State of the art, project context

Today’s agriculture is facing many challenges: to feed a growing world population, to reduce poverty and inequity, to adapt to climate change, to reduce negative externalities, to conserve natural resources, and to decrease chemical inputs (Griffon, 2013, Tittonell, 2015). The new paradigm called Agroecology is a rupture from artificializing productivist agriculture which mainstreamed science for over a century (Hainzelin, 2015). Many definitions and approaches have been proposed (Altieri, 1989; 2002; Gliessman, 1997; 2007; Francis et al., 2003; Wezel et al., 2009; Tittonell et al., 2012; Silici, 2014). Altieri (1995) defined principles for agroecological systems: (i) increase genetic and specific diversity at farm and landscape levels, (ii) integrate crop and animal, (iii) rely on biologically active soil organic matter-rich soils, (iv) rely on high biomass recycling rates and tight nutrient cycles, (v) rely on natural control mechanisms against pests, and (vi) optimize the use of space. Many advantages are expected at the plot level and especially increase in biomass production, carbon C sequestration in soil, fertility in the long term, resilience to stress and resistance to pests and diseases. Agroecology is also recognized to necessitate a multidisciplinary field of research. Farmers must be involved with scientists in the development of agroecological innovations both for local and scientific knowledges’ hybridation and facilitating future improved practices dissemination. As a consequence, agroecological practices must be based on indigenous and scientific knowledge, be economically and environmentally sound and based on local resources, be socially and culturally acceptable, and enhance farm productivity. Agroecological transition is particularly important for smallholders in tropical regions where rural societies and food systems generally face many challenges.

Scientists generally recognize the need for an ecological intensification of agricultural production by increasing biodiversity and complexity in agrosystems, to rely more on natural functions, biotic interactions and ecological processes, and to amplify the services provided by living organisms (Altieri, 1995; Barrios et al., 2015). Generally at field and farm levels, agroecological practices drive and optimize functional biodiversity aboveground, whereas soil (belowground) biodiversity and functions are rarely managed. The importance of soil functions in the performance of agroecological systems is widely recognized and their restoration (see annexed Lexicon) appears necessary (Altieri, 1999; Barrios 2007; Brussaard et al., 2010; Ratnadass et al., 2013; Bardgett et al., 2014; Clermont-Dauphin et al., 2015; Bender et al., 2016). Unfortunately, due to the complexity of soil functioning and the poor knowledge of its determinism, only poor consideration is still given to soil when designing agricultural systems. Four basic soil ecological functions are of interest when regarding plant functions: (i) soil organic matter dynamics, (ii) nutrient cycling,

(iii) maintenance of soil structure, and (iv) pest regulation (Lavelle et al., 2006; Brussaard et al., 2007; Kibblewhite et al., 2008). These soil functions directly or indirectly affect plant functions (De Deyn et al., 2004) and are provided by the activity of soil organisms (soil functional biodiversity).
In recent years, considerable attention has been given to the increase in soil biodiversity following the use of agroecological practices. Indeed, many studies have shown the positive effect of agroecological practices on soil biota (for instance, Blanchart et al., 2006; Henneron et al., 2015) contrary to conventional practices (Tsiafouli et al., 2015). Conversely, many fewer studies considered the restoration of ecological functions following agronomic or soil restoration practices. There is a great lack of knowledge regarding the effect of agricultural practices on soil function restoration (SFR). This scientific deficiency is serious knowing that soil ecological functions confer high degrees of resistance and resilience following disturbance, on the basis of sustained resource use within the ecosystem (Odum, 1969; Gliessman, 2004). As a consequence, the reduction in soil functional diversity within the agroecosystem causes the loss of its resilience and constant human-derived external inputs must be maintained (Gliessman, 1990; 2004). To reintegrate sustainability, the soil ecological functions have to be optimized (Barrios, 2007; de Vries et al., 2013; Coleman & Wall, 2015).
The increase in soil ecological functions is possible (i) indirectly by improving soil as a habitat for soil biota, and (ii) directly by inoculating soil organisms. Nevertheless the inoculation of soil biota seems difficult without a previous restoration of the soil habitat (Senapati et al., 1999; Lavelle et al., 2001; Singh et al., 2011; Thuita et al., 2012). The soil habitat as a suitable environment to intensify the ecological processes is at the basis of many agrosystem services. Some soil functions do not occur, not because the actors are absent, but because the favourable abiotic conditions are not met. Restoring soil functions firstly requires restoring the abiotic environment and providing energy to soil biota (Lavelle et al., 2001). Current farmer practices aiming to increase plant production likely to affect soil habitat and functions: (i) increase in soil organic matter content through inputs of residues, manure, composts, (ii) increase in nutrient availability through fertilizer addition and soil pH correction by ashes, lime or dolomite application, (iii) diversification of cropped plants through associations of plant species or varieties, and/or rotations. In the Highlands of Madagascar, one of the main agricultural constraints is the very low fertility of soil due to P deficiency; this led scientists to study and propose practices to improve P availability and crop production. Another knowledge gap is the genotype-specific response of plants to soil functions. This is a serious shortcoming knowing that the ability of cultivars to interact with soil organisms may be highly variable. For instance, the increase in rice nutrition by the presence of protists in the plant rhizosphere seems highly variable according to rice genotype (Somasumdaram et al. 2008). By selecting a genotype for a specific agronomic target, the agronomist may blindly select a set of plant traits likely involved in many other ecological interactions. This may result in the disappearance of specific plant functions. On-going greenhouse experiments in Madagascar by our consortium (unpub. data) showed that rice varieties respond differently to the presence of soil plant mutualists (SPM): their effect is either positive, negative or neutral on total biomass and shoot/root ratio. It thus appears central to take into account rice genotype variability when restoring soil and plant functions.
Understanding and manipulating the plant-soil interactions and feedbacks in agricultural transition is thus challenging (Van der Putten et al., 2013). A key question is what interventions are required for successful restoration of soil functions in agroecological agrosystems (Hobbs & Harris, 2001).

