Geosci. Model Dev., 14, 735–761, 2021 https://doi.org/10.5194/gmd-14-735-2021 © Author(s) 2021. This work is distributed under the Creative Commons Attribution 4.0 License. CoupModel (v6.0): an ecosystem model for coupled phosphorus, nitrogen, and carbon dynamics – evaluated against empirical data from a climatic and fertility gradient in Sweden Hongxing He1, Per-Erik Jansson2, and Annemieke I. Gärdenäs1 1Department of Biological and Environmental Sciences, University of Gothenburg, P.O. Box 460, Gothenburg 40530, Sweden 2Department of Land and Water Resources Engineering, Royal Institute of Technology (KTH), Stockholm 10044, Sweden Correspondence: Hongxing He ([email protected]) and Annemieke I. Gärdenäs ([email protected]) Received: 2 March 2020 – Discussion started: 14 April 2020 Revised: 2 December 2020 – Accepted: 12 December 2020 – Published: 3 February 2021 Abstract. This study presents the integration of the phos- the Coup-CNP to account for various feedback mechanisms phorus (P) cycle into CoupModel (v6.0, referred to as Coup- that have a significant impact on ecosystem C sequestration CNP). The extended Coup-CNP, which explicitly considers and N leaching under climate change and/or elevated N de- the symbiosis between soil microbes and plant roots, enables position. simulations of coupled carbon (C), nitrogen (N), and P dy- namics for terrestrial ecosystems. The model was evaluated against observed forest growth and measured leaf C=P, C=N, and N=P ratios in four managed forest regions in Sweden. 1 Introduction The four regions form a climatic and fertility gradient from 64◦ N (northern Sweden) to 56◦ N (southern Sweden), with Phosphorus (P) is an essential element for photosynthetic mean annual temperature varying from 0.7–7.1 ◦C and soil plants in terrestrial ecosystems, with the P cycle coupled to C=N and C=P ratios varying between 19.8–31.5 and 425– carbon (C) and nitrogen (N) fluxes through processes such 633, respectively. The growth of the southern forests was as decomposition of soil organic matter and nutrient uptake found to be P-limited, with harvested biomass representing (Lang et al., 2016; Vitousek et al., 2010). A steep increase the largest P losses over the studied rotation period. The sim- in the anthropogenic release of C and N to the atmosphere ulated P budgets revealed that southern forests are losing P, relative to P release has altered plant and soil nutrient sto- while northern forests have balanced P budgets. Symbiotic ichiometry, leading to new forcing conditions (Elser et al., fungi accounted for half of total plant P uptake across all four 2007; Penuelas et al., 2013). For instance, numerous moni- regions, which highlights the importance of fungal-tree inter- toring studies have revealed increasing N=P ratios in plants actions in Swedish forests. The results of a sensitivity anal- and soils, especially in forests from North America (Crow- ysis demonstrated that optimal forest growth occurs at a soil ley et al., 2012; Gress et al., 2007; Tessier and Raynal, 2003) N=P ratio between 15–20. A soil N=P ratio above 15–20 will and central and northern Europe (Braun et al., 2010; Jonard result in decreased soil C sequestration and P leaching, along et al., 2015; Talkner et al., 2015). Such trends are generally with a significant increase in N leaching. The simulations assumed to indicate that these ecosystems are shifting from showed that Coup-CNP could describe shifting from being being N-limited to either co-limited by both N and P or P- mostly N-limited to mostly P-limited and vice versa. The limited (Elser et al., 2007; Saito et al., 2008; Vitousek et potential P-limitation of terrestrial ecosystems highlights the al., 2010; Du et al., 2020). Human activities are expected need for biogeochemical ecosystem models to consider the P to continue increasing atmospheric N deposition; as such, cycle. We conclude that the inclusion of the P cycle enabled P availability and P cycle dynamics will become progres- sively more important in regulating the biogeochemistry of Published by Copernicus Publications on behalf of the European Geosciences Union. 736 H. He et al.: CoupModel (v6.0) terrestrial ecosystems and amplifying feedbacks relevant to trients (Smith and Read, 2008). Several studies have shown climate change, e.g. limiting the growth response of plants that the depletion zone around plant roots, which is caused to increased temperature (Deng et al., 2017; Fleischer et al., by plant uptake and the immobile nature of mineral P, in- 2019; Goll et al., 2017). creases when a plant interacts with mycorrhizal fungi (Bolan, Nevertheless, the P cycle is seldom incorporated into 1991; Schnepf and Roose, 2006; Smith, 2003). Global meta- ecosystem model structures. Incorporating the P cycle is es- analysis studies have highlighted that the symbiosis between sential to improving how global models can assess climate–C plants and soil mycorrhizal fungi strongly influences plant