ex post n°58

N° 58 October 2014

Agroecology: Evaluation of 15 Years of AFD SupportSummary of final report Laurent LEVARD, Aurélie VOGEL, Christian CASTELLANET (Gret) and Didier PILLOT (Montpellier SupAgro)

AFD Evaluation

Agence Française de Développement5, rue Roland Barthes 75012 Paris www.afd.fr

ExPost exPost

Disclaimer

The analyses and conclusions presented in this document are those of the authors.

They do not necessarily reflect the official views of Agence Française de Développement or its partner institutions.

Publication Director: Anne PAUGAMEditorial Director: Laurent FONTAINEISSN: 1962-9761Legal deposit: 3rd quarter 2014Cover photo: Rice sowing under Stylosanthes guianensis cover on a farmer’s plot in Midwest Madagascar in November 2013.© Cathy CLERMONT-DAUPHIN (GRET)

Layout: Marie EHLINGER

Authors:

Laurent LEVARD, Aurélie VOGEL, Christian CASTELLANET (GRET) and Didier PILLOT (Montpellier SupAgro)

Coordination AFD: Constance CORBIER-BARTHAUX, Evaluation and Capitalisation Unit – [email protected] from French by Eric ALSRUHE

The final report of this evaluation is available at the following address on internet:http://www.afd.fr/home/publications/travaux-de-recherche/publications-scientifiques/autres-collections

Agroecology: Evaluation of 15 Years of AFD Support – Summary of final report

1 AFD 2014 • ExPost exPost

Acknowledgements

Evaluation team

The work enjoyed the support of a monitoring committee

called the ‘reference group’, made up of around 20 members:

Patrice BURGER, (Director of CARI association, ‘reference

group’ chairperson); Khalifa SAOUSSEN (MAE); Hacina

BENAHMED (MAAF); Hervé SAINT-MACARY (CIRAD);

Sébastien TREYER (IDDRI); Jean-Luc CHOTTE (IRD);

Robert LIFRAN (Supagro/INRA); Stéphane BELLON (INRA);

Georges SERPANTIÉ (IRD); Sylvain BERTON (Agrisud);

Christophe CHAUVEAU (AVSF); Vera EHRENSTEIN (Ecole

des Mines); Laurent FONTAINE, Constance CORBIER-

BARTHAUX, Laure MONTCHAMP, Emmanuelle POIRIER-

MAGONA, Anne LEGILE, Claude TORRE, Marie-Cécile

THIRION, Veronika CHABROL, Nicolas ROSSIN, Hélène

WILLART, Alain HENRY, and Tiphaine LEMÉNAGER (AFD);

Didier SIMON (FFGE). We would like to give them special

thanks here.

We also thank all of our contact people, both in France and

in the countries where we have carried out site studies, for their

cooperation in making this evaluation possible.

The evaluation team, which was coordinated by Laurent

LEVARD (GRET) and Didier PILLOT (Montpellier SupAgro),

was also made up of Aurélie VOGEL (GRET), Christian

CASTELLANET (GRET), Cathy CLERMONT-DAUPHIN

(IRD), Joël COUDRAY (consultant agronomist) and Julie

SORÈZE (Montpellier SupAgro).

Philippe DEYGOUT (Institute for Research and Application

of Development Methods – IRAM), Laure MONTCHAMP

(AFD), Constance CORBIER-BARTHAUX (AFD), Albert

RAKOTONIRINA and Thierry RABARIJAONA (GRET) also

took part in certain stages of the evaluation.

Evaluation and Capitalisation Series • N°58

2 AFD 2014 •ExPost exPost

CONTENTS

Introduction 3

1. Overall evaluation of the projects based on OECD DAC criteria 7

1.1. Relevance and coherency 7

1.2. Effectiveness, results, efficiency 8

1.3. Impacts and sustainability 10

2. Key factors affecting DMC-related results 11

2.1. With regard to agricultural production systems 11

2.2. With regard to collective constraints and relations between systems of production 12

2.3. With regard to relations with the environment 12

3. Assessment of the tool/programme and of the learning approaches 13

3.1. How the tool/programme was structured 13

3.2. Programme management 13

3.3. Knowledge production 14

3.4. Scientific monitoring 14

3.5. Learning 14

4. Conclusions and recommendations 16

4.1. Conclusions 16

4.2. Operational recommendations 18

Acronyms and Abbreviations 20



Introduction

This document is a summary of the final report on the

external evaluation of 15 years of support actions by Agence

Française de Développement and the French Facility for

Global Environment – FFGE in agroecology (in practice, on

Direct Seeding Mulch-based Cropping systems – DMC). The

evaluation was carried out at the request of AFD, from July

2013 to May 2014, by Gret.

Direct Seeding Mulch-based Cropping, also called ‘con-

servation agriculture’, was first experimented and developed

in Brazil in the 1950s. The technique was taken up and promoted

by teams from the Centre de coopération internationale en

recherche agronomique pour le développement (French

Agricultural Research Centre for International Development

– CIRAD) in the 1990s and, from 2000, by French development

cooperation bodies (French Ministry of Foreign Affairs – MoFA,

AFD and the FFGE) as part of their Plan d’Action pour l’Agro-

écologie – (Agroecology Action Plan – PAA).

The PAA, officially launched in June 2000, was based on two

main parts:

• A group of actions to adapt and diffuse DMC techniques

in several representative countries with a diversity of agro-

climatic zones, as part of specific projects or aspects of research

and development (R&D) on rural development projects or

programmes. A ‘first circle’ of projects (Tunisia, Madagascar,

Cameroon and Laos) preceded a ‘second circle’ (Vietnam

and Cambodia) several years later. Other projects with DMC

components were also supported by AFD in Mali and, outside

the framework of the PAA, in Gabon and Morocco. In Viet-

nam, the experiments did not lead to significant diffusion

activities.

• The Programme Transversal d’Accompagnement (Cross-

cutting Support Programme – PTA) was launched in 2000,

followed by the Programme d’Actions Multi-Pays en Agro-

écologie (Multi-country Action Programme in Agroecology –

PAMPA), in 2007. The purpose of these cross-cutting tools was

to ensure coherency among the various PAA actions, the com-

plementary technical support actions, the communication and

exchanges among the various experiences, the capitalisation,

and the sharing of knowledge. For the PAMPA, a call for

proposals was launched for the ‘research’ subcomponent.

