Délivré par L`Ecole supérieure en Agronomie - Umr-System
Transcription
Délivré par L`Ecole supérieure en Agronomie - Umr-System
Délivré par L’Ecole supérieure en Agronomie-Montpellier SupAgro Préparée au sein de l’école doctorale SIBAGHE Et de l’unité de recherche UMR SYSTEM Spécialité : Fonctionnement et conduite des systèmes de cultures tropicaux et méditerranéens. Présentée par ROZA CHENOUNE Quels leviers pour promouvoir la production et la consommation du riz en Sierra Léone ? De la caractérisation à la simulation de la performance des ménages rizicoles. Soutenue le 19/11/2014 devant le jury composé de M. Marco ACUTIS, Professeur, Université de Milan Mme Karine DANIEL, Enseignant Chercheur, INRA, Nantes M. Sergio GOMEZ Y PALOMA, Docteur, JRC M. Jacques WERY, Professeur, SupAgro M. Alain CAPILLON, Professeur, SupAgro M. Hatem BELHOUCHETTE, Enseignant Chercheur, IAMM Rapporteur Rapporteur Examinateur Examinateur Directeur de thèse Co-encadrant 1 Délivré par L’Ecole supérieure en Agronomie-Montpellier SupAgro Préparée au sein de l’école doctorale SIBAGHE Et de l’unité de recherche UMR SYSTEM Spécialité : Fonctionnement et conduite des systèmes de cultures tropicaux et méditerranéens. Présentée par ROZA CHENOUNE Quels leviers pour promouvoir la production et la consommation du riz en Sierra Léone ? De la caractérisation à la simulation de la performance des ménages rizicoles. Soutenue le 19/11/2014 devant le jury composé de M. Marco ACUTIS, Professeur, Université de Milan Mme Karine DANIEL, Enseignant Chercheur, INRA, Nantes M. Sergio GOMEZ Y PALOMA, Docteur, JRC M. Jacques WERY, Professeur, SupAgro M. Alain CAPILLON, Professeur, SupAgro M. Hatem BELHOUCHETTE, Enseignant Chercheur, IAMM Rapporteur Rapporteur Examinateur Examinateur Directeur de thèse Co-encadrant 2 Merci Tout d’abord, un grand Merci aux personnes qui ont cru en moi et qui m’ont permis d’arriver au bout de cette thèse de doctorat. Je tiens à exprimer mes plus vifs remerciements à Alain CAPILLON, professeur en Agronomie (SupAgro Montpellier), qui fut pour moi un directeur de thèse exemplaire, pour ses conseils, disponibilité, sa patience, son esprit scientifique, et compétence m’ont beaucoup appris. J'ai beaucoup apprécié de travailler à vos côtés tant sur le plan scientifique que sur le plan humain. C’était une expérience très riche pour moi d’avoir eu la chance de partager des débats et de me faire partager son expérience. Merci beaucoup Alain ! J’exprime tous mes remerciements à Hatem BELHOUCHETTE, enseignent chercheur à l’IAMM qui a co-encadré cette thèse de doctorat de très près. Merci pour avoir accepté de m’accompagner pendant cette thèse en tant que tuteur pédagogique, de ta confiance et de m’avoir permis de découvrir la Sierra Léone avant la propagation du virus Ebola ! Merci pour tes précieux conseils quant à l’avancée scientifique de ce travail. Merci pour ta patience, ton soutien, ta grande générosité et le temps que tu as consacré à ce travail, et sans qui cette thèse n'aurait jamais vu le jour. Je le remercie très sincèrement. J’adresse toute ma gratitude aux rapporteurs : Karine DANIEL enseignante chercheur à l’université d’Angers et Marco ACUTIS enseignant chercheur à l’université d’Italie, qui ont accepté de juger se travail, et de m’apporter leurs expériences dans le domaine. Merci d’avoir accepté d’évaluer cette thèse et d’avoir contribué aux discussions. Mes sincères remerciements et ma gratitude vont à Sergio gomez y paloma pour avoir accepté de juger ce travail et d’en présider le jury de soutenance. Merci à jacques WERY professeur en agronomique Montpellier SupAgro qui a accepté de juger ce travail en tant qu'examinateur. Je lui adresse mes sentiments les plus respectueux. Je ne pourrai pas continuer mes remerciement sans cité Guillermo FLICHMAN, enseignant émérite à l’IAMM-CIHEAM qui a participé de près à la réalisation de ce travail. Merci Guillermo pour tés disponibilités, ton soutien, ta bonne humeur et surtout ton côté humain ! Un très grand merci à Myriam ADAM, Pablo TITTONEL et Anthony WITHBREAD pour leurs collaborations, relecture et disponibilités. Merci pour vos conseils, j’espère que nous travaillerons un jour ensembles ! 3 Je remercie également Thomas ALLEN, pour sa collaboration dans cette thèse, ses conseils, et sa disponibilité. Qu’ils trouvent ici ma profonde reconnaissance. Je dois beaucoup à ma famille, merci à mes parents, frères, sœur et beau-frère, pour leur contribution, leur soutien, leur encouragement et surtout leur patience ; et voilà un autre docteur dans la famille ! Merci à Nailoula qui m’a fait beaucoup rire durant cette aventure. Je remercie tous le personnel de l‘IAMM en particulier Martine Padilla responsable de la plateforme doctorale, pour m’avoir accueilli dans le laboratoire de recherche LAMES et m’a permis de travailler dans de bonnes conditions. Un très grand merci Mme ABDELHAKIM Tahani & Mr LE GRUSSE Philipe, enseignant chercheur à l’IAMM pour leur grande sympathie. D’autres aides m’ont été précieuses, je pense aux services informatiques particulièrement à Mapie, Cécile, Dominique et Raphaël. Merci à vous de m’avoir consacré du temps, votre bonne humeur et surtout de m’avoir débloquée techniquement. Je remercie toutes les personnes formidables que j’ai rencontrées par le biais de l’IAMMCIHEAM. Merci pour cette richesse méditerranéenne. Je pense particulièrement à mes collègues : Issam, Ledina, Adolpho, Edward et Guillaume. De manière générale, je remercie mes autres collègues de bureau à savoir : Imen kellou, Imen titouche, Imen souissi, Ouassila, Lamia, Karima, Rym, Guilia, Loubna, Meriem hammouda, Meriem trabelsi, Oussama, Rachid, Valter et Paolo. Mes vifs remerciements vont également aux trios, Saida, Amel et Meriem pour leur soutien, sympathie, encouragement et surtout les bons moments que nous avons passés malgré nos différences ! Cette aventure a été agrémentée par des moments très agréables partagés avec ma chère Radia DAHMANI-ACHOUR, Hakim ACHOUR, Mehdi et Raouf qui ont partagés cette aventure avec moi malgré la distance. Merci Radia pour ta disponibilité et le temps que tu m’as consacré durent cette aventure. Un très grand merci tout particulier aux enseignants de l’Ecole National Agronomique d’Alger (ENSA) qui ont contribué à ma formation. Je pense particulièrement aux enseignants du département de phythotechnique- production et amélioration végétale: Mr Bélarbi, Mr et 4 Mme Amirouche, Mr Reguig, Mr et Mme khelifi, Mme Yekhlef, Mr Abdelgherfi, Mr Ounane. Enfin, j'adresse mes plus sincères remerciements à tous ceux qui, de près ou de loin, ont contribué à la réalisation de cette thèse de doctorat. Une autre aventure qui commence ! 5 Sommaire Résumé ..................................................................................................................................... 11 Introduction générale ................................................................................................................ 15 Chapitre I : Contexte et problématique de l’étude ................................................................... 19 1. Contexte de l’étude............................................................................................................... 19 2. Le secteur agricole ............................................................................................................... 20 3. Les ménages agricoles à Bombali ........................................................................................ 23 4. Problématique....................................................................................................................... 30 Chapter 2 .................................................................................................................................. 36 Assessing the diversity of smallholder rice farms production strategies in Sierra Leone ........ 36 Summary .................................................................................................................................. 36 1. Introduction ....................................................................................................................... 37 2. Materials and Methods ...................................................................................................... 38 2.1 Description of the study area ............................................................................................. 38 2.2 Database ............................................................................................................................ 39 2.3 Levels of analysis and methods for assessing rice growing systems in Bombali ............. 41 3. Results ............................................................................................................................... 43 3.1 Description of cropping cycles for rice and oil palm ......................................................... 43 3.2 Comparative performance of rice per ecosystem .............................................................. 44 3.3 Comparative analysis by type and class of farm .............................................................. 45 3.3.1 General description of the types of farms surveyed ....................................................... 45 3.3.2 Typology: classes of farm .............................................................................................. 45 4. Discussion ......................................................................................................................... 61 5. Conclusion......................................................................................................................... 64 Chapter 3 .................................................................................................................................. 65 Finding pathways out of poverty: The case of oil palm as a cash crop enabling the intensification of subsistence rice based farming systems in Sierra Leone. ............................ 65 1. Introduction ....................................................................................................................... 66 2. Materials and methods ...................................................................................................... 67 2.1. Conceptual framework ...................................................................................................... 67 2.2. Case study: rice-growing households in Sierra Leone ...................................................... 70 3. Results ............................................................................................................................... 75 3.1. Stage 1: “target production vs. consumption” relationship. .............................................. 75 3.2. Stage 2: Farm typology: household categories and production strategy ........................... 76 3.3. Stage 3: Analysis of the performance and efficiency of the farm categories .................... 79 4. Discussion ......................................................................................................................... 84 5. Conclusion......................................................................................................................... 87 Chapter 4 .................................................................................................................................. 90 A household model to assess consumption-production-resources nexus in West Africa: The rice based farming systems in Sierra Leone. ............................................................................ 90 1. Introduction ....................................................................................................................... 91 2. The conceptual model ....................................................................................................... 93 3. Empirical application ........................................................................................................ 99 3.1 The case study area ............................................................................................................ 99 3.2 Choice of representative rice farming households ........................................................... 100 3.3 Farm type data specification............................................................................................ 103 4. Model calibration ............................................................................................................ 105 5. Scenario and indicator specification ............................................................................... 105 6. Results ............................................................................................................................. 107 6 7. Scenario analysis ............................................................................................................. 110 8. General discussion .......................................................................................................... 114 9. Conclusion....................................................................................................................... 117 Chapitre 5 : Discussion générale ............................................................................................ 118 1. Résumé des principaux résultats obtenus. ...................................................................... 118 2. Intensification des systèmes rizicoles ............................................................................. 124 3. Retour sur les choix de production.................................................................................. 128 Annexe_1 : Structure de la base de données .......................................................................... 143 Annexe_2 : les grandes étapes de l’itinéraire technique à Bombali. ...................................... 149 Annexe_3 : Modèle bio économique (Household model)..................................................... 152 Annexe_4 : Politique agricole et rurale en Sierra Leone ....................................................... 153 Annexe_5: Résultats des scénarios ........................................................................................ 158 7 Liste des figures Figure 1 : Carte de la consommation de riz par kg/capita et par an (OCDE, 2011). ............... 15 Figure 2 : Carte des principaux bassins de production de riz (OCDE, 2011). ......................... 16 Figure 3 : Carte de la Sierra Leone .......................................................................................... 19 Figure 4 : Carte économique de la Sierra Leone. Le jaune au niveau de la carte représente les zones où le riz est (ou peut-être) cultivé................................................................................... 21 Figure 5 : Evolution des productions des principales cultures entre 1980 et 2006. ................. 22 Figure 6 : Situation géographique du district de Bombali au nord de la Sierra Leone ............ 23 Figure 7 : Augmentation des surfaces cultivées au détriment des forêts. ................................ 29 Figure 8: Démarche générale de la thèse ................................................................................. 35 Figure 9: Location of Sierra Leone and the Bombali district (www.d-maps.com) .................. 39 Figure 10: Work schedules by type of crop system (rice and oil palm) and by ecosystem. (a) Upland rice; (b) Lowland rice and (c) Oil palms. .................................................................... 44 Figure 11: Median, maximum and minimum values, first quartile, third quartile and mean (values in brackets) for upland and lowland ecosystems for the following variables: yield, surface area and total labour. These variables were calculated from surveys and concern 126 rice fields .................................................................................................................................. 45 Figure 12a: Distribution of farms surveyed (n=81) by classes of farms as a function of PC1 and PC2 .................................................................................................................................... 48 Figure 13: Total surface area, standard deviation (value in brackets) and proportion of the surface area in upland rice, lowland rice and oil palm per class of farm. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). ...................................................................................................................................... 50 Figure 14: Average distance and standard deviation from household to farm by class of farm. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). .......................................................................................................... 51 Figure 15: Average distance and standard deviation from household to farm by type of farm. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). Averages and mean deviations are calculated on the basis of the 81 farms surveyed. ........................................................................................................................ 51 Figure 16: Total labour, standard deviation relative to total labour and proportions (%) of family labour per class of farm relative to the total amount of labour. This analysis was made on the basis of the 81 farms surveyed. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). ..................................... 52 Figure 17: Proportions of male and female labour relative to total labour by task and class of farm. This analysis was made on the basis of the 81 farms surveyed. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). .................................................................................................................................................. 53 Figure 18: Average rice yields per farm and per ecosystem (uplands and lowlands) and per farm as a function of seeding density. Average rice yield per farm is calculated as the weighted average for the surface area of rice grown for both ecosystems (upland and lowland). ................................................................................................................................... 56 Figure 19: Seeding density as a function of the quantity of seed stored per class. .................. 57 Figure 20: Average family size (number of family members) and standard deviation per class of farm as a function of the proportion (as a percentage of total production) of the quantities consumed, sold and stored. ...................................................................................................... 57 Figure 21: Quantity of rice stored as seed as a function of family size for farms not cultivating oil palms ................................................................................................................................... 58 8 Figure 22: Quantity of rice stored per farm as a function of the proportion of surface area allocated to oil palms per farm ................................................................................................. 58 Figure 23: Performance rice tray depending on the length of fallow. Farms are surrounded farms rice and palm oil pan (Rice_U_palm) in low yield (class 3.4) ....................................... 59 Figure 24: Quantity of Labour depending on the sowing fallow periods. Farms are surrounded farms rice tray and oil palm (Rice_U_palm) in low yield (class 3.4) ...................................... 59 Figure 25: Seeding based on the length of fallow. Farms are surrounded farms rice and palm oil (Rice_U_palm) in low yield (class 3.4). ............................................................................ 60 Figure 26: Criteria influencing the decisions of smallholder rice farms in Bombali ............... 63 Figure 27: Phases in explaining and analysing the performance and efficiency levels of ricegrowing systems. ...................................................................................................................... 70 Figure 28: Correlation between farm-level rice production and overall consumption per family. This correlation is shown in three ways: y1: farms growing oil palm, y2: all farms regardless of enterprise, and y3: farms growing no oil palm. .................................................. 76 Figure 29: Variation of the median, maximum and minimum values, first quartile and third quartile for yield, seeding density and total labour for the four categories of farms (C1, C2, C3, C4). .................................................................................................................................... 80 Figure 30: The total labour and seeding density for the surveyed farms (n= 81) according to the farm size. The solid bar indicates the standard deviation of the average value for each category of farm. ...................................................................................................................... 82 Figure 31 : box plots for fallow duration, rice consumption by member of family, seeding density and labour amount for the categories C1 and C2 dominated by the upland ecosystem. .................................................................................................................................................. 83 Figure 32 : Average and standard deviation of rice yield per farm (n = 81 farms) according to seed density and total labour efficiencies. The hatched lines indicate hypothetical limits for different zones of rice efficiencies (yield per seeding density and yield per the amount of labour) and rice yield. ............................................................................................................... 84 Figure 33 : Efficiency of total labour (rice yield to total labour) and seed density (rice yield to seed density) according to the share of lowland rice on the farm. C1, C2, C3, C4 represent the averages per farm category....................................................................................................... 87 Figure 34: Levels of performance and efficiency in a traditional rice-growing household in West Africa (adapted from Tittonell, 2013). ............................................................................ 89 Figure 35: Conceptual representation of the structure and material flows of the consumptionproduction-resources relationship of the rice farming households in West Africa. ................. 95 Figure 36: variation of calorie number and monetary value on the self-consumption by subsidization oil palm subsidization level for the 4 household types. ................................... 112 Figure 37 : variation of the consumption amount of different food products by oil palm subsidization level for the Uplandextensive farm type................................................................ 113 Figure 38 : Variation of rice consumption (a) and total calorie (b) for the 4 farm types, Guinea (average national values), Liberia (average national values) and the targets set by the Sierra Leone Government for the National Agricultural Strategy (2010-2030) and the FAO organization. ........................................................................................................................... 115 Figure 39 : Répartition mensuelle des principales tâches agricoles pour un ménage rizicole au Nord de la Sierra Leone. La courbe de pluie est une moyenne calculée sur la période 19791990. (élaboration personnelle) .............................................................................................. 128 9 Liste des tableaux Tableau 1 : Production de riz entre 1997 et 2009 en Sierra Leone .......................................... 22 Tableau 2 : Importation de riz en Sierra Leone (Mt), 2006-2008 ............................................ 23 Tableau 3 : Niveau de vulnérabilité de la population au nord de la Sierra Leone ................... 24 Tableau 4 : Rendement moyen des principales cultures en Sierra Leone (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). ................. 26 Table 5: Description of the variables used for this study, from Questionnaires A, B and C. The data were collected from farmers in Bombali between March and November 2009. They concern only the 81 farms chosen for the study. Yield is the only variable to have been calculated (total production per hectare) instead of being obtained directly from surveys in the field........................................................................................................................................... 40 Tableau 6: Number of rice farms and number of rice fields per ecosystem (upland or lowland) and per farm for 126 rice fields on the 81 farms selected and surveyed within the study ....... 42 Table 7: Average yield, seeding density, duration of fallow period, total labour and number of farms per class. Farms were classified on the basis of a PCA and HAC from a sample of 81 farms surveyed in Bombali. The total in the table gives the total number of farms and the weighted average and standard deviation corresponding to the weighted average and standard deviation by area of the following variables: average yield, seeding density, length of fallow period and total labour.............................................................................................................. 47 Tableau 8: Correlation values (R²) between yield and the amount of labour by ecosystem, task and type (family, hired, women, men). These values are expressed by class of yield: high rice yields (class 1), Fairly high rice yield (class 2), Fairly low rice yield (class 3) et Low rice yield (class 4) ........................................................................................................................... 54 Tableau 9 : Description of farm type and ecosystem type (upland, lowlands) from the sample of 81 farms identified and surveyed. ........................................................................................ 71 Tableau 10 : Description of the data collected from 81 farms in Bombali between March and November 2009. ....................................................................................................................... 73 Tableau 11: Variables for farms clustering according to structural factors, factors of production and factors of consumption. These variables concern 81 farms survived in the Bombali region. ........................................................................................................................ 75 Tableau 12 : Absolute values of the loadings of the classification variables with respect to the 5 principal components ............................................................................................................ 77 Tableau 13 : Typical households according to rice production per farm and the determinants of farm structure and the inputs used based on PCA. (n= number of farms per category) ...... 79 Table 14 : farm types by considering structural, production and consumption criteria. The table is adapted from (Chenoune et al., 2014) ........................................................................ 103 Table 15 : Observed (OBS) vs. simulated (SIM): crop pattern, production, total labour, consumption and total calories per activity and farm type. The estimated risk-aversion coefficients are 0.8, 1.5, 0.4 and 0.7 respectively for Uplandintensive, Uplandextensive Lowlandextensive, and Lowlandintensive households, “dif” indicates the difference in percentage between simulated and observed values. The grey values indicate a difference of more than 20%......................................................................................................................................... 109 Table 16 : comparison of socio-economic, consumption, total calorie and crop pattern for the three scenarios (Sseeds, Sop, Srice) in comparison to the baseline scenario (Sbaseline). ..... 114 10 Quels leviers pour promouvoir la production et la consommation du riz en Sierra Léone ? De la caractérisation à la simulation de la performance des ménages rizicoles. Résumé La Sierra Leone est un des trois pays les plus pauvres au monde. Cette situation est aggravée par une guerre civile qui a duré dix ans et qui a détruit l’essentiel des infrastructures du pays. Plusieurs initiatives et programmes tentent, depuis le milieu des années 2000, de relancer la production agricole, afin de subvenir aux besoins alimentaires, notamment ceux en riz, des populations rurales. Aujourd’hui, bien que l’agriculture contribue à presque 56% du PIB national, la Sierra Leone reste toujours tributaire des importations et de l’aide internationale pour subvenir à ses besoins en riz. Le nord du pays, et notamment le district de Bombali, est aujourd’hui la principale région rizicole. Néanmoins, cette zone présente des rendements très faibles et très variables par comparaison au potentiel pédoclimatique de la région. L’UE, via son programme STABEX, cherche à promouvoir la production agricole en mettant en place plusieurs initiatives d’ordre technique, mais également des mesures économiques de soutien direct aux riziculteurs. C’est dans le cadre de ce projet que nous avons fixé les objectifs suivants pour ce travail de thèse : i) caractériser la diversité des ménages rizicoles dans le district de Bombali. Cela nous permettra de comprendre, en nous basant sur une analyse de la performance et de l’efficience de la production élaborée dans le cadre de cette thèse, les facteurs de production qui affectent ces rendements, et ii) d’évaluer, via le développement et l’utilisation d’un modèle de ménage basé sur la programmation linéaire, la performance de trois initiatives (déclinées sous forme de scénarios), qui visent à améliorer la production et la consommation de riz (en termes quantitatifs mais également en nombre de calories). Pour réaliser ce travail, nous nous sommes appuyés sur une base de données qui regroupe des informations sur la production et la consommation, complétées grâce à des enquêtes auprès de 181 riziculteurs de la région. Le travail total engagé par parcelle de riz et par exploitation, les doses de semis ainsi que la durée de la jachère (et, par conséquent, la fertilité initiale du sol) expliquent en grande partie la faiblesse et la variabilité des rendements entre les différents types d’exploitation. Il ressort également de notre étude que le palmier à huile joue un rôle très important en tant que culture de rente, mettant ainsi à disposition du riziculteur du cash, indispensable pour intensifier la production de riz. Cette stratégie a également permis à certains riziculteurs d’agrandir leur parcelle de riz (sans forcément intensifier le riz), et, par conséquent, leur production totale au 11 niveau de l’exploitation. Cette option s’est souvent traduite par une baisse de l’efficience (rendement/facteur de production) de la production de riz. Ce résultat est dû essentiellement à une faible fertilité des sols (cas des exploitations dominées par l’écosystème plateau), à des doses de semis faibles (cas des ménages pauvres qui donnent la priorité à la consommation aux dépens du stockage de semence) ou à de faibles quantités de travail disponible (cas des ménages riches et intensifs). Il résulte aussi de notre étude que les ménages traditionnels à écosystème unique seront probablement non résilients face aux différents chocs du marché, aux variations climatiques mais surtout au retour massif des populations après la fin de la guerre. Ce travail a, par ailleurs, montré que la mise en place, préconisée par l’Etat, d’une subvention spécifique pour installer le riz sur les basfonds est l’option qui a le plus amélioré, par rapport à la situation actuelle, la production (de +53% à 78% selon les types de ménages), la consommation (de +15 à 34% selon les types de ménages) et le nombre total de calories par capita et par jour (de 7 à 21% selon les types de ménage). En ce qui concerne les deux autres scénarios (subvention des semences de riz et celle de la plantation de palmiers à huile), ils n’ont engendré, par rapport au premier scénario et aux prévisions des pouvoirs publics, que très peu de changement sur la production, et, par conséquent, sur la consommation de riz. Au final, ce travail a fait l’objet d’importantes réflexions d’ordre méthodologique, non seulement, par rapport à la caractérisation de la diversité des performances et l’efficience des systèmes rizicoles, mais également par rapport aux hypothèses de bases pour construire un modèle de ménage. L’ambition était de reproduire au mieux grâce à ce modèle le comportement des ménages, en termes de production et de consommation par rapport à des ressources limitées, mais aussi par rapport à leur prise de risque en considérant, à la fois, la variabilité climatique et celle du marché. Les méthodes développées et suivies dans cette thèse sont, à notre avis (au-delà des problèmes d’acquisition des données), facilement extrapolables, non seulement à d’autres régions de l’Afrique de l’Ouest, mais également à d’autres systèmes de culture et de production aussi importants que ceux du riz. Mots clés: Sierra Léone, système de production rizicole, analyse de performance et d’efficience, scénarios, modèle de ménage, consommation. 