Objectives

The importance of soil ecological processes and functions for plant growth and other ecosystem services make soil an essential component of sustainable agroecological systems. Soil (ecological) Function Restoration, i.e. the intensification of these ecological processes, during agroecological transition, is the core of our project. The overall objective of the SECuRE project is to provide Soil Function Restoration (SFR) practices based on local and scientific knowledges, in order to increase both agronomic, socio-economic and ecological performances of agroecological agrosystems in a tropical context. We hypothesize that innovative cropping practices improving SFR will promote major ecosystem functions, i.e. nutrient cycling, carbon storage, control of pathogens and resistance to climatic stresses, performed by soil biotic diversity and assemblages. SFR aims to optimize current farmer’s practices and propose innovative practices that will promote soil habitat in order to increase soil functional diversity and intensify associated soil and plant functions. It is part of agroecological restoration. For this purpose, optimized and innovative SFR practices could be:

• The use of original organic inputs with high agroecological performances such as vermicomposts
• An efficient combination of existing organic and mineral inputs promoting plant functions
• An increase in soil heterogeneity by providing various coupled organo-mineral substrates in a stratified way
• Biofertilization, i.e., inoculation of SPM to restore some soil functions
• The use of crop varieties that best respond to innovative SFR practices.

The SECuRE project will focus on rainfed rice cropping systems in the Highlands of Madagascar because of the importance of rice for Malagasy food and farmers' livelihoods. Productivity in the Highlands of Madagascar is very low due to the presence of nutrient-poor Ferralsols (Randriamanantsoa et al., 2013), the low access to external chemical inputs, and unsustainable practices (Rabeharisoa, 2004).

Specific objectives will be:

• to assess farmers' interventions in terms of soil restoration
• to improve our knowledge of the effect of SPM on diverse plant functions,
• to test various restoration practices based on farmers' and scientists' knowledge,
• to analyse the impact of soil restoration practices in terms of agronomic, socio-economic and ecological performances through an innovative original method,
• to support co-learning and dissemination of our results with stakeholders: restoration practices to farmers, knowledge on soil functioning in agroecology for students, indicators of soil functioning for scientists and users.
• SECuRE also aims at bringing generalizable knowledge on soil/plant functions in agroecology and more contextual outputs in the case of Malagasy family agriculture with “custom-made” cropping systems where farmers are at the centre of innovation systems.

Overal approach

The SECuRE project principally aims to provide/improve soil function restoration practices to small farmers in Madagascar. Four main questions arise from this objective: What are the existing traditional/local and scientific knowledge on soil restoration? How to combine both form of knowledge to propose innovative practices? How to evaluate the ecological, socio-economic and agronomic performances of such practices? How to produce co-learning and disseminate innovative practices and who to involve?

The project will be carried out in Madagascar where social, economic and biophysical constraints of agriculture are very strong, and where a multidisciplinary consortium is already working with farmers for an increase in crop production in family farms in the Highlands of Madagascar while preserving natural resources. The consortium SPAD (Highland production systems and sustainability in Madagascar) gathers six research and educational institutions: CIRAD, IRD, University of Antananarivo, FOFIFA, FIFAMANOR, AfricaRice (http://www.dp-spad.org).

Two sites in Madagascar will be investigated during the SECuRE project: Mid-West of Vakinankaratra, near Antsirabe and Itasy near Antananarivo. These sites are interesting as they present two contexts different in terms of:

• Climate: Itasy is 1300 m asl, 1300 mm Mean Annual Rainfall, 18°C Annual Temperature, Mid-West is 900 m asl, 1300 mm MAR, 23°C AT.
• Soil in the Mid-West is characterized as clay-loam Ferralsol with a clay-silt-sand composition of 35-40-25 % in the top layer. The soil is moderately deficient in nutrient and organic matter content, with 17.2 gC kg-1, 1.4 gN kg-1 and acid (soil pH H2O around 5.3 but strongly deficient in available phosphorus (2.5 mg kg-1 of available P). Soils in Itasy are red clay Ferralsols (40-20-40% clay-silt-sand composition) with very low carbon and nutrient availability: 20 gC kg-1, 1.2 gN kg-1, 1.6 mg kg-1 of available P, soil pH H2O 5.1 and CEC 1.68 cmol.kg-1.
• Farmers' crops and practices: In the Mid-West, farming systems is mainly practiced on smallholder farms, with dominance of mixed crop and livestock systems. Lowland rice production is the most important crop (for two-thirds of the farmers), for food self-sufficiency and income generation. In uplands area annual cropping systems are based on rice, maize and cassava crops in continuous or with short-fallow successions. Tree plantations are scarce and mostly concentrated around village houses. Most of the field labour is hand-based; only cattle draught is used during soil preparation before the sowing or planting time. In Itasy, population density is high (> 100 ind.km-2) due to the proximity of Antananarivo. Rice cropping systems are predominant and mainly for subsistence farming associated with market gardening crops and fruit trees. In both sites, recent development projects have diffused the use of agroecological practices such as the use of composts, agroforestry, cover crops, crop association, no-tillage, etc.

