). The bioavailability of these nutrients in soil is influenced by: (i) the mineralogical composition of the parent material; (ii) soil properties such as pH, redox potential, organic matter content, and texture; (iii) physicochemical interactions among chemical elements, including chelation, immobilization, and sorption/desorption processes; (iv) geophysical processes such as erosion and aeolian transport; (v) biotic mobilization (for example, decomposition, mineralization, and solubilization) and immobilization (for example, uptake) of elements at characteristic stoichiometric rates by organisms; and (vi) biotic redistribution (Sterner and Elser 2002; Schlesinger and Bernhardt 2020). The magnitude and rates of key ecosystem processes such as net primary production or soil C storage can, thus, be explained in relation to biogeochemical processes related to supply, stocks, and fluxes of chemical elements, as well as by alterations in stoichiometric relationships within the soil–microbe–plant–animal system (Sterner and Elser 2002; Schlesinger and Bernhardt 2020).

The global C, water, and nutrient cycles have been widely altered by human activities (Gruber and Galloway 2008; Schlesinger and Bernhardt 2020). For example, the concentration of CO2 in the atmosphere has increased from 250 parts per million (ppm) in pre-industrial times to over 400 ppm today and is predicted to reach 450–550 ppm by 2050 (Vargas Zeppetello and others 2022). This increase in CO2 in the atmosphere can have direct physiological effects on plants such as greater photosynthesis (often referred to as fertilization effect), both at the leaf level and at the global scale (Ainsworth and Rogers 2007; Leakey and others 2009; Walker and others 2021). Elevated CO2 can also have indirect effects on plants by altering the global climate which, in turn, can affect ecosystem nutrient cycling (de Graaff and others 2006). Moreover, these direct and indirect impacts of CO2 can have cascading effects on the productivity and composition of terrestrial ecosystems and, therefore, on the global C cycle and its role in climate regulation (de Graaff and others 2006). However, it is still unclear how combinations of global change drivers such as elevated (eCO2) and altered precipitation regimes acting simultaneously may drive the structure and functioning of terrestrial ecosystems. The role of changes in precipitation in combination with other global change drivers may be particularly relevant given the importance of water availability as a modulator of the response of ecosystems to increases in resource availability, including CO2 and nutrients (Piñeiro and others 2020).

The effects of eCO2 on photosynthesis induce, and are affected by, changes in plant chemistry and stoichiometry (Peñuelas and Estiarte 1998; Dong and others 2018; Wang and others 2021). These changes typically include greater C/nutrient ratios of plant tissues, particularly, but not limited to, greater C/N, due to the greater availability of C in relation to other elements, resulting in nutrient dilution (Loladze 2002; Sardans and others 2012; Du and others 2019; Welti and others 2020; Wang and others 2021). This can have implications for decomposition rates of plant material and organic matter mineralization, which are generally reduced when C/N/nutrient ratios are high (Wu and others 2020). This increase in C/N/nutrient ratios under eCO2 can also favor the production of fine roots, whose response to eCO2 may be additionally conditioned by the availability of soil resources, including N and water (Piñeiro and others 2017, 2020). Moreover, eCO2 can also alter the bioavailability of elemental nutrients in soil, at least in the short term. For example, a free-air CO2 experiment (FACE) carried out in a native Australian eucalypt woodland (EucFACE) found that higher CO2 concentrations increased plant-accessible soil N (+ 93%), NH4+ (+ 12%), and P (+ 54%), relative to ambient plots, during the first 18 months of CO2 fumigation (Hasegawa and others 2016; Ochoa-Hueso and others 2017). This response was attributed to changes in microbial enzyme activity in the rhizosphere, which may allow plants to take advantage of the higher concentrations of atmospheric CO2 (Finzi and others 2006; Ochoa-Hueso and others 2017; Piñeiro and others 2022). In contrast, Ochoa-Hueso and others (2019) showed that soil nutrient bioavailability did not clearly change in response to 18–30 months of fumigation, but instead the biogeochemical cycles of macro- and micronutrients became decoupled. Elemental decouplings, based on the strength of correlations among elements in absolute value, were particularly associated with K+, Mg2+, and Ca2+, because these are relatively mobile ions and are also limiting elements for plant growth in the low-fertility forest where the experiment took place (Crous and others 2015; Ochoa-Hueso and others 2019). However, whether eCO2 can contribute to a decoupling of plant chemical elements has not previously been evaluated, despite its potential role for plant growth and metabolism due to the importance of maintaining tight stoichiometric associations across a broad range of elements within plant tissues (Kaspari 2012; Tian and others 2019).