12 What levers to promote rice production and consumption in Sierra Leone ? From characterization to simulation of rice farming households' performance. Abstract Sierra Leone is one of the three poorest countries in the world. A ten-year-long civil war has made the situation worse and destroyed most of the country's infrastructures. A number of initiatives and programmes have attempted to boost agricultural production since the mid2000s – notably rice production – in order to meet the food requirements of rural populations. Nowadays, despite the fact that its agriculture contributes to nearly 56% of the national GDP, Sierra Leone still relies on import and international help to meet its rice requirements. Northern Sierra Leone, and especially the Bombali district, is now the main rice production region in the country. However, this area produces very low and variable yields in comparison with its pedoclimatic potential. The EU's STABEX programme seeks to encourage agricultural production via several technical initiatives, but also direct economic support measures for rice farmers. This project has led us to define the following objectives for this thesis : i) to characterize the diversity of rice farming households in the Bombali district. This mainly involves identifying household categories with different yields. This will then allow us to understand – based on a framework of performance and efficiency analysis of the production developed within this thesis – and ii) to assess – via the use of a household model based on linear programming which is specific to this thesis – the performance of three initiatives (presented as scenarios) that aim to improve rice production and consumption (in terms of quantity but also of the number of calories consumed). In order to carry out this work, we used a database which gathers information on production and consumption and is based on a survey among 181 local rice farmers. The low yields and their variability between the different farm types are mostly explained by the total amount of work undertaken in each rice plot and on each farm, the seeding density, as well as the fallow duration (and therefore the initial soil fertility). Our study also highlights the fact that oil palm plays a very important part as cash crop ; it provides the rice farmer with cash, which is essential for rice intensification. This strategy has also made it possible for a number of rice farmers to extend their rice plot (without necessarily intensifying rice production), and therefore to increase their total production at farm level. This strategy often saw a decrease in the efficiency (yield/factors of production) of rice production. This result is mainly due to low soil fertility (in the case of farms in mainly upland ecosystems), to low 13 seeding density (in the case of poor households which favour consumption over seed storage) or to low amounts of available work (in the case of rich and intensive households). This study also underlines the fact that traditional households with an exclusive ecosystem will probably be non-resilient when faced with various market and climate-related shocks, but above all with the mass return of populations following the end of the war. This work has also shown that the option which has most improved production (from +53% to 78% according to household types), consumption (from +15 to 34% according to household types) and the number of calories consumed per capita and per day (from +7 to 21% according to household types) compared to the current situation is the specific subsidization of rice plantation in the lowlands, as advocated by the State. As for the other two scenarios (rice seed subsidization, and the subsidization of oil palm plantation), they have induced very little change in rice production, and therefore consumption, in comparison with the first scenario and the predictions of public authorities. This work has thus led to important methodological reflections, not only regarding the characterization of the diversity of performance, and of the efficiency of rice farming systems, but also regarding the initial hypotheses in building a household model. The purpose of this paper was to elaborate a model that reproduced household behaviours fairly accurately in terms of production and consumption, and considering limited resources, but also the risks regarding the variability of the climate as well as the market. According to us (regardless of data acquisition problems), the methods carried out and followed throughout this thesis can easily be applied, not only to other parts of West Africa, but also to other cropping and production systems, as significant as rice. Keywords: Sierra Leone, rice production system, performance and efficiency, scenarios, household model, consumption. 14 Introduction générale Dans les pays côtiers de l’Afrique de l’Ouest, la consommation de riz approche celle des niveaux sud-asiatiques : plus de 60 kg par habitant et par an en Guinée, Guinée-Bissau, au Liberia, au Sénégal et en Sierra Leone. Dans ces pays, importateurs nets de riz, la part de cette denrée dans la consommation alimentaire globale des ménages les plus démunis est particulièrement importante (plus de 100 kg/capita et par an). La sécurité alimentaire de ces ménages dépend, par conséquent, de l’accès au riz qui est largement cultivé dans plusieurs zones de la région. Figure 1 : Carte de la consommation de riz par kg/capita et par an (OCDE, 2011). Référence : Enjeux Ouest-Africains. EOA, N°2. 2011 Le riz couvre, en Afrique de l’Ouest, une superficie supérieure à 5,5 millions d’hectares dont près de 2,4 millions sont situés au Nigeria. La Guinée est le deuxième plus grand bassin rizicole en termes de surfaces avec près de 1 million d’hectares et les bassins sierra-léonais (0,5 millions d’hectares), ivoirien (0,38 millions d’hectares) et libérien (0,2 millions d’hectares) se trouvent dans le prolongement du bassin guinéen (OCDE, 2011). 15 Figure 2 : Carte des principaux bassins de production de riz (OCDE, 2011). Référence : Enjeux Ouest-Africains. EOA, N°2. 2011 Malgré ces surfaces de riz importantes, l’Afrique de l’Ouest importe, aujourd’hui, 5,2 millions de tonnes de riz contre 1,7 au début des années 1990 et ne couvre que 60 % de ses besoins (OCDE, 2011). Lorsque les prix mondiaux du riz ont brusquement augmenté en 2008, l’impact a varié selon les pays en fonction du degré de dépendance et d’exposition au marché mondial: entre janvier et avril 2008, les prix sont multipliés par trois au niveau mondial, par deux au Sénégal, par 1,5 au Bénin et au Mali et par 3 en Sierra Léone (OCDE, 2011). A titre d’exemple, le Sénégal diminue ses importations de 16 % et la Sierra Léone a pu éviter la crise alimentaire grâce aux aides des organisations internationales. Pour réduire au mieux les risques de crise alimentaire, les pouvoirs publics multiplient, au niveau de la région, depuis cette crise (et même avant), les initiatives pour accroître la production qui s’opère souvent par extension des surfaces, logique qui se heurte déjà au manque de terres aménagées. Dans ce cadre, la région restera, d’après (OCDE, 2011), durablement tributaire d’un marché international structurellement marqué par des prix élevés et de plus en plus volatiles. Ce constat est d’autant plus vrai que les rendements sont également souvent faibles et très variables. On trouve les productivités les plus élevées dans les zones irriguées au Mali et au Sénégal, les rendements intermédiaires dans les zones 16 rizicoles aménagées du Ghana et du Nigeria. Enfin, le Bénin, le Liberia et la Sierra Léone ont les rendements les plus bas du fait de la prédominance du riz pluvial de montagne et de basfonds faiblement aménagés (WAF et al., 2011). Si les rendements se situent nettement en deçà des 5 à 7 t/ha enregistrés aux États-Unis, en Argentine, au Pérou, en Colombie ou encore au Vietnam, les rendements moyens du riz au Sénégal et au Mali, évalués à 3 t/ha, soutiennent la comparaison avec la Thaïlande. Le Liberia, le Bénin et la Sierra Léone présentent des rendements inférieurs à 1t/ha (Diallo B. et al., 2012). La Sierra Leone est l’exemple même de ce paradoxe avec des possibilités pédoclimatiques considérables mais elle reste toujours dépendante du marché international et des aides internationales pour subvenir à ses besoins en riz. Cette situation s’est aggravée avec une guerre civile qui a détruit l’ensemble des infrastructures, y compris agricoles. La propagation rapide du virus de l’Ebola qui a causé plus de 1000 morts en 4 mois, atteste actuellement de l’incapacité de ce pays, dépourvu de toute infrastructure efficace, de réagir face à des crises sanitaires ou alimentaires. Aujourd’hui, la Sierra Léone est l’un des pays les plus pauvres au monde et demeure incapable d’augmenter de façon significative sa production de riz, et ceci malgré les multiples initiatives européennes et internationales. Les défaillances du secteur agricole sont multiples : des rendements faibles, des filières mal organisées, une mauvaise qualité des semences, des unités de stockages insuffisantes et mal entretenues. C’est dans le cadre de son programme « sécurité alimentaire » que le gouvernement de Sierra Leone a sollicité le Fonds européen de développement (FED) pour promouvoir sa production rizicole. Ainsi, suite à cette demande, le programme STABEX (Stabilisation of Export Earning), qui vise à soutenir l’exportation du café et du cacao et à améliorer le rendement du riz, a vu le jour. Ce programme, qui a concerné plus de 9000 ménages, a duré deux ans (de 2007 à 2009) et était financé à hauteur de plus de 4 millions d’euros. C’est dans le cadre du prolongement de ce programme qu’en 2010 le JRC (Joint Research Center, à Séville) a chargé le CIHEAM-IAMM de développer un modèle de ménage afin de simuler le comportement des ménages rizicoles en Sierra Leone. Ma thèse de doctorat s’inscrit dans le cadre de ce projet. Elle cherche à caractériser la performance des ménages rizicoles dans le district de Bombali au nord de la Sierra Leone. Elle a servi également à proposer et à évaluer, via un modèle de ménage élaboré dans le cadre 17 de cette thèse, des incitations économiques pour relancer la production et la consommation de riz. La thèse est composée de 5 chapitres. Le premier précise le contexte général de la production de riz en Sierra Leone et plus particulièrement à Bombali. Elle détaille également la problématique et les objectifs de la thèse. Le deuxième chapitre présente une analyse des systèmes de culture à base de riz en insistant sur les déterminants de leurs rendements. Le troisième chapitre est réservé à l’analyse de la performance et de l’efficience des systèmes rizicoles. Cette analyse permettra également de comprendre les stratégies de production des ménages en considérant des critères de classification socio-économique, de structure, et de consommation. Le quatrième chapitre a permis de présenter les hypothèses et les modules qui composent le modèle de ménage. De même, une partie importante au niveau de ce chapitre a servi à évaluer des mesures incitatives pour l’amélioration de la production et de la consommation de riz. Le dernier chapitre sera l’occasion de revenir sur les principaux résultats obtenus et de discuter les enseignements tirés de notre étude en termes méthodologiques et par rapport aux attendus de la thèse. 18 Chapitre I : Contexte et problématique de l’étude 1. Contexte de l’étude La Sierra Leone est un petit pays de cinq millions d’habitants située en Afrique de l’ouest. Elle occupe une superficie de 72 300 kilomètres carrés (km2), elle est limitée au Nord et à l’Est par la Guinée, au Sud par le Liberia et à l’Ouest par l’Océan Atlantique. Ce pays est divisé en quatre provinces (Nord, Ouest, Est et Sud) et 12 districts dont Freetown est la capitale (figure 3). Freetown est la plus grande ville (avec une population d’environ 1 million d’habitants) mais c’est également le centre économique, financier et culturel du pays. La population totale, majoritairement jeune, représente 6.1 millions d’habitants, dont 61% sont des ruraux (Perrault F. et al., 2013). En 2013, cette population présente un taux d’accroissement annuel de 2.6% et un taux d’alphabétisation des adultes de plus de 15 ans de 42 %. Ce taux d’alphabétisation est la conséquence directe de la guerre civile (ponctuée par cinq coups d’Etat) que la Sierra Leone a subie entre 1991et 2002 et qui a vu le déplacement d’environ 2 millions de personnes. Figure 3 : Carte de la Sierra Leone Entre 1991 et 2002, la guerre civile a engendré de lourds dommages notamment sur la production agricole locale et les infrastructures rurales. Néanmoins, ce pays dispose encore d’un riche potentiel de terres agricoles, soit 5 millions d’hectares de terre arable, ce qui 19 représente 80% de la superficie totale du pays (AFDB, 2009). A cela, il faut rajouter un climat tropical très favorable à une diversification des cultures, notamment pendant la saison humide qui s’étale de mai à décembre avec un gradient de 1400 mm/an au nord à 2400mm/an au sud. Depuis 2002, l’économie en Sierra Leone est essentiellement assurée par les secteurs informels, l’agriculture, la pêche et les services (World Bank, 2009). Actuellement, même si des progrès ont été enregistrés du point de vue économique et politique, le pays a toujours besoin de l’aide internationale pour subvenir à ses besoins alimentaires. Il est d’ailleurs toujours cité comme étant l’un des pays les plus pauvres au monde avec seulement 480$ de revenu national brut (RNB) par habitant et par an (FAO, 2002b). Avec la fin de la guerre, une forte reprise de l’économie a été constatée. Elle est caractérisée par un PIB annuel de l’ordre de 5.5% en 2013, dont 53% sont représentés par l’agriculture qui assure 80% du total de la main d’œuvre active (20% d’hommes et 60 % de femmes) (FAOSTAT, 2013). Malgré cette croissance, 80% de la population agricole vit toujours endessous du seuil de pauvreté et le pays reste toujours dépendant de l’importation de ses principaux produits tels que le riz (riz blanc, riz cassé, farine de riz). A ce titre, la Sierra Leone a été identifiée par la FAO comme le huitième pays le plus vulnérable à la crise alimentaire (FAO, 2009); soit 16% de la population qui ne peut pas se permettre une alimentation de base. Cette situation de pauvreté s’explique par plusieurs facteurs, notamment, un revenu faible basé exclusivement sur une activité agricole très peu productive, aggravée par l’absence de subventions et de soutien politique. 2. Le secteur agricole L’Est et le Nord de la Sierra Leone sont les principales zones agricoles. Elles assurent 65 à 70% de la production agricole totale du pays (CORAF et al., 2014). Au Nord, la majeure partie de la surface est dédiée aux cultures vivrières représentées essentiellement par le riz. A l’Est, même si le riz est très présent, les agriculteurs cultivent essentiellement des cultures de rente destinées à l’exportation comme le cacao et le café (figure 4). Le riz est traditionnellement cultivé en association ou en rotation avec le manioc, le maïs, le millet, le haricot, la patate douce ou encore l’arachide. 20 (MAFFS et al., 2005) Figure 4 : Carte économique de la Sierra Leone. Le jaune au niveau de la carte représente les zones où le riz est (ou peut-être) cultivé. Après la guerre, la culture du manioc s’est développée très rapidement aux dépens d’autres cultures vivrières moins rentables. C’est une culture qui croît très vite et s’adapte facilement aux différents écosystèmes mais avec une valeur nutritive en terme d’apport calorique très limitée. Cela pousse aujourd’hui les agriculteurs à bien réfléchir à la place que doit occuper cette culture qui concurrence d’autres cultures, certes plus consommatrices en ressources (main d’œuvre, terre) et moins productives, mais qui restent très utiles pour l’équilibre alimentaire des ménages. Il faut également mentionner que la production des cultures à tubercule (principalement manioc et patate douce) excède les besoins nationaux (Akintayo I. et al., 2008). L’excédent de production est souvent détruit faute de moyen de stockage ou d’exportation. La figure 5 présente l’évolution des productions des principales cultures entre 1980 et 2006. Elle montre que la production des principales cultures vivrières, notamment le riz, le manioc et le maïs, a augmenté d’une façon importante dès la fin de la guerre au milieu des années 2000. 21 (Gomez y Paloma S. et al., 2012) Figure 5 : Evolution des productions des principales cultures entre 1980 et 2006. En ce qui concerne le riz, une nette amélioration de la production entre 1997 et 2009 est observée (tableau 1). Celle-ci est principalement due à l’augmentation des surfaces dédiées au riz et à l’amélioration du rendement qui passe en moyenne de 1.1 t/ha en 1997 à 1.4 t/ha en 2009. Dans ce registre, même si la population a augmenté de presque d’un million entre 1997 et 2008, la production globale en riz a crû plus rapidement permettant ainsi une amélioration du niveau de l’autosuffisance en riz (qui passe de 54% en 1997 à 72% en 2008). Tableau 1 : Production de riz entre 1997 et 2009 en Sierra Leone Year 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Area (ha) Yield (Mt/ha) Production (Mt) 360.789 1.14 411.300 320.517 1.16 371.800 212.137 1.17 248.200 166.387 1.19 198.000 258.850 1.20 310.620 343.142 1.23 422.065 356.506 1.25 445.633 426.772 1.27 542.000 427.556 1.29 552.000 422.556 1.33 562.000 432.356 1.36 588.004 475.592 1.43 680.097 22 Cependant, malgré cette amélioration en termes de rendement et de production, la Sierra Leone reste un pays nettement importateur pour ses besoins alimentaires, notamment celle de riz (tableau 2). En revanche, elle exporte essentiellement du cacao (85% du total exporté) et du café qui sont cultivés principalement dans la partie Est du pays. Tableau 2 : Importation de riz en Sierra Leone (Mt), 2006-2008 Année 2006 2007 2008 Importation 90680 99839 157942 Aides alimentaires 805 0 1218 Total 91485 99839 159160 (Bah A. A., 2013) 3. Les ménages agricoles à Bombali Le district de Bombali est situé au Nord du pays. Sa capitale, Makeni, est la plus grande ville au nord de la Sierra Leone. Les autres grandes villes de ce district sont Kamakwie, Kamabai, Karina and Binkolo (figure 6). Avec une surface de 7985 km² et une population estimée en 2010 à 435 milles habitants, Bombali est le deuxième plus grand district, à la fois en termes de surface et de population. Figure 6 : Situation géographique du district de Bombali au nord de la Sierra Leone Le dernier recensement effectué par le ministère de l’Agriculture montre que 70 à 80% de la population au Nord de la Sierra Leone, et plus particulièrement à Bombali, sont engagés entièrement dans l’activité agricole (Jalloh A., 2006). Cela rend cette population extrêmement vulnérable à la volatilité des prix du marché mais également aux aléas climatiques. En effet, 23 39 et 36% de la population agricole sont identifiés comme vulnérables respectivement pour le Nord (notamment Bombali) et l’Est de la Sierra Leone (tableau 3). Tableau 3 : Niveau de vulnérabilité de la population au nord de la Sierra Leone - Fonctionnement des ménages rizicoles à Bombali Les pratiques culturales des ménages rizicoles dans un village sont structurées selon le type et la taille de la famille. Le village se compose souvent de plusieurs grandes familles qui vivent principalement de la production agricole. Chacune de ces grandes familles dispose d’une surface qu’elle gère de façon autonome. La terre peut appartenir à la famille, avoir été reçue en héritage ou bien louée pour une courte durée (inférieure à 5 ans) contre une somme symbolique (sac de riz par exemple). Faute de document juridique prouvant la propriété de la terre, les chefs de familles sont souvent réticents vis-à-vis d’une location de longue durée. Cette situation entraîne un morcellement très important des terres agricoles. En effet, 60 à 70% de la population rurale possèdent une terre agricole comprise entre 0.5 et 2ha. A cela, il faut rajouter qu’environ 13% des ruraux pauvres ne possèdent pas de terres et vivent uniquement de la location. Le ménage rizicole à Bombali est composé en moyenne de 8 membres (chef, épouse, enfants). Notons que la décision du choix de la parcelle à cultiver chaque année est prise par le chef de famille. Cependant, les dates de semis et de récolte sont généralement fixées par le chef du village. La décision en terme de type de semence, de désherbage et de répartition de la main d’œuvre est prise par le membre de la famille à qui la parcelle a été attribuée par le chef de famille. Une partie de la vie quotidienne se passe sur l’exploitation agricole. Nous pouvons identifier deux périodes : la période sèche où on cultive principalement les cultures maraîchères et du 24 manioc en monoculture, suivie par la période humide où on retrouve le riz en culture dominante. La main d’œuvre est divisée selon la pénibilité du travail : les hommes pratiquent l’abatage, le brûlis, le défrichage et le semis, tandis que les femmes s’occupent principalement du désherbage, de la récolte et des activités de post-récolte (séchage, vannage, vente). Par ailleurs, chaque famille dispose d’un un jardin potager de petite surface (en moyenne 0,2 ha) adjacent à leur hutte destiné à la production de légumes. Une description des itinéraires techniques du riz sous différents écosystèmes est présentée en annexe 1. Parmi les parcelles dont la famille dispose, c’est le chef de famille qui décide du choix de celle à cultiver, et de celles à laisser en jachère. La durée de la jachère est déterminée en fonction du nombre de parcelles que l’agriculteur possède. En effet, plus la taille de l’exploitation, représentée par des parcelles souvent dispersées dans l’espace, est grande, plus la jachère est longue. L’outillage utilisé pour les différentes activités agricoles est très rudimentaire et pénible à manier. L’opération d’abatis-brûlis est réalisée en utilisant des haches qui servent aux opérations de défrichement et d’abattage. La préparation du lit de semis est réalisée en utilisant une houe. Les houes sont souvent fabriquées manuellement par les villageois, avec une durée de vie de 2 ans. La récolte du riz est faite manuellement en se servant d’un couteau, ce qui engendre souvent des pertes importantes en grain sur le champ. Enfin, à Bombali, le système rizicole est le système le plus dominant en termes de surface et de production. Cette culture vivrière représente pour l’agriculteur en moyenne 80% de la surface totale de l’exploitation. Ce système peut être cultivé de différentes façons selon les reliefs,: en monoculture dans les vallées marécageuses (bas-fonds), et en culture mixte sur les plateaux. Cette diversité, liée au type de sol, conjuguée aux différentes contraintes, impose à la culture des techniques de production adaptées procurant ainsi des rendements souvent faibles et variables. - Les contraintes des systèmes rizicoles à Bombali Le rendement du riz en Sierra Leone et plus particulièrement à Bombali est très faible et variable par comparaison, non seulement au potentiel de cette culture, mais également à la moyenne de la région (tableau 4). Ce constat s’applique également aux autres cultures vivrières. Plusieurs contraintes peuvent expliquer ce constat : 25 Tableau 4 : Rendement moyen des principales cultures en Sierra Leone (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). Average crop yield in metric ton per hectare Sierra Leone Africa Upland rice 0,7 Boliland rice 0,72 IVS rice 1,58 Riverain rice 1,56 Mangrove rice 2,61 All rice ecologies 1,43 2 Cassava (Upland sole crop) 6,37 10 Sweet potato (Upland second crop) 3,77 5--10 Groundnut (Upland second crop) 0,68 0,8 Maize (Upland second crop) 0,84 1-1,2 Cowpea (Upland sole crop) 0,49 0,1-0,4 - Préparation du sol La qualité et le choix du moment opportun pour effectuer la préparation du sol sont importants afin d’assurer un bon rendement en général et, plus particulièrement, pour le riz. La sensibilité du travail du sol sur une rizière de plateau est beaucoup plus prononcée que celle sur les bas-fonds (Sakurai S., 2006). Etant donné que le riz est cultivé sous deux régimes hydriques, variant entre les conditions pluviales (plateaux) et la profondeur d’immersion (basfonds), la mise en culture est différente selon l’écosystème. En effet, à Bombali, la préparation du lit de semences est faite uniquement sur les plateaux avec des outils très rudimentaires (houe). Cette opération reste difficile à maîtriser faute d’outil adapté et en raison du nivellement des parcelles. Dans ces conditions, le semis est réalisé sur des champs souvent envahis par les adventices (Saito K. et al., 2014). Sur les bas-fonds, le problème des mauvaises herbes au moment du repiquage est quasiment absent à cause de la présence en permanence d’une lame d’eau. Néanmoins, il reste important d’y faire passer un « rouleau » pour tracer les sillons et faciliter le repiquage. Cette opération 26 est réalisée par des outils fabriqués localement et peu performants créant souvent une mauvaise uniformité dans la distribution de la lame d’eau (Akabundo I. O., 1991). - Choix de la variété Le choix de la variété selon l’écosystème est également un facteur déterminant à Bombali. Les agriculteurs mélangent souvent des variétés qui possèdent différents cycles phénologiques selon leurs disponibilités sur le marché, ce qui rend la gestion, en termes de désherbage et de récolte, difficile (Abdelrasoul A. et al., 2013). Le mélange variétal nécessite également une disponibilité en main d’œuvre permanente, ce qui n’est pas toujours le cas et provoque ainsi une perte d’une partie de la récolte qui peut atteindre, pour des années favorables au bon développement du riz, 50% de la production totale au champ (NEPAD et al., 2009). Néanmoins, faute de moyen de stockage efficace, ce mélange variétal est parfois choisi par les agriculteurs comme une stratégie pour échelonner la récolte et avoir du riz sur le champ sur une longue période de l’année (Becker L. et al., 2004). Néanmoins, le choix du bon moment pour la récolte est également un facteur qui pourrait limiter de 5 à 15% la production totale (FAO, 2002b). Cependant, du fait des différences des propriétés physiologiques des variétés, il sera difficile de préconiser un critère pour déterminer le bon moment de la récolte (Akintayo I., Cissé B., Zadji D., 2008). - Les adventices Au niveau des plateaux, les adventices sont des concurrents permanents pour le riz et peuvent réduire selon les espèces d’adventices, la variété de riz et l’humidité ambiante, son rendement jusqu’à 84 % (Akintayo I., Cissé B., Zadji D., 2008). Une étude menée sur le riz dans le district de Bombali a montré que (RARC et al., 2010): i) sans fertilisation azotée un seul désherbage effectué deux semaines après le semis produirait les mêmes effets que 2 ou 3 désherbages. Ce constat est identique pour des variétés locales améliorées. En fertilisant le riz, au moins deux passages de désherbages manuels seront nécessaires, ii) les variétés locales types (Rok 16) sont plus résistantes aux mauvaises herbes que les variétés améliorées (type Nerica). Sans désherbage, les variétés locales types peuvent procurer jusqu’à 20% de plus de rendement par comparaison aux variétés améliorées types Nerica 1 ou Nerica 4 (RARC, SLARI, 2010) et iii) le riz plateau est plus sensible aux effets des mauvaises herbes (en moyenne 48% de chute de rendement à cause des mauvaises herbes) que celui des bas-fonds (28% de baisse relative) (Akabundo I. O., 1991) 27 - Les bio-agresseurs et les maladies Le riz pluvial est très peu favorable aux insectes qui se nourrissent de feuilles, par contre, il convient très bien aux ravageurs des sols, tels que les fourmis et termites (Niangado O., 2010). En revanche, les rizières de bas-fond sont très sensibles aux insectes aquatiques tels que les mouches du riz (Diopsis macrophthalma), les charançons aquatiques (Sitophilus oryzae) (CORAF, WECARD, 2014). A cela, il faut rajouter les attaques des oiseaux (Canirallus oculeus), au moment des remplissages des grains, qui peuvent anéantir jusqu’à 100% de la récolte. Une tâche importante, souvent assurée par les personnes âgées, les enfants et les femmes, est de chasser ces oiseaux de façon permanente, de la floraison à la récolte (MAFFS, NARCC, IAR, RRS, 2005). Cette tâche peut consommer jusqu’à l’équivalent de 50% de la main d’couvre totale réservée à la production de riz. Les maladies enregistrées sur le riz à Bombali sont souvent cryptogamiques telles que la maladie i) d'Udbatta (Ephelis oryzae). Il s’agit d’une pathologie qui se manifeste par un durcissement anormal de la panicule, et qui se traduit par des pertes souvent comprises entre 10 et 15% de la production totale. Elle est engendrée par les températures élevées et l'humidité abondante du sol et souvent transmise par la semence, et ii) la bigarrure jaune pâle qui provoque des pertes importantes en Afrique de l’Ouest et notamment à Bombali. Il s’agit d’une maladie virale qui se transmet de façon mécanique par les insectes du genre Chaetocem. En Afrique de l'Ouest, contrairement aux variétés améliorées, de nombreuses variétés locales comme ROK 16, Lac 23, Moroberekan, OS-6, 63-83 sont résistantes à la bigarrure jaune pâle. - Gestion de la lame d’eau Sa variabilité par écosystème reste également un facteur déterminant pour la production de riz à Bombali. Une hauteur très basse de la lame d’eau peut diminuer l’aptitude des plants de riz à taller. De même, une hauteur trop importante pour le tallage accélère la montaison (Bahan F. et al., 2012) et, par conséquent, affecte le rendement en grain. Sur les plateaux, l’arrivée tardive de la pluie ainsi que le stress hydrique en fin de cycle sont souvent cités comme étant les facteurs qui influencent le plus négativement le rendement du riz (Worou O. N. et al., 2012). 