Both sites are investigated by our consortium in different projects: Mid-West is the main site for the STRADIV project (http://ur-green.cirad.fr/projets/stradiv). Itasy is the site for the Mahavotra (http://www.agrisud.org/fr/projet-mahavotra-goodplanet-agrisud-etcterra/) and FRB-CAMMiSolE project. Many interactions will be developed between these project and SECuRE project. The project will thus involve scientists of different disciplines that are locally present: agronomists, plant scientists, plant breeders, soil scientists, soil ecologists, ethno-sociologists, agro-economists. Different approaches will be developed in order to reach the objectives within the timeframe of the project; they are described in each WP.

WP0: Project management (E. Blanchart, IRD Eco&Sols)

A management team will include the project coordinator and WP leaders. For each WP, we proposed a couple of leaders in order to maintain as much as possible an equilibrium between Malagasy and French partners, and between genders. This team will monitor and evaluate project progress within the time frame and for the proposed budget. Meetings will be frequent (every six months) and especially at the field level.

A steering committee will also be organised in order to fit the project with global/African considerations on agroecology and international initiatives. The steering committee will be composed of four recognised scientists covering different disciplines: Dr. Jean-Luc Chotte (IRD Eco&Sols, Montpellier), Dr. Hélène Joly (CIRAD Agap, Montpellier), Dr. Eric Scopel (CIRAD Aida, Montpellier), Dr. Bernard Vanlauwe (IITA, Nairobi). We plan to organize a visit of the steering committee to Madagascar during the kick-off meeting of the project.

WP1: Local based farmers' knowledge on soil function restoration (M. Razafimahatratra, FOFIFA & Patrice Autfray, CIRAD Aida)

This WP will be especially conducted by a mixed team composed of socio-economists and farming system agronomists in order to apply a holistic approach of the different farm activities. Three main approaches will be developed:  (i) a large survey of farmers (400) to describe and assess farmers' interventions for soil restoration, (ii) a specific permanent partnership with reference farms (40) to compare multiyear agronomic performance of their cropping systems, both traditional and innovative ones and (iii) a specific innovation platform on soil function restoration, linked with the general multi-stakeholder platform on biodiversified farming systems developed in the STRADIV project (System approach for the TRAnsition to bio-DIVersified agroecosystems project, Agropolis Fondation). The questions will be to determine: (i) what are the current practices of soil function restoration, (ii) farmers' constraints, (iii) the opportunities and willingness to improve them, and (iv) their reasons for testing improved technologies. The main outputs of this WP1 will be local knowledge on soil function restoration, based on empiricism and data recorded at farm level on real situations. Specific relevant information on some socio-economic determinants (labour and investment risks for instance), linked with labour and land productivity, will feed WP4 in terms of socioeconomic descriptors (e.g. socio-economic performance) while helping field trial implementation of WP3.

The large survey on soil restoration practices will comprise specific questionnaires of 200 farms per site which will describe interventions such as fallows, manure, organic matter inputs, mineral fertilization, dolomite or ash inputs, plant diversification, tree plantation, etc. Additionally, frequency and amount of inputs (mineral or organic) will be documented. The 200 farmers in two villages in each zone will be randomly selected. Plateau of tanety (slope 0%) and sloping tanety will be considered separately.

Data to be collected will also concern:

• household demographic information, current/former beneficiary or non-beneficiary of an agroecological  project implemented in the site; and some farm structure aspects like tanety area and livestock;
• inventory and history of current soil restoration practices (first adoption, technical itinerary, concerned area, visual indicator of soil degradation, reason of adoption…) and agroecological and economical performance indicators of performance of these practices (agroecological and economical performances) according to indicators detailed in WP4.
• retrospective analysis of the process of soil restoration in some plots determined with each farmer surveyed in order to highlight the evolution of soil restoration practices over the last 3 to 5 years, and to identify abandoned practices and reasons of drop-out.

The permanent partnership with 20 selected farms per site will permit both (i) to assess multi-year socio-economical and agronomical performances of traditional and improved restoration practices compared with non-restoration ones and (ii) to conduct participative working sessions using a focus group methodology inside the innovation platform. Each farm will be studied using a GIS tool in order to record field areas, technical choices, crop yields, labour time, incomes and input supply. All data will allow to estimate land and labouring productivity. In the Mid-West this network of 20 farms already exists within the STRADIV project. It could be thus reinforced and focused on this theme facilitating methodology transfer to the Itasy site.

The innovation platform on soil function restoration will be based on regular focus group sessions in each site, combining the scoring of relevant criteria using radar representation, open discussions and role-playing games. While describing sociotechnical constraints at farm level, other criteria will be identified at the regional level related to the economic and institutional landscape for soil restoration-related decisions, such as access to market, previous experiences as source of learning, content of local extension and advisory services, etc.