Most FACE studies have been carried out in Europe or the USA on young forest plantations where N is typically the main limiting nutrient for plant growth (Finzi and others 2007; Liberloo and others 2007; Norby and others 2010). The consensus is that increases in CO2 are usually beneficial to plants (that is, it enhances photosynthesis and growth) as long as they have an extra supply of water and/or N from the soil (Terrer and others 2019). In contrast, relatively little is known about the effect of P availability as a modulator of the response of terrestrial ecosystems to eCO2 (Fleischer and others 2019; Terrer and others 2019; Jiang and others 2020a, b). And we know even less about the role of other elements that are equally important for plant metabolism such as K, S, Ca, or trace elements (Sardans and Peñuelas 2015). Phosphorus, an essential macronutrient in the control of key physiological processes such as those related to energy metabolism, cell regulation, and protein formation (Elser 2012), is particularly limiting in the old soils of the southern hemisphere and in the tropics (Lambers and others 2011; Crous and others 2015; Cunha and others 2022). In this context, the availability of essential nutrients such as P is expected to become an increasingly important modulator of the response of plant communities to rising atmospheric CO2 concentrations (Fleischer and others 2019; Terrer and others 2019; Jiang and others 2020a, b; Cunha and others 2022), a concept that has not been extensively tested.

In this study, we evaluated the effect of an increase in the atmospheric concentration of CO2 (400 ppm vs. 650 ppm) on the concentration of 14 chemical elements that are plant nutrients (C, N, P, K, Mg, Ca, sodium [Na], S, Mn, Fe, Cu, Zn, silicon [Si], and chlorine [Cl]) and their degree of elemental coupling (Ochoa-Hueso and others 2021b) in 64 experimental plant communities grown in pots under controlled greenhouse conditions. Communities were grown from a soil seed bank collected from a Eucalyptus forest in southeastern Australia. Our definition of coupling followed the framework of Ochoa-Hueso and others (2021b), who defined it as the multiple ways in which the components of ecosystems, in this case chemical elements, are orderly connected across space and/or time. Operationally, plant elemental coupling was evaluated as the degree of association among elements based on Spearman correlations in absolute value (Ochoa-Hueso and others 2021b). In addition, we studied the modulating role of P and water availability on the effects of eCO2 on tissue chemistry and on the degree of elemental coupling.

We tested the hypotheses that (H1) altering plant resources will lead to altered plant tissue chemistry and (H2) differential uptake rates, altered demand, and nutritional imbalances will, in turn, result in changes in the coupling of chemical elements in plant tissues. We specifically predicted that (H3) the bioavailability of P will be the most important factor controlling tissue chemistry, and (H4) plant element coupling due to the low P nature of local soils (Ochoa-Hueso and others 2021a) and previous results from this study showing that P was the most important factor controlling biomass production (Piñeiro and others 2023). The effects of P are predicted to operate directly due to the role of P as limiting nutrient and also through interactions with other nutrients. For example, it has been shown that increases in P can favor the absorption of other nutrients such as Mg, Ca, and Mn (Lynch 2011), a phenomenon that has also been termed as N-P catalysis (Kaspari and others 2021). In turn, we predicted that (H5) water may also be particularly important to modulate the effects of eCO2 on plant nutrient acquisition, allowing the incorporation of greater concentrations of elements into plant tissues (Cai and others 2017). We also predicted that (H6) eCO2 will result in an overall elemental decoupling of plant tissues due to the inability of plants to extract all essential elements in adequate proportions to meet their increased nutritional demands in these poor-nutrient soils, particularly when combined with lower supply of the other two resources (that is, P and water). Finally, we predicted that (H7) the coupling of individual elements in plant tissues would be linked to properties of those elements such as atomic mass, valency, or ionization energy, which have been previously shown to control the coupling of soil elements across global topsoils (Ochoa-Hueso and others 2021c). These atomic properties are relevant because they control the mobility and reactivity of elements and, thus, their ability to interact with each other in the form of highly ordered living systems.

Materials and Methods

Experimental Design

To carry out the experiment, we collected two types of native soil from a Eucalyptus forest located in the vicinity of the Hawkesbury Institute for the Environment (Western Sydney University, Richmond, NSW, Australia) (33° 36′59″ S, 150° 44′17″ E). This forest is located near the EucFACE experimental site (Jiang and others 2020b). The climate is characterized by an annual average rainfall of approximately 800 mm and the local soil, of the Claredon formation, is nutrient-poor (Crous and others 2015). The undergrowth of this forest is highly diverse, with at least 70 different species, among which we find C3 and C4 grasses such as Microlaena stipoides and Eragrostis curvula or shrubs such as Breynia oblongifolia (Crous and others 2015; Hasegawa and others 2018; Ochoa-Hueso and others 2021a).