28 - La jachère La jachère longue est une vielle pratique appliquée en Afrique de l’ouest. En l’absence d’une fertilisation significative, cette pratique a pour objectif de régénérer et d’améliorer la fertilité des sols. Après la fin de la guerre et le retour massif des populations, les riziculteurs ont adopté deux stratégies pour augmenter leur production : intensifier le riz en raccourcissant les jachères, qui sont passées en moyenne de 15 à 20 ans dans les années 70 à 3-5 ans actuellement (Sene M., 1996), ou accroître les surfaces cultivées aux dépens des forêts comme le montre la figure 7. La Sierra Leone a des niveaux élevés de déforestation, estimée à 3.000 hectares par an. Cette déforestation est principalement due, en dehors de la forte demande pour le bois de chauffage, la construction des maisons, les feux de brousse utilisés pour l'élevage et la chasse, et l'exploitation minière, à l’intensification de l’agriculture au détriment des forêts et au raccourcissement des jachères. Néanmoins, de longues jachères ne sont pas toujours synonymes de rendement élevé pour le riz. En effet, à Bombali, l’absence d’une main d’œuvre suffisante pour nettoyer convenablement les parcelles et les faibles doses de semis anéantissent partiellement ou complètement les effets positifs d’une fertilité initiale importante des sols (Farooq M. et al., 2006; Farooq M. et al., 2009). Dans ce cadre, une récente étude menée dans la station expérimentale de SLARI à Bombali montre qu’avec une fertilité du sol non limitante le rendement du riz augmente de 20%, lorsque la dose de semis passe de 45 à 90 kg/ha (IRRI, 2009). 2003 2004 2005 2006 2007 Figure 7 : Augmentation des surfaces cultivées au détriment des forêts. 29 4. Problématique En raison de la forte croissance démographique et d’une décennie de guerre civile qui s’est terminée en 2002, la pauvreté reste très répandue en Sierra Leone. L’agriculture, principale activité des ménages locaux, reste très traditionnelle avec des systèmes de culture, à base de riz ou de cultures pérennes, caractérisés par des rendements faibles et très variables (0,7 t/ha à 2,61 t/ha). Cette variabilité considérable des niveaux de rendement est constatée même dans des systèmes de culture et des environnements biophysiques semblables. L’écart entre les rendements peut être justifié par plusieurs critères : biophysique (sol, climat), technique culturale (travail du sol, variété, densité de plantation), socio-économique (taille de la famille, main d’œuvre, prix des intrants et des machines), politique (crédit, association) et technologique (compétence, connaissance et qualification). Aujourd’hui, plusieurs projets et initiatives (Addoxbioenergy, Biopalm energy, STABEX) (CORAF, WECARD, 2014) ont vu le jour en Sierra Leone. Ces projets visent à analyser la structure des ménages rizicoles en termes de production et de consommation afin d’identifier les leviers nécessaires pour promouvoir la production du riz, et, par conséquent, sa consommation. Néanmoins, tous sont confrontés à l’absence, non seulement de données récentes (post-guerre) mais également à celle d’études détaillées décrivant les stratégies de production des agriculteurs pour satisfaire leurs besoins alimentaires, notamment en riz. C’est le cas du district de Bombali qui reste, malgré son importance en termes de production rizicole et en nombre de ménages, à la fois pauvre et incapable de subvenir à ses besoins alimentaires, très peu étudiés (FAO, 2002a). Dans ce district, où les ménages produisent essentiellement pour l’autoconsommation, l’aliment de base est le riz, ainsi 40% des calories végétales sont apportées par cette céréale. Le reste est assuré par les huiles végétales et les graisses animales (15.7%), suivi par les céréales autres que le riz (10.6%) et les racines (10.1%). A cela, il faut ajouter les viandes, les sucres et le lait qui assurent respectivement 10.2%, 2.9% et 10.6% (FAOSTAT, 2009). Pour renforcer la production de riz et, par conséquent, sa consommation, l’Etat en Sierra Leone a élaboré, en 2010, une politique agricole et rurale qui vise à remédier, via des incitations mais également de la sensibilisation, à deux principales contraintes (le détail de ces mesures sont en annexe 1, (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012) : 30 Le faible pouvoir d’investissement des riziculteurs qui se manifeste essentiellement par un niveau quasi-inexistant, voire totalement absent de mécanisation. En conséquence, les terres de bas-fond, par exemple, malgré leur richesse en matière organique, sont souvent très peu cultivées à cause de leurs battances. À cela, s’ajoutent des problèmes de faible trésorerie (cash) ne permettant pas aux ménages l’achat de semences et d’intrants (notamment de la main d’œuvre). Ce contexte pousse le ménage à adopter une agriculture basée sur de faibles apports d’engrais, soit en moyenne 4kg N/ha, ce qui engendre une dégradation très rapide de la fertilité des sols (NSADP, 2009). Le faible niveau de technicité et d’infrastructure D’après une étude menée par la FAO en 2005, les agriculteurs ont une connaissance très limitée en techniques de production. Cela est principalement dû au manque de vulgarisation et au rôle que doit normalement jouer l’Etat. Il existe d’autres défaillances, de type accès limité aux services financiers agricoles ou aux installations de micro-crédits accordés par les ONG pour contribuer à la production de riz, de palmier à huile, etc.; des infrastructures inexistantes telles que les routes, les lieux de stockage et de multiplication de semences sont également mentionnées par la FAO. Cet accès limité aux infrastructures de stockage engendre des pertes post - récolte importantes pour le ménage (FAO, 2008). Ainsi, bien que l’agriculture soit en théorie le secteur économique le plus important de la Sierra Leone (IMF, 2008), elle continue à être confrontée à la précarité, la pauvreté et à la faiblesse des revenus. Cette situation paradoxale est résumée par l’inadéquation entre les potentialités réelles qu’offre ce pays pour une meilleure production agricole, et la faiblesse des revenus et de la consommation conditionnée par des stratégies de production inadaptées. Pour faire face à ces problèmes, le ministère de l’Agriculture en Sierra Leone, en collaboration avec le programme STABEX de l’Union Européenne, a lancé un appel d’offre pour définir et évaluer, d’une façon ad-hoc, l’effet de certaines mesures incitatives pour la production de riz. C’est dans ce contexte que le projet « FSSIM-AFRICA » a été élaboré en 2010 conjointement par l’IAMM et le JRC (Joint Research Centre). Mon rôle dans ce projet était surtout d’apporter quelques éléments de réponse aux trois objectifs suivants : 31 i) Porter la réflexion sur les déterminants qui expliquent les faibles performances (principalement le rendement) et les différences d’efficience entre les systèmes de culture à base de riz actuels. ii) Comprendre et expliquer les stratégies de production des riziculteurs en tenant compte des contraintes et des objectifs de production (notamment pour répondre à des besoins alimentaires) auxquels les ménages rizicoles sont confrontés. iii) Proposer et évaluer, en adéquation avec la politique agricole et rurale et le programme STABEX de l’EU, des mesures incitatives permettant de promouvoir la production et la consommation de riz. La réponse à ce troisième objectif doit s’effectuer en tenant compte de trois éléments de contexte essentiels : - Pour plusieurs auteurs, l’écosystème bas-fond est l’avenir de la riziculture en Afrique de l’Ouest et plus particulièrement à Bombali (plus largement le nord de la Sierra Leone) (CARD, 2009). Aujourd’hui, on estime les surfaces bas-fonds non cultivées en Sierra Leone à 85% par rapport à la surface totale basfond disponible (Bah A. A., 2013). Pour des raisons économiques mais également de production, l’Etat pousse les agriculteurs à cultiver le riz prioritairement sur les bas-fonds au lieu des plateaux. En effet, les pouvoirs publics encouragent les ménages à réserver le palmier à huile, une culture très rentable, aux plateaux (SLIEPA, 2012). A cela, il faut ajouter que le riz bas-fond aura, à cause d’une fertilité initiale plus élevée, en moyenne un rendement plus élevé que sur les plateaux et un besoin en main d’œuvre plus faible (Delarue J., 2007). Néanmoins, il existe trois limites importantes à l’installation de cette culture sur les bas-fonds : en l’absence de mécanisation, l’hyper battance des pluies associée à la texture fine des basfonds, provoque dans la plus part des cas des croutes de battance difficile à travailler, et entraine par conséquent des faibles levées (Bal A. B. (ed), 2002), une toxicité fréquente par les phosphates (Dixon C. A. et al., 2001), des capacités de stockage faibles aggravant ainsi les risques de pénurie pendant les périodes de soudure (FAO, 2002b) et, enfin, une incapacité à cultiver sur ce type d’écosystème, des légumes, indispensables à l’équilibre alimentaire des ménages, pendant la saison humide. - Il est historiquement connu qu’un des moyens importants pour intensifier les cultures vivrières de base est d’introduire, au niveau de l’exploitation, des cultures à forte valeur ajoutée. C’est le cas, par exemple, des poivrons dans les systèmes rizicoles en Malaisie (Cram,1993) ou les arbres fruitiers au Malawi dans les systèmes à base de maïs (Franke A. C. 32 et al., 2014). Cette stratégie est également suivie en Sierra Leone où l'huile de palme n'est pas seulement un aliment de base important, mais aussi une source importante de revenus pour les ménages ruraux. Ce revenu est souvent exploité pour intensifier les cultures vivrières et, par conséquent, augmenter leur rendement. Ce type d’investissement se confronte essentiellement à deux types de problèmes: les faibles ressources par exploitation où le palmier à huile vient souvent en concurrence en termes de surface et de main d’œuvre aux cultures vivrières (FAO, 2003) et le coût initial important pour planter des palmiers à huile (Acet, 2010). - Un très grand nombre de projets se développent actuellement en Sierra Leone (Biopalm energie, BHB Gmbh) (terres G. A. d., 2012) dont l’objectif est de tester et promouvoir de nouvelles variétés de riz plus productives. D’autres proposent, de façon occasionnelle ou permanente, la distribution gratuite des semences. Aujourd’hui, les pouvoirs publics réfléchissent à systématiser la distribution des semences pour combler partiellement ou complètement les besoins en semence de riz des ménages. Cette initiative émane de deux constats majeurs : i) la faible capacité de stockage et les mauvaises conditions de stockage qui entraînent parfois la perte de plus de 50% de la récolte (NEPAD, FAO, 2009), ii) les faibles récoltes de l’année génèrent obligatoirement des quantités stockées faibles et, par conséquent, des surfaces emblavées pour l’année suivante réduites (NEPAD, FAO, 2009). Ce cercle vicieux est souvent durable à cause de la faible capacité des riziculteurs à acheter des semences dont le prix est souvent élevé et changeant (Niangado O., 2010). Cette thèse apportera, au-delà des questions stratégiques autour de la production rizicole en Sierra Leone, une réflexion d’ordre méthodologique. Afin d’analyser les systèmes de culture et de production en vue de proposer et d’évaluer des solutions appropriées et applicables aux contextes de la Sierra Leone, au moins deux questions méthodologiques s’imposent : - Comment peut-on et quels critères faut-il choisir pour représenter les diversités des systèmes de culture et de production à base de riz ? Cette caractérisation et analyse de la performance de ces systèmes doit tenir compte des éléments de différents domaines et disciplines (agronomie, socio-économie, consommation). - Quels outils choisir ou construire pour évaluer la performance des ménages rizicoles en termes de production et de consommation ? Répondre à ces deux points requiert souvent une analyse intégrée des systèmes de culture et de production afin de tenir compte de la complexité des systèmes analysés. Cette approche (analyse intégrée) s’appuiera sur au moins trois étapes (Therond O. et al., 2009): i) phase 33 diagnostic : il s’agit de caractériser et d’analyser la performance des systèmes de production actuels. Cela permettra surtout de mieux comprendre les forces et les faiblesses de ces systèmes, ii) phase construction de scénarios : proposer, en se basant sur ce diagnostic et après concertations avec les acteurs locaux (agriculteurs, décideurs politiques…), des alternatives (techniques ou politiques) capables d’améliorer la performance des systèmes actuels, et iii) phase évaluation : évaluer et analyser, en calculant des indicateurs de durabilité, la performance des alternatives qui se déclinent en scénarios. Cette approche repose souvent, pour le calcul d’indicateurs, sur l’utilisation d’une chaîne de modèles (qualitative ou quantitative) qui représente les différents processus qui se déroulent à l’échelle de la parcelle (fertilité du sol, rendement, etc.), de l’exploitation (marge brute, main d’œuvre, etc.) et de la région (Belhouchette et al., 2011). C’est cette approche qui a été suivie dans notre travail de thèse et que nous déclinons en trois principales étapes comme suit (figure 8) : 1- Caractériser et analyser la performance des systèmes de culture à base de riz (chapitre 2). Cette tâche a été réalisée en nous appuyant sur une enquête de terrain réalisée entre 2009 et 2010 par le JRC et dont la structure et présentée en annexe 1. Il s’agit essentiellement de comprendre les déterminants expliquant les performances des systèmes de culture actuels (notamment en termes de rendement). 2- Décrire, caractériser et analyser la performance et l’efficience des ménages rizicoles (Chapitre 3): Il s’agit, sur la base de plusieurs critères biophysiques (écosystème, fertilité initiale, etc.), de structure et de ressources disponibles (SAU, main d’œuvre, semence, etc.) et de préférence de consommation (autoconsommation, stockage, consommation, etc.), de construire des types d’exploitations (assez) homogènes en termes de logique de production. L’objectif de cette étape est de pouvoir évaluer la performance de ces exploitations. L’enquête qui a servi pour identifier et analyser les systèmes de culture dans l’étape 1 a permis également de caractériser les systèmes de production rizicoles. 3- Proposer et évaluer trois mesures économiques qui visent l’amélioration de la production du riz et, par conséquent, sa consommation (chapitre 4). Ces mesures sont tirées de la nouvelle politique agricole et rurale de la Sierra Léone, et qui sont jugées par les Pouvoirs Publics comme prioritaires. Il s’agit, dans cette étape, surtout de construire un modèle de ménage basé sur la programmation linéaire capable de simuler le revenu global des ménages. Ce dernier concerne à la fois le revenu agricole et la part réservée à l’autoconsommation. Ce modèle est établi sur la base d’un modèle conceptuel qui reproduit la logique de production d’un ménage rizicole tel qu’il a été décrit au niveau du chapitre 2. Les effets des trois mesures 34 (déclinées sous forme de scénarios) sur la performance des ménages en termes de production et de consommation sont évalués en calculant des indicateurs agronomiques (type de culture, ressources mobilisées par culture, rendement, etc.), socio-économiques (revenu agricole, main d’œuvre, coût de production, etc.) et de consommation (autoconsommation, consommation par produits, calories, etc.). Etapes Etape 1 Caractériser et analyser la performance des systèmes de culture. Etape 2 Comprendre les déterminants expliquant les performances et l’efficience des systèmes de culture actuels. Etape 3 Proposer et évaluer des mesures /incitations économiques pour améliorer la performance des systèmes de culture Objectifs Outils - Enquêtes - Typologie - Analyse de corrélation Caractériser les déterminants expliquant les performances et limites des systèmes de culture - Enquêtes - Typologie /critères biophysiques de structure/ressources disponibles préférence de consommation - Modèle conceptuel - Modèle de bioéconomique ménage Chapitre 2 Evaluer la performance et l’efficience des systèmes de culture afin de comprendre la Chapitre 3 stratégie de production des ménages agricoles Quelle place pour les cultures de rente et les cultures vivrières ? Quel compromis avec le riz (notamment plateau) avec des ressources Chapitre 4 limitées ? Est-il possible de développer une activité rizicole uniquement sur les bas-fonds comme le suggère l’Etat ? Figure 8: Démarche générale de la thèse 35 Chapter 2 Assessing the diversity of smallholder rice farms production strategies in Sierra Leone Chenoune, R (a ,b,c)., Belhouchette, H (a,c)., Capillon, A (b). (a) IAMM-CIHEAM, 3191 Route de Mende, 34093 Montpellier. Cedex 5, France (b) Montpellier SupAgro, UMR System, 2 Place Viala, 34060 Montpellier, France. (c) IAMM- UMR System, 2 Place Viala, 34060 Montpellier, France. Journal submission: NJAS- Wageningen journal of Life Sciences Summary In Sierra Leone, several international organizations are trying to help the government improve the productivity of its rice farms, which currently have the lowest rice yields in West Africa. However, the various programmes attempting to increase rice production, and consequently rice self-sufficient food production, are handicapped by an absence of thorough studies explaining the way rice farmers take the available socio-economic, technical and natural production factors into account when making their decisions. The purpose of the current article is to assess rice production performance on smallholder rice farms in Sierra Leone. To achieve this goal, an agronomic and socio-economic survey was carried out among 180 rice farmers in the district of Bombali in Northern Sierra Leone. The survey, combined with a specific statistical analysis, made it possible to assess production strategies for rice farms according to various discriminant parameters (family size and composition, fallow duration, seeding density, labour availability, ecosystems, share of oil palm, distance from field to farm…). This analysis revealed that the rice smallholder farms that perform best are those growing rice under two ecosystems together with oil palm. Those farms have more income to purchase rice seed in years when production is low or if they have large families to feed. However, traditional rice farms with one exclusive ecosystem will probably not be sustainable and they will not be able to satisfy their households’ future rice needs. Keywords: smallholder rice farms, ecosystems, yield, farm typology, Sierra Leone 36 1. Introduction Twelve years after the end of its civil war (1991-2001), Sierra Leone is still ranked as the third poorest country in the world (Bangura Z. H. et al., 2012). Agriculture is an important aspect of its economy, providing 56% of its GDP and employing 2/3 of the rural population (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). In this context, rice farmers mainly produce rice for their own consumption (Sékou D. et al., 2011). This means that smallholder rice farms are very vulnerable to fluctuations in the amounts of rice produced. Rice yields by small farmers in Sierra Leone are the lowest and most variable of all the countries in West Africa (Becker M. et al., 1999; Sammeth F. et al., 2010). There even is considerable variability between rice farmers using similar growing practices (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). In Sierra Leone, rice is generally grown without either fertilisers or irrigation (Bumb B. L. et al., 2011). Field-clearing, sowing, transplanting, weeding and harvesting are all carried out without mechanization (BAD, 2010). This means that rice production is very vulnerable to the availability of labour and the initial fertility of the soil (CRISTO, 2010; Harding S. S. et al., 2012; Mbetid-Bessanee E., 2004; Melendez C. J., 2006). The question of soil fertility is of particular importance for the upland ecosystem which is highly dependent on the duration of the fallow period (Sene M., 1996). Other determining factors, such as the availability and quality of seed, could also explain low and variable yields (Deen S. et al., 2010). It has often been noted that the proportions of rice grains stored as seed, sold, or consumed by the growers vary according to the structure of the farms and households (size of family, needs in terms of vegetables, storage capacity for seed, etc.) (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). Traditionally, farms have often consisted of a single plot of upland rice (Fall A. A., 2000). Since the end of the war and the massive return of displaced populations, there is no longer enough uplands for all, and rice farmers find themselves compelled to grow in the lowlands to satisfy their family’s rice needs (FAO, 2011). Other rice farmers branch out into oil palm, a fairly profitable crop compared to rice (SLIEPA, 2012). Recently, several national and international projects have been launched to help Sierra Leone restructure its agriculture, and its rice production in particular (e.g. the European STABEX fund from 2007 to 2009). However, these projects are handicapped by the lack of any 37 thorough characterization of current rice growing systems and their productivity by ecosystem (upland or lowland) and type of farm (with or without oil palm) (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). Such characterization is necessary in order to understand how rice farmers make their choices depending on the main production factors, the different ecosystems (upland or lowland) and the surface area of each farm allocated to oil palms. The purpose of the current article is to assess the performance of smallholder rice farms in northern Sierra-Leone. Its aim is to show how rice farmers make their choices depending on the ecosystems cultivated, the main production factors and the surface area allocated to oil palms. This study was carried out as part of the FSSIM-Africa (2009-2012) project aimed at developing a decision-support tool to help boost rice production in Sierra Leone (Louhichi K. et al., 2013). 2. Materials and Methods 2.1 Description of the study area The Bombali district covers 7985 Km², or 40% of the total surface area of the northern part of Sierra Leone (Bangura Z. H., Louis M., 2012). The area receives considerable rainfall with average precipitation of about 2500 mm/year (Jalloh A., 2006). The rainy season lasts from June to November. The district has an estimated population of 435,000, each household having an average of seven people, and an average of 2.5 men and 4.5 women. Agricultural land is made up mostly of uplands (60 to 80%) and lowlands (20 to 40%) (NASDP, 2009). Agricultural activity in the district mainly consists of rice production. Bombali produces more rice than any other district in Sierra Leone and actually supplies rice for the entire country (Figure 9). 38 Figure 9: Location of Sierra Leone and the Bombali district (www.d-maps.com) In the two main ecosystems found in the district, soils are relatively homogenous in terms of organic matter and texture, which is often clayey. Upland soils are fairly shallow (ultisols and oxisols) with an average organic matter content of about 1% (MAFFS et al., 2004). Oil palm and rice are the main upland crops. The lowlands have deeper soil that also contains more organic matter (1.3% on average) (Wakatsuki T. et al., 2008). Unlike upland rice, which is cultivated with very variable fallow periods (from 3 to 20 years), lowland rice is often cultivated as a monoculture for several years running (Jalloh A., 2006). 2.2 Database The study was carried out as part of the FSSIM-Africa (2009-2012) project to develop a decision-support tool to propose and assess policies to support traditional smallholder rice farms in Sierra Leone (Louhichi K., Sergio Gomez y P., Belhouchette H., Allen T., Fabre J., Blanco M., Chenoune R., Acs S., Flichman G., 2013). Traditional smallholder rice farms account for the largest number of farms in Sierra Leone, the largest farming area and the largest contributor to national rice production (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). The database used was compiled from surveys of Bombali rice farmers conducted between 2009 and 2012 by the Joint Research Centre of Seville (JRC) and the Institut Agronomique Méditerranéen de Montpellier (IAMM) (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012; Louhichi K., Sergio Gomez y P., Belhouchette H., Allen T., Fabre J., Blanco M., Chenoune R., Acs S., Flichman G., 2013). The surveys 39 involved 180 farmers from seven villages in the Bombali district: Kagbere, Magboema, Manonkor, Pelwalah, Robanka, Rogbonko, Rokontha. However, out of these 180, our study was based on 81 smallholder rice farms that will benefit from the support policies of the next EU initiatives. The data were collected from rice growing households by means of three questionnaires: A, B and C. Questionnaire A investigated the socio-economic characteristics of the household (composition, age of its members, farming activities, sources of income, etc.). Questionnaire B considered the farms already surveyed by Questionnaire A and looked in more detail at each household’s farming activities (Table 5). The aim was to describe the crop cycles of rice and oil palm by ecosystem. A second aim was to identify the main tasks involved in growing rice and oil palm as well as the breakdown of labour (for men and women) by type of task. Questionnaire C sought structural data to describe the farms growing both rice and oil palm: usable agricultural area, land use (crops grown), amount of time per agricultural task, proportion of family labour, proportion of hired labour, proportion of each task performed by men and women, length of fallow period for each ecosystem, and yields per ecosystem (upland, lowland) and by farm (Table 5). Table 5: Description of the variables used for this study, from Questionnaires A, B and C. The data were collected from farmers in Bombali between March and November 2009. They concern only the 81 farms chosen for the study. Yield is the only variable to have been calculated (total production per hectare) instead of being obtained directly from surveys in the field 40 Variables Total rice production (t/farm). Cultivated area and crop rotation per farm (ha) Seeding density (t/ha) Total work carried out on farm (d/ha) Family size per farm Description The total production per farm is broken down for each farm into total production per type of crop and by ecosystem. It corresponds to the quantity harvested, before drying and winnowing. This area is broken down by crop and by ecosystem. This is the quantity of rice seed used per hectare and per ecosystem. - This is the amount of male and female labour actually used on the farm. The amount of labour was calculated after discussions with several growers on the basis of 6h/d. The total amount of labour per farm is calculated on the basis of the total amount of male and female labour, whether family or hired, per task, per crop (rice or oil palm) and per ecosystem. - For the uplands, the amount of work for sowing includes land clearance, burning and actual sowing. - For the lowlands the amount of planting work includes field clearance and transplanting. - We decided not to take child bird-scaring labour into consideration as there was considerable uncertainty about the quality of the data. This variable concerns the total number of adults and young people per family. Proportions consumed, sold or stored per farm These proportions are percentages of the entire production per farm, of the quantities consumed by the family, sold at the local market or stored as seed for the following year. Quality of plot and difficulty of working the soil This qualitative item describes the quality of the field that has been left to lie fallow, after clearance and burning, and the difficulty of working the soil as a result of its texture. The possible answers were “Good” or “Bad” regarding its post-fallow condition and “Difficult” or “Not difficult” regarding the difficulty of working the soil. Calculated by taking the production per crop and per ecosystem divided by the surface area of each crop per ecosystem. Yield (t/ha) 2.3 Levels of analysis and methods for assessing rice growing systems in Bombali The study followed a procedure at two levels: - at crop per ecosystem level, to describe cropping cycles for rice and oil palm, mainly based on Questionnaire B. It will also be a matter to compare for all rice smallholder farms, the rice 41 crop performance per ecosystem (upland or lowland) (Table 6). This involves comparing the surface area, yield, amount of work for smallholder rice farms per ecosystem (upland or lowland), in order to be able highlight the diversity of rice systems in Bombali. - at farm level, to help us understand and explain rice yields depending on farm type. As mentioned above, very few rice farms in Sierra Leone are mechanized or use artificial inputs. In this context, average rice yield per farm and per ecosystem essentially depends on these three production factors: the total amount of work invested in the farm, the length of the fallow period (which determines the level of soil fertility) and seeding density. Based on this observation, a principal component analysis (PCA) was performed to classify the farms by their level of average yield, taking into account these production factors. A Hierarchical Ascendant Classification (HAC) was then applied to group the rice farms into homogenous classes (Tittonell P. et al., 2010). Tableau 6: Number of rice farms and number of rice fields per ecosystem (upland or lowland) and per farm for 126 rice fields on the 81 farms selected and surveyed within the study Frams number of rice plot / ecosystem Total Total farms Upland Lowland Rice_U Rice_U_palm Rice_L Rice_L_palm 14 6 11 5 14 6 0 0 0 0 11 5 Rice_U_rice_L 21 21 21 Rice_U_L_palm 24 24 24 81 126 In a second phase, a more detailed analysis of each type of farm was carried out so as to characterize their behaviour based on all the variables surveyed and listed in Table 5 (Section 2.3). 42 3. Results 3.1 Description of cropping cycles for rice and oil palm - Upland rice: upland rice is cultivated mostly from June to November (Figure 10). The field can be prepared for sowing from December to March, just before the arrival of the first rains. This includes tree-felling and field-clearing by slash-and-burn. Rice is sown in mid-June, the density of seed varying considerably depending on the labour available and the quantity of rice seed stored since the previous harvest. Rice is often sown in association with other crops such as beans. Weeding is usually done two weeks after sowing, during the second half of July. Harvesting may be staggered depending on the family’s needs. Concerning the distribution of labour, women only play a marginal role in sowing but do a considerable part of the weeding. Men, meanwhile, do most of the tree-felling, field-clearing by slash-and-burn, and sowing. Men may occasionally participate in weeding and harvesting. - Lowland rice: preparing the field for lowland rice always begins with field-clearing. This operation (removing weeds) often begins in February but may continue into March (Figure 10). This is done entirely manually. Plants, often obtained by barter, are prepared in nurseries from May onwards, and transplanted in July; the harvest follows at the end of November, sometimes extending into December. - Oil palm: Oil palms are ideally planted in May and the first crop of fruit can be harvested from the third year after planting. The harvest begins in January and usually coincides with the sale of rice. The extraction of palm oil is an exclusively manual practice that starts in March and can extend over two months. This is often a crucial period for the preparation of rice fields (Figure 10). In Bombali, no maintenance tasks (such as weeding or pruning) are carried out on oil palms. They are only grown in the uplands. Before the war, oil palms were cultivated together with upland rice, but most farmers now allot separate fields to this crop. This choice is driven by the profitability of the crop, which tends to be planted at higher and higher densities, thus preventing any association with other crops. 