Box 1. Direct Seeding Mulch-based Cropping systems (DMC): The three principles

DMC systems are part of a huge group of systems and techniques promoted within the framework of agroecology. They

offer a theoretical technical alternative to the fertility crisis observed in many agrarian systems, following the reduction of

fallow systems or the intensification of agriculture based on the Green Revolution. DMC systems are based on both the

absence of tilling and the permanent cover of the soil by plants. The latter work simultaneously to enrich the soil with

organic matter, vitalise its biological life and stop the development of weeds. Finally, crop rotation makes it possible both

to optimise organic and mineral fertilisation and better control weeds and parasite cycles.

Context and objectives of the evaluation



IRD-INRA-CIRAD created a ‘consortium’ to respond to it, and

its proposal, named the ‘initiative Réponse Intégrée Multi-

Équipes’ (Multi-team Integrated Response initiative – RIME)

was selected.

The financial amount of actions supported by AFD and the

FFGE that are specific to DMCs or that include a DMC com-

ponent is estimated at EUR 123.2 million, with projects prior

to the PAA excluded.1 Of this total, AFD funded EUR 88.7 million

(72%), the FFGE EUR 5.8 million (5%) and the French MoFA

EUR 2.0 million (2%). If we consider only DMC-related actions,

the amount is estimated at EUR 56 million, including 10%

( EUR 5.5 million) for the cross-cutting programmes.

The external evaluation sought to a) measure and assess

the actions supported by AFD and the learning approaches at

each project level and for the system as a whole, b) to describe

the factors of success and failure of the actions, and c) to draw

lessons from them in order to work out recommendations for

AFD’s future actions in the agricultural domain.

The methodology used for the evaluation combined the

following: desk research, interviews in France with the AFD

project officers and the CIRAD researchers involved, several

field visits (especially including interviews with farmers and

partner institutions). It was based on:

a) The study of DMC components at 7 sites in 6 countries

(technical content of actions, operating methods and institu-

tional systems). For each of the sites visited, a specific report

(‘site study’) was written, sent to the partners for comments,

and then finalised;

b) A comparative synthesis of the site studies, which acted as

a basis for the overall report;

c) The assessment of successive cross-cutting programmes;

d) The working out of recommendations based on the lessons

of the previous phases.

The work enjoyed the support of a monitoring committee

called the ‘reference group’, made up of around 20 members

coming from AFD, the French MoFA, the French Ministry of

Agriculture, Agrifood and Forestry, and around 10 French

researchers from different organisations and all disciplines.

The evaluation team met four times with the reference group,

making it possible to fine-tune the preparation for the studies

of the sites and to discuss the analyses, syntheses and re-

commendations.

Two specific meetings were also organised with the CIRAD

teams involved, before and after the site studies, for exchanges

on the hypotheses and methodology in the first meeting, and

on the initial conclusions at the second.

Methodology of the evaluation

1 The following are not included: the PPI Farafangana and PPI Manakara projects in Madagascar, the DGPT project in Cameroon, PRODESSA in Laos, PDRI Kef and Siliana in Tunisia, the Hevea Village and SAM projects in Vietnam.



The sites studied and visited (see Table 1 below) are located

in Madagascar (Lake Alaotra and Vakinankaratra), Cameroon

(cotton production area), Laos (South Sayaburi, Plain of Jars),

Cambodia (Kampong Cham and Battambang provinces),

Gabon (Estuaire Province) and Morocco (Middle Sebou). The

independent evaluation conducted in Tunisia (Siliana-Kef and

North-West) by Iram for the FFGE was also taken into account

in the overall evaluation.2

The sites studied

2 In all, the evaluation team devoted 45 days (including by 2-person teams) to these field studies, the main objective of which was to evaluate the ex-post impact of DMC projects (which had often been finished by the time of the mission) and to analyse the factors explaining why the techniques proposed by the farmers were adopted or not.

Table 1. The sites and the projects studied

PRODESSA

PTA

PAMPA

Previous projects with DMC activities

‘First circle’ of projects

‘Second circle’ of projects

Source: Authors.

PHASE 3: Farm and terroir

approach PHASE 2: Transition

PHASE 1: PTA and launching pilot

projects

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Madagascar

Cameroon

Laos

Tunisia

Cambodia

Morocco

Gabon

Cross-cutting

Agrobiological Management of Soils

BV Lac1 BV Lac 2

BV PI – SE HP

ESA 1 DPGT (incl. PTA) PSC – ESA 2

BVLac = Lake Alaotra watershedBVPI SHP = Watersheds and Irrigated Areas Project in the Southeast and the High Plateaus

PRONAE NUDP

PROSA

PHASE 2Family-scale

Rubber Growing

PHASE 3Rubber

DMC FFGE

PADAP

PTA

PHASE 1Family-scale

Rubber Growing

DMC FFGE 2

PASS

PRONAE – transition

phase

PRODESSA

PADAC

CANSEA

PMH-Middle Sebou II

PRODIAG

PAMPA

DGPT = Farm Development and Land Management ProjectESA = Water, Soil, Tree projects

PRODESSA = Rural development project in the South of Xayaburi ProvincePASS = Point of Application in Southern XayaburiPRONAE = National Agroecology ProgrammePROSA = Sectoral Agroecology Programme

PADAC = Cambodian Agricultural Development ProjectCANSEA = Conservation Agriculture Network in South East Asia

PADAP = Project to Support the Development of Periurban AgriculturePRODIAG = Project for Agricultural Development and Investment in Gabon

PMH = Small and Medium-scale Water Project



All the projects were based on a standard intervention model

designed in three theoretical phases:

• Phase A: experimentation in research stations. The proposed

DMC systems were perfected and experimented in a research

station. Different forms of associations between the plant

grown and the cover plant were tested there, as were a range

of formulas of fertilisation, sowing methods and rotations.

