Recent initiatives, such as the United Nations declaring 2015 as the International Year of Soils and the French « 4 per 1000 » initiative call attention on soils and on the importance of maintaining and increasing soil organic matter stocks for soil fertility and food security, and for climate change adaptation and mitigation. We stress that soil organic carbon storage (i.e. an increase of soil organic carbon stocks) should be clearly differentiated from soil organic carbon sequestration, as the latter assumes a net removal of atmospheric CO 2 . Implementing management options that allow increasing soil organic carbon stocks at the local scale raises several questions, which are discussed in this article: how can we increase SOC stocks, at which rate and for how long; where do we prioritize SOC storage; how do we estimate the potential gain in C and which agricultural practices should we implement? We show that knowledge and tools are available to answer many of these questions, while further research remains necessary for others. A range of agricultural practices would require a re-assessment of their potential to store C and a better understanding of the underlying processes, such as no tillage and conservation agriculture, irrigation, practices increasing below ground inputs, organic amendments, and N fertilization. The vision emerging from the literature, showing the prominent role of soil microorganisms in the stabilization of soil organic matter, draw the attention to more exploratory potential levers, through changes in microbial physiology or soil biodiversity induced by agricultural practices, that require in-depth research.

The exchange of carbon between soil organic carbon (SOC) and the atmosphere affects the climate and-because of the importance of organic matter to soil fertility-agricultural productivity. The dynamics of topsoil carbon has been relatively well quantified, but half of the soil carbon is located in deeper soil layers (below 30 centimetres), and many questions remain regarding the exchange of this deep carbon with the atmosphere. This knowledge gap restricts soil carbon management policies and limits global carbon models. Here we quantify the recent incorporation of atmosphere-derived carbon atoms into whole-soil profiles, through a meta-analysis of changes in stable carbon isotope signatures at 112 grassland, forest and cropland sites, across different climatic zones, from 1965 to 2015. We find, in agreement with previous work, that soil at a depth of 30-100 centimetres beneath the surface (the subsoil) contains on average 47 per cent of the topmost metre's SOC stocks. However, we show that this subsoil accounts for just 19 per cent of the SOC that has been recently incorporated (within the past 50 years) into the topmost metre. Globally, the median depth of recent carbon incorporation into mineral soil is 10 centimetres. Variations in the relative allocation of carbon to deep soil layers are better explained by the aridity index than by mean annual temperature. Land use for crops reduces the incorporation of carbon into the soil surface layer, but not into deeper layers. Our results suggest that SOC dynamics and its responses to climatic control or land use are strongly dependent on soil depth. We propose that using multilayer soil modules in global carbon models, tested with our data, could help to improve our understanding of soil-atmosphere carbon exchange.

Pulse-labelling of trees with stable or radioactive carbon (C) isotopes offers the unique opportunity to trace the fate of labelled CO(2) into the tree and its release to the soil and the atmosphere. Thus, pulse-labelling enables the quantification of C partitioning in forests and the assessment of the role of partitioning in tree growth, resource acquisition and C sequestration. However, this is associated with challenges as regards the choice of a tracer, the methods of tracing labelled C in tree and soil compartments and the quantitative analysis of C dynamics. Based on data from 47 studies, the rate of transfer differs between broadleaved and coniferous species and decreases as temperature and soil water content decrease. Labelled C is rapidly transferred belowground-within a few days or less-and this transfer is slowed down by drought. Half-lives of labelled C in phloem sap (transfer pool) and in mature leaves (source organs) are short, while those of sink organs (growing tissues, seasonal storage) are longer. (13)C measurements in respiratory efflux at high temporal resolution provide the best estimate of the mean residence times of C in respiratory substrate pools, and the best basis for compartmental modelling. Seasonal C dynamics and allocation patterns indicate that sink strength variations are important drivers for C fluxes. We propose a conceptual model for temperate and boreal trees, which considers the use of recently assimilated C versus stored C. We recommend best practices for designing and analysing pulse-labelling experiments, and identify several topics which we consider of prime importance for future research on C allocation in trees: (i) whole-tree C source-sink relations, (ii) C allocation to secondary metabolism, (iii) responses to environmental change, (iv) effects of seasonality versus phenology in and across biomes, and (v) carbon-nitrogen interactions. Substantial progress is expected from emerging technologies, but the largest challenge remains to carry out in situ whole-tree labelling experiments on mature trees to improve our understanding of the environmental and physiological controls on C allocation.

The international 4 per 1000 initiative aims at supporting states and non-governmental stakeholders in their efforts towards a better management of soil carbon (C) stocks. These stocks depend on soil C inputs and outputs. They are the result of fine spatial scale interconnected mechanisms, which stabilise/destabilise organic matter-borne C. Since 2016, the CarboSMS consortium federates French researchers working on these mechanisms and their effects on C stocks in a local and global change setting (land use, agricultural practices, climatic and soil conditions, etc.). This article is a synthesis of this consortium's first seminar. In the first part, we present recent advances in the understanding of soil C stabilisation mechanisms comprising biotic and abiotic processes, which occur concomitantly and interact. Soil organic C stocks are altered by biotic activities of plants (the main source of C through litter and root systems), microorganisms (fungi and bacteria) and 'ecosystem engineers' (earthworms, termites, ants). In the meantime, abiotic processes related to the soil-physical structure, porosity and mineral fraction also modify these stocks. In the second part, we show how agricultural practices affect soil C stocks. By acting on both biotic and abiotic mechanisms, land use and management practices This synthesis of the CarboSMS French consortium's first seminar was already published in French: Derrien D, Dignac M-F, Basile-Doelsch I, Barot S, Cécillon L, Chenu C, Chevallier T, Freschet GT, Garnier P, Guenet B, Hedde M, Klumpp K, Lashermes G, Maron P-A, Nunan N, Roumet C, Barré P (2016) (choice of plant species and density, plant residue exports, amendments, fertilisation, tillage, etc.) drive soil spatiotemporal organic inputs and organic matter sensitivity to mineralisation. Interaction between the different mechanisms and their effects on C stocks are revealed by meta-analyses and long-term field studies. The third part addresses upscaling issues. This is a cause for major concern since soil organic C stabilisation mechanisms are most often studied at fine spatial scales (mm-μm) under controlled conditions, while agricultural practices are implemented at the plot scale. We discuss some proxies and models describing specific mechanisms and their action in different soil and climatic contexts and show how they should be taken into account in large scale models, to improve change predictions in soil C stocks. Finally, this literature review highlights some future research prospects geared towards preserving or even increasing C stocks, our focus being put on the mechanisms, the effects of agricultural practices on them and C stock prediction models.

Fractionation of soil organic carbon (SOC) is crucial for mechanistic understanding and modeling of soil organic matter decomposition and stabilization processes. It is often aimed at separating the bulk SOC into fractions with varying turnover rates, but a comprehensive comparison of methods to achieve this is lacking. In this study, a total of 20 different SOC fractionation methods were tested by participating laboratories for their suitability to isolate fractions with varying turnover rates, using agricultural soils from three experimental sites with vegetation from C3 to C4 22-36 years ago. Enrichment of C4-derived carbon was traced and used as a proxy for turnover rates in the fractions. Methods that apply a