43 Harverst Plantation (c) (b) Trans* (a) Seeding March Apr Mai Jun Oil palm Aug Sep Oct Wet season Dry season Rice_L Clearing + burn Harvest Weeding July Clearing Harverst Nov Dec Jan Fev March Dry season Rice_U * Transplantation Figure 10: Work schedules by type of crop system (rice and oil palm) and by ecosystem. (a) Upland rice; (b) Lowland rice and (c) Oil palms. 3.2 Comparative performance of rice per ecosystem As shown in Figure 11, rice growing practices in lowland ecosystems give higher average yields than in exclusively upland systems. Average yields are 0.29 t/ha (median yields 0.26 t/ha) and 0.34 t/ha (median 0.27 t/ha) for upland rice and lowland rice respectively. Conversely, the total amount of labour required per hectare is much higher for upland rice than for lowland rice. Rice farmers invest an average of 121 d/ha (median 110 d/ha) for upland rice and 90 d/ha (median 79 d/ha) for lowland rice. The distribution of labour between men and women differs for each ecosystem. Women provide approximately half of the total amount of labour for each type of ecosystem, specifically 45% and 41% for upland rice and lowland rice respectively (data not shown). The average size of upland rice fields is larger (average 0.99 ha; median 1.11 ha) than that of lowland rice (average 0.66 ha; median 0.61 ha). A frequency analysis shows that more than 80% of lowland fields have a surface area of less than 1 ha (and 50% have less than 0.5 ha). By contrast, almost 70% of upland fields have a surface area of between 0.5 ha and 1.1 ha, and on average 10% of farms have a surface area greater than 1.1 ha (data not shown). In fact, as shown in Figure 11, irrespective of the variable analysed (yield, amount of work invested, etc.), there is considerable variability in both the average and the median values of these variables, demonstrating the wide diversity in the practices of these rice farmers. The same variability is found for upland rice as for lowland rice. For example, the standard deviations of yields for upland rice and lowland rice are 0.13t/ha and 0.2t/ha respectively for average yields of 0.29t/ha and 0.34t/ha. 44 (0.29 t/ha) (0.34 t/ha) (0.99 ha) (0.66 ha) (121 d/ha) (90 d/ha) Figure 11: Median, maximum and minimum values, first quartile, third quartile and mean (values in brackets) for upland and lowland ecosystems for the following variables: yield, surface area and total labour. These variables were calculated from surveys and concern 126 rice fields 3.3 Comparative analysis by type and class of farm 3.3.1 General description of the types of farms surveyed An analysis of the sample group based on our surveys of 81 farms in the Bombali district reveals six types of rice farms, depending on the prevailing ecosystem and the presence or absence of oil palm (Table 7): 1- Rice_U: Farms with only one rice field in an upland ecosystem. 2- Rice_L: Farms producing only rice in a lowland ecosystem. 3- Rice_U_palm: Farms growing both rice and oil palm in an upland ecosystem. 4- Rice_L_palm: Farms with two crops: rice in the lowland ecosystem and oil palm in the upland ecosystem. 5- Rice_U_Rice_L: Farms producing rice only, but in both upland and lowland ecosystems. 6- Rice_U_Rice_L_palm: Farms with three types of crop: rice and oil palm cultivated (separately) in an upland ecosystem and rice cultivated in a lowland ecosystem. 3.3.2 Typology: classes of farm Smallholder rice farms were grouped into two clusters considering, as explained in section 2.3, the three main production factors (total labour, fallow duration and seeding density) 45 represented by proxy indicators derived through PCA analysis. The analysis of the absolute value of the loadings (in %) of the production factor variables with respect to the two PC’s shows that the seeding density has the greatest loading with PC1. PC2 is alternatively associated with total labour and fallow duration (data not shown). In the final analysis, the two axes (PC1 and PC2) explained 92% of the variance (data not shown). These PCs were used as classificatory variables in the Hierarchical Cluster Analysis (HCA) allowing the identification of 4 classes of farms as illustrated in figure 12a and figure 12b. In detail (table 7). 46 Table 7: Average yield, seeding density, duration of fallow period, total labour and number of farms per class. Farms were classified on the basis of a PCA and HAC from a sample of 81 farms surveyed in Bombali. The total in the table gives the total number of farms and the weighted average and standard deviation corresponding to the weighted average and standard deviation by area of the following variables: average yield, seeding density, length of fallow period and total labour Class of farm High rice yields 0.69 0.51 Seeding density (t/ha) 0.33 0.32 Lenght of fallow period (year) - * 6 Rice_L_palm Rice_U_rice_L 8 3 1 0.54 0.11 0.52 0.39 0.32 0.14 0.3 0.04 6 2 10 97 27 182 193 Rice_U_rice_L_palm 8 0.42 0.25 5 174 Rice_U_palm 2 0.48 0.24 5 192 14 0.22 5 178 Number of farm Average yield (t/ha) Rice_L Rice_U_rice_L_palm 2 6 Farm system Total Average Standard deviation Fairly high rice yield Total Total labour (d/ha) 46 106 Average - 0.43 Standard deviation - 0.13 0.12 1 27 9 11 4 5 1 30 - 0.29 0.29 0.33 0.27 0.2 0.07 0.07 0.19 0.11 0.13 8 6 8 5 57 79 81 69 78 0.3 0.07 0.11 0.06 7 2 74 18 Rice_L_palm Rice_ U_rice_L Rice_ U _rice_L_palm 2 9 6 0.23 0.23 0.21 0.06 0.08 0.12 9 13 138 106 129 Rice_ U Rice_ U _palm 9 3 29 - 0.25 0.13 0.05 0.1 9 6 131 95 0.23 0.06 0.08 0.04 9 4 119 24 Fairly low rice yield Rice_L Rice_U_rice_L Rice_U_rice_L_palm Rice_ U Rice_ U _palm Total Average Standard deviation Low rice yield Total Average Standard deviation *The minus sign (-) in the “length of the fallow period” column indicates lowland farms that do not use fallow periods Farms with high rice yields: Class 1 contains the farms with the highest rice yields (average yields of 0.54 t/ha), high seeding density (0.32 t/ha), short fallow periods (6 years) and a fairly low total amount of work (97 d/ha). This class is mostly made up of farms 47 growing rice in both ecosystems (upland and lowland) together with oil palm (Rice_U_rice_L_palm). Farms with fairly high rice yields: Class 2 mostly contains farms cultivating rice in one or both ecosystems but always associated with oil palm (Rice_L_palm, Rice_U_rice_L_palm, Rice_U_palm) with the exception of a single farm with both ecosystems but no oil palm (Rice_U_rice_L). This class has fairly high rice yields (0.43 t/ha), medium seeding density (0.22 t/ha), fallow periods similar to those of Class 1 (5 years) but the highest total amount of labour of all four classes (178 j/ha). Farms with fairly low rice yields: Class 3 contains farms with fairly low rice yields (0.30 t/ha), lower seeding density than the first two classes (0.11 t/ha), longer fallow periods than the first two classes (7 years) and the lowest total amount of labour of all the classes (74 d/ha). It contains 30 farms covering all types of ecosystems. Farms with low rice yields: Class 4 contains farms with the lowest average rice yields (0.23 t/ha) and sowing densities (0.08 t/ha) of all the classes. The farms in this class use very long fallow periods compared to the other classes (9 years) and a fairly large amount of labour (119 d/ha). This class contains 29 farms. Figure 12a: Distribution of farms surveyed (n=81) by classes of farms as a function of PC1 and PC2 48 High Short Fallow duration Upland rice Low land rice 0,8 Low land rice Yiel (t/ha) C4 0,4 Upland rice Low land rice OP C2 C3 C1 Upland rice 0 0 100 200 300 Total labour (day/ha) Figure 12b: variation of rice yield for the 4 types of household rice farms (C1to C4) by taking into account the rice ecosystems (upland, low land and the oil palm (OP)) and fallow duration. For a better comprehension of the behaviour of the four farm classes we also compared the: - Surface area by class of farm and by ecosystem: An analysis of Figure 13 shows that the farms in all classes have almost the same average surface area (2.2 ha), with the exception of those in Class 3 (1.3 ha). As a proportion of our sample, the farms in Classes 3 and 4 (rather low and low yields) are the most numerous, with 30 and 29 for the two classes respectively (Table 3, Section 3.3.2), compared to Classes 1 and 2 (fairly high and high yields) with 8 and 14 farms respectively. The farms in Class 1 (high yields) have an average surface area presenting similar proportions of lowland rice and upland rice, respectively 28% and 32% of the farms’ total surface area. The remainder (40%) is allocated to oil palm cultivated in the upland ecosystem. This distribution in terms of crops (rice and oil palm) and ecosystem (upland and lowland) is also found in the farms in Class 2. For the farms in Classes 3 and 4 (fairly low and low yields), the proportion of oil palm is considerably higher (53% of the total farm area) than the proportion of upland rice or lowland rice, respectively 13% and 26% for Class 3 and 35% and 22% for Class 4. 49 farm area (ha) 2.2 ha (1.2*) 1.3 ha (0.7) 2.2 ha (0.7) 1.9 ha (1.05) Share by class of farm (%) 100% 80% 60% 40% 20% 0% Class 1 Class 2 Class 3 Class 4 class of farm (*) Standar deviation upland rice lowland rice oil palm Figure 13: Total surface area, standard deviation (value in brackets) and proportion of the surface area in upland rice, lowland rice and oil palm per class of farm. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). - Distance from household to farm by class of farm and by ecosystem: Farms in Classes 3 and 4 are located fairly close to the rice farmers' households, with average distances of 1.26 km and 0.83 km respectively (Figure 14). The farms in Classes 1 and 2 (fairly high and high rice yields) are respectively 1.76 and 2 km from the rice farmers' households. A more detailed analysis of these distances shows that irrespective of the type of class, farms with an exclusively upland or lowland ecosystem are closest to the rice farmers' households (average distance of less than 0.1 km) (Figure 15). Farms where oil palms are cultivated (Rice_U_rice_L_palm farms or Rice_U_palm farms) are the most distant. The latter (Rice_U_palm farms) have the lowest yields (Table 3, Section 3.3.2). 50 Figure 14: Average distance and standard deviation from household to farm by class of farm. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). Figure 15: Average distance and standard deviation from household to farm by type of farm. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). Averages and mean deviations are calculated on the basis of the 81 farms surveyed. 51 3.3.3 Impacts of production factors on rice performance. - Effect of labour use on yield An analysis of the labour factor per class of farm shows, as previously explained in Section 3.2, a considerable difference in the total amount of labour invested per class of farm. This variability can also be observed in the proportions of total family and hired labour (Figure 16). The proportion of total family labour in total labour is higher in farms belonging to Class Percentage of family labour out of total labour (%) 1 (40%) than in other farms (20% on average). 40% class 1 30% class 3 class 4 class 2 20% 10% 0% 0 100 200 Total labour (d/ha) 300 400 Figure 16: Total labour, standard deviation relative to total labour and proportions (%) of family labour per class of farm relative to the total amount of labour. This analysis was made on the basis of the 81 farms surveyed. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). The same type of result is found in lowland ecosystems with a larger proportion of family labour (as a percentage of total labour) for Class 1 (52% of total labour) compared to the other classes, for which the average is approximately 45% (data not shown). The distribution of the amount of labour between men and women per task and per class of farm shows that men do more of the sowing while women do more of the weeding and harvesting. Nevertheless, in the high yield class (Class 1), the proportion of female labour in weeding and harvesting is much higher than in the other classes (Figure 17). The same type of result is found for farms in lowland ecosystems where female labour is more widespread during harvesting for Class 1 farms than for the other classes (data not shown). In both cases, female labour is mostly family labour (data not shown). 52 Share of total men and women labour (%) 100% 80% 60% 40% 20% 0% Class Class Class Class Class Class Class Class Class Class Class Class 1 2 3 4 1 2 3 4 1 2 3 4 Seeding Weeding man labour Harvesting women labour Figure 17: Proportions of male and female labour relative to total labour by task and class of farm. This analysis was made on the basis of the 81 farms surveyed. Class 1 (high rice yields), Class 2 (fairly high rice yields), Class 3 (fairly low rice yields) and Class 4 (low rice yields). Table 8 shows that for the four classes of smallholder farms, the yield is only slightly correlated, except for class 1 which has an R² of 0.49, with the total amount of work done on the farm. Nevertheless, a closer analysis of this variable (total work) shows that the levels of correlation with the yield vary greatly according to the cropping tasks per ecosystem and per type of manual labour (family, hired, male, female). 53 Tableau 8: Correlation values (R²) between yield and the amount of labour by ecosystem, task and type (family, hired, women, men). These values are expressed by class of yield: high rice yields (class 1), Fairly high rice yield (class 2), Fairly low rice yield (class 3) et Low rice yield (class 4) Coefficient of correlation (R²) Upland Lowland Farm Seeding Weeding Harvesting Transplanting Harvesting Family labour 0.76 0.07 0.12 0.02 0.15 0.71 Hired labour 0.84 0.64 0.54 0.48 0.03 0.11 0.79 0.16 0.52 0.59 0.46 0.51 Female labour 0.39 0.83 0.52 0.02 0.15 0.42 Labour/task 0.84 0.67 0.6 0.47 0.28 - Classe 1 Man labour Total labour 0.89 0.49 Family labour 0.37 0.25 0.08 0.20 0.05 0.2 Hired labour 0.21 0.05 0.34 0.20 0.43 0.2 0.24 0.00 0.28 0.25 0.42 0.1 Female labour 0.40 0.16 0.33 0.20 0.05 0.02 Labour/task 0.4 0.16 0.53 0.23 0.37 - Classe 2 Man labour Total labour 0.48 0.31 0.04 Family labour 0.52 0.25 0.24 0.34 0.15 0.09 Hired labour 0.33 0.29 0.52 0.17 0.32 0 0.49 0.30 0.58 0.24 0.39 0 Female labour 0.21 0.30 0.34 0.34 0.15 0.03 Labour/task 0.51 0.49 0.89 0.36 0.4 - Classe 3 Man labour Total labour Classe4 0.1 0.03 0.52 0.01 Family labour 0.01 0.00 0.06 0.17 0.04 0.06 Hired labour 0.39 0.35 0.58 0.38 0.44 0.21 Man labour 0.36 0.02 0.44 0.45 0.54 0.12 Female labour 0.00 0.26 0.00 0.17 0.04 0 Labour/task 0.38 0.31 0.30 0.40 0.40 - Total labour 0.48 0.45 0.05 54 For the classes with high yields (classes 1 and 2), the rice yield for the upland ecosystem appears to be more sensitive to the amount of work per task than that of lowland rice. Indeed, classes 1 and 2 respectively show correlations of 0.89 and 0.48 for the upland areas and 0.1 and 0.31 for the lowland areas. The result is entirely the opposite for class 3 smallholder farms, with correlations of 0.52 and 0.03 respectively for the lowland and upland ecosystems. For class 4 farms, the correlation between the yield variable and the total amount of work is practically the same for both ecosystems (0.48 and 0.45). In upland areas, the correlations between yield and total amount of work per task (seeding, weeding, harvesting) varied according to the class of smallholder farm. The highest correlations were found for class 1 with an R² of 0.84, 0.67 and 0.6 respectively for seeding, weeding and harvesting. On the other hand, the lowest correlations were recorded for class 4 with an R² of 0.38, 0.31 and 0.30 respectively for seeding, weeding and harvesting. The table also shows that, depending on the tasks and type of labour (hired, family, male, female), the correlation between the amount of work and the yield vary greatly. For example, for class 1, the correlations between the amount of hired work and the yield are 0.84, 0.64 and 0.54 respectively for seeding, weeding and harvesting. These correlations are lower for class 4, namely 0.39, 0.35 and 0.58 respectively for seeding, weeding and harvesting. We also observed that the strong correlation between family labour and yield was only recorded for class 1 and class 3 and in particular for the seeding task (0.76 and 0.52 respectively for class 1 and class 3). In lowland areas, the correlations between the total amount of work per task and per type of labour and the yield are fairly low when compared to those of the upland areas. In this context the most significant correlations were observed for the amount of hired labour and the amount of work done by men, particularly for transplanting and harvesting. - Effect of seeding density on yield Figure 18 shows that the average rice yield per farm is strongly correlated with seeding density (R² = 0.8). This correlation is closely linked to the strong correlation between seeding density and rice yields for upland ecosystems (R² = 0.86) and lowland ecosystems (R² = 0.6). From this figure we can also conclude that the high yields in Classes 1 and 2 are due to a simultaneous increase in rice yields in both upland and lowland ecosystems. We find the same type of result for yields in farms belonging to Classes 3 and 4 (fairly low and low rice yields) 55 where the low yield is explained by a simultaneous drop in yields in both ecosystems at the same time. average rice yield / class of farm = 1.3266 x 0.1185 (R ² = 0.7924) lowland rice yield / classe of farm = 1.4995 x 0.1284 (R ² = 0.8575) yield upland rice / class of farm = 0.9109 x .1468 (R ² = 0.5836) 0.6 Average yield (t/ha) Class (1,2) Class (3,4) 0.4 0.2 0.0 0.0 0.1 0.2 0.3 0.4 Seeding density (t/ha) farm lowland ecosystem upland ecosystem Figure 18: Average rice yields per farm and per ecosystem (uplands and lowlands) and per farm as a function of seeding density. Average rice yield per farm is calculated as the weighted average for the surface area of rice grown for both ecosystems (upland and lowland). Seeding density itself depends heavily on: - the quantity of rice stored by each farm (figure 19). This only concerns farms with low rice yields in Classes 3 and 4 (fairly low and low rice yields). The larger the quantity stored, the higher the density sown. For farms with high yields (Classes 1 and 2) this correlation is very low. Figure 17 shows that the proportion of the harvest stored as seed is much higher in Class 1 than in Classes 2, 3 and 4. This seems to correlate with family size, which is lower in Class 1 than in the other classes (figure 20). Figure 20 also shows how the proportion consumed increases from Class 1 to Class 4. In quantitative terms, this corresponds to an annual consumption per person of 0.03, 0.016, 0.015 and 0.014 tonnes for Classes 1, 2, 3 and 4 respectively. For all classes, the proportion of the crop sold is almost the same, irrespective of the class of farm. 56 Figure 19: Seeding density as a function of the quantity of seed stored per class. Family size (number of family menbers) 6 (2*) 10 (4) 10 (4) 9 (5) Percentage of rice consumed, stored and sold /class of farms of farms (%) 100% 80% 60% 40% 20% 0% Class 1 Class 2 consumed Class 3 class of farm stored Class 4 sold (*) standard deviation Figure 20: Average family size (number of family members) and standard deviation per class of farm as a function of the proportion (as a percentage of total production) of the quantities consumed, sold and stored. 57 - The correlation between the quantity stored and the family size expressed as the number of family members is particularly evident in farms that do not cultivate oil palm, i.e. mainly farms in the low yield classes (Classes 3 and 4). As expected, there is a negative correlation between the quantity of rice stored (and consequently seeding density) and family size (Figure 21). - For farms where oil palm is cultivated, the quantity of rice stored as seed for the following year is negatively correlated with the proportion of oil palm on the farms (Figure 22). Figure 21: Quantity of rice stored as seed as a function of family size for farms not cultivating oil palms Figure 22: Quantity of rice stored per farm as a function of the proportion of surface area allocated to oil palms per farm - Effect of the length of fallow periods on yield This analysis only concerns rice yields in upland ecosystems for the following farms: Rice_U, Rice_U_rice_L, Rice_U_palm and Rice_U_rice_L_palm. 58 An analysis of figure 23 showing the variations in yield as a function of the average duration of fallow periods for upland ecosystems per farm shows fairly heterogeneous behaviour for the different classes of farms. In general, yield declines in a linear fashion as the length of fallow period increases. Overall the correlation is 0.2 but this becomes 0.8 when the two outliers (the farms circled in figure 23) are excluded. These two farms grow upland rice in association with oil palm and have short fallow periods (5 to 6 years) and low yields, mainly caused by low seeding densities (figure 23). The lowest yields are associated with the longest fallow periods, mainly for farms in Classes 3 and 4. yield of upland rice (t/ha) 0.6 y = -0.0311x + 0.5456 R² = 0.7675 0.4 0.2 0.0 0 5 10 15 Fallow duration (year) Figure 23: Performance rice tray depending on the length of fallow. Farms are surrounded farms rice and palm oil pan (Rice_U_palm) in low yield (class 3.4) Figure 24: Quantity of Labour depending on the sowing fallow periods. Farms are surrounded farms rice tray and oil palm (Rice_U_palm) in low yield (class 3.4) 59 As has been mentioned by several authors (e.g. Gleave, 1996), the length of the fallow period, and consequently its impact on soil fertility and the degree of invasion by bushes and weeds, may influence the choices of rice farmers in terms of seeding density and labour investment during the sowing period. Figures 24 and 25 illustrate the way these two parameters (seeding density and amount of work invested during sowing) vary as a function of the length of the preceding fallow period. With the exception of the two exclusively upland farms that also cultivate oil palm (Figures 24 and 25), there is a negative correlation between seeding density and the length of fallow periods. However, the amount of work needed for sowing (“sowing” here involves tree-felling, slashand-burn operations and the sowing itself, as explained in Table 1, Section 2.4) correlates positively with the length of fallow periods, except for farms in Class 2. Regarding the farms in Class 2, the considerable amount of work needed for sowing can be explained by the length of the fallow period and the condition of the rice field prior to sowing, as well as soil quality and the difficulty in working with that soil (DATA not shown). Figure 25: Seeding based on the length of fallow. Farms are surrounded farms rice and palm oil (Rice_U_palm) in low yield (class 3.4). 60 4. Discussion In Sierra Leone, rice farms are almost entirely non-mechanized and with low input, which largely explains the low average rice yields and their variability from farm to farm. Yield variability can often be accounted for by the types of overall decisions made by farmers. The results of the current study suggest that the principal decisions concern seeding density, surface areas for rice per ecosystem and the surface area allotted to oil palms (figure 26). Nevertheless, before making these decisions, rice farmers must take into consideration a number of production factors such as the availability of land per ecosystem, the length of fallow periods in the uplands, and family size (number of family members). This last factor (family size) determines the family’s rice needs and the availability of family labour. In practical terms, average rice yield by farm, and consequently total farm production, will be affected by: - Labour availability: for several authors (Mbetid-Bessanee E., 2004; Toulmin C. et al., 2003) the proportion of family labour to total labour primarily depends on family size (number of family members). The results of the current study contradict these findings. It was found that large families are often less involved in the activities of their own farms. According to Maclean J. L. et al. (2002) large families are often either poor or new arrivals in Sierra Leone (returning after the war, or immigrants from neighbouring regions) who may well have to change their place of residence and occupy new land every year (which would explain the short distances from the homes of this type of grower to their farms). They are often sharecroppers, forced to hand over a small proportion of their production. Indeed, it seems that growers from this group work for other farmers more than on their own land, in order to bring in as much income as possible. This income often takes the form of sacks of rice or meals for the whole family (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). For several authors the low rice yield in Sierra Leone is mainly related to the amount of labour used at the weeding stage (Rodenburg J. et al., 2013; Yamah A., 2002). In the current study it was shown that depending on the farm types the yield is differently correlated with the amount of labour engaged in each work schedule. In addition, the highest correlation between rice yield and the amount of labour employed is observed for farms with high rice 61 yields (class 1). For the other classes this result is less evident, thus proving that other limiting factors affect yield more negatively than the labour engaged. - the quantity of seed stored: for CORAF/WECARD et al. (2003) the quantity of rice seed stored from the previous season is a predominant factor affecting seeding density, and consequently production, for the following year. This situation reflects the real difficulty for rice farmers in deciding, each year, how much to keep for the year’s consumption and how much to store as seed for the following year. To summarize, the quantity stored as seed heavily depends on family size (the larger the family the larger the quantities consumed compared to the quantities stored), on the year’s production (the larger and more fertile the fields cultivated, the higher the production) and on the area allotted to oil palms. On this subject, it seems that the more oil palms are cultivated per farm, the less seeding density depends on the year’s production or on the family’s rice needs. According to Carrere R. (2010) the income from the sale of palm oil is an important resource to break the interdependence of quantities stored and quantities consumed. This is because rice farmers can use the income from the sale of palm oil to buy rice seed or rice for consumption and can therefore decide what quantities of rice to store as seed irrespective of the year’s production and their needs for household consumption. - length of the fallow period: this factor has a significant effect on soil fertility, especially in upland ecosystems. Several authors found that the longer the fallow period was, the more fertile the soil and the higher the yield were (Gaiser T. et al., 2011; Serpantié G. et al., 2001). On the farms in the current study, however, yields after long fallow periods (more than 10 years) were rather low, compared to medium-length fallow periods (5 to 7 years). Because of the lack of any mechanization, the biomass produced during long fallow periods is not always burnt off correctly. As a result, larger items (trunks) remain on the ground and rot, causing several diseases (Delarue J., 2007). In this context, rice-growers consider it important for a farm to combine upland and lowland ecosystems to ensure a minimum production, independently of the length of the fallow period in upland fields. 62 Fallow duration Labour availability Familly / Hired Crop partner Yield * Upland rice Lowland rice Oil palm Farm area Farm production Oil palm effect Family size Familly rice need Rice consummed Rice stored Rice sold Rice loss during storage Seeds density Figure 26: Criteria influencing the decisions of smallholder rice farms in Bombali Outlook for rice farmers in Sierra Leone In the light of the current results, traditional rice farms with exclusively upland ecosystems will probably not be able to satisfy their households’ future rice needs, especially those with large families. This result concurs with the findings of (NEPAD, FAO, 2009), who explain that this type of farm is tending to disappear because of low rice yields. Combining lowland with upland rice can increase average yield compared to that of exclusively upland farms. This type of farm, which is very common in northern Sierra Leone, and given the lack of efficient rice storage facilities (Kargbo A. B., 2002), reflects the real desire of growers to extend their rice production over as long a period as possible (from November to March). According to Toulmin C., Guèye B. (2003), this combination of ecosystems enables growers to reduce both climate-related and socioeconomic risks (access to seed, availability of labour, etc.), which can significantly affect one or other of the two ecosystems in a given year. In view of the results obtained, the best-performing farms (high average yield and low total amount of labour invested per farm) all combine two ecosystems and cultivate both rice and 63 oil palm. Many leading agricultural authorities believe that it is vital to promote this type of farming in the years to come. Combining the two ecosystems makes it possible to spread production over several months, thus reducing climate-related and socioeconomic production risks in each ecosystem, and disassociating the proportion to be stored as seed from the proportion to be stored for household consumption. However, it will be difficult to achieve this balance between the two types of ecosystems and between rice and oil palm, because of increasing population pressure in the uplands concomitant with political stability and the arrival of immigrants from neighbouring regions (DIC, 2007). In addition, some politicians would like to lease the uplands to multinational developers for the intensive production of palm oil (NASDP, 2009). 