• Phase B: experimentation by farmers. The systems selected

were set up by reference farmers on their own plots, which

also act as a training tool and for visits for other farmers.

The reference farmers enjoyed certain advantages (remune-

ration, free supply of inputs, subsidised credit). The systems

can evolve to take into account feedback from the farmers.

• Phase C: actual diffusion beyond the reference farmers.

In practice, these three phases could be overlapped more

or less over time, with variations according to the country.

Operating method

Box 2. Variations on the three phases of the intervention model

In the beginning, all the PAA projects were designed to

implement development research in three phases.

However, the projects quickly sought to get by without

the first phase, by using the references obtained from

research at other sites, adapting them, and having them

tested directly by reference farmers of a new site. This

way, the second generation of projects, like those of

Vakinankaratra in Madagascar, of Battambang in

Cambodia or of Morocco was based on the references

previously created on the Malagasy high plateaus, the

red soils of Kampong Cham in Cambodia, or in Tunisia.

These shortcuts sometimes turned out to be detrimental,

as the ecological and social situations of second regions

are rarely the same as those of the first. In Battambang,

for example, it took two whole years to realise that the

soil pH, which was significantly more alkaline than

that of the red soils, required the use of totally different

varieties of cover plants.

Varied operational systems were set up. The projects

were, for example, implemented a) by a public or para-

public institution; b) by an ad hoc organisation in which

CIRAD played an important role; c) by CIRAD directly, in

possible cooperation with other stakeholders; d) by other

various operators, sometimes with various actions being

delegated by the main organisation in charge of implemen-

tation; or e) by a producers organisation.

Operational systems

Photo 1. Cambodia, Kompong Cham (Cambodia): plot of cassava under vegetal cover

Photo credit: Laure Montchamp.



Box 3. The interest of techniques other than DMC – examples in Laos and Madagascar

In Laos, in the Plain of Jars, the knowledge now acquired

about the environment suggests that the most promising

possibilities for intensification are in extensive livestock

production. Yet, this is where the Laotian national training

centre on DMC was established. In Vakinankaratra, in

Madagascar, hedging and fodder crops, as well as manure

management, now seem at least as promising in terms

of innovation for farmers as the DMC models initially

promoted by the project.

Many farmers around the world, especially in difficult

environments, are dealing with an ecological crisis (erosion,

decrease in fertility management, water management) or the

incapacity to sustainably increase production in a context

of demographic growth. The negative environmental limits

and impacts of the conventional model of agricultural inten-

sification are increasingly obvious. Furthermore, the planet’s

food security requires roughly a doubling of production in

the next half century, and this increase will have to take

place with less use of deforestation, less water and fewer

inputs than in the past. In this context, research for alterna-

tive or complementary solutions to the dominant model of

agricultural intensification, among which agroecology, seems

highly relevant in view of the great challenges of sustainable

development.

On the other hand, the appropriateness of the initial

programme focusing on only DMC practices must be put into

question. This is because other agroecological techniques

(sometimes not yet qualified as ‘agroecology’) were, at the

time of launch of the programme, known and often imple-

mented by many farmers, and these techniques also made

it possible – potentially or in practice – to respond to the

same challenges. In reality, analysis in the field confirms

that DMC does not always correspond to the most suitable

solutions to agroecological intensification. However, we can

observe an opening up of certain projects to other agro-

ecological practices from 2008.

1.1. Relevance and coherency

1. Overall evaluation of the projects based on OECD DAC3 criteria

3 Organisation for Economic Co-operation and Development - Development Assistance Committee



This focusing on DMC systems can be found in the

reasoning and plans of action of the intervention. The

underlying hypothesis was that as long as there were

adaptations to local agro-climatic conditions, support from

competent technicians and measures for relieving the

‘constraints’ to their ‘adoption’ (lack of access to inputs, tools

and credit; common grazing land; 4 land insecurity, etc.), the

DMC systems would be ‘adopted’ by the farmers because

they would make it possible to ultimately improve their income.

Participation by farmers in the research and development

process was of course provided for – but not beforehand,

for defining the technical systems of intensification, but rather

afterwards – in order to adapt the latter to local contexts

and to perfect them. What was sought was thus not to respond

foremost to the problems experienced by farmers, but to

promote the diffusion of DMC.

The implementation of these systems was generally

hindered by the absence of long-term perspectives and

continuity in the funding. A certain continuity was nonetheless

provided by the succession of projects on a single piece of

land, which made it possible to pass on knowledge and

operational networks. But it may be difficult to grasp the

coherency of these successions of projects. Furthermore,

the periods without funding between two projects often

became a decapitalisation that was harmful for the action,

especially from the scattering of the trained human resources.

The monitoring and evaluation systems quite often turned

out inadequate for providing hindsight for projects and for

reorienting their strategy if needed. This would have required

not only to recording the results of the diffusion of the technical

model, but above all understanding the mechanisms that

are behind it. However, in reality, the monitoring and evalua-

tion measures were often limited to collecting quantitative

‘adoption’ indicators (DMC surface area or number of farmers

concerned). There was a lack of more qualitative studies,

which would have made it possible to better understand

the choice of farmers. It must be pointed out that the project

logical frameworks did not encourage more analytical

approaches to monitoring-evaluation.

Photo 2. Madagascar, Vakinankaratra: hedging with tephrosia hedge

Photo 3. Vakinankaratra (Madagascar): dried dung powder, used as organic manure on rice plot

Photo credit: Cathy Clermont-Dauphin.

1.2. Effectiveness, results, efficiency

4 Common pasture land: Collective right authorising herders to graze their animals on certain land, particularly on stubble fields and crop residues on all the plots that have been harvested.



By adding up the surface area cultivated in DMC at the

end of the projects in the countries for which these data are

available (Cambodia, Cameroon, Madagascar, Morocco,

Tunisia), we obtain a total (approximate and dispute) 5 of

10,700 ha. The diffusion of ‘full’ DMC (i.e., including no tilling,

permanent coverage and crop rotation) is very variable from

one site to another, and is everywhere very inferior to the

objectives. There are also high drop-out rates after peaks

of adoption, which are either due to opportunistic behaviour

by the farmers (who ‘adopt’ to benefit from the services of

the project), or show the difficulty in carrying out the invest-

ments required. At the same time, we often see a spontaneous

process of innovations, which take up elements from the

DMC system and recombine them in original systems that

respond better to the economic logic or the capacities of

the producers. The monitoring-evaluation systems have not

systematically included the monitoring of these adaptations,

or have done so too late.

