Discussion

We examined P dynamics in the plant, soil and leachate in a humid tropical forest exposed to 7-year of continuous field warming. Contrary to most studies which show that short-term warming increases ecosystem P deficiency in arid ecosystems or large scales (Jonasson et al., 2004; Yuan & Chen, 2015), we found that warming by 2.1°C increased plant P uptake and decreased foliar N:P, indicating warming drove sustained plant P demand and may alleviate P deficiency and/or limitation of enhanced plant growth and contribute to sustaining plant C fixation in these tropical forests. It is worth pointing out that the sustained plant P demand could not be explained by traditional views in accessing P acquisition (through fine root and mycorrhizal fungi; Rosling et al., 2016) and supply (through P availability), because neither fine root biomass nor arbuscular mycorrhizal fungi abundance were affected by 6-year warming as they were by short-term warming (Lie et al., 2021), and soil soluble P had no response to warming.
We observed a enhanced P redistribution capacity and soil P mobilization to support increasing plant P demand without altering litter P mineralization and leachate P, which led to a net transfer of P from soil to plant regulated by biogeochemical processes (Fig. 5): (1) warming increased plant P resorption by 4.7%, which enhanced plant P recycling and soil P inputs; (2) warming elevated soil P mineralization by 15.4% and dissolution by 6.8% in the topsoil, thereby improving soil P supply; (3) reduced P loss in surface runoff under warming facilitated P retention for topsoil and shallow roots. Our findings improved the understanding of biogeochemical controls on plant P acquisition under warming, which are crucial to project the future C sequestration potential of tropical ecosystems that are mostly limited by P (Fig.1; foliar N:P > 20) (Güsewell, 2004; Cleveland et al., 2013).
Plant P resorption increased under warming, similar to observations at the global scale that plant P resorption was positively correlated with temperature (Yuan & Chen, 2009; You et al., 2018). Plant growth periods strongly drive plant nutrient resorption (Sun et al., 2016). A warming experiment conducted in subtropical forests showed that foliar growth periods were extended by an average of 5–18 days at 2°C of warming (Gunderson et al., 2013). Thus, warming potentially delays foliar senescence (Chung et al., 2013) and consequently increases foliar plant P resorption (Estiarte & Peñuelas, 2015). Moreover, it is noted that plant P resorption increased with P supply derived from soil (i.e.,enhanced P mineralization and dissolution) under warming, which contradicts the view on a trade-off between nutrient resorption and soil nutrient supply (Vergutz et al., 2012; Gerdol et al., 2019). We found that increased P resorption was likely driven by warming-induced drying soil (Fig. S5), which is consistent with a study in a dry ecosystem (Ren et al., 2018). P resorption increases P-use efficiency in plants via higher P productivity and longer residence time of P, while new roots may be limited in the capacity to take up P from litter in drier soil. Thus, an increase in plant P resorption may be a more economical P acquisition strategy than increased investment in roots. These findings extend the application of resorption theory to P cycling, and also highlight the key P source from resorption for plants under warming.
Most studies reported that P mineralization in humid forest soils was enhanced under warming (Rui et al., 2012; Zhou et al., 2013; Zi et al., 2018; Zuccarini et al., 2020). Consistent with these studies, warming elevated soil P mineralization by increasing soil potential acid phosphatase activity and decreasing moderately-available organic P in topsoil. The elevated soil P mineralization could be driven by joint increases in plant demand for P and microbial activity (e.g., higher soil respiration (Li et al., 2016)). The increase in soil acid phosphatase activity could be stimulated by microbes to compensate for the reduction of P diffusion in warming-induced drying soil (Allison et al., 2010) and the limitation of microbial P utilization during litter decomposition with higher litter C:P (Table S3; Sardans et al., 2006). Our data thus showed that more P was retained in plants than returned to soil in each litter cycle under warming, which led to a net transfer of P from soil moderately-available organic P to the plant.
At our site, a large fraction of the moderately-available P in the soils was in inorganic form (>58%), suggesting that inorganic P dissolution may be especially important for P supply. In line with our hypothesis, warming enhanced P dissolution, as evidenced by lower Fe-Pi in the topsoil. However, the Al-Pi and pH in the three soil depths were not affected by warming, suggesting that the decrease in Fe-Pi was not caused by soil pH, which served as a critical factor of P dissolution (Navratil, et al., 2009; Devau et al., 2011). Rather, the lower Fe-Pi was likely related to sensitive changes of soil Fe phases under warming. In topsoil, warming led to decreases in redox potential and increases in both Fe(II)HCl:FeHCl and Feca concentrations, indicating higher Fe reduction. Our finding challenges the prevailing expectation that warming drives more oxidative soil, due to increased soil porosity by warming-induced drying soil (Wang et al., 2017). The higher Fe reduction probably was due to the enhanced microbial activity, root respiration coupled to higher P uptake capacity and thus oxygen depletion under warming at our site (Li et al., 2016). Compared with Al minerals, the reduction of Fe minerals is considered to effectively reduce geochemical sequestration of easily-available P to increase soluble P and even alleviate P limitation in humid tropical soils (Chacón et al., 2006; Lin et al., 2018; Gross et al., 2020). Some studies in subtropical wetland systems have reported similar declines in redox potential and increases in P release from Fe minerals under warming (Zhang et al., 2012; Wang et al., 2013). Accordingly, the increases in P release can elevate soil soluble P, as observed in incubation experiments with higher soluble P under reducing conditions (Peretyazhko & Sposito, 2005; Wang et al., 2017). However, we did not find changes in soil soluble P, which might suggest a rapidly attained equilibrium, in which soluble P induced from the dissolution of Fe-Pi was rapidly taken up by plants in this P-poor system (Liptzin & Silver, 2009). The rapid plant uptake could prevent potential P re-immobilization and loss by leaching. Overall, the increase in P release from Fe minerals provides evidence for greater soil P supply under warming, which was expected to facilitate plant P utilization.
The losses of P deserve to be noticed under warming. P in terrestrial ecosystems is primarily lost through leachate that makes it difficult its return to the system. Currently, global P mineable reserves are gradually being lost from ecosystems due to improper use of P fertilizers in forest and farmland managements. Greater plant P uptake generally promotes P fixation in plants and thereby potentially reduces P loss. Intriguingly, we noted that the increased plant P uptake does not imply less P losses, because the net P losses were counterbalanced by the decreased P losses in surface runoff and the increased P losses in underground water. These results suggest that warming shifts P loss from surface to subsurface, and emphasizes the necessity to assess effects of warming on P dynamics at the ecosystem level.
Our study provides new insights into understanding tropical P cycling processes and modelling forest P limitation under warming. The findings of the sustained plant P demand driven by increasing P supplies through multiple biological and geochemical processes within the ecosystems (e.g., plant P resorption, soil P mineralization, and P release from Fe minerals) under warming extend the traditional assessment method (measure P supply by available/soluble P) and the current paradigm states (assess P conversation only by soil P mineralization). This suggests that previous studies may have overestimated the impacts of P limitation on ecosystem C stock under at least short-term or medium-term future warming (Wieder et al., 2015). Our study also provides additional information on P dynamics that can be incorporated into earth system models which require further quantification (Reed et al., 2015; Sun et al., 2017).