5. Conclusion The purpose of the current article was to identify and characterize the main factors affecting average rice yield per farm in Bombali. The study found that two types of farms could be distinguished depending on the ecosystems cultivated, the area allotted to oil palm, the length of fallow periods, and seeding density. In rice farms with one or both ecosystems but no oil palm, farmers cultivate rice primarily for their household consumption. These are mainly farms with low or very low yields, cultivating rice after long fallow periods and with low seeding density. In this group, there often is a positive correlation between the average farm yield and an increase in lowland rice area. The second group of farms is made up of those cultivating land in one or both ecosystems and including oil palm. The yields in this group vary greatly (from very low to very high). Growers in this group cultivated rice after medium to long fallow periods. In these cases, a positive correlation was found between the average rice yield per farm and an increased seeding density, shorter fallow periods and greater areas allocated to oil palm per farm. In this context of low and variable rice yields, several international organizations are currently working with the government of Sierra Leone to help rice farmers boost production (WASNET, 2005). In an agricultural context that is likely to remain largely unmechanized and low-input for several years, the current findings suggest that these initiatives should consider the problems of low seeding density and soil fertility (associated with the duration of fallow periods) as priorities for any action aiming to improve rice yields. 64 Chapter 3 Finding pathways out of poverty: The case of oil palm as a cash crop enabling the intensification of subsistence rice based farming systems in Sierra Leone. Chenoune, R (a ,b)., Belhouchette, H (a,c)., Adam, M (d)., Whitbread A(e,f)., Capillon, A (b). (a) IAMM-CIHEAM, 3191 Route de Mende, 34093 Montpellier. Cedex 5, France Montpellier SupAgro, UMR System, 2 Place Viala, 34060 Montpellier, France. (c) IAMM- UMR System, 2 Place Viala, 34060 Montpellier, France. (d) CIRAD, UMR AGAP, Route de Mende, 34093 Montpellier Cedex 5, France. (e) Crop Production Systems in the Tropics, Georg-August-Universität Göttingen, Grisebachstraße 6, 37077 Göttingen, Germany. (f) International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Patancheru, Andhra Pradesh, India. (b) Abstract In West Africa, the production of rice as a staple food crop by subsistence households is often found to be unstable and highly inefficient. Several studies have explained this poor performance and low efficiency by analysing the effect of input use on rice yield performance. The aim of this paper is demonstrate the use of an analytical framework that more broadly analyses household level rice yield performance by accounting for, not only production factors (inputs), but also household production targets and farm structure in a large rice-producing area in northern Sierra Leone. The results showed that farmers produce rice essentially for their own consumption. In this context, rice production is linked closely to the availability of labour, initial soil fertility and the availability of seeds. Furthermore, four types of household reflecting two different production strategies were observed. The first involved households with a low target production (< 0.3 t/household) characterised by small-scale farms (< 1.4 ha/farm) dominated by a single ecosystem (upland or lowland) with a low level of intensification (seed planting density <0.02 t/ha) due to the absence of a cash crop (in this case oil palm). The second type comprised of households with larger target production (> 0.8 t/farm) obtained through increased use of inputs (seed density >0.05 t/ha) enabled by the combination of oil palm and rice. This study also showed that the levers for improving the target production of rice farms are very limited and are currently limited to increasing the areas under cultivation or via intensification strategies enabled by the cash crop, oil palm. The study also revealed that the different levels of efficiency observed can be explained by 65 ecosystem type, soil fertility status, the existence of a cash crop, or the duration of the fallow on the upland area. Because of higher initial soil fertility, the lowlands derive greater value from labour than labour investments in the uplands. However, in both types of ecosystem, the intensification is synonymous with increased yield (on average from 0.25t/ha to 0.5 t/ha) with a fall in efficiency in relation to the quantities of seed used. This analytical framework, applied to explain the performance and efficiency of rice production in Sierra Leone, can be extrapolated to other rice-producing regions in West Africa and to other types of food crop. Key words: household typology; rice performance; efficiency; intensification; analytical framework. 1. Introduction More than 50% of agricultural food products worldwide are produced by subsistence farming households with little or no inputs (CRDI, 2013; Gollin D. et al., 2014). Long considered a basic staple food in Asian countries, rice has become one of commodities most in demand in African households as well. Outstripping production, the demand for rice in African countries is experiencing a drastic increase. Demand in West Africa is increasing at a rate of 6% per year, which is higher than population growth and greater than anywhere else in the world (Eklou A. S., 2013), while domestic production, more than 80% of which is provided by subsistence households covers only 60% of the population’s demand (Resimao, 2007). There are many possibilities for increasing west-African rice production significantly. The agroecological conditions favourable to rice farming are indeed very diverse in West Africa with its rainfall and numerous waterways, although the region also demonstrates extremely low and variable yields and indicative of highly variable levels of efficiency (Tittonell P., 2014). The main aim of analysing the efficiency of crop and farming systems is to identify the factors limiting production to suggest solution/ identify potential action to improve the systems performance (Carberry P. S., 2004; Haefele S. M. et al., 2013; Keating B. et al., 2010). Several studies have attempted to analyse and understand the performance and efficiency of rice farms in West Africa (Ceesay M. M., 2004; Diagne A. et al., 2010). They are usually focused on the impacts, in various biophysical conditions, of different type of rice practices and input amounts on yield performance and efficiency (Saito K. et al., 2010). Within these 66 studies, they define envelope curve by attempting to isolate the biotic and abiotic factors limiting potential yield (Van Ittersum M. K. et al., 2013). This type of analytical approach gives very little importance to reasons behind the farmer decision making in terms of resource allocation among the main cultivated crops and their respective performances and efficiencies (Ignacio A. C. et al., 2012; Sultan B. et al., 2005). Norman D. W. et al. (1995) cited at least three main factors should be considered to explain farmer decisions in term of resources allocation: resource endowments (land, labour, capitals), production goals and aspiration, and the farm production intensification levels (defined by the amounts of fertilizer, seeds, irrigation, etc.). Eleven years after the end of the war (1991-2002), Sierra Leone is still ranked as the third poorest country in the world (Bangura Z. H., Louis M., 2012). Within this context, agriculture and more particularly rice production, represents an important resource for the economy, accounting for 56% of the country’s GDP and employing two-thirds of the rural population (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). Recently, several national and international projects have been launched to help Sierra Leone restructure agriculture, in particular the rice production sector (e.g. the STABEX European fund from 2007 to 2009). To date, there remains no assessment of the performance and efficiency of the current rice-growing systems (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). Such an assessment should be available as a basis for understanding rice farmers’ production choices in light of the main biophysical and socio-economic factors of production. The aim of this study was to explain and interpret the performance (rice yield) of the rice farms in the Bombali, Northern Sierra Leone, by applying a conceptual analytical framework that includes factors of production, resource endowments and production targets. This framework enables to define the main determinants of farmer production choices and thus the levels of performance and efficiency (yield per factor of production) observed. 2. Materials and methods 2.1. Conceptual framework To explain and justify the different performance and efficiency levels among subsistence ricegrowing households in West Africa, we designed the following framework consisting of three stages (Figure 27), and in agreement with Norman D. W., Worman J. D., Siebert E, Modiakgotla (1995), explicit discrimination of factors explaining farmer decision making. 67 Phase 1: Establish the production aim: the notion of “target production” level Within the context of traditional rice-growing households in West Africa, production is primarily used for own consumption (Jalloh A., 2006) in order to satisfy a need (demand) for rice per family which depends on the size of the household. This consumption per household is closely linked to production (Von braun J., 1988) which itself is a direct function of the average rice yield per farm and the size of the farm. Economic anthropologists believe that West African households essentially target minimum survival needs (non-market product) and do not attempt to maximise income (Iversen V., 2002). Due to a high level of market price instability of inputs and even the complete lack of a genuine agricultural inputs market, ricegrowing households prefer to rely on their own labour endowments and resources to satisfy household requirements. There is therefore, a strong correlation between rice production and consumption per household. This leads us to define the notion of “target production” for a household which indicates the minimum production that the household targets to satisfy a minimum food need for a given product. Again conceptually, this also means that the quantity required of each factor of production and farmed area must be directly determined by the level of production targeted. Phase 2: Realize a farm typology to identify factors influencing farm performances To understand performance and efficiency of the rice farms (figure 27), a farm typology can group farm households with similar, biophysical and resource endowments characteristics (Bidogeza J. C. et al., 2009). Typological studies can therefore be of great importance for exploring factors explaining farm strategies for resource allocations (Kostrowicki J., 1983a; Tittonel P. et al., 2013). Multivariate statistical techniques offer the means of creating such typologies, such as principal components analysis (PCA) and cluster analysis (CA), for identifying farm household, methodology applied before by (Hardiman R. T., 1990; Madry W. et al., 2005; Usai M. G. et al., 2006). However, before using this classification, a question must be answered concerning the factors/criteria to be considered during the analysis (stage 2.1). In an unmechanized/ low intensive production context in West Africa, three important factors should be considered, as suggested by Norman D. W., Worman J. D., Siebert E, Modiakgotla (1995): o Factors explaining resource endowment: represented at farm level by: the environmental potential, as indicated by i) the type of ecosystem comprising a subsistence farm as well as the total area of the farm and the area reserved for each crop, and 68 ii) the initial soil fertility which, in the absence of fertiliser, is often dependent on the duration of the fallow; o the availability of financial resources from non-farming income or the presence of a cash crop on the farm (e.g.: oil palm, cassava, etc.). This financial resource can contribute to obtaining factors of production (e.g. purchasing seeds, hiring labour, etc.) (Raphael et al., 2010). Factors explaining production goals: these are the factors which reflect the demand for rice to be eaten in each household. In a consumption model essentially based on own consumption, this demand for rice is primarily expressed through the global production of the farm (target production) which itself depends on the total area of rice per farm and the average rice yield per farm on the one hand and on the size of the household (number of people per household) on the other (Dixon C. A., Johnson S. D., Fomba S., 2001). Factors explaining production intensification levels: the first two structural criteria can have a considerable impact on the quantities of the non-substitutable (seed, fertiliser, irrigation) or substitutable (labour) factors of production (inputs) to be called on per household (Bossa A. Y. et al., 2012). Phase 3: Analyse the performance and efficiency of farm categories. The multivariate (household clustering) analysis enables more homogenous household categories to be defined in relation to the criteria adopted. For each household category, the analysis of the potential relationship between the environment and the resources available for the quantity of factors of production used will help to understand the farmers’ choices with regard to the target production quantities (yield and farm area) and the different levels of efficiency observed. The efficiency is defined here as the farmer success in producing as large as possible an output (rice production) from given set of inputs (Keating B., Carberry P., Bindraban P., Asseng S., Meinke H., J. D., 2010). 69 Figure 27: Phases in explaining and analysing the performance and efficiency levels of ricegrowing systems. 2.2. Case study: rice-growing households in Sierra Leone Description of the study zone The study zone is located in the district of Bombali, which covers 40% of the total surface area of northern Sierra Leone (Larbi A., 2011) and accounts for approximately 80% of domestic rice production (Spencer D. et al., 2009). More than 90% of rice-growing households in this district are households with almost no agricultural income which essentially produce rice for own consumption purposes (Braima J. et al., 2010; Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012). The farming land is primarily located on upland (60%) and lowlands (20%), with the remainder occupied by highly marginal and relatively unfarmed ecosystems (NASDP, 2009). Table 9 presents the composition of our sample described by the number of farms surveyed and their distribution according to the rice field and ecosystems present. In contrast to the lowlands, where rice farmers primarily grow continuous rice over several years, upland rice is normally farmed in rotation with a fallow period (duration ranging from 70 3 to 25 years) and increasingly combined with oil palm, a profitable cash crop (Jalloh A., 2006). While rice-growing households are traditionally represented by a single upland ecosystem, they increasingly consist of mixed ecosystems (upland and lowlands) in order to satisfy the growing demand for rice. Within this district, rice is farmed without the use of inputs (fertiliser, irrigation) or mechanisation (Bumb et al., 2011). Rice production is therefore linked closely to the availability of labour and initial soil fertility (Harding S. S., Taylor D. R., Jalloh A. B., Mahmood N., Dixon C. A., Johnson S. D., 2012; Mbetid-Bessanee E., 2004; Melendez C. J., 2006). Poor availability of good quality rice seed is a further impediment to production (Deen S., Ngaujah A. S., 2010) Tableau 9 : Description of farm type and ecosystem type (upland, lowlands) from the sample of 81 farms identified and surveyed. Type of farm A single upland ecosystem farmed with rice A single upland ecosystem farmed with rice and oil palm A single lowland ecosystem farmed with rice A upland ecosystem farmed with oil palm and a lowland ecosystem farmed with rice A upland ecosystem farmed with rice and a lowland ecosystem farmed with rice A upland ecosystem farmed with rice and palm oil and a lowland ecosystem farmed with rice Total Number of farms 14 6 11 5 Number of rice plots/ecosystem Upland Lowlands 14 0 6 0 0 11 0 5 21 21 21 24 24 24 81 126 - Survey data Extensive data from surveys of rice farmers in Bombali conducted between 2009 and 2010 by the Seville Joint Research Centre (JRC) (Gomez y Paloma S., Acs S., Saravia Matus S., Lakoh A., Baudouin M., Hites G., Sammeth F., 2012; Louhichi K., Sergio Gomez y P., Belhouchette H., Allen T., Fabre J., Blanco M., Chenoune R., Acs S., Flichman G., 2013) were the basis of this study. The surveys involved 81 farmers from seven villages in the Bombali district: Kagbere, Magboema, Manonkor, Pelwalah, Robanka, Rogbonko, Rokontha. Quantitative and qualitative data were collected from smallholders through a questionnaire and face to face interviews that were realised during multiple visits. Interviewers were trained at Njala 71 University and administered the questionnaire between June and November 2009. Interviews mostly focussed on ongoing socio-economic issues. There is no data reflecting interviewees' conditions prior to 2009. Therefore, the data analysis chiefly provides an in depth picture of smallholders' situation in terms of household composition, description of farm based-rice product ion and rice household production and consumption in the second half of 2009. Data collection took place in a context of limited research networks and written sources for local information. Consequently, most of the data presented in this report is primary data (combining survey and interview material), gathered through the research project itself. In details the collected data concern: - Socio-economic characteristics of the household: family size and composition per household (women, men, children), age of its members and sources of income (agricultural income, offfarm income). - Household rice production and consumption: each intertwined farmer has given the measured value of the rice production by ecosystem and farm. It corresponds to the quantity harvested, after drying and winnowing. The total rice production is estimated from the number of rice bags harvest. Each bag weighs 0.05 tons of rice. The rice yield was calculated (total production per hectare) instead of being obtained directly from surveys in the field. From the production, farmer has given an estimation of the proportion (as a percentage) that was reserved for consumption, soled or stocked as seeds. - Structural data to describe the farms growing both rice and oil palm: usable agricultural area, land use (crops grown), working time per agricultural task, proportion of each task performed by men and women, proportion of hired labour by task. - household farming activities: The aim was to describe, in details at each household, the type of dominant crops by ecosystem (upland, lowland), the crop cycles of rice and oil palm by ecosystem, the main tasks involved in growing rice and oil palms as well as the breakdown of labour (for men and women) by type of task, length of fallow period for each ecosystem, the seeding density for rice. The later variable (seeding density) is estimated using the quantity of rice stored as seed minus losses, which fixed as 50% of the total quantity stored (Gomez y Paloma t al., 2013). 72 Tableau 10 : Description of the data collected from 81 farms in Bombali between March and November 2009. Variables Total rice production (t/farm). Description The total production per farm is disaggregated for each farm into total production per type of crop and by ecosystem. It corresponds to the quantity harvested, after drying and winnowing. The total rice production is estimated from the number of rice bags harvest. Each bag weighs 0.05 tons of rice. Cultivated area and crop rotation per farm This area is disaggregated by crop and by ecosystem. (ha) Seeding density (t/ha) Rice seed quantity used per hectare and per ecosystem. This density is estimated using the quantity of rice stored as seed minus losses, which fixed as 50% of the total quantity stored. The proportion of rice stored from total production was estimated by farmer. The total amount of labour per farm is calculated on the basis of the total amount of male and female labour, whether family or hired, per task, per crop (rice or oil palm) and per ecosystem. Total work carried out - For the lowlands the amount of planting work includes field clearance and transplanting. on farm (d/ha) - Because of uncertain data quality, labour allocated to birdscaring, usually provided by children, was not included. Family size per farm Total number of adults and young people per family. Household These proportions, as expressed by each farmer during the consumption of farm survey, are percentages of the entire production per farm, of the grown produce quantities consumed by the family. Yield (t/ha) Calculated by taking the production per crop and per ecosystem divided by the surface area of each crop per ecosystem. 73 Application of the analysis framework to the case study As shown in Figure 28, the analytical framework was applied in three stages. Stage 1 of application checks the relationship between the production aim and the rice needs per family, based on the size of the household. The stronger the correlation between target production and the demand for rice, the more the household operates in a context of minimum rice production. Based on the 8 variables in table 3, a farm typology analysis was then conducted (phase 2) in order to categorise households using three types of criterion: 1- the resource endowment represented by: i) the potential of the location expressed by the ratio of lowland area to upland area and the duration of the fallow on the upland, and ii) the availability of financial resources represented, in the absence of non-farming income, by the ratio of oil palm land to total farming land, 2) the production goal expressed by family size indicating the minimum rice requirement per household and 3) the farm production intensification level represented by the type and quantity of inputs used to ensure rice production. Here, this means the seed density and the total quantity of labour used. Concretely, the farm household typology was constructed by using two multivariate statistical techniques, respectively Principal Component Analysis (PCA) and Cluster Analysis (CA). The purpose of PCA is to transform linearly an original set of variables, as mentioned in table 11, into a substantially smaller set of uncorrelated variables that represents most of the information in the original set (Bidogeza J. C., Berentsen P. B. M., 2009).The number of factors to be retained, according to Kaiser’s criterion (Mokuwa A. et al., 2014), should present an eigenvalues greater than 1. A small set of uncorrelated variables is much easier to understand and use in cluster analysis than a larger set of correlated variables (Jolliffe I. T., 1986). Next, factors retained from the PCA, were used in cluster analysis. Cluster analysis seeks to typify entities (farm households) according to their (dis) similarity in terms of their attributes represented by the variables chosen (Aldenderfer M. S. et al., 1984; Alfenderfer M. et al., 1984; Everitt B., 1993). Entities within a certain group (cluster or categories) should be very similar to each other and entities belonging to different classes should be very dissimilar (Bidogeza J. C., Berentsen P. B. M., 2009). 74 Tableau 11: Variables for farms clustering according to structural factors, factors of production and factors of consumption. These variables concern 81 farms survived in the Bombali region. Explanatory factors/variables 1- Factors of resource endowment Location potential Farm size (ha) Share of lowlands/upland (%) Duration of the fallow (years) Financial availability Share of oil palm (%) 2- Factors of production goal Family size (number) Rice consumption/family (t/family) Mean Std. deviation 1,71 43 6 1,03 34 4 19 24 8 0,14 4 0,11 0,03 105 0,03 44 3- Factors of production intensification level Seed density (t/ha) Quantity of labour (days/ha) These variables were used in phase 3 to explain the area of rice and oil palm farmed per ecosystem, the rice target production in each household, and thus the rice yield performance and efficiency per farm. The yield efficiency will be expressed by calculating for each category of farms the labour use efficiency (rice yield divided by the total labour used by hectare) and the seeding density efficiency (rice yield divided by the seeding density used by hectare). 3. 3.1. Results Stage 1: “target production vs. consumption” relationship. High correlations (R² = 0.64) between farm-level rice production and overall consumption per family. This strong correlation thus confirms that these households produce primarily for own consumption, thereby justifying the subsequent stages of the analysis mentioned in the analytical framework. On average, every tonne of rice demand for consumption purposes must be matched by an average production of 2t. The quantity of rice corresponding to the difference between rice needs and production must essentially serve as rice seed. However, this correlation increases if we consider households which only rice farm (R²= 0.85) 75 compared to those farming rice and oil palm (R²= 0.68). For one tonne of rice demand, the difference between production and consumption is 2.5 and 1.9 t/ha respectively for mixed farms (rice and oil palm) and whole-rice farms. The difference in rice production vs. consummation in the case of mixed farms should serve as rice seed and therefore intensify the rice production or increasing the rice area by exploring new lands. Figure 28: Correlation between farm-level rice production and overall consumption per family. This correlation is shown in three ways: y1: farms growing oil palm, y2: all farms regardless of enterprise, and y3: farms growing no oil palm. 1 Production by farm (t/farm) 2 3 Consumption per family (t/family) Without oil palm With oil palm 3.2. Stage 2: Farm typology: household categories and production strategy - Principal component analysis results The PCA analysis was undertaken on the 8 variables as shown in table 4and the 81 surveyed farms. 5 principal components, absolute values of the loading expressed as (%), with eigenvalues greater than 1 (Bidogeza J. C., Berentsen P. B. M., 2009) have been retained for cluster analysis (Table 12). These new variables explaining 89% of total variability in the original data set. 76 Tableau 12 : Absolute values of the loadings of the classification variables with respect to the 5 principal components Explanatory factors Loadings (%) PC1 PC2 PC3 PC4 PC5 Farm size Share of lowlands/upland 37 1 27 -56 2 11 15 17 8 0 Duration of the fallow 6 56 17 2 6 Share of oil palm Seed density 63 64 6 9 3 6 8 3 3 6 Quantity of labour 37 20 1 3 29 Factors of production goal Family size Rice consumption by family Eigenvalues 1 8 2.62 14 0 2.06 64 77 1.11 15 10 1.09 4 0 1.0 Cumulative explained variance (%) 33 59 73 81 89 Factors of resource endowment Factors of production intensification level Variables Location potential For each of the 5 principal components identified, it is possible to define each component according to the variables with which it is most strongly associated. To make it easier to identify relatively large loadings, correlations above 50% are in bold (Table 4). The first component (PC1), which explains 33% of variance, is positively correlated with share of oil palm and seed density. The second (PC2) component is almost as important as the first component, explaining 26% of the variance. This second component is mostly negatively related to the ratio of lowland/upland ecosystems and positively correlated to the duration of the fallow on the upland ecosystem. The last component (PC3), which explains 14% of the total variance, is mainly correlated to the family size and the total rice consumption per family. - Cluster analysis results First, the five components were analysed using Ward’s technique (Ward J. H., 1963). The dendogram, resulting from this technique, illustrated the sequence in which farm households were merged into the clusters and including four cutting lines. A key issue in generating such diagrams is where to “cut” the tree in order to arrive at an appropriate number of clusters which best fit the data set. The number of retained clusters must be realistic with respect to empirical situation in order to be accepted as meaningful classification. Following that line of 77 reasoning, 4 categories based on the partitioning method were defined as appropriate as this seem to be most representative of farm household within the Bombali region (data not shown). These categories, based on the data presented in Table 13 are described qualitatively as follows: - Category 1 (medium yield (0.4t/ha), large farming area (3ha) and thus large target production (1.8t/farm)). This category has a high seed density (0.05 t/ha), labour quantity (145d/ha) and share of oil palm (45%). It is dominated by upland ecosystems (63%) with a relatively short fallow (6.7 years) compared to category 2. - Category 2 (low yield (0.2t/ha), medium farming area (1.4ha) and thus low target production (0.3 t/farm)). This category is characterised by a very low seed density (0.02 t/ha) labour quantity (99 d/ha) and share of oil palm (8%). It is dominated by upland ecosystems (74%) with a longer upland fallow period (8.4 years) than the first category. - Category 3 (low yield (0.3 t/ha), small farming area (0.7 ha) and thus low target production (0.2 t/farm)). The farms in this category have the lowest seed density (0.02 t/ha) and labour quantity (59 d/ha) of all the categories. These farms are dominated by lowland ecosystems (91%) with a highly marginal presence of oil palm (3%). - Category 4 (high yield (0.6 t/ha), medium farming area (1.3 ha) and thus medium target production (0.8 t/farm)). These farms have high levels of seed density (0.09 t/ha) and labour quantity (121d/ha). They are dominated by lowland ecosystems (87%) with a high presence of oil palm (33%). 78 Tableau 13 : Typical households according to rice production per farm and the determinants of farm structure and the inputs used based on PCA. (n= number of farms per category) Cat. 1 (n=20) Cat. 2 (n=43) Cat. 3 (n=11) Cat. 4 (n=7) Mean Std.deviation Mean Std.deviation Mean Std.deviation Mean Std.deviation Duration fallow (years) SAU (ha) Share lowlands (%) Rice yield (t/ha) Share oil palm (%) Seed density (t/ha) Total labour (d/ha) Rice production (t/farm) Size of family Rice consumption (t/family) 6.7 3.0 37 0.4 45 0.05 145 5 0.87 22 0.15 15 0.03 39 8.4 1.4 26 0.2 8 0.02 99 3 0.63 23 0.08 24 0.01 30 0.5 0.7 91 0.3 3 0.02 59 1 0.35 30 0.09 30 0.02 30 0.7 1.3 87 0.6 33 0.09 121 2 0.65 22 0.1 14 0.03 67 1.18 0.3 0.3 0.19 0.2 0.12 0.8 0.28 10.2 5 8.3 4 6.8 4 5.6 1 0.18 0.13 0.12 0.08 0.1 0.05 0.13 0.12 3.3. Stage 3: Analysis of the performance and efficiency of the farm categories Ecosystem impact on yield Proportionally to the dominance of the oil palm by class of farms, the more the farm area is dominated by the lowlands, the greater the aggregate rice yield per farm (Figure 29). As reported by several authors, this result is mainly due to higher soil fertility and less acute water stress at the end of the cycle in the lowlands than on the upland. The envelope curve (data not shown) proves that average farm yield increase by 0.0037 t/ha for each percentage point increase of lowlands at the expense of the upland ecosystem. However, this result becomes improper when the class of farms are compared without considering the share of oil palm. As an example, the class C1, for which the ecosystem upland presents 63% of the total area, has a higher rice yield (median of 0.375 t/ha) than the C3 (median of 0.27 t/ha) overlooked by the ecosystem lowland at 91%. Similarly, proportionally to the share of oil palm by class of farm, the greater the lowland area per farm in relation to upland area, the lower the quantities of labour used (figure 29). This reflects higher levels of labour demand on the upland compared to the lowlands (e.g. an average total labour of 0.4t/ha and 0.15t/ha respectively for exclusively upland or lowland farming, data not shown). However, the seeding density seems to be more related to the share of oil palm than the type of ecosystem (figure 29). 