First, we collected approximately 200 kg of mineral soil (3 kg per pot) from the nearby Eucalyptus forest. This soil was collected from a depth below 2 cm and was used as mineral substrate for the pots. Secondly, we collected seed bank soil adjacent to the experimental rings of the EucFACE experiment; soils were collected from a depth of 0–2 cm and subsequently homogenized. This seed bank soil was placed on top of the mineral substrate in all 64 pots. Pots were transferred to the greenhouse and kept under controlled environmental conditions, allowing the development of the plant community from the native seed bank. This means that each pot developed its own community, introducing some level of variability in our experimental setup. The main advantage of this approach is that it allowed the community to develop naturally, establishing different types of interactions between the microbial community and the native plants in this Australian soil, while closely monitoring and controlling the growing environment (Ochoa-Hueso and Manrique 2010; Ochoa-Hueso and others 2021a).

The eight experimental treatments were a combination of three factors, each with two levels: (i) ambient CO2 [400 ppm] and eCO2 [650 ppm]. For this, pots were randomly distributed among four chambers with different concentrations of CO2, two under ambient CO2 and two under eCO2. On irrigation days (2–3 times per week, see below), the pots were swapped among chambers with the same CO2 treatment. (ii) High water availability (target of 90% water holding capacity on irrigation days, simulating SWC during a wet spring), and lower water availability (herein after referred to as control; target of 65% water holding capacity on irrigation days, simulating SWC in an average growing season). To control the availability of water, pots were initially weighed. These pots were weighed again every 2–3 days to determine their water deficit and then re-watered to target levels. (3) High P availability (+ 30 mg kg soil−1) and low P availability (no P added). Phosphorus was added only at the beginning of the experiment, mixed with the soil as crushed triple superphosphate (Ca(H2PO4)2 * H2O) (Ochoa-Hueso and others, 2021a).

Harvest

After 4 months of experiment, pots were harvested. Our experimental communities comprised a total of 38 species, with graminoids being the dominant functional group (average cover across all treatments of 60–70% by the end of the experiment) (Ochoa-Hueso and others 2021a). Microlaena stipoides represented 35% of total biomass and 52% of grass biomass across all pots by the end of the experiment, while the most abundant forb species were Commelina cyanea, Gamochaeta pensylvanica, Hydrocotyle sibthorpioides, Hypochaeris radicata, Oxalis perennans, Poranthera microphylla, Viola betonicifolia, and Wahlenbergia gracilis (Ochoa-Hueso and others 2021a). In previous studies, we found that P addition increased total plant cover and biomass (Ochoa-Hueso and others 2021a; Piñeiro and others 2023). In contrast, plant cover was not affected by eCO2 and water addition (Ochoa-Hueso and others 2021a), while eCO2 increased aboveground biomass by 10% (Piñeiro and others 2023). In addition, water, phosphorus, and, to a lesser extend, eCO2 shifted the composition of plant communities in our pots (for a more detailed description of these results, see Ochoa-Hueso and others 2021a). To measure plant chemical composition, the aerial part of the plant communities was cut at the ground level. Plant samples were oven-dried at 65 ºC, weighed, and then ball-milled to fine powder.

Macro- and Micronutrients in Plant Tissues

The total concentrations of C and N in aboveground plant tissues were measured using a CHN elemental analyzer (Costech Analytical Technologies, Inc., Valencia, CA, USA). The concentrations of the other macro- (P, S, Mg, K and Ca) and micronutrients (Na, Mn, Fe, Cu, Zn, Cl, Si) were measured by X-ray fluorescence (XRF; Epsilon 3XLE, PANalytical BV, Almelo, Overijssel, The Netherlands).