One assessment of the effectiveness/cost ratio was carried

out, but it does not seem very significant in such R&D projects,

all the more so because we do not yet have the advantage of

hindsight. It was nevertheless possible to make approximate

calculations of the DMC-related expenses per hectare of DMC

on several sites. These give approximate results ranging from

EUR 1300/ha (in Cameroon) to EUR 12,000/ha (in Laos and

Madagascar).

5 In Lake Alaotra, for example, the project’s capitalisation document estimates the DMC surface area for the 2012-2013 season at 2,601 ha. But this figure includes the farmers who adopted it in the ‘first year’, even though several studies conducted on the site estimate that more than 60% abandoned the project after this first year.

Box 4. Monitoring-evaluation: The need to combine qualitative with quantitative – examples in Madagascar and Laos

For the BVLac project in Madagascar or on the PASS project sites in Laos, the project technicians recorded the statistical

data that gave information on the models of adoption finalised by the research. However, they never really acquired the

skills enabling them to perform critical analysis of the results they observed among the farmers they followed. They thus

had difficulty understanding the results they obtained, especially the drop-outs, which were experienced as ‘failures’, even

though these were often only perfectly rational adaptations by the farmers concerned.

At the same time, an impressive number of more qualitative analyses and diagnostics were carried out on the perimeter

of Lake Alaotra. These were often very interesting in terms of understanding the dynamics at work, but they did not have

much impact on the technicians’ practices in the field.

Photo 4. Middle Sebou (Morocco), Direct Seeding (wheat after wheat)

Photo credit: Julie Sorèze.



Summary data on the technical and economic impacts of

setting up DMC systems among smallholders is lacking, even

for the first-circle projects. Indeed, the first adoptions of the

systems are less than six years old, and the continuity of DMC

‘adoption’ can be judged only over the long term. Furthermore,

it may be misleading to single out the ‘DMC effect’ strictly

speaking from the effect of supply in chemical fertiliser,

which was promoted at the same time as DMC. Generally,

yield increases are fairly low and delayed in time. At the

Vakinankaratra site, four to five years after adoption, rice

yields increased 3 to 13% and maize yields around 50%.

Several DMC environmental services were revealed within

the framework of the cross-cutting RIME research (carbon

sequestration, diversity of plant cover, limitation of trickle-down

and erosion), though with a certain variability. The RIME results

nevertheless also show that the herbicides often used in DMC,

especially glyphosate and its metabolite AMPA, can persist

in the field for at least one year and migrate into sediments

and into trickle-down water at doses much higher than those

tolerated for drinking water.

The impacts in production and diffusion of knowledge are

sometimes appreciable. They include:

• The publication of scientific references as part of the RIME

project, but more limited from the experimentation sites. The

references are moreover often difficult to use in operational

terms.

• Training and awareness-raising actions linked to DMC-

related themes, targeting a diverse public (smallholder farmers,

technicians, decision-makers, etc.).

With regard to institutional impacts, when the projects were

implemented by stable institutions, they helped reinforce the

latter. Impact was weak when they were implemented by an

ad hoc organisation or by operators not designed to work in

the same areas on an ongoing basis.

The projects’ exit strategies or strategies for continued actions

are variable. They include the existence of another project that

takes over (for instance the succession of BV Lac 1 and BV

Lac 2 projects in Madagascar), transfer to existing institutions

(SODECOTON in Cameroon, and IGAD in Gabon) or the

setting up of ad hoc systems (the ‘Maize Fund’ in Laos), with

more or less success.

1.3. Impacts and sustainability

Box 5. After project completion, what relays?

In Laos, the PASS actions were always carried out

with the help of the agricultural extension service of

Sayaburi Province. Currently, several years after termi-

nation of the project, this service is maintaining a small

team on-site, which is continuing to monitor groups

using DMC and is developing value chains for impor-

ting material for cultivation. They have also set up an

original funding mechanism (the ‘Maize Fund’) based

on voluntary contributions from export traders.

In Cameroon, SODECOTON, a powerful public institution

in charge of developing the cotton value chain, was also

able to integrate elements from the agroecology projects

into its own R&D system.

In Gabon, it was the Institut Gabonais d’Appui au Dé-

veloppement (Gabonese Institute for Development

Support – IGAD), an association founded by Agrisud,

the Gabonese state and Elf Gabon that developed DMC.

All these configurations enable the activities to continue

at least minimally when the projects are interrupted. In

other cases, the local institutions managing the projects

are too weak to maintain activities – beyond the formal

end of the project – that require long-term support.



DMC innovations are complete systems that shake up the

entire productive organisation of the farms concerned, and

they must be considered as medium-term investments. They

therefore fuel complex interactions between the economic and

ecological environment on the one hand, and the processes

of capital accumulation and productive organisation of

producers within their societies. There is thus no simple

typology of ‘favourable’ or ‘unfavourable’ agrarian situations,

but a range of factors that encourage or, on the contrary,

impede the adoption of DMC.

2. Key factors affecting DMC-related results

2.1. With regard to agricultural production systems:

• Gains in yield and income are often real, but too limited

to provide significant intensification. Furthermore, as they

are scattered over time, the DMC actions represent an invest-

ment with deferred profitability. The interest for DMC then

depends on the farmers’ capital accumulation strategy, which

in turn depends on their resources, other opportunities for

investment and their perception of the risk linked to DMC.

• The choices made regarding use of labour force and of

liquid assets. The choice of farmers largely depends on two

elements, namely a) the implications of DMC on the farmer’s

labour calendar and finances, and b) opportunity costs

between labour force and available finances. However, DMC

may generate savings in labour as much as new peaks in

work. In motorised systems, the drop in costs linked to the

absence or reduction of tilling nonetheless represents a

recurring factor of success.