Finally the results on both sites will be shared and discussed inside a two-site platform to provide mainly results for the WP5.

WP2: Scientific knowledge on soil function restoration (B. Rabary, FOFIFA & J. Trap, IRD Eco&Sols)
Objectives: This WP has two main objectives: (1) identify organo-mineral assemblages that maximize soil and plant functions in order to provide enough information to design the field trials (WP3) and (2) deepen scientific knowledge on the effect of soil restoration on soil and plant functions. We thus intend to better understand processes by which various restoration practices influence plant functions (plant nutrient acquisition, plant growth, plant tolerance against pests (phyto-parasitic nematodes and blast disease) and support the activity of SPM (mainly earthworms and microbivorous nematodes). Finally, the roles of SPM and plant genotypes crossed with the innovative restoration practices on nutrient cycling and plant responses will be investigated in short-term laboratory experiments.

The ‘functional dissimilarity theory’ as a conceptual framework to design innovative restoring practices: The amount and quality of mineral-organic material input have a strong impact on carbon, nutrient cycling and thus plant nutrition and growth (Asai et al., 2009; Grandy et al., 2002; Tejada et al., 2006, Tester 1990), especially in nutrient-poor soils such as in Madagascar (Raboin et al., 2016). This input effect depends on the relative contribution of the materials within the assemblage and their characteristics to the overall pool (Hättenschwiler et al., 2005; Petchey & Gaston, 2006). The literature reveals that rates of community processes may deviate substantially from those expected from single-litter-species decomposition of component species because of non-additive interactions among litter species (Fanin et al., 2012; Hättenschwiler et al., 2005; Heemsbergen, et al., 2004). These results lead to the formulation of the ‘functional dissimilarity theory’ that suppose that the decomposition and the release of nutrients from a mixture of organic materials increase with increasing functional (chemical) dissimilarity among the materials. Here, we thus hypothesize that the functional dissimilarity of the organic assemblage will be the best attribute that fits linearly the soil and plant functions.

A 6-step sequential approach as a method to design innovative restoring practices: In Madagascar, farmers usually amended the soil with a single type of material, either a mineral one such as ashes or dolomite, or an organic one such as compost or manure. The combination of materials within an assemblage is rare and thus constitutes an innovative restoring practice per se. However, the effects of combining materials into an assemblage on soil and plant functions are unknown and difficult to predict because of the huge total number of possibilities in combining materials: with an initial number of 14 organic materials, the total number of possible assemblages is 16383. Testing all organic assemblages is obviously not feasible and no empirical data are thus available. Fourteen materials is a good compromise between a huge number of combinations and a good representativeness of the material diversity in Madagascar. It is important to take into account the diversity in restoring practices in order to increase our chance of identifying practices that optimize agronomic and ecological performance of rice systems. We thus propose a 6-step sequential approach including a pre-test experiment, a screening greenhouse experiment and several short-term laboratory experiments that will be designed according to results from previous steps.

• Step 1: Preparation of materials

At the beginning, a total of 14 organic materials will be selected from our ‘organo-library’ including different types of manures, composts, vermicomposts, etc.): 5 crop residues, 1 manure, 1 compost, 5 vermicomposts, 1 tree leaf litter, 1 wood fragment. All materials are already available at the LRI. Two mineral materials will be used: dolomite and eucalypt ash. All organic materials have been dried, ground and sieved at 100 µm, and stored at ambient temperature.

• Step 2: NIR spectra acquisition

Each organic material will be scanned in the NIR region between 1100 and 2500 nm at 2 nm intervals in order to determine its reflectance spectrum. For each organic material, twenty subsamples will be scanned and averaged. No spectrum pre-processing methods (derivations or smoothing) will be performed. Using NIR signature is a relevant integrative method of organic material chemistry without the need to analyse the chemical composition of organic matters, reducing hence the time required for this step and the cost.

• Step 3: Functional Dissimilarity (FD) assessment

Here, the objective is to compute the functional dissimilarity (FD) for each organic material assemblage. NIR signatures will be plotted on the factorial plane of a discriminant analysis based on partial least square analysis (PLS-DA) using the function “plsDA” from the package ‘DiscriMiner’ in R. From the outputs, we will calculate the FD-value for each assemblage using the multidimensional Euclidian distance (for the 2-material category) or area (for the 4- and 6-material categories). PLS-DA is a useful test analysing NIRS data to counteract the colinearity problems among variables. Other tests are available to discriminate the organic materials, such as the non-metric multidimensional scaling (NMDS) as well as different similarity index (Bray-Curtis distance, Rao’s dissimilarity coefficient, etc.).