Numerical Calculations and Statistical Analysis

Statistical analyses were performed using R v3.6.0 (R Core Team 2017). We used linear models to evaluate the effects of CO2 and the availability of P and water on the concentration of elements within the plant communities, for which we used the ‘lm’ function from the stats package. The coupling of the elements in the plant tissues for each of the eight experimental treatments was calculated based on the correlation of the different elements in absolute value (14 elements, n = 8 replicates per treatment), using the Spearman correlation coefficient (Ochoa‐Hueso and others 2021). Coupling was then compared against a randomly generated null model based on 999 permutations of the same dataset. Within this framework, plant element coupling can be described as being in one of three coupling states: (i) coupled (above the two-tailed 95% null model envelope), implying that the mean of system-level correlations is greater than what is expected by chance; (ii) decoupled (within the two-tailed 95% null model envelope); and (iii) anticoupled (below the two-tailed 95% null model envelope) (Ochoa-Hueso and others 2021b). Anticoupling means that the mean of system-level correlations is lower than that obtained from a random dataset. Importantly, this property only emerges after considering correlations in absolute value, which is a unique feature of this coupling approach. To evaluate the effects of treatments on coupling, we also used linear models. In this case, pairwise interactions within treatments were considered as replicates. We also analyzed the degree of elemental coupling and coupling states of each of the elements evaluated separately, focusing on correlations involving those specific elements. Finally, we evaluated the relationships between the coupling of individual elements and atomic properties like mass, the abundance of elements in the crust reported in Yaroshevsky (2006), and the coupling of elements in global topsoils reported in Ochoa-Hueso and others (2021b).

Results

Chemical Variation and Elemental Composition of Plants

Consistent with our hypotheses (H1 and H3), altered resource availability altered tissue chemistry, with P addition having the largest impact on plant tissue chemistry, increasing the concentrations of P (+ 543%), Mg (+ 38%), Ca (+ 44%), and Mn (+ 58%) and reducing the concentrations of C (− 2%), N (− 36%), S (− 34%), Na (− 33%), and Cu (− 9%) (Figures. 1 and 2, Table 1). Greater water availability resulted in increased concentrations of P (+ 41%), Ca (+ 19%), and S (+ 18%) and lower concentrations of C (− 1%) (Figure. 1, Table 1). Elevated CO2 resulted in lower concentrations of Zn (− 11%) and N (− 6%) but marginally greater P (+ 12%) (Figures. 1 and 2, Table 1). The effects of eCO2 on Na and, to a lesser extent, on Cu and Mn (marginally significant effects) were dependent on P levels, while the effects of eCO2 on S, Cu, and Zn were dependent on water levels (Figures. 1 and 2, Table 1), as evidenced by significant interactions.

Figure 1
figure 1

Effects of CO2, water, and P on primary and secondary macronutrients. aCO2 = ambient CO2; eCO2 = elevated CO2. Error bars are SE.

Full size image
Figure 2
figure 2

Effects of CO2, water, and P on micronutrients. aCO2 = ambient CO2; eCO2 = elevated CO2. Error bars are SE.

Full size image
Table 1 Effects of CO2, Water, and P on Plant Element Content
Full size table

Plant Element Coupling

As predicted (H2), altering resource availability resulted in changes in coupling. Six of the eight treatment combinations had stronger elemental coupling than what would be expected by chance and two were decoupled, while none of the experimental treatments resulted in an anticoupled pattern (Figure. 3). All four communities under aCO2 showed coupled elements, while two communities under eCO2 became decoupled. We also found a significant interaction between CO2 and P and between water and P, on plant element coupling (Figure. 3, Table 2). In contrast to our hypothesis (H4) and the results of tissue chemistry, adding P was not the single factor that most affected coupling; instead, the role of P on coupling operated via interactions with water and CO2. For example, the most elementally coupled communities were found under eCO2, high P, and low water conditions. In contrast, the weakest elemental coupling was found under eCO2, low P, and low water and under aCO2, low P, and high water conditions (Figure. 3). Elevated CO2 reduced the coupling of plant P (P = 0.06), particularly under high water conditions, while the coupling of Cu increased in the high water treatment (P = 0.09) (Figures. 4 and 5, Table 2). Overall, plant elements were coupled in 82 out of 112 instances, decoupled in 23 instances, and anticoupled in 7 instances (Figures. 4 and 5). Anticouplings involved six elements, including P, Ca, Cu, Zn, Mn, and Cl (Figures. 4 and 5).

Figure 3
figure 3

Effects of CO2, water, and P on plant element coupling. aCO2 = ambient CO2; eCO2 = elevated CO2. Black dots indicate that coupling (that is, the mean of pairwise correlations) is significantly different from the null model (P < 0.05), while white dots indicate no significant difference. Error bars represent the 95% CI of all possible pairwise correlations. Gray dots in the background represent single pairwise correlations.