Box 6. Farmers’ interest in DMC highly dependent on labour time and calendar

On the tanety of Lake Alaotra in Madagascar, the labour

savings gained from no longer needing workers is more

than offset by the difficulty in manually controlling the

cover plants.

In recently cleared land in Battambang, Cambodia,

the most dynamic pioneers highly appreciate DMC,

insofar as it makes it possible to take advantage of

a mechanised sowing service that helps avoid use of

paid labour, which is expensive in frontier zones (see

Photo 5).

In Morocco and Tunisia, the main motive for farmers’

interest in DMC is the related cost savings, as all

the mechanised work is often carried out by outside

services.



• Whether or not there is competition with livestock, or,

on the contrary, synergies. Using a cover plant may act as

an opportunity for fodder, but keeping biomass as ground

covering conflicts with its use for feeding.

Grazing rights 6 often limit the use of DMC. However, individ-

ual or collective solutions sometimes make it possible to

overcome these constraints.

Photo 5. Cambodia: direct sowing equipment

Photo credit: Laure Montchamp.

2.2. With regard to collective constraints and relations between systems of production

• The agroecological conditions influence both the easiness

of DMC implementation (for example, good rainfall furthers

biomass production and thus the setting up of DMC) and its

agronomic impacts (e.g., the impact of mulch 7 depends on

the rainfall).

• The level of soil degradation and land pressure. Farmers’

effective perception of a fertility crisis largely determines

their interest in DMC. This interest is stronger when there

is not much land available and thus when land pressure

is high (see Box 7). On the other hand, in frontier areas, the

dynamics of land conquest or the comparative advantages

of long-fallow systems do not encourage farmers to intensify

their crop systems as would be required by DMC.

• The existence of favourable market conditions. When they

can market their products at good price, farmers are able to

make the new investments profitable.

• Land tenure. Good security for access to land stimulates

farmers to invest in the long term in the fertility of their soils.

In this context, in which a range of factors that encourage or,

on the contrary, impede the adoption of DMC, the existence

of overall analyses and the implementation of interactive and

flexible action plans are decisive. The top-down model of action

advocated was a strong constraint to the effective diffusion

of the recommended techniques. However, there was an

evolution in systems from PAMPA’s launching in 2007.

2.3. With regard to relations with the environment

Box 7. Land pressure, crucial for maintaining DMC – an example in Laos

In Botene (Laos), the DMC systems were continued

more than elsewhere after the PASS project was over.

As this is where land pressure is highest in Sayaburi

Province, the farmers have no other choice than to

intensify their practices in a way that does pose the

risk of soil loss due to post-ploughing erosion.

6 Grazing rights: collective right authorising breeders/rearers to graze their animals on certain pieces of land, in particular once the stubble and crop residue has been gathered.

7 Mulch: Material, such as straw, decaying leaves, bark, strubble, compost, or plastic cover, spread around or over a plant to reduce evaporation and erosion, to suppress weeds and protect roots from excessive temperatures



The PAA was structured in an original way, with an ambitious

objective of the simultaneous development of practices and

knowledge related to DMC. This was to allow large-scale

learning, with the use of a French body of skills and research,

focused on DMC development.

Certain characteristics of the system furthered such learning,

in particular:

• Its length (14 years), with several successive phases,

but continuity as much in the field as in the cross-cutting

programmes;

• The combination of pilot projects in various contexts and

a cross-cutting programme of exchange and capitalisation

of knowledge, enjoying flexible resources;

• Integration of research and development as much in

institutional terms as at the level of planned activities.

While the programme was an occasion for significant learn-

ing, it also revealed limits that were only partially corrected as

the projects went along. For example, it remained restricted

to an enthusiastic but too small a team for a long time, as much

on the research side (CIRAD) as in management by AFD.

The running of the programme remained rather restricted.

The joint management by the two cross-cutting programmes

was insufficient and without real improvement over time:

absence of regular reports, deficient monitoring-evaluation,

insufficient formalisation of the sharing of responsibilities

between project owner and project manager and of

responsibilities of the partners, and absence of specific

scientific steering. Several recommendations from the

final evaluation of the second phase (PAMPA) thus remain

pertinent: opening up to new teams, increase in resources

for its information management, and improvement of the

scientific monitoring.

3.1. How the tool/programme was structured

3.2. Programme management

3. Assessment of the tool/programme and of the learning approaches



Knowledge production was weak during the PTA but

improved noticeably during the PAMPA. The RIME initiative

made it possible to accumulate knowledge about DMC, its

impacts on the environment and the constraints to their

adoption according to types of environment and agrarian

systems. It furthered gradual change in the method of action

and a broadening of its field of action. The research carried out

by the multi-disciplinary teams was shown to be worthwhile.

However, the extensive documentation stemming from the

actions or dealing with them (various studies, abundant grey

literature) has remained not very accessible or usable and

insufficiently summarised.

AFD had trouble ensuring the programme’s scientific

monitoring and considered that CIRAD should take care of

it. However, this had not been formalised between AFD and

CIRAD through a framework agreement, for example. CIRAD’s

mechanisms for evaluating its researchers and research

units focus above all on the scientific quality of their publi-

cations; therefore they alone do not enable evaluation of

the programme as a whole (i.e., a set of finalised research

actions, or of action research, that must both present criteria

of standard scientific quality and respond to the development

questions at the heart of the programme). Finally, it would

have been desirable to entrust the scientific evaluation to

peers from outside the institution in charge of the research,

in order to avoid institutional conflicts of interest and to

guarantee a plurality of points of view and disciplines.

The learning capacity of AFD and its partners over time

was manifested by evolutions in the following approaches:

• Incorporation of the ‘farm’ approach, in particular with

analysis of the socio-economic constraints within the farms

with regard to DMC adoption;

• Broadening of the range of proposed innovations: ‘ICSs’

(‘innovative cropping systems’) that no longer match the

strict definition of DMC, use of joint farming and livestock

production, agroforestry practices, erosion control;

• The taking into account, at the local (terroirs) level, of

practices and collective rights that can encourage or, on

the contrary, slow down technical change (grazing rights, land

tenure);

• Involve farmers and smallholder farmer organisations

ornetworks morer in designing experimentation actions and

in the diffusion of innovations.