• Step 4: Pre-test experiment

There are different techniques to provide the organic material into the soil, either as mulch, mixed in the topsoil or in thin layers to increase heterogeneity. Recent experiments in Madagascar showed that the application of mulch at the soil surface promotes plant growth more than when residues are incorporated into the soil (Ratsiatosika et al., submitted). By increasing soil heterogeneity, the application of organic matter in layers within the soil promotes SPM-driven functions (Bonkowski et al., 2001). In the same way, plant responses to organic amendments are known to vary according to rice genotypes. These two parameters (localization of organic materials and genotypes) have to be fixed before the screening experiment in order to limit the total number of microcosms during step 5. Here we aim to identify the best (high effects) method to provide the organic materials and the rice genotype that exhibits the largest response to amendments in terms of growth and nutrition. Four organic materials will be tested, two of them exhibiting high FD-value and two of them exhibiting low FD-value. The soil will be a Ferralsol collected from bozaka (natural savannas as reference ecosystems). No mineral materials (ashes, dolomites, lime) will be provided. The total number of pots is 4 (organic materials) × 3 (material localization) × 4 (rice genotypes) × 3 (replicates) = 144 microcosms. A restricted number of ecological and agronomic indicators described in WP4 will be measured at the end of the experiment (2 months of growth).

• Step 5: Screening greenhouse pot experiment

We will select a limited number of organic material assemblages following this procedure: (a) Three categories of assemblage will be computed in order to create a richness gradient of organic materials within assemblages: low material richness (2 materials among 14; 91 possible combinations), intermediate material richness (4 materials among 14, 1001 possible combinations) and high material richness (6 materials among 14, 3003 possible combinations). (b) We will compute the FD-values from the three categories of organic assemblages using the PLS-DA procedure. (c) We will enclose the FD-values for each assemblage category between 0 and 1 using a homothetic transformation; (d) We will cluster assemblages into three FD-groups: low (0-0.2), intermediate (0.4-0.6) and high (0.8-1) in order to create a functional dissimilarity gradient; (e) We will rank each assemblage within FD-groups according to the social indicators from WP1 in order to select assemblages that exhibited high socio-economic values (increasing their probabilities of adoptability by famers); (f) We will select the first four assemblages within each FD-group and within each assemblage category. We will thus select 3 × 3 × 4 = 36 assemblages that have high social performance. Each organic material (14) will be tested separately (mono-assemblage pots). We thus obtain 14 + 36 assemblages that will be tested without mineral material (native soil), with eucalypt ashes or with dolomite, leading thus to a total of 150 pots. The experiment will be carried out in a greenhouse during three months. At the end of the experiment, the agro-ecological performance of each assemblage will be characterized using the indicators described in WP4.

• Step 6: Mechanistic experiments

To better understand the ecological mechanisms behind the diversity effect of the assemblage, we propose four complementary experiments.

(1) Laboratory experiment 1: Phosphorus (P) cycling using isotopic labelling. P is the first limiting nutrient is these soils. The ability of the restoring practices to increase the P plant-availability in soil is crucial. Using 32P, we will label the soil and monitor soil gross P mineralization and plant P uptake after amendment in order to better identify the mechanisms by which organic amendments impact soil P cycling.

(2) Laboratory experiment 2: The role of soil organisms (bacterial-feeding Cephalobidae nematodes and endogeic earthworms – Pontoscolex corethrurus) on the diversity effects of organic assemblages. During the screening experiment, no fauna will be inoculated in the pot. However, the diversity effect of an assemblage is known to be driven by soil fauna (Coulis et al., 2014; 2015). Here, the objective is to characterize the interactive effects of soil fauna inoculation and organic amendments. Again, a restricted number of organic assemblages will be selected from step 5.

(3) Greenhouse experiment 3. The role of plant genotype is crucial. A recent experiment made by the consortium showed that interactions occurred between soil organisms, amendments and plant genotype. Here, the effect of earthworms, microbivorous nematodes and organic amendments on plant responses of eight rice cultivars previously selected by the FOFIFA will be tested in the laboratory.

(4) Greenhouse experiment 4. Silicon seems an interesting matter to be tested as its plant acquisition is increased by earthworms (Bityutskii et al., 2016) and increases rice resistance to blast disease (Cai et al., 2008). Fertilization with soluble silicon compounds coupled with earthworm inoculation will be tested in a pot experiment.

WP3 Soil function restoration field trials (T. Razafimbelo, LRI & P. Salgado, CIRAD Selmet)
Following WP1 and WP2 results and combining local and scientific knowledge on soil function restoration, we plan to set up two field trials (one at each site) designed to test innovative SFR practices on rice functions in field conditions. Only the ‘best’ SFR practices resulting from WP2 will be tested in the field. The trial will be conducted after Bozaka (natural grass vegetation) removal. Rice will be the main crop. Climatic data will be collected all along the experiment at both sites. Individual plots will be 4x4 m2, separated by 1 m with the next plot.

The field experiments will be split into two trials:

• Soil habitat restoration: SFR practices based on organic and mineral amendments will be tested. Eight to twelve SFR practices (treatments) will be tested in this first trial. We plan to set up randomized block designed experiments with five replications. Practices will be applied to the soil in the same way (same materials, same localization, same amount, and at the same time as sowing as in the laboratory experiments. The length of the experiment will be 2 years. SFR practices will be compared to a negative control (native soil without any amendments).
• Soil functional diversity rehabilitation: inoculation of SPM (earthworms, nematodes and mycorrhizal fungi together). The two ‘best’ SFR practices from WP2 will be tested in this second trial. These treatments will be replicated 10 times and monitored for 2 years. At the beginning of the second cropping year, SPM will be inoculated (5 replications) or not (5 replications). The objective of this trial is to test the restoration of the soil habitat before inoculation of key soil organisms and to observe whether the introduction of such organisms leads to synergistic effects of organo-mineral amendments and improves plant functions.