79 Yield (t/ha) 0,8 C2 C3 C1 C4 0,6 0,4 0,2 Seeding density (t/ha) 0 0,5 0,3 0,2 0,0 Labour (day/ha) 250 200 150 100 50 0 C2 high share of upland C3 high share of lowland Low share of oil palm C1 high share of upland C4 high share of lowland High share of oil palm Figure 29: Variation of the median, maximum and minimum values, first quartile and third quartile for yield, seeding density and total labour for the four categories of farms (C1, C2, C3, C4). 80 - Impact of financial availability on yield (oil palm effect) Table 5 and figure 30 show that the highest rice yields are found in categories 1 (0.4 t/ha) and 4 (0.6 t/ha). In these two cases, oil palm accounts for an area of 33% and 45% in categories 4 and 1 respectively. This is combined with higher seed densities and labour quantities than those observed in categories 2 and 3 (figure 30). In fact, the presence of oil palm induces an important increase in seeding density, as well as, the total quantity of labour. Concerning the seeding density the average difference reaches 0.07 t/ha (78%) and 0.03 t/ha (60%) when the C4 (dominated by oil palm with high level of oil palm) is compared to C3 (dominated by low land with low level of oil palm) and C1 (dominated by upland with high level of oil palm) is compared to C2 (dominated by low land with high level of oil palm) respectively. Concerning the amount of total labour this difference attempt 62 d/ha (31%) and 56 d/ha (51%) respectively when C4 is compared to C3 and C1 to C2. It should also be noted that, respectively to the dominance of ecosystem by farm, the presence of oil palm has caused an increase in rice-farming land. The average difference is equal to 0.4ha/farm and 1.6 ha/farm when the C4 is compared to C3 and the C1 is compared to C2 respectively. In addition, irrespectively to the presence of the oil palm, the farms dominated by the ecosystem lowland are smaller (C3 and C4) in comparison to those dominated by the ecosystem lowland (C1 and C2). In this type of subsistence household, this result proves the means of increasing production is always accompanied by an increase in farmed area land by exploring mainly upland ecosystem. In the absence of other stresses (envelope curve), each additional percentage point of oil palm at the expense of rice generates an increase in inputs for rice of 0.0028 t/ha and 1.54d/ha for seed density and labour quantity respectively used on the farm. Furthermore, the area of oil palm observed doesn’t exceeds 80% of the total area devoted to rice, otherwise, the rice production could be insufficient to satisfy household rice needs (data not shown). 81 Figure 30: The total labour and seeding density for the surveyed farms (n= 81) according to the farm size. The solid bar indicates the standard deviation of the average value for each category of farm. Impact of fallow on yield The variation in average yield per farm shows that the longer the fallow period in a upland ecosystem, the lower the rice yield per farm. In fact, the medians for C1 (average fallow duration of 6.7 years) and C2 (average fallow duration of 8.7 years) are respectively of 0.37 and 0.23 t/ha (Figure 31). This result – relatively surprising in light of the literature which indicates higher rice yields for long fallow periods in comparison to short fallow periods (Langyintuo A. S. et al., 2005) – can essentially be explained by a lower demand for rice for consumption purposes on farms dominated by long fallow periods (more than 8 years). As shown in Figure 31 this is reflected by lower seed densities (medians of 0.06 t/ha and 0.017 t/ha respectively for the fallow duration of 6.7 and 8.4 years) and labour levels (medians of 138 and 90 respectively for the fallow duration of 6.7 and 8.4 years) employed (i.e. a lower level of intensification) during long fallow periods in comparison to shorted periods. 82 0,06 Consumption by menber (t) 0,08 0,04 0,04 0,02 0 0 250 12 Fallow duration (year) Total labour (day/ha) Seeding density (t/ha) 0,12 200 150 100 8 4 50 0 0 C1 (duration fallow of 6.7 years) C2 (duration fallow of 8.4 years) C1 C2 (duration fallow of 6.7 years) (duration fallow of 8.4 years) Figure 31 : box plots for fallow duration, rice consumption by member of family, seeding density and labour amount for the categories C1 and C2 dominated by the upland ecosystem. Impact of factors of production on yield The comparison between the performance of rice (yield) with the efficiency of use of inputs (seed, labour) gives hypothetically rise to 4 behaviours as shown in Figure 32. Concerning labour, C4 is the most performing and efficient category, while C2 is the less performing and efficient. C1 and C3 are respectively performing but not efficient and not performing but efficient. This statement is different for the seeding density where only 2 behaviours were observed; performing but not efficient farms (C1 and C4) and not performing but efficient farms (C2 and C1). To understand these results, Figure 32 represents the variation of labour and seeding efficiencies with the share of low land area by farm. This figure shows that categories C3 and C4 are more efficient in terms of total labour and that this efficiency is more marked in lowland ecosystems than in upland systems. This result is different for seed density, which appears to be unaffected by the type of ecosystem. In contrast, less intensive systems in terms of seed density (C2 and C3) appear to be more efficient than those displaying a high seed density (C1 and C4). In the case of primarily lowland farms (C3 and C4), this difference can be explained by restrictive labour in the case of high densities. As shown in Figure 30, the 83 difference in primarily upland-based farms (C1 and C2) can be explained by a combined effect of the fallow period and labour. It is true that the greater the see density, the greater the labour requirements and the shorter the fallow period thereby giving rise to a fall in the fertility of the upland and thus the efficiency of the system. High yield, low efficiency High yield, high efficiency High yield, low efficiency High yield, high efficiency C4 C4 C1 C1 C3 C2 Low yield, high efficiency C2 Low yield, low efficiency Low yield, low efficiency C3 Low yield, high efficiency Figure 32 : Average and standard deviation of rice yield per farm (n = 81 farms) according to seed density and total labour efficiencies. The hatched lines indicate hypothetical limits for different zones of rice efficiencies (yield per seeding density and yield per the amount of labour) and rice yield. 4. Discussion - General farm characteristics Varying between 0.06 and 0.74 t/ha, the rice yield per farm in Bombali district is very low compared to the global average of 1.43t/ha in West Africa (Sammeth F., Lakoh A., Baudouin M., Hites G., Gomez y Paloma S., 2010). In addition to an absence of mechanisation, irrigation and fertiliser, this low production level can primarily be traced to low seed densities (0.02t/ha) and labour volumes (118d/ha), which are nevertheless comparable to the average observed in West Africa (e.g. 100d/ha for labour, Fontan, 2008). However, such low yields demonstrate a certain variability, for the most resulting from differing levels of intensification. This is compounded by different combinations of ecosystem (upland, lowlands) on each farm. On average, lowland ecosystems use fewer inputs than upland ecosystems but produce a higher yield per hectare. This difference can be between 14% and 31% in terms of seed density and rice yield respectively for lowland ecosystems compared to upland systems. The desired production levels are dictated by household rice demand. Increasing this demand will force rice farmers to shorten the fallow period (from 8 to 6 years 84 on average, Figure 33) and to farm more oil palm to ensure better intensification. In this case, oil palm can account for up to 66% of farm area. The results obtained in this study agree with those obtained by Tittonell P. (2013), showing 4 types of behaviour in West African households when the performance of the production system (well-being, income, production, etc.) per household is represented according to the resources available (human, natural, economic). Building on the concept defined by Tittonell P. (2013) we replaced the resources available by the efficiency of the production systems (Figure 33). Depending on the efficiency of use of production resources, these 4 households reflect different production strategies: i- Low-performance, inefficient households with few financial resources (point A, category 2 in our typology): these are traditional household which used to be very widespread (Jalloh A., 2006), in relatively infertile ecosystem, farming only rice on upland with a relatively short fallow period, and no cash crops (Saito K., Azoma K., Sokei Y., 2010), and generating very little rice production per family member (e.g. 0.04 t/ha in our application). ii- Low-performance but efficient households with few resources (point B, category 3 in our typology): these are also traditional households although highly marginal in fertile ecosystems (Saito K., Azoma K., Sokei Y., 2010), primarily in the lowlands, farming only rice and with very few financial resources. This type of household generates very little rice production per household member (0.03 t/person in our application). The high efficiency of this group derives from the initial fertility of the lowlands and the limited need for factors of production compared to the upland. iii- High-performance households with many resources and good efficiency (point D, category 4 in our study): this type of household is also marginal in West Africa. The availability of financial resources (due to cash crop) and the initial fertility of the ecosystems comprising this type of farm enable these households to combine performance and efficiency. This type of household generates a very low level of rice production per household member (0.12 t/person in our application). iv- High-performance households with many resources but low efficiency (point C, category 1 in our application): these are households in ecosystems with a poor initial fertility. According to Ceesay M. M. (2004), it is the most widespread household model and involves mixed ecosystems (upland, lowlands) dominated by low-fertility upland resulting from a short fallow period. This production model has gradually been replaced (for example after the end of the war in Sierra Leone) the single-ecosystem farms. According to the works of Dingkuhn M. et al. (2006), these farms are increasingly large, but primarily for the upland share of the farm. 85 According to, Ceesay M. M. (2004) the presence of financial resources (shown in our application by the presence of oil palm) leads to an improvement in the yield (due to intensified production) and overall production per farm, but not in production efficiency. This type of household generates a high level of rice production per household member (0.14 t/person in our application). They are likely to step out of farming and have more off-farm activities. In short, this analysis reveals three important levers to increase rice production in West Africa. First, the region is faced with limited land resources, particularly fertile land (Jalloh A., 2006), and limited financial resources for intensification. Second, the plantation of cash crops or non-farming income leads to intensification. In figure 33, this concerns the shift from point A to D and B to C. As noted by Yemadje R. H. et al. (2012), this raises the issue of the area to be occupied by the cash crop without compromising rice production as well as the pressure exerted on the upland which continues to shorten the fallow periods and thus reduce rice yield (Becker M. et al., 2001b). Third, farming highly fertile ecosystems (lowland in our case) shown by a shift from B to A or from C to D. This strategy is observed in several West African countries such as Cameroon, Burkina Faso, Guinea, Benin and Sierra Leone, where political powers encourage the installation of rice in the lowlands (Rafflegeau S., 2008; Tanaka A. et al., 2013). However, the most common strategy adopted by rice farmers at present to increase rice production is to explore the upland (Kent R. et al., 2001; Rodenburg J., J. Zwart S., Kiepe P., Narteh L. T., Dogbe W., Wopereis M. C. S., 2013). The lowlands have three major disadvantages: the difficulty in working the land (Kayombo B. et al., 1993); the toxicity, in places, caused by iron (Dixon C. A., Johnson S. D., Fomba S., 2001) and the inability to produce vegetables and rice in this type of ecosystem (Erenstein O. et al., 2006). Thus there is a need to combine to upland rice with cultivation of cash crop to be able to invest in endowment of the farming system (C1 in our study). 86 Figure 33 : Efficiency of total labour (rice yield to total labour) and seed density (rice yield to seed density) according to the share of lowland rice on the farm. C1, C2, C3, C4 represent the averages per farm category. Broader application The analysis framework developed in this study can be adapted to other regions in West Africa and for other subsistence crops. This framework enables to consider not only biophysical aspect but also some household strategy aspect, combination important to understand better decision making process of farmers. However, other more intensive and more productive systems, other factors such as biotic and abiotic stresses, strongly present in the rice-growing system, must be taken into consideration to explain the performance and efficiency of the rice-growing systems as indicated by Tsujimoto Y. et al. (2013). Even though they are present in the rice-growing system in Sierra Leone, these two factors do not significantly reduce the already low yields of the system which are severely limited by soil fertility and low input levels (Smaling E. M. A. et al., 2006). 5. Conclusion The aim of this work was to better understand the performance and efficiency of rice-growing households in Sierra Leone. This study showed that rice yield is both low and variable. Production efficiency is variable with regard to the production factors of seed quantity and total labour according to the type of ecosystem, the presence of a cash crop and the duration of the fallow period on the upland. This analysis also showed that intensification is followed by an increase in production due to enhanced yield and the increased area planted with rice. 87 This improvement is nevertheless accompanied by a fall in efficiency, primarily due to the quantity of labour available. This analysis proved also that rice households are highly vulnerable, and therefore poorly resilient to face disruptions (socio-economic, climatic…). With often very little intensification means, many households may, as a result of a shock affecting directly or indirectly seed storage or the availability of family labor ( illness, death, old age ... ), switching from a high productive position (points D or C in Figure 34 ) to a low productive position (point B or A at Figure 34) . 88 Household performance (well-being, expected production, Ecosystem effect: income, self - sufficiency…) (natural fertility) C Stepping up High Asset D A B Hanging in Low Asset Resource use efficiency Ecosystem effect: (natural fertility) Figure 34: Levels of performance and efficiency in a traditional rice-growing household in West Africa (adapted from Tittonell, 2013). To this is added a very rapid degradation of soil fertility due to the absence of fertilization and the limited availability of new fertile farming lands. In this context, it would be appropriate to rethink agricultural policy by initiating incentive measures for diversifying the current farming systems by cultivating, in addition to rice, more cash crops such as cassava and trees (cacao, coffee). This strategy may improve the resilience of the current rice households by reducing the risk around the rapid degradation of the soil fertility (by making fertilization), as well as, the availability of labor and seeds. 89 Chapter 4 A household model to assess consumption-production-resources nexus in West Africa: The rice based farming systems in Sierra Leone. Chenoune, R.(a,b), Allen, T.(d), Gomez y Paloma, S.(e), Yigezu, Y. (f), Flichman, G.(a), Capillon, A (b)., Belhouchette, H.(a,c) (a) IAMM-CIHEAM, 3191 Route de Mende, 34093 Montpellier. Cedex 5, France. Montpellier SupAgro, UMR System, 2 Place Viala, 34060 Montpellier, France. (c) IAMM- UMR System, 2 Place Viala, 34060 Montpellier, France. (d) Bioversity International, Parc Scientifique Agropolis II, F-34397 Montpellier cedex 5, France. (e) European Commission, Joint Research Centre (JRC), Institute for Prospective Technological Studies (IPTS), Edificio Expo. c/ Agriculture and Life Sciences in the Economy Unit (Agrilife), Edificio Expo. C/ Inca Garcilaso 3, 41092 Seville, Spain (f) International Center for Agricultural Research in the Dry Areas (ICARDA). (b) Abstract Most programmes initiated by national and international authorities in order to boost agricultural production and ensure food security for developing countries are often assessed qualitatively, and on large scales that ignore the diversity of households across a territory. This paper presents a household model which simulates strategies of production and of consumption preferences for rice farming households, with non-separability between production, consumption and available resources, and taking into account the risks linked to production and price variability. This model, applied to 4 households representative of rice farming households in the north of Sierra Leone, has made it possible to assess the effects of three policy scenarios seeking to improve rice consumption as well as the total vegetable calorie assessment, by calculating production, consumption and socio-economic indicators. This study highlights the fact that extensive lowland-based households reveal the lowest rates of rice consumption (0.04 t/an) and caloric intake (963 Kcal/day) in comparison with other household types. The scenario that considers subsidizing rice plantations in the lowlands compared to the baseline scenario (business as usual) is the only one to have led to a significant improvement in rice consumption for all households, i.e. an increase between 16 and 35% depending on the different household types. Besides, regardless of the scenario, the variation in terms of calories is less significant than that of consumption, thus expressing relatively balanced consumption preferences between the different products (vegetable, 90 cereal, oil, sweet potato). The improvement in rice consumption is mostly due to an increase in self-consumption, which itself is linked to a rise in rice production. Depending on the different types of scenario, the increase in rice production is due to an increase in the surface of upland rice, to its intensification, to an increase in cash or a specialization in lowland rice combined with a cash crop. This study, applied to rice cropping systems, could stretch to other production systems, other socio-economic or technical scenarios, but it could also concern other developing countries. Key words: self-consumption, consumption preferences, household model, risks, rice cropping systems, policy scenarios, indicators. 1. Introduction Food security is a very important issue for national and international authorities (Soule, 2012; Spiertz, 2012). Ensuring sufficient and accessible food supply that is of good nutritional quality in an uncertain political, socio-economic and climatic context is a constant challenge for most countries, including those of West Africa (Katic P.G et al., 2013). West African countries are essentially confronted with very low means of investment, population growth that is among the highest in the world (OECD/SWAC, 2013) and insufficient and unstable agricultural production (Spiertz, 2012). Several programmes have been established in order to boost rice production in those countries (Abdelrasoul et al., 2013; NASDP, 2009). The main actions carried out to boost this agriculture involve setting up innovation platforms, or subsidizing the use of inputs and mechanization (Fenske, 2011). However, these actions are often carried out at a national level without real consideration for the highly contrasting needs of agricultural households across a territory. This also implies the use of more specific approaches and methods that can quantify agricultural production while taking into account the diversity of cropping systems and of household food needs (Herrero et al., 2014). Political decision-makers and international organizations are constantly looking for alternative solutions in order to increase the production of rice, a main staple food for West African populations (CAADP, 2010). According to Chenoune et al., (2014) and Herrero et al., (2014), in such a poor context, alternative solutions essentially come down to cash crop plantations (cassava, oil palm, vegetable, etc.) which generate cash in order to intensify rice production, to keep lowlands – a relatively fertile ecosystem compared to others – for rice plantations, or 91 to extend agricultural land to new initially very fertile forestland. In a context where households mainly produce for self-consumption, these solutions must first be assessed on their ability to increase rice production without jeopardizing the production of other crops (vegetables, oil palm, tubers, etc.) which are essential to household food balance (Word food Programme et al., 2010). Household models based on linear programming are very useful tools for such an assessment (Husin, 2012; Louhichi and Gomez y Paloma, 2014; Seyoum, 1998). Utility-maximizing theory is the most common approach when decisions concerning consumption and production are interdependent and non separable, as is the case with rice crops in West Africa (Deybe, 1998; Louhichi and Gomez y Paloma, 2014). According to this theory, farm households are assumed to maximize the utility derived from the consumption of all available commodities (i.e. home produced goods, market-purchased goods…), subject to full income constraints (Iversen, 2002). Several applications use this type of approach, but their modelling assumptions often only take into account part of the West African context; for example: i) considering consumption as a constraint by reducing the utility of the household to income generated by the sale of non-consumed products (Herrero et al., 2014). It makes it possible to identify different production strategies but with consumption levels determined ad-hoc, without a real resources-production-consumption relationship, ii) simultaneously maximizing several goals (income, consumption, preservation/degradation of natural resources) by giving a weight for each utility component. This type of modelling is used to determine the effects of the different levels of preference attributed to each utility component (consumption, income, preservation of resources), on the socio-economic and environmental sustainability of households (Saha, 1994). In such cases, consumption is also determined ad-hoc, with no real income-resources-consumption interaction, iii) expressing the essential household food consumption needs as a minimum income that households should guarantee (Hazel et Norton, 1986). This minimum income can be expressed as a way of considering household risk aversion by assuming that farm households exhibit ‘safety first’ behaviour, i.e. in every ‘state of nature’ household income should at least attain a minimum level (Low, 1974). In that case, consumption remains at a minimum level regardless of the production levels, iv) In major low income developing countries, market imperfections and low intensification levels respectively imply high market prices and production variabilities between years. Contrary to most wellknown household models which are linear optimization based, the utility should therefore be simulated by considering different risks faced by households (Louhichi et Gomez Y Paloma, 92 2014), and v), expressing household consumption behaviours as a constant or a free variable instead of being considered as a function of prices and incomes (Holden et al., 2004; Louhichi et al., 2013; Strauss, 1984). The aim of this paper is to present a household model based on mathematical programming, and that is specific to West Africa. This non-linear optimization model must highlight the production strategies of rice farming households with non-separability between production, consumption, and available resources. The model is applied to four rice farming households in the north of Sierra-Leone, which is known for its significant rice production, but also for its low and variable yields and consumption levels that are among the lowest in the world. It also defines and assesses political alternatives encouraging agricultural production, and therefore the improvement of food consumption. -From the conceptual model to the numerical model 2. The conceptual model Figure 35 shows the structure and strategies of production and consumption of rice farming households in West Africa. This structure consists of three main components: - Resources, which essentially depend on the availability i) of agricultural land. According to Chenoune et al. (2014) and (Mutoko et al., 2014), households often resort to the exploitation of new forestland in order to increase agricultural production. Without sufficient fertilization, the initial productivity of agricultural land mainly depends on the type of cultivated ecosystem (Holden and Otsuka, 2014), ii) of labour, which may be family, hired or communal labour, and which is split according to gender and to the tasks within the technical itinerary (sowing, planting, slash-and-burn, weeding, etc.) (Spencer et al., 2009), iii) the inputs, essentially represented by seeds in a context where the extent of mechanization is very low (Theriault and Tschirley, 2014). In an unstable and unorganised market, households mainly use self-produced rice seeds (Kargbo, 2002). According to several authors, (AfricaRice, 2010; Kargbo, 2002) the amount of rice from the previous yield that is stocked as seed greatly influences the sowing density, and therefore the production, of the current year. This situation shows rice farmers' real difficulty in deciding, each year, how much to keep for the year's consumption, and how much to stock as seed for the following year (Kargbo, 2002), and iv) of the cash for the purchase of inputs (fertilizer, seeds, irrigation, mechanization...) or consumption goods. In a 93 very poor context and with very little support from the State, investments as well as purchased food products essentially rely on off-farm income, or on farm income provided by cash crops such as oil palm, cotton, cocoa and cassava (Andersen and Gamanga, 2009; Theriault and Tschirley, 2014). However, this raises a question as to the space which these crops can occupy in terms of land resources and inputs without jeopardizing food crops (Yiridoe et al., 2006). - The agricultural household, which is divided into two closely linked sub-systems : i) the production system which consists of the activities observed at field level. There are three main crop categories: cereal – mainly represented in West Africa by rice or maize – as staple food, legumes that are rich in protein (bean, black-eyed pea...) and perennial crops such as palm trees containing lipid-rich oil. These crops are grown in ecosystems of different fertility (lowlands, uplands, mangrove, etc.) and according to more or less long fallow periods that depend on the intensification level, and ii) the household system which essentially consists of a family head along with men, women and children, all very involved in agricultural work at home, but also in other households in the village (Toulmin C., Guèye B., 2003). The dual role of many households in West African countries as producer and consumer raises the question of the interdependence between production, resource allocation (labour, land, seeds…) and consumption decisions. Firstly, agriculture is still characterized by self-consumption, partly because local markets may fail to satisfy demand or to provide outlets for production. Secondly, the opportunity cost of employing members of the family on the farm may be very low given the scarcity of land (Louhichi et al., 2013). The decision variables, which are numerous in such a system. They concern the level of expected production, which itself often depends on the choice of crop rotations that are affected by the type of ecosystem present, on the type of fallow and on the expected level of intensification (Chenoune et al., 2014). This arbitration will depend on the relative prices of inputs (including labour) and on the market price of the products in relation to their production cost, but also on the household's food consumption preferences. Consumer’s preferences are represented by a Rotterdam demand function (Theil, 1965; Barten, 1964), with estimated parameters retrieved from the literature. Production and consumption levels are thus computed simultaneously, resulting with estimates for stocks, seeds and sales. 