Full size image
Table 2 Effects of CO2, Water, and P on the Coupling of Each Plant Element Separately
Full size table
Figure 4
figure 4

Effects of CO2, water, and P on coupling for primary and secondary macronutrients. aCO2 = ambient CO2; eCO2 = elevated CO2. Black dots indicate that coupling (that is, the mean of pairwise correlations) is significantly different from the null model (P < 0.05), while white dots indicate no significant difference. Error bars represent the 95% CI of all possible pairwise correlations. Gray dots in the background represent single pairwise correlations.

Full size image
Figure 5
figure 5

Effects of CO2, water, and P on coupling for micronutrients. aCO2 = ambient CO2; eCO2 = elevated CO2. Black dots indicate that coupling (that is, the mean of pairwise correlations) is significantly different from the null model (P < 0.05), while white dots indicate no significant difference. Error bars represent the 95% CI of all possible pairwise correlations. Gray dots in the background represent single pairwise correlations.

Full size image

Averaged across all treatments, K, Ca, and Fe were the most coupled elements in plant tissues, while P, Mn, and N were the least coupled ones (Figure. 6). Elemental coupling involving C represented the median of all elements (Figure. 4). The coupling of individual elements was positively related to the abundance of those elements in the Earth’s crust (R2 = 0.23; P value = 0.05; Figure. 6) and also to the coupling of those elements in global topsoils (R2 = 0.44; P value < 0.01; Figure. 6) while, opposite to our predictions, we did not find any evidence of a relationship between plant element coupling and atomic properties (P value > 0.05 in all cases).

Figure 6
figure 6

Coupling of all elements measured in plant tissue across experimental treatments ordered from least to most coupled (a), and in relation to the abundance and coupling of such elements in the Earth’s crust (b) and in topsoils (c). Black lines represent the fitted regression line and the shaded area is the SE of the fitted line.

Full size image

Discussion

In a P-impoverished soil, a seed bank grown over four months under greenhouse conditions revealed that soil P availability was the most important resource controlling plant biomass production (Piñeiro and others 2023), elemental composition of plant tissues, and the associated coupling of chemical elements. The effects of P on elemental coupling were particularly evident under eCO2 and varied depending on water availability. Moreover, the most significant effects of eCO2 only occurred under high P, which is consistent with the previously reported role of soil P as a determinant of the response of mature eucalypt forest to increases in CO2 concentrations (Jiang and others 2020a, 2020b; Ochoa-Hueso and others 2021a; Piñeiro and others 2023). This low P availability, and thus low responsiveness to eCO2. has been widely associated with the ancient nature of Australian soils, which are highly depleted in mineral nutrients (Orians and Milewski 2007; Kooyman and others 2017), thus reinforcing the role of the biogeochemical cycling of P in a changing environment (Jiang and others 2020a).

Elemental Variations

Phosphorus addition was the main factor affecting plant elemental content, increasing the concentrations of P, Mg, Ca, and Mn and reducing the concentrations of C, N, S, Na, and Cu, while the effects of water and CO2 were minor compared to those of adding P. This is similar to the findings of (Reinbott and Blevins 1994) in wheat. Several biochemical mechanisms may play a role in explaining this pattern. The synergies between P and Mg can be associated with the potential role of Mg as a carrier of P (Adams 2015) and the role of Mg in photosynthesis (Levitt 1954). Calcium, in turn, is responsible for capturing P and forming inorganic deposits of this element in the form of calcium phosphate (apatite) inside the mitochondria, suggesting that the increase of soil P may be accompanied by an increase of Ca (Raven 2023). However, triple superphosphate, the form in which P was added to this experiment, contains Ca, thus simultaneously increasing the bioavailability of this element. This makes it impossible to discern the operating mechanisms of the effects of P on Ca. Manganese is another element linked to P through the photosynthetic process, being a cation that activates enzymes such as phosphotransferase, dehydrogenases, or phosphomutases (Burnell 1988). Manganese is also involved, together with Fe, in the synthesis of chlorophyll and in photosynthetic oxidation–reduction systems, thereby contributing to higher photosynthetic rates (Burnell 1988). Consistent with our results, previous investigations have shown that P and Mn are often taken up together and are strongly correlated in leaves (Lambers and others 2015). The lower concentration of other elements such as C, N, S, Na and Cu in response to P addition could, in turn, be attributed to both dilution effects and reduced uptake rates. Importantly, the significant increase in leaf P content in response to eCO2 likely implies that more soil P is being mobilized from immobile, mineral sources, which may be attributed to enhanced rhizosphere priming (Dijkstra and others 2013).