 For all trials, agronomic and ecological performance indicators will be measured at the beginning and after each cropping year (see description in WP4) and climatic conditions (rainfall and temperatures) will be monitored.

WP4: Evaluation of soil function restoration (L. Bernard, IRD Eco&Sols & R. Randriamanantsoa, FOFIFA)
Objectives:

This WP proposes an original method to evaluate agrosystem performances. We are aware that agroecological systems must be multifunctional in the sense that many ecological functions must be performed in order to provide ecosystem services, not only food provisioning but also regulation services such as climate change mitigation and carbon sequestration, resistance to erosion, water quality, etc. Due to the importance of soil into the provision of services, we propose to represent the performance of agroecological systems as the combination of an agronomic performance, a (soil) ecological performance and a socio-economic performance (Fig. 5). Compared to traditional, conventional systems (i.e., rice monoculture, soil tillage, very low inputs), we hypothesise that agronomic, ecological and socio-economic performances will increase when developing agroecological practices/systems. Each performance will be characterized by a set of descriptors.
Performance descriptors:

• Agronomic performance will be described using classical agronomic parameters: yield, plant biomass, empty/full grains, grain and plant nutrient and carbon contents, photosynthesis efficiency (using leaf PRI and NDVI), Specific Leaf Area, and Leaf dry matter content. Field trial and farm plots will permit to compare the positive or negative effects of practices on water stress using the PRI (photochemical reflectance index). PRI is indicative of photosynthetic light use efficiency, the rate of carbon dioxide uptake, and is a reliable water-stress index (Gamon et al., 1992).
•  Social-economical performance will be described by the main classical parameters: Cost price of rice, Easy geographical access of inputs, Gross margin per ha, Labour productivity, investment return and Level of technicality.
•  Regarding soil ecological performance, we propose to focus on soil functions resulting from ecological processes (and following Kibblewhite et al., 2008 in order to assess effects on nutrients, carbon, soil structure, pests and pathogens). Each of the following parameters will be evaluated a posteriori on its reactivity (variation range) to the different agroecological practices. Only best descriptors will be kept in the final radar plot. Consideration will be given to adapt some of the best descriptors of ecological functions to be performed in the field and easily transferable to end-users.
• Carbon storage via the biochemical pathway: (1) bait-lamina test (Römbke, our consortium is currently improving this technique with a paper in preparation) and tea bag index (Keuskamp et al., 2013), (2) extraction of organic carbon by permanganate (Hurisso et al. 2016), (3) organic matter fractionation, and (4) carbon-linked enzymatic decomposition of specific substrates of BIOLOG ecoplates (Frac et al. 2012).
• Nutrient recycling and storage: (1) measurement of nutrient stocks and availability, (2) measurement of soil respiration (Hurisso et al. 2016) and priming effect (projet CAMMiSolE), (3) ‘Microbial loop indicator’ developed in the frame of the INDICES project (J. Trap, IRD, coordinator, Agropolis foundation), and (4) nutrient-linked enzymatic decomposition of specific substrates of ecoplates.
•  Pest regulation: density of white grubs and plant-feeding nematodes (major pests in rice crops in Madagascar).
• Soil structure: (1) soil aggregate stability, (2) water holding capacity.
•  In parallel, soils will be described by more classical physicochemical and biological parameters in order to understand which drivers/biota control the different ecological functions. Therefore a set of classical measurements will be performed: texture, pH, CEC, density and diversity of macrofauna and nematofauna, microbial density and activity (microbial Phosphorus), microbial structure (fungal/bacterial ratio measured by qPCR), functional microbial diversity by applying different diversity indices to substrate degradation profiles provided by BIOLOG ecoplates.

 Evaluation of performance in farms, trials and laboratory experiments:

Agronomic, ecological and socio-economic descriptors will be assessed in a set of 20 selected farms/plots in both studied regions (WP1). In the different field trials (WP3, at the beginning and after one and two agricultural seasons), only agronomic and ecological descriptors will be assessed. Solutions to eventually improve the social performance of such new practices will be discussed with end-users as part of the innovation platform. A database will be built integrating climatic and soil parameters, agronomic practices, agronomic, socio-economic and ecological performances. It will allow scientists to highlight the most interesting parameters to characterize agroecological systems.

WP5: Networking, co-learning and dissemination of knowledge on soil function restoration (S. Audouin CIRAD & S. de Tourdonnet Montpellier SupAgro)

Enabling and supporting new SFR practices leads us to consider learning processes, which are highlighted as crucial for any innovation process to address challenges and opportunities facing farmers (World Bank, 2006). This WP aims at sharing SECuRE data and knowledge with scientists, students, public, technicians, policy-makers and farmers. It will promote the dissemination of SFR practices to enhance the agroecological transition through dissemination, training and capacity building.
A multi-stakeholder and multi-scale dialogue will be enhanced, based partly on the focus groups organized under WP1 and on other participatory workshops with other stakeholders (farmers, farmers’ organisations, technicians, training organisations, extension and advisory services, research). This dialogue part is considered as an arena of information sharing and co-learning, allowing knowledge and learning processes between relevant stakeholders, based on the amount of resources they are prepared to commit. It will result in the generation and uptake of different types of scientific and non-scientific knowledge useful to learn lessons for experiences (Douthwaite et al., 2001). Part of the results of WP 1 (diagnostic) and WP 4 (performance), will be discussed in the arena. Stumbling blocks, barriers and levers for adoption of such restoration practices will also be identified and action plans elaborated. Action plans will include activities (such as multiplier groups), type of stakeholders involved and planning in order to support dissemination. These activities will also contribute to address scale issues by involving actors at different scales (from local, regional to national scale).