3.3. Knowledge production

3.4. Scientific monitoring

3.5. Learning



This evolution was nevertheless uneven depending on the

project, with countries where it occurred faster (Cameroon and

Madagascar) and others later or… not at all.

The evolutions observed nonetheless did not put into question

either the priority given to DMC as agricultural advice, or the

traditional approach based on experimental research aiming

for the ‘diffusion’ of technical solutions. Therefore, there was no

research site where an intervention methodology was applied

that would a) be based on identifying the objectives, problems

and needs of farmers based on a participative process; b) give

priority to exchanges among farmers to find solutions; and c)

determine objectives for the research and agricultural advice

according to these objectives, problems and needs.

Photo 6. Lake Alaotra (Madagascar): plot of vetch used as organic manure

Photo credit: Aurélie Vogel.



A courageous and ambitious change in paradigm…

Fourteen years of actions by AFD and its PAA partners

represent considerable effort. At the beginning of the 2000s,

the importance of working to develop sustainable solutions

to the intensification of agriculture was indeed far from being

appreciated by all, and AFD, the French MoFA and FFGE

were able to be precursors. This was a change in paradigm

after decades of research and development focused on the

Green Revolution. Furthermore, even though AFD did not

have a mandate for funding the research, the choice of

combining research and development within a joint

programme with the French MoFA and FFGE, and then of

adding a cross-cutting dimension of reflection, search for

support and capitalisation, was ambitious, courageous and

appropriate.

Today, hindsight allows us to see how far we have come,

to draw lessons from the experience – including and above

all the areas where insufficiencies have been identified –

and to seize on this capital to continue to build the future.

… but a questionable approach and system

While the programme initiators were far-sighted regarding

the idea and the principle, the way in which they went about

it seems more questionable. The choice of restricting them-

selves to one particular technical model, DMC, confined

the programme to the promotion of a pre-established

model. However, agroecological solutions are based on

the interaction of biological parameters and are naturally

dependent upon the diversity of environments. The eminently

variable socio-economic conditions in which the farmers

live also determine the possibilities not of adoption but of

integration of these solutions in the pre-existing production

systems and agrarian systems.

There can therefore be no pre-established universal solution,

differently from the Green Revolution model, which was largely

free from the constraints of local diversities by having artificial-

ised the environment by dint of irrigation, chemicalisation and

genetic simplification.

The programme remained deaf for a long time to the alerts

that had been given by both the initial evaluations and the

research sectors (including CIRAD) or the development world,

which over the same period was exploring many other agro-

ecological alternatives (joint use of farming and livestock

production, agroforestry, organic farming, etc.). The DMC model

did, of course, enjoy the double benefit of:

• Being able to concentrate efforts on one solution that, as

long as there were adaptations, seemed ‘ready-made’ and

that, compared to existing references, gave hope for significant

results within a brief period of time; and

• Mobilising AFD’s resources at a time when agriculture was

not a priority.

But due to the use of a single research team, which

moreover is not always inclined to cross-cutting collabo-

rations or to put into question its principle and methods of

intervention, and on a single technical model, the programme

was isolated instead of being put at the heart of society.

The initial projects, for example, all applied the same method-

ology and largely ignored a certain amount of experience

that had been obtained in development: system approach,

importance of prior agrarian analysis and participation by

farmers beforehand in defining the objectives and forms of

intervention, need to take into account local realities, and

absence of a single miracle solution (even if adaptations of

DMC systems had been proposed to different contexts), etc.

4.1. Conclusions

4. Conclusions and recommendations




All these projects created many reference techniques –

which is a significant gain. The research on DMC confirmed

that the latter made it possible to restore fertility to degraded

soil, to fight erosion, and therefore improve potential

yields after several years. But they failed overall in providing

solutions diversified enough to meet the needs of the farmers

in the regions concerned.

Diffusion of DMC thus remained at a low level, despite the

mobilisation of significant resources. The high drop-out rate

for DMC after the end of the projects shows the low level

of lasting quality of the systems proposed, although the

short duration of the projects relative to the objective sought

after (change in modes of production) also contributed to

this drop-out. There are nevertheless exceptions, in agrarian

situations where certain categories of farmers have a direct

interest in investing in DMC and in which collective mobilis-

ation makes it possible to adopt common rules favourable

to this technical change.

Furthermore, the running of the programme remained

rather restricted, and AFD and its partners encountered

great difficulties in managing the programme as a whole as

well as the applied or finalised research that was an integral

part of it.

Evolutions and initiatives that are limited but promising for the future

From 2007, the internal evolutions at both AFD and CIRAD

opened up the way to more fertile research and capitalisation

initiatives, especially the RIME initiative, which was a true

success. However, its capitalisation, syntheses and the com-

munication of results to development stakeholders meant that

appreciation for the programme did not match the level of

knowledge created.

In the field, the putting into question of the top-down model

lasted longer and was more limited. Some projects never-

theless opened up in terms of contents and methods, in

particular by incorporating the ‘farm approach’ and then the

‘terroir approach’, often following local inducement (‘mavericks’

at the Madagascar centre and SODECOTON in Cameroon).

However, these projects are reaching completion before

these new orientations have really been able to be evaluated.

However, despite the top-down model of diffusion, these

projects generated innovations that were not initially

planned. The smallholder farmers often seized on technical

elements promoted by the projects, especially the cover

plants, and reintegrated them their own way into their

production systems: mulch on intensive market gardening,

green fertiliser crops, new fodder combinations, hedging,

or direct seeding on crop residue. These farmer innovations,

which at the beginning were largely ignored by the research

and development bodies, also contributed to the evolution

of the projects.

All these initiatives represent a potential capital upon which

it would certainly be possible to build the third generation

of a large-scale cross-cutting programme, which would

undoubtedly be of interest. Indeed, France cannot remain

on the sidelines of an international sustainable agriculture

movement that is going to intensify both in the North and

the South. The success of such slogans as ‘Doubly Green

Revolution’ or ‘Ecologically Intensive Agriculture’ shows

that the research and development world has advanced to

a stage where it puts sustainability and the agriculture-

environment relationship at the heart of its concerns.



