Context

Soils represent the largest natural carbon reservoir on the Earth's surface. Of the 6 billion tonnes of carbon released annually by human activities, at least one quarter is reabsorbed by terrestrial ecosystems. However, this carbon sink may be disturbed by the climate changes underway and their repercussions on soil carbon dynamics. The anticipated effects are as follows:

• changes in temperature and water balance will influence the rate at which microbes break down organic material,
• the response of vegetation to new climate conditions will modify organic inputs to the soil, both their quantity and biochemical nature,
• increased availability of CO2 for photosynthesis will change the requirements for organic nitrogen, whose availability could become a limiting factor in the soil and have feedback effects on microbial activity,
• indirect effects of climate change (evolving farming practices: irrigation, fertilisation; changes in land use, etc.) will likely impact soil carbon dynamics in a manner still poorly defined.

There is currently significant uncertainty surrounding the flux of soil carbon to the atmosphere and the response of soil carbon reservoirs to climate change. This uncertainty is linked to existing tools, which are not accurate enough to simulate dynamic changes in soil carbon. This in turn compromises ecosystem models, which are based on an overly "simplistic" model of soil carbon dynamics.

Objectives

Against this backdrop, the CARBOSOIL project aims to improve the dynamic modelling of soil carbon by integrating recent knowledge on the decomposition of soil organic material. This will involve bridging the gap between the two consortium communities, which usually work independently: researchers studying the dynamics of soil organic material (Bioemco) and researchers who integrate this dynamic in ecosystem models (ESE, LSCE).
The teams will seek to answer four questions they have identified as obstacles in the current understanding and modelling of soil carbon:

1) What are the impacts of temperature and humidity on soil decomposition rates and how should they be modelled?

• Scientists understand the theoretical effect of temperature on microbial respiration. However, in situ soil experiments point to a degree of acclimatisation to global warming which needs to be understood and modelled.
• Water content in soil, which influences the rate of carbon transformation, is generally well integrated in the models. It also appears necessary to include short-term variations in the soil's water balance during wetting-drying cycles, which are not currently considered.
• Interactions between temperature and humidity, which appear to have non-linear effects on decomposition rates, must also be taken into account.

2) How should microbial decomposers and their biological regulation be taken into account?

• Current models of organic material decomposition are based on first-order kinetics, according to which the abundance and physiological state of the microorganisms do not limit breakdown. These models leave out several important phenomena:
- maintenance needs of microbial populations, their stoichiometric requirements, and variations in assimilation efficiency,
- priming effect: over- or under-mineralisation of the soil organic material when fresh organic material is added, resulting from the preferential choice of certain substrates by microorganisms and/or the interactions between decomposers.

• Furthermore, the models used to analyse long-term carbon dynamics do not consider the availability of other inorganic elements (primarily nitrogen, but also phosphorus and sulphur) in carbon transformation, or do so in a simplified or even a simplistic way.

Based on these observations, it is now critical to integrate nitrogen availability and the physiology of microbial decomposers in the global models.

3) What are the effects of ploughed land and organic material accessibility on decomposition rates and how should they be modelled?

Ploughing has a significant effect on carbon stocks, which it tends to diminish. A portion of the soil carbon is protected from breakdown by its location, making it inaccessible to decomposers. Ploughing makes this organic material vulnerable by periodically breaking up the soil aggregates and exposing bare soil to rain. However, neither ploughing nor variations in carbon accessibility - related to soil type - are explicitly taken into account in models of soil carbon evolution. Further research is needed on this process.

4) How should the effects of vertical soil carbon distribution on decomposition rates be represented?

Despite having a lower carbon concentration than surface horizons, deep horizons play a major role in soil carbon storage. Their equilibrium can nonetheless be disturbed by environmental changes. Consequently, deeper root development (in response to hydric stress, for example) can lead either to carbon mineralisation (labile carbon input causing compounds to break down) or an increase in carbon stocks (carbon stabilisation favoured at root level rather than in above-ground portions of the plant). Vertical discretisation of the carbon dynamic models is necessary for taking the specific characteristics of deep soil horizons into account and for integrating vertical transfers of soluble carbon.

Methodology

Part 1: Summary of existing knowledge and conceptual models

Research on the processes of soil organic material dynamics is developing rapidly, due in particular to the growing use of powerful tools based on isotopic biogeochemistry. Reviewing existing knowledge will involve making an inventory of new concepts that move beyond current approaches, which are too generic and not mechanistic enough.