cycling interactions (Reed et al., 2015). Most of the process- P availability and subsequently affects plant growth (Terrer based models that can simulate P cycling were specifically et al., 2016, 2019). Previous research has shown that my- developed for agricultural systems and focus on the soil pro- corrhizal fungi can receive between 1 % and 25 % of plant cesses, e.g. EPIC (Jones et al., 1984, Gassman et al., 2005), photosynthates and constitute as much as 70 % of the total ANIMO (Groenendijk et al., 2005), and GLEAMS (Knisel soil microbial biomass; thus, it is clear that this symbiont has and Turtola, 2000). A few catchment-scale models that focus a major impact on soil C sequestration (Averill et al., 2014; on surface water quality, e.g. SWAT (Arnold et al., 2012), Clemmensen et al., 2013; Staddon et al., 2003). Even though HYPE (Arheimer et al., 2012), and INCA-P (Jackson-Blake there is a well-established link between mycorrhizal fungi et al., 2016), aim to simulate how crop management influ- and plant P nutrition (Bucher, 2007; Read and Perez-Moreno, ences P leaching and thus consider processes such as nu- 2003; Rosling et al., 2016), this factor is seldom included in trient retention, leaching, and transport. The C response to ecosystem models (Smith and Read, 2008). To the best of P limitation has recently been studied through several em- our knowledge, only Orwin et al. (2011) have presented an pirical and field studies (Van Sundert et al., 2020; Du et ecosystem model that considers C, N, and P together with al., 2020). For example, Van Sundert et al. (2020) showed symbiotic fungi. They found that considering organic nutri- that the productivity of European beech (Fagus sylvestris) ent uptake by symbiotic fungi in an ecosystem model can sig- forests is negatively related to soil organic carbon concen- nificantly increase soil C storage, with this effect being more trations and mineral C=P ratios. Several global vegetation pronounced under nutrient-limited conditions. In this model, models have included the P cycle to study how it affects the organic nutrient uptake reflects a pathway through which C cycle (Goll et al., 2012, 2017; Wang et al., 2010; Yang plants can utilise organic nutrients by biochemical minerali- et al., 2014; Zhu et al., 2016; Thum et al., 2019). These P- sation, either in symbiosis with mycorrhizal fungi or via root enabled models differ in how they describe soil P dynamics, exudates (e.g. Schachtman et al., 1998; Gärdenäs et al., 2011; i.e. implicitly or explicitly through symbiotic relationships Richardson et al., 2009). However, plant growth is static in with mycorrhiza and other soil microbes, plant P use, and the model presented by Orwin et al. (2011); as such, plant– acquisition strategies, ultimately leading to considerable un- soil or plant–environment interactions are largely ignored. certainty in the C response (Fleischer et al., 2019; Medlyn et Our model (Eckersten and Beier, 1998; He et al., 2018) also al., 2016; Reed et al., 2015). Medlyn et al. (2016) applied six includes a shortcut for nutrient uptake that relies on rhizo- global vegetation models – including two coupled Carbon– sphere processes. The assumption is that nutrients released Nitrogen–Phosphorus (CNP) models (CABLE and CLM4.0- by biochemical mineralisation are instantly taken up by sym- CNP) – to study how the C cycle of the Eucalyptus-Free biotic microbes and/or the plants, thereby bypassing the soil Air CO2 Enrichment experiment responds to elevated CO2 matrix solution. He et al. (2018) integrated the MYCOFON (eCO2) levels. The results demonstrated notable variations model (Meyer et al., 2009) into CoupModel v5 to ensure that in predicted net primary productivity ranging from 0.5 % to the symbiosis between plant roots and mycorrhiza would be 25 %. The CNP models that explicitly considered the P de- sufficiently considered and compared the results with a previ- pendency of C assimilation predicted the lowest eCO2 re- ous implicit representation of N uptake in forest ecosystems sponse. Yu et al. (2018) included the P cycle in the ForSAFE with limited N availability. CoupModel v5 assumes that car- field-scale biogeochemical model to study the P budget of bohydrates provided by plants are the primary driver of my- a southern Swedish spruce forest site. They concluded that corrhizal responses to N availability and that fungal uptake internal turnover from mineralisation of soil organic mat- of N will influence host plant photosynthesis. We argued that ter affects the P supply more than weathering. Fleischer et terrestrial ecosystem models that explicitly consider mycor- al. (2019) demonstrated that four CNP models, when applied rhizal interactions should also take into account P cycling due to the Amazon forest, provide up to 50 % lower estimates to the significant role of symbiont mycorrhiza for P uptake in of the eCO2-induced biomass increment than the 10 coupled P-limited environments.