Some principles drawn from the experience of the two

previous phases could give structure to such a programme.

A technical, methodological and operational synthesis

To start with, it would be important to round out the RIME

capitalisation with a synthesis of the experiences gained

from the whole programme, combining analysis of scientific

publications, available reports and statements from experts.

The idea is to develop an informed opinion on the concrete

issues related to DMC and to produce a reference document

intended for students, people in the development world,

decision-makers and non-specialised researchers. The

projects supported by AFD would provide the raw materials

for this synthesis, but any project having contributed to

creating references should also be taken advantage of.

Continue the capitalisation research with a cross-cutting programme, with a mandate enlarged threefold

Next, it is essential to conserve the linkage between a

cross-cutting capitalisation programme and a group of field

actions that fuel it and that can welcome and develop

research or development experiments. It is just essential to

enlarge the current scope of action clearly and transparently

in three different directions:

• At the technical level, all the agroecological systems

should be considered;

• The capitalisation should cover the issue of forms of

intervention. This implies supplementing the adaptive diffus-

ion developed up to now by a broader set of practices:

smallholder farmer schools, ‘farmer-to-farmer’ processes,

‘R&D terroir’-type approaches, etc;

• It is important to cover the field of agricultural and

development-assistance policies that are likely to promote

the ecological transition in agriculture. The experience gained

from the PAA demonstrates the difficulty that smallholder

farmers in insecure situations can encounter when it comes

to implementing agroecological solutions. It should be pos-

sible to deal with the issue of subsidising this transition.

Internationalise the network of reference stations

It is necessary to continue the work of creating references

thanks to a network of stations, as CIRAD had done for

DMC, but on a long-term basis and with a broader range of

technical systems. Above and beyond the results (yields,

margins, environmental impact, etc.), these experiments

should deal with the study of processes leading to these

results, because it is this understanding that makes it pos-

sible to use the models in diverse situations.

This being the case, the creation of agroecology refer-

ences is of global public interest, and it would be incoherent

for each donor to fund the development of its own network

of references. We thus recommend starting up high-level

discussions, especially at the European level, pooling efforts

in this direction and considering supplementary support

from the European Commission.

A broader range of projects

The range of projects considered should be much broader

than now, by opening it up to innovative (and often smaller

and more flexible) projects implemented by NGOs, consult-

ancies, agricultural organisations, and research networks

or institutes, as long as these operators have reference

terms and clear responsibilities with regard to objectives,

expected results and outcomes, and are connected to the

cross-cutting component. The current projects should be

continued, by reorienting them according to the gains in

experience. Generally speaking, it is important for all the

projects to be able to receive support over the long term;

this is an essential condition for effectiveness in the agro-

ecological domain. Stricter monitoring-evaluation mechan-

isms should be implemented, in relation with feedback from

stakeholders (researchers, development bodies, farmers,

smallholder farmer organisations), the external scientific

monitoring (agronomic and socio-economic) and the monitor-

ing by donors.

4.2. Operational recommendations



While it is not the role of AFD to fund pure research projects,

this is not the case for projects that put into relation exper-

iments in smallholder farming, monitoring-evaluation of

changes, and more large-scale development of innovative

techniques. Such projects, in which the themes of experiment-

ation and research would be determined with participation

by smallholder farmers and their organisations, could be of

interest to and call on conventional research for more funda-

mental work on specific themes.

From the diffusion of a model to support for change

Above and beyond the diversification of technical solutions,

renewal of producer-support approaches is indispensable.

This involves shifting from an approach of diffusion of a

model to an approach of support for change. Experiences

exist, outside of the scope of DMC, that make it possible to

bring together method references. The programme will have

to incorporate the lessons learned from DMC:

• Agroecological systems are sensitive to environmental

conditions, which are themselves diversified. They thus require

great capacity of adjustment to these conditions;

• They can work only by mobilising knowledge, which is

broadly distributed between the farmers themselves and a

variety of stakeholders, and not concentrated at the research

and agricultural extension levels;

• Agricultural innovations originate not from linear diffusion,

but from processes of deconstruction and reconstruction

of the systems proposed. The spontaneous dynamics of

change thus contribute to creating useful references.

The consequences of these upheavals in paradigms, in

terms of project organisation and training of contributors

to the projects (and farmers) will have to be reflected on

specifically within the framework of the capitalisation

envisaged at the start of the programme.

Outsource the lead contracting of the cross-cutting programme

The governance of the cross-cutting programme should

avoid both its monopolisation by a single institution and the

lack of monitoring by AFD. The lead contracting could be

provided by a research and development institution that is

not likely to carry out the work directly, but is capable of

conducting active cross-cutting activities (of an Agreenium

or Agropolis International type). This governance, which

should associate equally actors of development (including

NGOs), decision-makers and researchers, would decide on

the actions to carry out, with an explicit definition of the

terms of reference and the expected results and outcomes

for each type of these actions. It would evaluate the results

of the actions and the progress of the programme. Compet-

itive calls for tender or proposals would be set up, open to

all. The research financed by AFD should be finalised, but

other financial mechanisms of an Agence Nationale de

Recherche (French National Research Agency – ANR)

type could be used to fund more long-term and more open

research. A scientific research-evaluation committee made

up of internationally recognised key figures should be set

up, along with sufficient resources to carry out an evaluation

of the programme’s scientific actions and publications, and

provide recommendations on its strategy and orientations.

If France were endowed with such a system and wanted

to rise to the challenges, it could then negotiate with major

foundations, or with CGIAR (Consultative Group on Inter-

national Agricultural Research), to multiply the system that

it would have thus initiated.