• On each of the scientific questions defined above, the CARBOSOIL group will conduct a critical review of the current state of knowledge and modelling: knowledge assessment, identification of gaps, inventory and analysis of existing conceptual models of organic material dynamics. If necessary, a conceptual model of the study question will be developed, as the basis for building a suitable code. The methodology will be based on dedicated workshops bringing together experts from within and outside the project. The goal will be to write one summary article per question, reviewing the knowledge and gaps identified.

• The main models of carbon and nitrogen dynamics will be compared, based on the analysis coordinated by P. Smith in 1996 (McGill, W.B., 1996, Review and classification of ten soil organic material models. In: D. Powlson, P. Smith and J. Smith, Evaluation of soil organic material models, NATO ASI Series, No. 38). Various models of increasing complexity will be taken into account. Each approach will be evaluated based on how well it simulates existing measures and its relevance for the new concepts needed to optimally integrate the interactions between organic material dynamics and the climate.

• The teams involved in the project have access to a remarkable set of experimental structures and long-term trials, which will be reviewed and summarised:

	Long-term agronomy trials at the Grignon experimental farm (AgroParisTech) Déhérain's plots were set up in 1875 to study long-term changes in soil organic material fertility according to the type of fertilisation used in large-scale farming. The trials have been used in studies of carbon and nitrogen dynamics. At the same site, trials were launched in 1973 to study the effect of simplified soil ploughing under large-scale farming conditions, as compared to conventional ploughing practices.
	Very long-term bare fallow plots at Versailles and Grignon In Versailles, 42 plots were set up in 1929 to study the effect of various fertilisers and amendments on soil structure (INRA/French National Institute for Agricultural Research). At Grignon, 36 plots were set up in 1959 to study the evolution of straw in soil as a function of nitrogen availability and/or straw composting (AgroParisTech). Since these plots have been cultivated and bare for several decades, their soil carbon has received no organic input and is thus comparable to the stable soil carbon compartment.
	Soil carbon labelling chronosequence using maize at the Closeaux site in Versailles This trial, launched in 1993, makes it possible to measure carbon dynamics based on the natural abundance of carbon-13. It involves a series of plots planted with continuous wheat or maize crops over periods of 6 to 15 years (INRA).

Aside from these three experiments, to which the project members have special access, the consortium teams are involved in other trials that will produce input data for the simulations, provided that ad hoc experimental programmes are developed in parallel:
- La Cage trial (INRA Versailles): comparison and evaluation of farming systems.
- Observatoire de Recherche en Environnement-Prairies, Cycles Biogéochimiques et Biodiversité (ORE-PCBB), INRA Lusignan: tracking changes in prairie ecosystems in response to human influence and monitoring the environmental consequences.

Part 2: Development of a complete one-dimensional mechanistic model

The development of a complete model will incorporate the priorities that emerge from analysing the processes and models created in Part 1.

• One or more mechanistic model(s) will be developed, taking into account:

- the effect of soil carbon decomposers and carbon accessibility (complex formation, microbial population growth),
- breakdown into progressively less energetic compounds and the interactions between the resulting reservoirs,
- the impact on soil carbon decomposition of vertical temperature and humidity gradients and their seasonal and interannual variations,
- interactions between carbon and nitrogen, modulated by plant organic nitrogen requirements and by inputs from the litter,
- vertical carbon transfers,
- spatial changes (soil type, depth) and temporal changes (ploughing, land use) in organic material accessibility.
Finally, all interactions, non-linear effects, and co-dependencies between edaphic (related to soil characteristics), climate, and agronomy factors will be considered.

• The models developed in the first phase will then be calibrated and their ability to represent soil organic material dynamics will be evaluated, using data from the experiments and long-term trials. The most suitable model, i.e. the one that responds to the four scientific questions raised by the CARBOSOIL team, will be incorporated in the spatialised ecosystem model ORCHIDEE.

Part 3: Integrating the soil module in ORCHIDEE

This part of the project will start with a comparison, based on previously described experimental data, of the outputs from the standard ORCHIDEE model (using a soil module such as CENTURY) and the outputs from the selected soil model, in order to characterise the differences and enhancements resulting from a more mechanistic, detailed approach. The standard soil module of ORCHIDEE will then be replaced by the mechanistic model, and a new series of simulations will be conducted to evaluate the difference between the mechanistic model and the new ORCHIDEE model: ORCHIDEE SOL+.
The final step will involve validating the new soil organic material outcomes, to achieve increasingly realistic carbon balances. For this purpose, the results of several simulations will be compared with ORCHIDEE SOL+:
- study of the interannual variability in net carbon flux over a short timescale (1 to 2 decades),
- study of spatialised carbon balances on the scale of France,
- study of carbon flux and storage evolution at global scale in the context of climate change, based on two or three contrasting scenarios for the 21st century.