A website will be developed based on project objectives and outcomes, and dedicated to SFR in agroecology. It will target a large audience with all outputs of the project being stored and shared even after the end of the project. The website also aims at strengthening existing scientific and/or practitioner networks on agroecology, at national and international levels. The project coordinator and WP5 leaders will be responsible for updating the website. In order to accentuate the dissemination and networking objectives, results will be shared within the FAO platform on agroecology (www.fao.org/agroecology/en).

• Scientists: The WP will reinforce existing networks between French and Malagasy partners. Results will be published in peer-reviewed journals or communicated in congresses on soil ecology, soil-plant interactions, agronomy, social sciences, etc.
• Students: The project will develop specific tools for student capacity building such as training sessions for students and young post-doctoral scientists, and specific e-learning and MOOC modules on soil function restoration, in relation with the PARMI project (Agropolis Foundation). Training will be ensured by a multidisciplinary team, based on continual interactivity with socio-economic, and biophysical sciences, on methodological and case study issues. Interactions between Master students from the University of Antananarivo (Higher School of Agronomy ESSA) and Montpellier SupAgro will be reinforced. The writing of a general booklet on soil ecological descriptors (WP4) will be jointly managed by students from both Schools in an interacting process and will target worldwide scientists.
• Users: all outputs of the project will be made available for all public and especially individuals and institutions interested by the development of sustainable agriculture specifically in Madagascar, or in sub-Saharan Africa, even worldwide. In particular, the booklet on soil ecological descriptors will be of great interest for scientists and practitioners; it will describe each descriptor: objective, advantages and disadvantages, method, measurement, baseline and interpretation, scientific references. Each descriptor will be accompanied by a short movie describing the method. Brochures on local SFR practices will target users and will be an educational and scientific tool, allowing knowledge sharing between scientist, students and practitioners. Videos on agroecology will also be produced with the help of the University of Antananarivo. In a former Agropolis Foundation’s project, CARIM, two short movies on agroecology were realised http://www.umr-ecosols.fr/index.php/fr/recherche /projets/52-carim). Four movies are planned to complete this series with a specific focus on soil functions.
• Farmers and farmers’ organisations: Dissemination of agroecology requires constant farmer participation, and technical solutions have to be considered as legitimate and culturally acceptable. The approach of the project, which promotes a multi-stakeholder dialogue aims at addressing this challenge. Collaboration with local training organisations, extension and advisory services, farmers’ organisations will be developed so as to ensure their relevant participation to knowledge and learning processes within the arena. Their involvement at different stages of the project will contribute to a large dissemination of the results. At the end of the project, brochures on the main and most interesting SFR practices will be published in Malagasy and communicated to relevant national networks. French and English versions will be posted on the website.

Major outputs and outcomes

Different outputs will be provided by our project and specifically by each WP.

WP1 will firstly update information on farmers' practices and knowledge on soil function restoration with a holistic and a participatory approach. Reports will be produced from the surveys on farmers' practices and the results coming from focus groups sessions. This WP will also permit the training of Master students involved in this WP and provide mainly inputs for the WP5 which aims to mix local and scientific knowledge produced by the WP2, WP3 and WP4. It also will provide local socio-economic data that will thus be used in the WP2 for ranking each innovative SFR practices. The researchers involved hereby will be associated to those of the other WP in providing scientific data of the agro-socio-economic context and results of the participative approaches for publishing in peer-reviewed journals.

WP2 will improve our knowledge on the soil-plant interactions following different restoration practices. Malagasy Master students and post-doc scientists will be involved in this WP in order to increase capacity building. Results from these experiments, beside student reports, will be published in peer-reviewed journals (dealing with plant-soil relations, or sustainable agriculture) and/or communicated in international conferences. Finally, this WP will also provide an innovative way to select specific SFR practices according to the farmers' needs and possibilities.

WP3 corresponds to a complex field trial that will last many years, if funds are available. During the frame of the SECuRE project, two cultivation years will be possible (from December 2017 to April 2018 and from December 2018 to April 2019). A PhD student will be involved on the trial. A precise description of the trial will be given on the website. Results on agronomic and ecological performances (linked with WP4) of each year will be gathered in a report (end of 2018 and end of 2019). Results will also be published in peer-reviewed journals and communicated in conferences. The trial will also be dedicated to farmer visits organized each year and to student’s teaching.

WP4 will propose an original method to characterize the joint agronomic and ecological performances of agrosystems and more specifically on SFR practices. This will be performed on our field trial and in selected farmers’ plots. The obtained results, gathered in a database will allow us to characterize and disseminate the ‘best’ restoration practices, ‘best’ meaning the ones improving both agronomic and ecological performances. The method will be published in a peer-reviewed journal as a new way to characterize the agro-ecological performance of agrosystems.