Acronyms and AbbreviationsAFD Agence Française de DéveloppementANR Agence nationale de recherche (National Research Agency)AVSF Agronomes et vétérinaires sans frontières (Agronomists and Veterinarians without Borders)BV Lac Lake Alaotra watershed (project)BVPI Watershed irrigated area (project)CARI Centre d’actions et de réalisations internationales

(Centre for International Action and Achievements)CGIAR Consultative Group on International Agricultural ResearchCIRAD Centre de coopération internationale en recherche agronomique pour le développement

(Agricultural Research Centre for International Development)CSI Centre de sociologie de l’innovation (Centre for the Sociology of Innovation)DAC Development Assistance CommitteeDGPT Farm Development and Land Management Project DMC Direct Seeding Mulch-based CroppingFFGE French Facility for Global EnvironmentIDDRI Institut du développement durable et des relations internationales

(Institute for Sustainable Development and International Relations)IGAD Institut gabonais d’appui au développement (Gabonese Institute for Development Support)INRA Institut national de la recherche agronomique (National Institute for Agricultural Research)IRAM Institut de recherches et d’applications des méthodes de développement

(Institute for Research and Application of Development Methods)IRD Institut de recherche pour le développement (Development Research Institute)MAAF Ministry of Agriculture, Agrifood and Forestry MoFA Ministry of Foreign AffairsNGO Non-Governmental OrganisationOECD Organisation for Economic Co-operation and Development PAA Agroecology Action PlanPAMPA Multi-country Action Programme in Agroecology PASS Application Point in Southern Sayaburi (project)PDRI Integrated Rural Development ProjectPPI Small irrigated areaPRODESSA Rural development project in the South of Sayaburi ProvincePTA Cross-cutting Support Programme RIME Multi-team Integrated Response (initiative)SAM Mountain Agrarian Systems SCI Innovative Crop System

Dernières publications de la série Les numéros antérieurs sont consultables sur le site : http://recherche.afd.fr

Previous publications can be consulted online at: http://recherche.afd.fr

N° 57 Évaluation des interventions de l’AFD dans les secteurs sanitaire et médico-social en Outre-merN° 56 Évaluation des activités de Coordination SUD dans le cadre de la convention AFD/CSUD 2010-2012N° 55 Évaluation et impact du Programme d’appui à la résorption de l’habitat insalubre et des bidonvilles au MarocN° 54 Refining AFD’s Interventions in the Palestinian Territories – Increasing Resilience in Area CN° 53 Évaluation des lignes de crédit de l’Agence Française de Développement octroyées à la Banque ouest-africaine de

développement (2000-2010)N° 52 Évaluation stratégique de projets ONG dans le domaine de la santé (Mali, Burkina Faso et Cambodge)N° 51 Secteur de l’hydraulique pastorale au Tchad – Évaluation et capitalisation de 20 ans d’intervention de l’AFDN° 50 Réhabilitation des marchés centraux – Les leçons tirées des projets de Ouagadougou, Mahajanga et Phnom PenhN° 49 Bilan des évaluations décentralisées réalisées par l’AFD en 2010 et 2011N° 48 Étude sur la facilité d’innovation sectorielle pour les ONG (FISONG)N° 47 Cartographie des prêts budgétaires climat de l’AFDN° 46 Méta-évaluation des projets « lignes de crédit »N° 45 Bilan des évaluations de projets réalisées par l’AFD entre 2007 et 2009N° 44 Impacts des projets menés dans le secteur de la pêche artisanale au SénégalN° 43 L’assistance technique résidente – Enseignements tirés d’un appui au secteurde l’éducation en MauritanieN° 42 Évaluation partenariale des projets d’appui à la gestion des parcs nationauxau MarocN° 41 AFD Municipal Development Project in the Palestinian TerritoriesN° 40 Évaluation ex post de 15 projets ONG à MadagascarN° 39 Analyse croisée de vingt-huit évaluations décentraliséessur le thème transversal du renforcement des capacitésN° 38 Étude des interventions post-catastrophe de l’AFDN° 37 La coopération française dans le secteur forestier du Bassin du Congosur la période 1990-2010N° 36 Suivi de la réalisation des objectifs des projets de l’AFD : état des lieuxN° 35 Cartographie des engagements de l’AFD dans les fonds fiduciairessur la période 2004-2010N° 34 Addressing Development Challenges in Emerging Asia:A Strategic Review of the AFD-ADB Partnership Final Report, Period covered: 1997-2009N° 33 Capitalisation des démarches pour la mise en oeuvre des projets de formation professionnelle :

cas de la Tunisie et du MarocN° 32 Bilan de l’assistance technique à la Fédération des paysans du Fouta Djallon(FPFD) en Guinée :

15 ans d’accompagnementN° 31 Adapter les pratiques opérationnelles des bailleurs dans les États fragilesN° 30 Cartographie de portefeuille des projets biodiversité Analyse sur la période 1996-2008Cartography of the AFD

Biodiversity Project Portfolio:Analysis of the Period 1996-2008N° 29 Microfinance dans les États fragiles : quelques enseignements de l’expérience de l’AFDN° 28 Un exemple d’amélioration de la gouvernance locale à travers le partenariat AFD / coopération décentralisée : capitalisation du projet de réhabilitation des marchés de MahajangaN° 27 Pratique de l’aide sectorielle : enseignements et perspectives pour l’AFD Sector Program Support in Practice:

Lessons and Perspectives for AFDN° 26 L’appui à l’hévéaculture familiale : capitalisation sur l’expérience AFD Developing Smallholder Rubber Production :

Lessons from AFD’s ExperienceN° 25 Évaluation rétrospective du projet FFEM d’efficacité énergétique dans la construction en Afghanistan Ex-post

Évaluation of the FGEF Energy Efficiency Project in the Construction Sector in AfghanistanN° 24 Évaluation des “Cadres d’Intervention Pays” (CIP)N° 23 Études d’évaluation de la société immobilière de Nouvelle-CalédonieN° 22 Les collaborations opérationnelles entre l’AFD et les ONG 2010 2010 Évaluation prospective • Projet Urban IV • N° 21 Cartographie des projets d’efficacité énergétiques et d’énergies renouvelables AFD et FFEMN° 20 Évaluation de l’usage de la concessionnalité dans les interventions de l’AFD en Afrique du Sud (1995/2005)

ex post n°58

Documents

evaluation possible

overall evaluation

laurent levard gret

constance corbierbarthaux

laure montchamp afd

christian castellanet

aurlie vogel gret

capitalisation series