94 Figure 35: Conceptual representation of the structure and material flows of the consumptionproduction-resources relationship of the rice farming households in West Africa. - The farm household model The household model developed in this study is based on a non-linear optimization in which the production, consumption and resources mobilised are interdependent and non-separable. It is a static annual model with a utility based on the full income approach, which includes both consumption and farm income (Strauss, 1984). However, even if the model is static, the crops are specified according to crop succession in order to take into account the effect of the previous crop on yield and on the evolution of soil fertility (Belhouchette et al., 2011). The mathematical structure of the model which is mainly based on the conceptual model described in figure 1 is as follows : 95 Where U is the value of the utility function, V is the matrix of agricultural products (pr) sold, i.e grain, oil, tuber, etc., Pv is the matrix of market agricultural product prices, A is the matrix of the amounts of products consumed (self-consumption), Pa are the shadow prices of agricultural products kept for self-consumption, Q is, for each activity, the total labour cost expressed by gender (mo) and period (pe), i.e. sowing, weeding and harvesting, Pl are the daily labour prices expressed by gender, C are the total input costs (other than labour) expressed by crop (c), ecosystem type (s), intensification level (t) and fallow duration (j), is the risk aversion coefficient and standard deviation of farm income by considering market price and yield variability in the study area. The Risk is treated for different “states of nature” in the model using a method that combines Freund's approach with the Target MOTAD method (Tauer, 1983). The maximization of the utility is considered by taking into account 4 essential constraints: i) resource constraints expressed by equation 2: Where ai,j are the requested resources for each activity (i) and resource (j) (land, labour, seeds…). This matrix is often established from a field survey or from expert knowledge; Xi is the area kept for each activity and bn the available resources at farm level. However, it is important to specify, in accordance with the conceptual representation in figure 1, that: - In many countries, the total available area kept for agricultural activities by households is often determined and known in advance. However, in West Africa the farmed area is closely related, in addition to resources available, to the fallow area. The longer the fallow duration, the more the farmed area is significant (Sakane N. et al., 2014). This constraint is specified as follows in the model: Where: X is the simulated area for each activity (i), and d the fallow duration which can vary from 1 to n years according to the type of crop and the ecosystem. For example, rice crop could be cultivated in an upland ecosystem with a fallow duration that can vary from 3 to 15 years. However, in lowlands the rice is often cultivated in a yearly rotation (Chenoune et al., 2014) 96 ii- the total production by activity can be partially or totally consumed, sold or stocked as seed. Equation 4 expresses this equality: Where: X is the simulated area by activity, Y is the yield by product (grain, tuber…) and activity, and ST, SD and SC are respectively the quantities stored, sold and self-consumed. iii- two more equalities must also be considered: i) the total consumption by product is equal to the total quantity of self-consumed and purchased products and, ii) the total quantity of seeds used by activity is equal to the total amount of seeds stored and purchased (equations not showed). iv- cash constraint states that the total value of inputs (including labour) and goods for consumption that a household purchases is constrained by its total cash income generated from production market sales plus off-farm incomes. v- In addition to the production resources and constraints a household food consumption component was also specified. A fifth constraint needs to be added to capture household food preferences, which might impact resource allocation and production decisions. The demand function introduced in the household model was first formulated by Theil (1965) and Barten (1964), and is known as the Rotterdam model. This approach is based on the total differential of the Marshallian demand functions and does not require the specification of any utility or cost function. It can be seen as a first order approximation of any demand system and one of the “flexible functional forms”. The total differential of the demand function is: The log-linearisation of the total differential equation1, and using the Slutsky equation, results in the seminal demand function (Theil, 1965): 1 By multiplying equation (5) and using the identity . 97 With the Divisia volume index: Where: share); is the budget share (weighting each demand equation by the expenditure the marginal propensity to consume good i; total expenditure. the Slutsky coefficients; and m is the . The expenditure elasticity of demand is equal to .. The different price elasticities of demand are: - Cournot price elasticities (un-compensated): - Slutsky price elasticities (compensated): - Frisch price elasticities: With as the inter-temporal substitution elasticity (inverse of the income flexibility) Demand theory imposes adding-up, homogeneity, symmetry and negativity restrictions. These restrictions are not independent. Adding-up, homogeneity, and symmetry are usually imposed in estimation, and the negative semi-definiteness of the matrix alone is empirically tested. Concretely the Rotterdam system model was transcribed in the household model as follows: 98 Where: are the buying market prices; share observed in the dataset; and is the full farm income; the initial expenditure the parameters of the demand function recovered from the econometric literature (See section 2.3). These two coefficients are treated as constants. This version of the Rotterdam model is thus linear in the parameters. Although the (variation of the logarithm of) full farm income is introduced, the model is written in terms of a conditional demand system, i.e. it only models demand for food depending on food prices and total food budget. Firstly, the optimized income does not include income from off-farm activities. Secondly, the demand equations are in firstdifference and logarithmic form. It thus assumed that the food budget increases proportionally with (the share of the total household income captured in) the farm full income. Elasticities estimated within such a subsystem are conditional on the goods included in that system. 3. Empirical application 3.1 The case study area Sierra Leone is considered as one of the poorest countries in the world by the United Nations, with an estimated average income of 1.25 US $/day and per person (United Nations Development Programme, 2012). In this precarious economic context, agriculture played a very important role by providing the equivalent of 48% of the total GDP and employing 75% of the labour force (Abdelrasoul et al., 2013; Johnson et al., 2011). Yet Sierra Leone remains a net importer for its food needs, including rice, that is to say 2.107 Mt in 2013 (United States Department of Agriculture, 2013). This paradox is mainly due to very low but also very variable production levels by rice farming households in comparison with other countries in West Africa (Sammeth et al., 2010). In order to maintain its self-sufficiency in rice, the State, along with international authorities and organizations, constantly seeks to boost rice production. Since 2005, the European Union has significantly contributed to the restructuring of the rice sector, by means of the STABEX scheme (stabilization of expert Earnings) (Perrault et al., 2013). The projects within that scheme have mainly served to improve rice yield via specific financial and technical measures (Gomez y Paloma et al., 2012). Three major measures of this plan to boost rice production have been chosen to be assessed by the household model developed in this application and will be described in paragraph 4. 99 3.2 Choice of representative rice farming households The study zone concerns the Bombali district which occupies 40% of the total area of northern Sierra Leone (SSL, 2004) and which provides about 70% of national rice production (Saravia Matus et al., 2011). More than 90% of rice farming households in the district essentially produce rice for self-consumption and have no other off-farm income (Spencer et al., 2009). Arable lands mainly consist of uplands (60%) and lowlands (20%), the rest consisting of very marginal ecosystems where very little is grown (NASDP, 2009). Contrary to lowlands where rice farmers mainly practice rice monoculture over several years, upland rice is often grown alternately with very variable fallow periods (from 3 to 25 years), and more and more in association with oil palm, a very profitable crop (Jalloh, 2006). In this district, rice is grown without input (fertilization, irrigation) nor mechanization (BAD, 2010; Bumb et al., 2011). This characteristic makes this activity very vulnerable to labour availability mainly, and to the initial fertility of soils (CRISTO, 2010; Emmanuel, 2005; Harding et al., 2012; Melendez, 2006). Furthermore, the non-availability of rice seeds could also explain low and variable yields (AfricaRice, 2010; Deen and Ngaujah, 2010). The household model was applied to four rice farming households representing the diversity observed at the Bombali district level. These four households were selected by Chenoune et al. (2014) on the basis of a sample of 181 farms whose data was collected between 2009 and 2010 by the Seville Joint Research Centre (JRC) (Gomez y Paloma et al., 2012; Louhichi and Gomez y Paloma, 2014). In detail the collected data concern: i) socio-economic characteristics of the household: family size and composition per household (women, men, children), age of its members and sources of income (farm income, off-farm income), ii) household agricultural production and consumption: each interviewed farmer gave the measured value of each crop production by ecosystem. For rice, it corresponds to the quantity harvested, after drying and winnowing. For each crop, yield was calculated (total production per hectare) instead of being obtained directly from surveys in the field. Furthermore, the share of total production dedicated to food consumption, by household and product, was also estimated by the farmer at the interview. This data was used as guesstimates of total on-farm food or self-consumption. Given the lack of proper consumption data, total food consumption figures were restated on the basis of FAO 2009 Food Balance Sheet (FBS) data for Sierra Leone. Based on secondary information and 100 assumptions, adjustments were made to the national average to reflect the dietary habits and preferences of the study area. Observed deviations from self-consumption average (after conversion into kcal and per capita) were then applied to the local adjusted average to retrieve approximations of total food consumption per household, iii) structural data to describe current cropping systems: usable agricultural area, land use (crops grown), working time per agricultural task, proportion of each task performed by men and women, proportion of hired labour by task, iv) and household farming activities: The aim was to describe, in detail for each household, the type of dominant crops by ecosystem (upland, lowland), the crop cycles of rice and oil palm by ecosystem, the main tasks involved in growing each crop as well as the breakdown of labour (for men and women) by type of task, length of fallow period by ecosystem and the seeding density for rice. Using that data base, Chenoune et al., (2014) selected 4 household farms by considering three types of criterion: 1- the resource endowment represented by: i) the potential of the location expressed by the ratio of lowland area to upland area and the duration of the fallow in the uplands, and ii) the availability of financial resources represented, in the absence of nonfarming income, by the ratio of oil palm land to total farming land, 2) the production goal expressed by family size indicating the minimum rice requirement – as main staple food – per household and 3) the farm production intensification level represented by the type and quantity of inputs used to ensure rice production. This involves seed density and the total quantity of labour used. Concretely, the farm household typology was constructed by using two multivariate statistical techniques, respectively Principal Component Analysis (PCA) and Cluster Analysis (CA) (For more details see Chenoune et al., 2014). These categories express different production strategies that could be divided into two classes (Table 14): Class 1: high rice consumption households. This concerns the Uplandintensive and Lowlandintensive farm types for which rice consumption per capita is higher than 0.075 t/capita/year and the total calories consumed per day and capita are higher than 1300 calories/capita/day. However, those two rice households showed two different production strategies: i) Uplandintensive includes high-performance households (rice yield of 0.4t/ha) with many resources but with poor initial soil fertility (share of upland of 18%) resulting from a short fallow period (average fallow duration of 6.7 years). The presence of financial resources guaranteed by the presence of oil palm (36% of total farm area) leads to an improvement in 101 yield (due to intensified production) and overall production per farm, and ii) Lowlandintensive composed of high-performance households (rice yield of 0.6 t/ha) with many resources (seeds and labour). The availability of financial resources (due to oil palm cash crops which represent 26% of total farm land) and the initial soil fertility of the ecosystems, ensured by the presence of the lowland ecosystem (44%), including this type of farm, make it possible for these households to combine performance and efficiency. - Class 2: low rice consumption households. This class regroups Uplandextensive and Lowlandextensive farm types for which the average rice consumption per capita and year is less than 0.057 t/capita/year and the total consumed calories are 900 and 843 respectively for categories 3 and 4. Here also, those categories revealed two different production strategies : i) Uplandextensive is composed of low-performance (rice yield of 0.2 t/ha), inefficient households with few financial resources. These are traditional households which used to be very widespread (Jalloh et al., 2012; Saito and Futakuchi, 2009), in relatively infertile ecosystems (uplands represent 89%), growing only rice in the uplands with a relatively long fallow period (8.4 years), and no cash crops (oil palm represents only 5%), and Lowlandextensive farm types regrouping low-performance but efficient households with few resources. These are also traditional households although highly marginal in fertile ecosystems (Saito and Futakuchi, 2009), primarily in the lowlands, farming essentially rice (oil palm represents only 4%) and with very few financial resources. The high efficiency of this group derives from the initial fertility of the lowlands and the limited need for factors of production compared to the uplands. 102 Table 14 : farm types by considering structural, production and consumption criteria. The table is adapted from (Chenoune et al., 2014) Class 1 : high rice consumption household uplandintensive (n=20) lowlandintensive (n=7) Class 2: low rice consumption household uplandextensive (n=43) lowlandextensive (n=11) Mean Std.deviation Mean Std.deviation Mean Std.deviation Mean Std.deviation land use (ha) Upland rice 1.04 0.42 0.17 0.31 0.88 0.39 0.04 0.12 Lowland rice 0.65 0.43 0.84 0.30 0.36 0.37 0.66 0.37 Oil palm 1.32 0.67 0.49 0.43 0.17 0.34 0.04 0.12 Cassava 0.35 0.54 0.26 0.38 0.20 0.46 0.13 0.28 Beans 0.21 0.60 0.11 0.26 0.08 0.20 0.00 0.00 Sweet potato 0.02 0.11 0.08 0.30 0.06 0.22 Total farm area (ha) 3.7 0.87 1.92 0.65 3.14 0.63 0.94 0.35 Duration fallow (years) 6.7 5 0.7 2 8.4 3 0.5 1 Share lowlands (%) 18 22 44 22 11 23 90 30 Share oil palm (%) 36 15 26 14 5 24 4 30 Rice seed density (t/ha) 0.05 0.03 0.09 0.03 0.02 0.01 0.02 0.02 Rice yield (t/ha) 0.4 0.15 0.6 0.1 0.2 0.08 0.3 0.09 Rice production (t/farm) 1.18 0.3 0.8 0.28 0.3 0.19 0.2 0.12 Rice total labour (d/ha) 145 39 121 67 99 30 59 30 Size of family (member) 10.2 5 5.6 1 8.3 4 6.8 4 Rice consumption (t/capita/year) 0.075 0.013 0.094 0.012 0.057 0.008 0.054 0.005 Total calorie /cap/day (kcal) 1338 180 1739 850 900 169 843 695 3.3 Farm type data specification In order to run the household model, three types of data were specified for each farm type: - Farm resource data such as available farmland per ecosystem and total labour sorted by gender. - Input-output coefficients needed to describe the associated total inputs and outputs by activity (association of crop, ecosystem type, intensification level characterized by the combined levels of seeds and labour use, and fallow duration). In total 5 crops (rice, legume (groundnuts, bean, etc.), cassava, potatoes, fallow), 2 ecosystems (lowlands, uplands) and 3 fallow duration types (1 year, short and long) were specified and divided into 18 activities with the following inputs data: the total amounts of labour by gender and period (sowing, weeding and harvesting) and seeding density. In fact, for rice three seeding density by ecosystem were specified. This information is also expressed as costs by specifying variable market average price for each input. Other costs data (nursery, planting,…) were also collected and specified by activity. 103 Additional input data such as observed land use by activity and farm type, the amounts of consumed goods, were also collected and used to assess the model's ability to simulate the observed data correctly. The output variables mainly described the activity yield (average and variability). For the calculation of risk, the activity yield variability was estimated based on local expert interviews (for the years 2009-2010 and 2011) and the price variability by goods were calculated based on an FAO data base for the years 2008-2012 (FAOSTAT, 2012). - Food consumption parameters: these parameters were hierarchically recovered from econometric literature. Estimates from the USDA International Food Consumption Patterns report (Muhammad et al., 2011), from the 2005 International Comparison Program (ICP) data, were used. This dataset contains estimates of total and marginal budget shares and income and price elasticities for nine broad consumption groups and eight food subgroups (Bread and cereals, Meat, Fish, Dairy products, Fruits and vegetables, Oils and fats, Beverages and tobacco, Other food products) across 144 countries, including Sierra Leone. Firstly, the marginal propensities to consume were recovered from the reported Sierra Leone national expenditure elasticities using the observed estimates for budget shares and the elasticities’ formula. Secondly, approximations of the Slutsky (compensated) price elasticities of demand were found from the reported Frish elasticities2 building on the theoretical relationships between Slutsky elasticities, expenditure elasticities and the inter-temporal substitution elasticity (also reported in the USDA data). Finally, the Slutsky coefficients, parameter of the demand function, were sorted out using the usual equation for the Slutsky elasticities. These elasticities can be understood as conditional elasticities3 and may differ from the unconditional elasticities estimated within a system including all goods4. 2 Please note that the Slutsky (and Cournot) price elasticities were not reported for the food subcategories in the USDA report. Given that the price parameter is constant and invariable across countries, the estimated own-price Slutsky (and Cournot) elasticities would increase as the share of the budget (dedicated to that good) decreased. For goods such as cereals, it would lead to results contrary to theory, where wealthier countries would have larger own-price elasticities. 3 Please note that the USDA report estimated elasticities through a two-stage demand system, thus computing conditional and unconditional elasticities. 4 Please note that in the USDA report, the unconditional and conditional Frisch own-price elasticities are equal. 104 4. Model calibration The calibration of the model was achieved by adjusting the risk aversion coefficient to fit the observed data. Hazell and Norton (1986) suggested that a reasonable range of the risk aversion parameter lies between 0 and 1.65. The challenge was to identify a single riskoptimal farm plan for each household situation for further analysis. Ideally, the identification method for the risk-optimal household plan for a sampled farmer would have been to elicit his or her utility function and determine the tangency point between his function’s expected utility curve and the relevant risk-efficient frontier (Elamin and Rogers., 1992). An alternative identification method for the risk optimal farm plan was to present the farmer with all plans contained in the risk-efficient frontier and let the farmer choose the preferred plan. However, both criteria seemed irrelevant for farmers in rural Sierra-Leone with high rates of illiteracy and ignorance of income probabilities and their associated distributions. Consequently, the procedure used in this study was considered the most appropriate for the risk aversion coefficient used for each type of farm by varying the risk aversion coefficient with an interval of 0.1 and within a range of 0 to 1.65. Adjustment stopped when further modification of aversion coefficient parameter generated little or no improvement on the basis of the relative difference between observed and simulated crop patterns (Holt C. A. et al., 2014). By calibrating the household model, 4 additional variables/indicators (other than activity use) were then compared to assess the model's ability to simulate observed data: farm production by activity, total labour by activity, quantity consumed by product, capita and day and total calories consumed by product, capita and by day. 5. Scenario and indicator specification Once the model was calibrated, it was used to simulate three policy scenarios. Those scenarios were established based on the recommendations of the two main initiatives of the STABEX projects(SLIEPA, 2012) and the later Strategy of Agricultural of Sierra Leone (CAADP, 2010): - Subsidizing the purchase of rice seeds scenario (Sseeds): Several initiatives and programmes have tried in the past – but also recently – to subsidize the purchase and free distribution of rice seeds to households in West Africa (CARD, 2009). This policy is a major measure of the new agricultural strategy in Sierra Leone (Perrault et al., 2013a; The Government of Sierra 105 Leone, 2013). It aims to subsidize the distribution of rice seeds partially or completely in order to boost its production, and therefore its consumption. This scenario (Sseeds) seeks to assess the impact of a gradual covering of rice seed purchase cost per hectare, varying its value from 0% to 100%, with a 20% pitch. - Subsidizing the plantation of oil palm scenario (Sop): oil palm cash crops must enable households to have more cash. This cash must partly serve to intensify rice production by buying rice seeds in the market, and therefore to be less dependent on seeds stocked from the previous yield (Chenoune et al., 2014). For this scenario (Sop), several levels of subsidization are also applied to cover the cost of oil palm plantations per hectare gradually, varying from 0% to 100% (with a 20% pitch). - Subsidizing rice plantation in lowland ecosystems scenario (Srice): according to several observers, the future of rice production in West Africa lies in the lowlands, more fertile than the uplands (Srivastava et al., 2012). Nowadays non-cultivated lowland areas are estimated at 85% of available land (Bah, 2013). For that reason, several initiatives have planned to subsidize rice production in the lowlands (Bah, 2013). There again, several levels of subsidization must be tested, in relation to the gradual covering (from 0 to 100% with a 20% pitch) of the cost per hectare of lowland ecosystem conversion to rice production. This plantation cost was estimated at an average of 5926740 Leone/ha in 2013 (entertains avec agricultures et les acteurs du minsteres; 2009). The initial conversion cost – which makes it possible to exploit lowlands for rice cultivation – is considered by public authorities to be a real brake on the development of this crop in that ecosystem (SAKURAI, 2006). The effects of these scenarios on household behaviours in terms of rice production and consumption were assessed by comparing the impacts of each scenario with the current situation (Baseline scénario : Sbaseline). This assessment was mainly based on three types of indicators : - Socio-economic indicators : this concerns farm income, the monetary value of self-consumption and total hired labour, - Consumption indicators : this concerns the amounts consumed per product and the number of calories consumed per capita and per day, 106 - Production indicators: this concerns crop pattern and the levels of rice intensification expressed in terms of seeding density and total amount of work per hectare. 6. Results - The household model evaluation Table 15 compares the results observed and simulated by the household model. Three main conclusions: The model simulates all observed variables correctly with – in most cases – a difference by comparison with variables inferior to 15%. This difference is smaller for crop rotation, production and total work variables by comparison with consumption and number of calories, for which the difference sometimes reaches 40%. Nevertheless, this only concerns marginal activities for each household (represented in grey in table 15). It is the case for example of Lowlandextensive exploitation which grows rice essentially in the lowlands (0.64ha). In this context, even if the model overestimates the upland rice area (33%), this will only slightly affect global rice production. Indeed, upland rice only provides 0.01t/household; i.e. the equivalent of 3.5% of lowland rice production (0.3t/household). This type of analysis could also be extrapolated to all variables represented in grey in Table 15. The model reproduces rice farmers' production strategies fairly accurately. The presence of oil palm in large areas (Uplandintensive et Lowlandintensive farms) involves intensification levels (labour and seed amounts) (data not shown), and therefore more significant rice productions than in Uplandextensive and Lowlandextensive farms characterised by smaller oil palm areas. This is also expressed by a shorter fallow period, i.e. 6 years, in the upland ecosystem of the Uplandintensive exploitation, by comparison with that of the Uplandextensive exploitation (8 years) (data not shown). - Lowland-based households show lower risk-aversion coefficients (0.7 et 0.4 respectively for Lowlandextensive et Lowlandintensive farms) than those that are upland-based (0.8 et 1.5 respectively for Uplandintensive et Uplandextensive households). This result is due – in accordance with the work of (Worou O. N., Gaiser T., Saito K., Goldbach H., Ewert F., 2012) – to more variable rice yields in uplands than in lowlands, essentially caused by a more pronounced water stress in uplands than in lowlands, especially at the end of a cycle. Furthermore, 107 households with similar ecosystems show lower risk-aversion coefficients for intensive systems by comparison with extensive systems. This is the case when Uplandintensive et Lowlandintensive intensive farms are respectively compared with Uplandextensive et Lowlandextensive farms. This result is relatively in accordance with several studies carried out in West Africa. 108 Table 15 : Observed (OBS) vs. simulated (SIM): crop pattern, production, total labour, consumption and total calories per activity and farm type. The estimated risk-aversion coefficients are 0.8, 1.5, 0.4 and 0.7 respectively for Uplandintensive, Uplandextensive Lowlandextensive, and Lowlandintensive households, “dif” indicates the difference in percentage between simulated and observed values. The grey values indicate a difference of more than 20%. uplandintensive uplandextensive lowlandextensive lowlandintensive Oil palm Upland rice Low land rice Cassava Beans Sweet potato Total Oil palm Upland rice Low land rice Cassava Beans Sweet potato Total Oil palm Upland rice Low land rice Cassava Beans Sweet potato Total Oil palm Upland rice Low land rice Cassava Beans Sweet potato Total OBS 1.32 1.04 0.65 0.35 0.21 0.02 3.59 0.17 0.88 0.36 0.20 0.08 0.00 1.69 0.04 0.04 0.66 0.13 0.00 0.06 0.93 0.49 0.17 0.84 0.26 0.11 0.08 1.95 Crop pattern (ha) SIM 1.43 1.06 0.65 0.33 0.21 0.02 3.70 0.15 0.78 0.36 0.21 0.07 0.00 1.57 0.04 0.03 0.66 0.14 0.00 0.06 0.93 0.48 0.15 0.84 0.27 0.10 0.08 1.91 DIF 8 2 0 -6 1 -9 3 -13 -13 0 5 -8 -7 0 -33 0 10 -2 0 -2 -13 0 4 -12 -5 -2 OBS 0.34 0.37 0.29 1.06 0.51 0.07 0.05 0.21 0.16 0.61 0.15 0.00 0.01 0.01 0.30 0.39 0.00 0.25 0.13 0.08 0.50 0.80 0.37 0.31 - Production (t) SIM 0.37 0.36 0.29 0.98 0.52 0.07 0.04 0.27 0.16 0.62 0.14 0.00 0.01 0.02 0.30 0.43 0.00 0.22 0.13 0.07 0.50 0.80 0.33 0.29 - DIF 8 -3 0 -8 2 -1 -16 22 1 2 -8 0 42 -1 10 -12 0 -14 0 0 -12 -7 - OBS 50 128 146 27 75 66 492 6 108 81 15 75 0 286 2 7 148 10 0 66 232 19 28 188 20 75 66 395 Total labour (d) SIM 55 130 146 25 74 66 496 6 96 81 16 81 0 280 2 5 148 11 0 66 232 18 25 188 21 84 66 402 DIF 9 2 0 -8 -1 0 1 -8 -13 0 5 7 -2 24 -40 0 10 0 0 -3 -12 0 6 11 0 2 Consumption (t/household/year) OBS SIM DIF 0.15 0.17 12 Calories per product (Kcal/capita/day) OBS SIM DIF 351 401 12 0.96 0.77 -25 919 730 -25 0.60 0.09 0.07 0.05 0.54 0.09 0.08 0.04 -11 0 13 -25 178 42 15 1505 175 157 42 20 1350 119 -11 0 13 -11 -25 0.47 0.43 -9 566 516 -9 0.61 0.05 0.05 0.02 0.62 0.04 0.08 0.02 2 -25 38 0 225 31 15 1013 93 229 23 25 913 93 2 -15 38 -11 0 0.33 0.40 18 460 570 19 0.89 0.01 0.18 0.06 0.82 0.02 0.15 0.06 -9 50 -20 0 392 8 70 1022 279 359 14 56 1070 279 -9 50 -20 4 0 0.60 0.64 6 1052 1116 6 0.79 0.14 0.15 - 0.80 0.12 0.13 - 1 -17 -15 - 422 124 70 1946 429 102 59 1967 1 -17 -15 1 7. Scenario analysis The comparison of the effect of the three scenarios aiming for the following subsidization: purchase of rice seeds (Sseeds), plantation of oil palm (Sop) and rice in lowland ecosystems (Srice), shows relatively contrasted behaviours between the different types of households, in terms of rice production and consumption. In detail: - Rice seeds scenario (Sseeds) The scenario for the subsidization of the purchase of rice seeds has only generated a slight and marginal improvement of economic indicators. Indeed, for all households the farm income as well as the monetary value of self-consumption have not exceeded an 11% increase (maximum) when scenario Sseeds with 100% subsidizing is compared with Sbaseline (table), at the four households' level. At production level, rice seed subsidization has not caused any changes for lowland rice (table3), which is very profitable by comparison with upland rice. Concerning upland rice, 100% rice seed subsidization has mainly induced two behaviours : i) for upland rice based households (Uplandintensive, Uplandextensive), for both farm types a more significant rice intensification which expresses itself as a 100% increase from density 2 to density 3 (data not schown), and an 18 and 29% increase in labour respectively for Uplandintensive and Uplandextensive. Nevertheless, at crop level, and concerning Uplandintensive households, a slight decrease in rice production area (-8%) as well as in bean production area (-8%) and fallow area (-8%) was observed. This decrease in rice production, beans production and fallow areas came with an increase in cassava (+46%) and palm oil (+10%) cash crops areas. On the contrary, an increase in upland rice production (+23%) and bean production (+23%) areas was observed for Uplandextensive farms, at the cost of oil palm, and ii) no intensification was observed for lowland-based households (Lowlandextensive, Lowlandintensive), but only an increase in upland rice and beans areas at the cost of oil palm (table 16). At consumption level this scenario (Sseeds) has generated an improvement of rice consumption of 13, 8, 2 and 5 % respectively for Uplandintensive, Upland extensive, Lowlandextensive and Lowlandintensive households, compared with the baseline situation (Sbaseline). For all households except Uplandintensive with +6% rice purchased for consumption when scenario Sseeds is compared with Sbaseline, this systematically involves an increase of the proportion kept for selfconsumption (data not shown). However, in terms of calorie numbers this scenario (Sseeds) has only increased the number of calories consumed per capita and per day for intensive households (+7% for uplandintensive, +12% for lowlandintensive). 