WP5 is the WP where outputs will be diffused and disseminated to different audiences. Main outputs will be the website, academic training sessions, e-learning and MOOCs for students, a booklet on the descriptors of soil ecological functioning for users, brochures on ‘best’ SFR practices for farmers, movies on ecological methods and on agroecology, participatory workshops to build action plans, and a final workshop in Madagascar for stakeholders (including policy-makers, developers, funders, NGOs). Collaboration with local training organisations, extension and advisory services, farmers’ organisations will be developed so as to ensure their relevant participation to knowledge and learning processes within the innovation platform.

In consequence, the SECuRE project will have important outcomes for different audiences. Diffusion and co-learning based on our results with stakeholders and farmers will facilitate the dissemination of innovative sustainable practices to smallholder farmers at the scale of Malagasy Highlands, and at a larger scale to improve food security and famer livelihoods in sub-Saharan Africa where soil and farmers' constraints are similar to Malagasy Highlands. The presence of institutions for rural development in our consortium (FOFIFA) and our links with NGOs (AgriSud International, GSDM) will facilitate the diffusion of innovative practices to farmers. The fact that our innovative approaches include traditional knowledge and dialogue within an innovation platform may reinforce the adoption of these techniques by farmers. Policy-makers are also awaiting scientific expertise to develop agricultural and development policies. The strong links between our consortium with the Malagasy Ministries of Agriculture and of Higher Education and Scientific Research will ensure the transfer of research to end-users. One of the four master plans produced by the Ministry of Research and Higher Education in 2016 relates to food security, another one to climate change. Our results may impact both priorities as soil function restoration may impact plant growth and other regulation services, especially carbon sequestration.

Our activities will reinforce the link between Malagasy and French scientists, especially scientists from the LabEx Agro in Montpellier. Our links with CGIAR (AfricaRice in SPAD consortium and IITA in the steering committee) is a guarantee to see our works diffused at larger scales than Madagascar.

Originality and innovativeness

1. One of the most innovative points is the development of an original method to evaluate both the agronomic, socio-economic and ecological performances of agroecological agrosystems. This method will allow to efficiently analysing the agroecosystem performances and to identify the dynamic functional trajectories of restored agroecosystems. Importantly, each performance and each indicator can be weighted according to farmer’s possibilities in order to better fit practices with farmer’s requirements. This method will be applied both in field experimental trials and in farm’s plots and will be disseminated in the scientific sphere.
2. A second innovation is the possibility to propose complex organo-mineral amendments using a sequential approach based on the functional dissimilarity theory. The diversity of agroecological inputs in Madagascar is huge and until now no method was proposed to design new amendments combining them in an efficient way. Using ecological concepts directly in agronomy is fundamental to initiate an agroecological transition promoting sustainable plant production.
3. A third original perspective is the taking into account of the soil functional biodiversity as a way to manage agroecosystem functioning and not only as an environmental consequence of practice shifts. Indeed, usually soil functional ecology is not managed in agroecological transition while aboveground biodiversity is. Direct actions on soil are generally restricted to a reduction of soil tillage and to organic matter inputs. Reboud (2014) analysed 2,500 scientific references dealing with Agroecology between 1975 and 2012: the word “soil” was only associated with ‘Soil organic matter’. This confirms the weak integration of the soil component in agroecological transition.
4. An important original aspect is the presence of many scientists from different but complementary disciplines in the same consortium and around the same objectives of proposing innovative practices. Indeed, the consortium includes different disciplines already working together (www.dp-spad.org).
5. Our project also proposes original outputs favouring dissemination and capacity building such as the publication of a booklet on soil functional descriptors or courses (e-learning and MOOC) on SFR practices, a multi-stakeholder dialogue for co-learning and dissemination of the results.

Multidisciplinarity

The SECuRE project is an interdisciplinary project covering two scientific fields: especially the field ‘Agro-ecosystems, agri-environmental innovations and resource management’ and at a lower degree ‘Integrated crop protection, plant pests and diseases, symbiotes, population ecology’. It is also related with other scientific fields with interest given to plant breeding, innovation processes, and participatory approaches. The achievement of our objectives needs strong collaborations between various scientific disciplines covering the broad field of agronomy: ecology, soil science, plant breeding and science, systemic agronomy, social sciences, economics, transfer and teaching. Those disciplines are present in our consortium gathering institutions from Madagascar and France (5 units from Labex Agro: Eco&Sols, Aida, Selmet, Agap, Innovation). One of the main challenges of the project is to link scientific knowledge on soil ecology with plant functions and more widely on agroecological innovation processes. Researchers in ecological/biological sciences and researchers in agronomy and innovation processes will strongly interact to enhance the flow of knowledge between actors. The structure of the project is designed to facilitate the exchange between disciplines, with co-leaders of each WP belonging to different disciplines.

In fine, our project is at the interface between agronomy and ecological restoration with explicit incorporation of soil ecological knowledge and co-learning, which acknowledges interactions among the principal components of the soil system as well as feedbacks between the aboveground and belowground ecosystem compartments, into agroecological restoration (Aronson et al., 1993; Wardle & Peltzer 2007; Heneghan et al., 2008).