Decoupling of Biogeochemical Cycles

As a general rule, most treatment combinations showed coupled patterns of plant chemical elements, particularly under aCO2, while eCO2 decoupled plant elements in two out of four cases. In contrast to the response of biomass (Piñeiro and others 2023) and plant element concentration (this study), P addition alone was not the main single driver of plant element coupling patterns. In turn, the strongest effects of P occurred under eCO2 and varied depending on water levels, as evidenced by significant interactions. Particularly noteworthy was the effect of eCO2 on plant P coupling. Our findings may be interpreted as the result of insufficient P to satisfy the increased nutritional demands of plants for this highly limiting element (Jin and others 2015). This decoupling of P turned into anticoupling (that is, P was less coupled than expected by chance) under eCO2 and high water and P conditions, presumably because of greater plant competition for this element under these conditions and a poor stoichiometric adjustment of P in relation to the rest of elements. In this sense, the eCO2, high water, and high P treatment was also the treatment that had the greatest number of elements becoming decoupled (C, Mg, Ca, and Si) or anticoupled (Ca, P, Mn and Cl), which is again most likely associated with greater competition for essential plant elements under a scenario of insufficient multiple nutrient supply to meet such increased demands. In the situation that most closely reflects soil resource availability under the local field conditions (that is, low water and low P), several elements also became decoupled (C, P, Mg, S, and Mn) and anticoupled (Ca, and Cu) in response to eCO2. This pattern is surprisingly similar to findings by Ochoa-Hueso and others (2019) that, under eCO2, the bioavailabilities of Ca, Mg, K, and Cu became consistently decoupled, while elements like P, Zn, and S became decoupled or anticoupled in a season-dependent manner, in both cases most likely due to increased plant demand (Rumpel and others 2015).

Interestingly, we found that the coupling of individual elements in plant tissues was significantly correlated with the abundance of those elements in the Earth’s crust reported by Yaroshevsky (2006). This is, to our knowledge, the first time that such association has been reported. This implies that the tight stoichiometric control of living organisms on their cells and tissues may have been the result of a natural selection process that favored a greater homeostatic control over those elements that are more abundant (such as those forming part of feldspars) and that are, therefore, likely to be more easily obtained. However, the fact that the coupling of P and Mn slightly deviated from this trend may imply that these two elements, which are frequently taken up and found together, are particularly relevant for the biochemistry of life and so such natural selection process could not, at least solely, operate based on their abundance (Westheimer 1987; Lambers and others 2015). Our results also suggest that metabolic dis-adjustments in one of these two elements may also result in dis-adjustments of the other. Moreover, the fact that P and Mn were the most decoupled elements may be also attributed to the potentially luxurious storage of P and Mn in vacuoles for periods of low availability and high demand (Yang and others 2017; Alejandro and others 2020). Also very interestingly, we found that the coupling of individual elements in plant tissues was significantly correlated with the coupling of those elements in global topsoils reported by Ochoa-Hueso and others (2021c). In this case, we attribute this association to the role of plant tissues in controlling the coupling of soils through the incorporation of dead matter. When organisms die, they leave a coupling fingerprint on soils that will be gradually erased as their bodies become decomposed and elements are redistributed, implying a highly dynamic feedback mechanism controlling the coupling of plants and soils (Rumpel and Chabbi 2019).

Conclusions

In this work, we have demonstrated the importance of P as a limiting factor for the development and growth of understory plants in an Australian Eucalyptus forest. Phosphorus was also the main driver altering the elemental chemical composition of plant tissues, including the decoupling of some elements due to insufficient bioavailability of P and other nutrients. Our data also strongly suggest that evaluating the coupling of specific plant chemical elements is a very powerful tool to detect elements that may be particularly relevant for plant nutrition under specific environmental conditions. Interestingly, none of the elements became decoupled/anticoupled under high water, low P, and aCO2 conditions, most likely indicating that plant nutritional requirements were satisfactorily met. The establishment of P as a fundamental limiting chemical element is not something new and, as Westheimer (1987) posited, may be the result of a process of natural selection, for which we add further evidence in the form of the relationship between plant element coupling and the bioavailability of elements in the crust. That is the reason why P, with its low bioavailability in the natural environment and its high demand by living organisms, influences the ecology of all ecosystems and thus may become the ultimate limiting factor (Elser 2012). Finally, the ability of P to act as a modulator of the changes in the production and biogeochemical cycling and coupling of different elements indicates that the role of P, and perhaps also Mn, may become increasingly important in the context of global change and terrestrial feedbacks on the global C cycle.