110 Concerning extensive households, on the contrary, the increase in rice consumption at the cost of energy products such as oil palm and cassava leads to a slight decrease in the total number of calories consumed per capita and per day, i.e. -2% and -13% respectively for Uplandextensive et Lowlandextensive households. - Palm oil scenario (Sop) In the same way as the first scenario, scenario Sop has generated very little change for the economic indicators (table 16) of all households in comparison with scenario Sbaseline, with the exception of Uplandextensive household. Indeed, Uplandextensive household has shown a distinct increase in farm income (+52%) and a distinct decrease in the monetary value of selfconsumption (-30%) when scenario Sop is compared with scenario Sbaseline. More specifically, the monetary value of self-consumption first decreased starting from a 20% subsidization, to reach its lowest level at 40%, before it rose again and reached a 100% subsidization 1.1 Million Leone value (figure 34). The same behaviour was also observed for the global consumption of the different products as well as the number of calories consumed per capita and per day (figure 35). Indeed, contrary to other households which have maintained the same consumption levels per product regardless of the subsidization level, Upland extensive household saw its consumption per product decrease before it increased with a 60% subsidization level (figure 34). Above all, global rice consumption has gone from 0.4t/household and per year to 0.55t/household and per year respectively for Sbaseline and Sop with a 100% subsidization. Indeed, at low subsidization levels households prefer to purchase rice rather than produce it even if it involves lowering consumption levels overall. In this context purchasing seems less risky and more profitable than producing with low rice seeding densities. This phenomenon is relatively well known in literature as explained by (Fleuret and Fleuret, 1980; Lunven, 1982). In addition, as stated by Immink and Alarcón (1991) that the household's vulnerability to food insecurity and dietary inadequacy may be increased, particularly when household food availability does not change much in response to higher household income (such as the case of Uplandextensive farm type in this application) from cash crops. Household labour inputs per hectare are often higher for production of cash crops than for basic food crops; thus the household's daily energy requirements may be raised. By increasing the value of the subsidy dedicated to oil palm, households continue to purchase but also to take up rice production again but with a more significant intensification which expresses itself by a higher amount of work (+43%) and seeding density by comparison with 111 Sbaseline. That is how the equivalent of 0.2 ha of rice grown under density 2 (0,017kg/ha) in scenario Sbaseline are replaced, at 100% subsidization Sop scenario, with density 3 (0,05kg/ha) (data not shown). Similarly to scenario Sseeds, scenario Sop induced no change in lowland rice production. Indeed, at the level of lowland rice-based households (Lowlandextensive and Lowlandintensive), the small areas occupied by upland rice alone have been replaced by more profitable oil palm (table 16). This choice of production systems therefore favoured oil palm consumption; but in the end this has not upset the caloric balance of households compared to the baseline situation (Sbaseline scenario). Calorie (Kcal/cap/day) 2000 1600 1200 800 400 0 Self-consumption (Million Leone) 2.5 2.0 1.5 1.0 0.5 0.0 0% Uplandintensive 20% 40% 60% Oil palm subsidization level Uplandextensive Lowlandextensive 80% 100% Lowlandintensive Figure 36: variation of calorie number and monetary value on the self-consumption by subsidization oil palm subsidization level for the 4 household types. 112 Figure 37 : variation of the consumption amount of different food products by oil palm subsidization level for the Uplandextensive farm type - Lowland scenario (Srice) This scenario (Srice), in comparison with Sbaseline, and for all households, has generated a significant progress in farm income and monetary value of self-consumption. As an example, by comparing the two scenarios, farm income has progressed by 21, 57, 29 and 9% respectively for Uplandintensive, Uplandextensive, Lowlandextensive and Lowlandintensive households. For all households, this development in economic terms comes with an increase in rice consumption, i.e. 19, 34, 22 and 15% respectively for Uplandintensive, Uplandextensive, Lowlandextensive and Lowlandintensive households (table 16). This change in consumption preference was followed by a decrease in oil palm and beans consumption for upland-based households and only in beanss for lowland-based households. For all households except Uplandintensive households, this change has induced an increase in the number of calorie per capita and per day; i.e. 14, 21 and 7% respectively for Uplandextensive, Lowlandextensive and Lowlandintensive households. In concrete terms, with this scenario (Srice) households have become more specialised by increasing lowland rice areas (70%, 84%, 42% and 69% respectively for Uplandintensive, Uplandextensive, Lowlandextensive and Lowlandintensive) and one single cash crop (cassava or oil palm) (table16). Upland rice and beans have completely disappeared from crop pattern (table 16). This strategy has made it possible for households to increase their rice stock (data not shown) and above all to become net bean purchasers. 113 Table 16 : comparison of socio-economic, consumption, total calorie and crop pattern for the three scenarios (Sseeds, Sop, Srice) in comparison to the baseline scenario (Sbaseline). Uplandintensive Uplandextensive Sbaseline Seeds(*) Sop Srice Sbaseline Seeds Sop Lowlandextensive Srice Sbaseline Seeds Lowlandintensive Sop Srice Sbaseline Seeds Sop Srice Socio-economic Farm income (Million Leone/ha) 3.6 9 0 21 1.9 9 52 57 2.7 6 7 29 4.0 5 0 9 Self consumption (Million Leone/ha) 2.2 11 1 34 1.5 0 -30 25 1.3 -3 -1 31 1.4 5 -2 36 304.1 18 15 -95 168.6 29 43 -95 138.4 -6 -1 23 181.9 -5 2 -58 Rice 0.8 13 1 19 0.4 8 22 34 0.4 2 0 22 0.6 5 -3 15 Oil 0.2 0 0 -94 0.0 -75 0 -100 0.0 0 75 0 0.1 34 30 0 Beans 0.1 0 0 -77 0.0 54 54 -100 0.1 0 0 -72 0.1 0 0 -77 Sweet potato 0.1 0 0 0 0.1 0 0 0 0.3 0 0 0 0.1 0 0 0 Cassava 0.5 0 0 0 0.6 0 0 0 0.4 -33 -75 33 0.8 0 0 0 1347 7 0 -15 889 -2 16 14 1032 -13 4 21 1882 12 1 7 Rice upland 1.0 -8 -22 -100 0.8 23 -61 -100 0.0 41 0 -100 0.1 32 -20 -100 Rice Lowland 0.7 0 0.4 0 0 84 0.7 0 0 42 0.8 0 0 69 Beans 0.2 -8 -22 -100 0.0 23 -61 -100 0.0 41 0 -100 0.0 32 -20 -100 Oil palm 1.4 10 49 -78 0.2 -75 97 0 0.0 -79 90 0 0.5 98 26 -1 Cassava 0.5 46 20 -6 0.2 0 0 -9 0.1 0 0 63 0.3 0 0 79 Sweet potato 0.0 0 0 70 0.0 0 0 84 0.0 0 0 42 0.0 0 0 57 Fallow 6.2 -8 6.3 -2 -67 -100 0.2 41 0 -100 0.9 32 -20 -98 Farm labour (d/ha) Consumption (t/household/year) Total calorie (calorie/capita/day) Crop pattern (ha) 0 70 -22 -100 (*) For each indicator the value corresponding to the Sbaseline is the absolute value. However, for each scenario (Sseeds, Sop and Srice) the value for each indicator is expressed in percentage in comparison to the baseline scenario by calculating the following formula (value_Sscenario- valueSbaseline)/Sscenario*100. 8. General discussion The main purpose of this paper was to develop a household model based on a conceptual model representing the behaviour, in terms of production and consumption, of rice farming households in West Africa. The goal was also – with the help of that model – to assess the impact of the three scenarios that aim for the improvement of rice production, and therefore its consumption, in Sierra Leone. This study leads to two main conclusions: - The effects of the 3 scenarios on lowland-based households are very low in terms of rice consumption. Lowlandextensive households show the lowest amount of rice consumed per capita and per year in comparison with other households. Scenario Srice alone has induced a significant rise in rice consumption. With this scenario (Srice) rice consumption for all households has reached the same level as that of the two bordering countries : Guinea and Liberia, but far away from the consumption planning hoped for by the National Sustainable Agriculture Development Plan 2010-2030 (figure 38). In terms of calories, the Lowlandintensive 114 households alone are at the FAO's recommended minimum level but far from Guinea which shows an average number of calorie intake of 2500 capita/day (figure 38). The other households, especially extensive ones, show very low caloric intakes. Similarly, regardless of the type of household, the scenarios have only had a slight effect on the number of calories consumed per day and per capita. This seems logical, insofar as the preference for rice consumption – which expresses itself by an increase in rice consumption – happened at the cost of other products (for example oil palm or cassava), since purchasing and planting power remain low overall. These results are in contradiction with State policies, which essentially rely on the distribution/subsidization of rice seeds and oil palm plants to boost rice production and consumption (The Government of Sierra Leone, 2013). The scenario for the subsidization of rice plantation in lowland ecosystems seems to be the most relevant in the current context of Sierra Leone, but it is also the most costly in the short term by comparison with other scenarios. This study also highlights the fact that lowland-based households with low intensification means (Lowlandextensive) barely respond to the scenarios. More and more of this type of household can be found – while they hardly existed in the past – because of the mass return of populations following the end of the war and the saturation of uplands very sought after for agriculture or urbanization (Sanoh N., 2013). Target (a) Guinea Guinea (b) Liberia Target Liberia Figure 38 : Variation of rice consumption (a) and total calorie (b) for the 4 farm types, Guinea (average national values), Liberia (average national values) and the targets set by the Sierra Leone Government for the National Agricultural Strategy (2010-2030) and the FAO organization. The construction of the household model, first conceptual then numerical, has made it possible – beyond the assessment of current production and consumption systems – to detect a 115 certain number of significant points concerning: i) the model : this household model is one of the rare models that can simulate production and consumption (including self-consumption) separately. The separation of quantities stocked, sold and self-consumed also gives a more realistic representation of food crops in West Africa. Nevertheless, this model has made it possible to identify two major limits. The first concerns the association of a fictitious price with self-consumption. An indigenisation of prices (including that of self-consumption) according to supply and demand at the level of countries with low-income economy would be possible and desirable according to the approach of Louhichi et al (2014), and ii) The dataset : the dataset suffers from severe drawbacks when it comes to consumption estimates. In the future more detailed consumption data ought to be collected to model household behaviours both as producers and consumers. A proper food questionnaire, such as a budget or a consumption survey, could be administered to provide better estimates of food intake at the household level. Interviews are usually carried out with an enumerator asking one or more household members to recall expenditures made over a reference period. This data is plagued by the typical reporting biases faced by all interview methods (i.e. recall errors, misreporting, etc.), however it has the advantage to come directly from the location where behaviours are to be modelled. For instance, the Burkina Faso National Institute of Statistics and Demography (INSD) conducted several household surveys using the Standardized Questionnaire on the Basic Indicators of Wellbeing (QUIBB) protocol, which collected food expenditure of 8,500 burkinabé households in 425 enumeration area in 2003 (Allen et al., 2011). To differentiate between household members and further fine-tune the model, surveys should be conducted at the individual level through surveys such as either a 24-hour recall or a multiple-day dietary record (usually 3- to 7-day diaries). As the purpose of the current model is not to analyse individual nutrient intake, including a comprehensive set of food groups is enough. Ideally, this data should be collected several times during the year to control for seasonal variations. The same observation can be made about production for which information concerning all abiotic (soil fertility, water...) and biotic (pests, birds...) factors and their effects on yield and externalities (mainly soil fertility) must be collected for a better definition of current activities. Based on that, this study can be used to specify the type of surveys that needs to be carried out and of data to be collected with this kind of approach. Moreover, it could be used to specify the modification needed to adapt several relatively standardized exhaustive surveys (such as the survey used in this study but also as example the Household Consumption and Expenditures Surveys “HCES”, Fiedler, 2012; A multi-topic, nationally and regionally representative household survey “GLSS5”, Quiñones and Diao, 2012) that have already been 116 carried out across developing countries by several international organizations (IFPRI, USDA, FAO...) in order to assess consumption and agricultural production behaviours and strategies. 9. Conclusion The purpose of this paper was to elaborate a household model based on the theory of nonseparability between production, consumption and production resources. This model was applied at the level of rice farming households in order to test the effects of the three scenarios, and served to show that certain initiatives can improve rice consumption without necessarily upsetting current rice farming systems as was considered by several programmes (Delarue J., 2007). The scenario that contemplated subsidizing rice plantation in lowland ecosystems has been identified as the most efficient by comparison with other scenarios. This result is relatively coherent with the numerous initiatives that contemplated this type of investment (SLIEPA, 2012). Nevertheless, this scenario has also generated a rice-cash crop specialization which could raise questions as to the resilience of such systems within very unstable markets, and with generally quite low means of investment. Incidentally, this type of behaviour with problems of resilience/durability has already been observed concerning other systems such as the introduction of technology to reduce fallow periods and introduce animal breeding in Gana (Yiridoe et al., 2006), or water management and the introduction of a species (Dile et al., 2013; Mertz et al., 2005), or improving land work conditions (Makurira et al., 2011). Besides, the effects of these scenarios remain below the expectations of public authorities. The structure of the current model based on the notion of activity would make it possible without any major changes – beyond the matter of data availability – to test other scenarios strongly contemplated by several initiatives such as the introduction of irrigation (Burney and Naylor, 2012), of fertilization (Chen et al., 2011), the improvement of rice crops (Saito and Futakuchi, 2014) or of mechanization (Seck et al., 2013). Finally, it must be observed that a regional version of this model is under way in order to simulate the behaviour of agricultural households, but taking into account resources (especially labour and land) and consuming goods exchanges between the different households. 117 Chapitre 5 : Discussion générale Dans ce dernier chapitre, nous résumons, dans un premier temps (partie 1 de la discussion), les principaux résultats obtenus dans cette thèse. Dans la deuxième partie nous revenons sur la démarche mise en œuvre pour caractériser et évaluer la performance des systèmes agricoles au Nord de la Sierra Leone. Nous discuterons également, dans cette partie, les atouts et limites de la méthode utilisée et, ensuite, les principales leçons tirées de ce travail de caractérisation. Dans la troisième partie, nous revenons sur l’approche suivie pour modéliser le comportement des ménages en termes de production et de consommation. Enfin, nous présenterons les limites et les apports de cette thèse concernant l’évaluation de l’impact des mesures incitatives sur les performances des systèmes agricoles, et la place que doit occuper le riz par rapport à des besoins alimentaires croissants et des moyens de production limités. Mais, avant de développer l’ensemble de ces points, il est important de rappeler ici quelques éléments de contexte. Il faut observer que l’agriculture de subsistance en Afrique de l’Ouest est majoritairement caractérisée par une faible utilisation d’intrants (matériel, fertilisation, irrigation..) (Chen J. et al., 2011), une mauvaise gestion des pratiques culturales (Worou O. N., Gaiser T., Saito K., Goldbach H., Ewert F., 2012), une dégradation des ressources en terre, conséquente à une intensification non contrôlée, une inadaptation des politiques locales (Katic P. G., Namara R. E., Hope L., Owusu E., Fujii H., 2013), et une défaillance du marché représentée par des prix souvent élevés et instables. Tous ces éléments, auxquels il faut ajouter l’incapacité de cette agriculture à subvenir aux besoins alimentaires des populations ouest africaines, ont toujours fait de cette agriculture, malgré de multiples études menées au niveau de cette région, un axe de recherche majeur. Il en va de même pour les acteurs politiques qui se font une préoccupation permanente d’assurer une alimentation durable pour les petits ménages agricoles (Babu S. C. et al., 2014; Graef F. et al., 2014; Katic Pamela G. et al., 2013; Webber H. et al., 2014). Cette préoccupation a atteint son apogée lors de la crise rizicole en 2008 (Katic P. G., Namara R. E., Hope L., Owusu E., Fujii H., 2013; Timmer C. P., 2010). C’est dans ce cadre que notre travail de thèse s’inscrit. 1. Résumé des principaux résultats obtenus. Notre travail a fait ressortir deux principaux résultats qui concernent les systèmes agricoles étudiés au niveau du district de Bombali. 118 - Globalement, le rendement moyen du riz pour l’écosystème plateau (0.29 t/ha) est plus faible que celui de l’écosystème basfond (0.34 t/ha) (chapitre 2 et 3). Dans ce cadre, le rendement moyen de riz par exploitation dépend essentiellement, en l’absence de fertilisation, de mécanisation et d’irrigation, de la quantité de travail totale engagée, de la dose de semis, de la qualité de l’écosystème cultivé, mais également de l’objectif de production (défini par rapport aux besoins en riz par ménage). En ce qui concerne l’écosystème, notre étude a montré que le rendement moyen par exploitation augmente en moyenne de 0.0037 t/ha pour chaque pourcentage d’augmentation de la surface de basfond au détriment de celle de plateau. Cet accroissement du rendement s’est accompagné d’une baisse de la quantité totale de travail engagée par hectare de riz. En effet, un hectare de riz utilise en moyenne 90 et 121 jour/ha respectivement pour les écosystèmes basfond et plateau. La présence du palmier à huile comme culture de rente joue également un rôle important sur les rendements de riz. En effet, pour chaque hectare de palmier à huile cultivé aux dépens d’un hectare de riz, cela entraîne une hausse moyenne de 0.0028 t/ha et 1.54 jours/ha, respectivement de la dose de semis et de la quantité totale de travail engagée pour le riz. La présence de palmier à huile a aussi provoqué une augmentation de la surface moyenne par exploitation, et, par conséquent, de la surface réservée au riz. De même, il ressort de notre étude que le rendement moyen de riz par exploitation est négativement corrélé à la durée de la jachère. En effet, nous avons enregistré des rendements moyens riz de 0.37 et 0.23 t/ha respectivement pour des jachères de 6.7 et 8.7 ans. Ce surprenant résultat, par rapport à la bibliographie, est dû à des moyens de production mobilisés (principalement dose de semis et quantité de travail par hectare) moins importants sur les jachères courtes par comparaison à celles plus longues. Cela s’explique, d’une part, par des demandes en riz pour la consommation moins importantes (familles peu nombreuses ou familles pauvres de retour après la fin de la guerre) sur les exploitations qui utilisent des jachères longues, et, d’autre part, par l’absence de culture de rente sur ce type d’exploitation. - Seul le scénario Srice subventionnant l’installation de riz sur l’écosystème bas-fond a permis une augmentation importante de la consommation de riz (chapitre 4). Avec ce scénario (Srice), cette dernière a augmenté au niveau de tous les ménages d’au moins 15% pour atteindre son maximum au niveau des ménages extensifs à dominance plateau (+34% d’augmentation). Ce résultat s’est accompagné d’une hausse du nombre de calories consommées qui varient, selon 119 les types de ménage, de 7 à 21%. Ce résultat est obtenu grâce à une spécialisation des riziculteurs (par rapport à une situation de base plus diversifiée) qui s’est manifestée par une production de riz exclusivement sur les bas-fonds, et par la réserve des plateaux pour les cultures de rente (palmier à huile ou manioc selon les types de ménages). Le scénario subvention des semences (Sseeds) a entraîné, quant à lui, une augmentation moins importante (de + 2 à 13% selon les types de ménages) de la consommation de riz et du nombre de calories (+7% à 12% selon les types de ménages) par comparaison au scénario Srice. Dans certains ménages, la subvention des semences de riz a même généré une baisse du nombre de calories consommées (cas des ménages extensifs). Ce résultat est dû à une intensification marginale du riz ou à une augmentation limitée de la surface riz par exploitation accompagnée souvent par une baisse de la consommation des légumes ou de l’huile de palme. En ce qui concerne le scénario qui a prévu la subvention de la plantation de palmier à huile, les résultats en termes de consommation de riz et de nombre de calories sont globalement assez similaires à ceux du scénario Sseeds. Néanmoins, avec ce scénario, nous avons retrouvé un comportement de consommation assez attendu et décrit en bibliographie. En effet, avec des niveaux faibles de subvention réservée à la culture de rente, la consommation des cultures vivrières baisse (en l’occurrence le riz dans notre cas) avant d’augmenter de nouveau avec des niveaux de subvention plus élevés (+60% de subvention réservée au palmier à huile dans notre cas). A de faibles niveaux de subvention, le riziculteur a préféré investir dans la plantation de palmier à huile au détriment de celle de riz et de légumes. A des niveaux plus importants de subvention pour le palmier à huile, non seulement la surface de riz a augmenté, mais son niveau d’intensification a également été plus élevé (dose de semis, quantité totale de travail). Que retenons-nous comme facteurs de diversité pour évaluer la performance et l’efficience des systèmes rizicoles ? Retours sur la méthode utilisée et les principales conclusions d’analyse de performance (Chapitre 2 et 3). La caractérisation de la diversité des systèmes de culture et de production à base de riz dans notre étude s’est basée, en utilisant plusieurs critères de structure, de production et de consommation, sur la création de plusieurs ménages agricoles types (typologie). Cette méthode nous a permis d’identifier plusieurs stratégies de production par ménage à l’échelle d’un district. Cette approche en soi n’est pas nouvelle. D’ailleurs, elle a été largement 120 appliquée au niveau de plusieurs études comme l’attestent les travaux de Brouwer J. et al. (1993) et Tittonell P., Muriuki A., Shepherd K. D., Mugendi D., Kaizzi K. C., Okeyo J., Verchot L., Coe R., Vanlauwe B. (2010) pour caractériser le comportement des agriculteurs en termes d’irrigation et de fertilisation en Afrique de l’Ouest. Néanmoins, notre approche présente deux innovations majeures : - C’est la première étude qui s’intéresse à la caractérisation et à l’analyse des stratégies de production des riziculteurs en Sierra Leone d’après-guerre (post 2002). Elle vient répondre à une commande de l’UE afin d’aider les riziculteurs à promouvoir la production et la consommation de riz. Elle vient notamment compléter plusieurs initiatives qui se sont focalisées sur la caractérisation, d’une façon globale, des défis socio-économiques mais également institutionnels que doit surmonter la Sierra-Leone. - Nous avons élaboré un cadre d’analyse spécifique pour évaluer la performance et l’efficience des systèmes rizicoles en Afrique de l’Ouest appliqué à la région nord de la Sierra-Leone. Ce cadre considère des critères socio-économiques, de structure et de consommation pour expliquer la stratégie de production des riziculteurs. Cette analyse, qui se base sur une analyse de la composante principale (ACP) suivie par une classification hiérarchique (CH), fait ressortir plusieurs types de comportements traduisant des stratégies de production différentes dictées par des objectifs de production (pour satisfaire des besoins alimentaires) et des ressources (notamment en main d’œuvre, en terre et en semence) assez contrastées d’un riziculteur à l’autre. De cette analyse, trois principales conclusions peuvent être tirées : i- Pour ce qui concerne les systèmes de production rizicoles : globalement, en l’absence de moyens d’intensification significatifs, le riz présente toujours un rendement moyen très faible (exemple 0.7t/ha au nord de la Sierra Leone) par comparaison aux régions de l’Afrique de l’Ouest. Néanmoins, cette moyenne cache des inégalités très importantes en termes de performance et d’efficience entre les différents ménages. Le type d’écosystème, et notamment la fertilité initiale des sols, mais également les objectifs de production expliquent en grande partie les moyens mobilisés (main d’œuvre, dose de semence) et, par conséquent, les différences de rendement observées entre les ménages. Ce type de résultat est assez commun avec d’autres régions en l’Afrique de l’Ouest comme le montrent les travaux de Whitbread A. M. et al. (2010) et Saito K. et al. (2009). Il ressort également que, pour augmenter la production et satisfaire les besoins croissants des populations de retour en Sierra à Leone 121 après la fin de la guerre, deux options se présentent. Pour les plus riches, cela passe par une intensification relative des systèmes actuels en cultivant des cultures de rentes garantissant aux riziculteurs un cash qui permet d’augmenter à l’hectare les quantités de main d’œuvre engagée et les doses de semis. A cela s’ajoute également un agrandissement de la surface de l’exploitation, permettant ainsi de cultiver le riz sur deux écosystèmes différents (plateaux, bas-fonds). Pour les riziculteurs pauvres, l’augmentation de la production passe essentiellement par l’agrandissement de l’exploitation au travers de l’exploration de nouvelles forêts (notamment des plateaux). Quelle que soit l’option choisie, l’augmentation de la production est suivie par une baisse de l’efficience de production (mesurée dans notre application comme étant le rendement par quantité d’intrants utilisés). La baisse de l’efficience est due, en fonction du type d’écosystème, à la faible fertilité des sols (cas des plateaux cultivés avec une jachère courte), aux faibles doses de semis (cas des petits ménages dont les parts consommées prennent la priorité sur les quantités de riz stockées comme semence pour la saison suivante), ou à une main d’œuvre insuffisante, notamment pour effectuer le désherbage (cas des grandes exploitations intensives). Ces résultats montrent, comme le soulignent les travaux de Aker J. C. et al. (2010), que, sans un changement radical au niveau des systèmes de culture actuels, les productions resteront à la fois faibles et variables. Ce changement passe souvent par l’adoption de l’irrigation, de la fertilisation ou/et de l’utilisation des semences améliorées. Néanmoins, l’histoire de l’évolution des agricultures de par le monde montre que le passage d’une agriculture familiale, peu productive et à faible efficience, à une agriculture intensive et efficiente ne peut se faire qu’en jouant à la fois sur des critères techniques mais également socio-économiques et institutionnels (Kouassi B., 2008). ii- En ce qui concerne les bases de données mobilisées. Pour la réalisation de la typologie des ménages, nous avons mobilisé, pour cette étude, une base de données constituée d’enquêtes de terrain réalisées par le JRC. Elle renferme, par ménage, des informations caractérisant la structure des exploitations, les activités agricoles présentes et leurs caractéristiques technico-économiques ainsi que la composition du ménage et ses préférences alimentaires. Même si ce type d’enquête (assez formaté), qui a pour ambition de caractériser à la fois la production et la consommation, est très répandu, il reste très utile notamment pour une description assez globale des habitudes alimentaires en relation avec la production au niveau d’une grande région ou d’un pays. Ce type d’enquête, même exhaustif, souffre de 122 plusieurs limites pour des études de caractérisation du comportement qui se veulent à l’échelle d’un ménage : - Les détails concernant les quantités d’intrants utilisés (main d’œuvre, dose de semis) par tâche et par sexe ne concernent que la culture du riz. Pour les autres cultures, on ne dispose que des quantités totales engagées. L’absence de données détaillées pour les autres cultures n’est pas uniquement due au manque de moyens (temps et financier), mais surtout témoigne du fait que dans l’esprit des porteurs du projet (Pouvoir Public et certains scientifiques) que les autres cultures ne jouent que très peu en termes d’assolement et de revenu. D’ailleurs, dans le guide d’enquête ces cultures sont marquées comme « marginal crops ». Néanmoins, ces cultures sont d’une importance capitale pour l’équilibre alimentaire des ménages rizicoles ou elles représentent, pas moins de 60% du total calorique apporté par la partie végétale (le reste est apporté essentiellement par le riz). - Il manque pour l’année d’enquête (y compris pour le riz) les dates de réalisation des différentes tâches. Cela ne nous a permis de décrire les stratégies de production des riziculteurs que d’une façon globale (à l’année) sans tenir compte des contraintes au cours de l’année. Or, ces contraintes peuvent être d’une importance capitale, notamment lorsqu’il s’agit d’arbitrer la répartition de la main d’œuvre entre les différentes tâches. Cet arbitrage se fait en fonction des besoins mais également des disponibilités qui sont fonction de l’importance de la tâche et des objectifs de production. A titre d’exemple, lorsque la production de riz obtenue sur les plateaux (vers le mois d’avril) est importante, les ménages sont souvent contraints d’abandonner une partie de la récolte pour pouvoir cultiver du riz sur des surfaces bas-fonds. Cette stratégie permet aux ménages (faute de moyen de stockage efficace) de se disposer d’une production du riz tout au long de l’année (Hoogh I. D. et al., 2011). Cette même analyse autours de la compétition pour la main d’œuvre peut se faire lors de la cueillette et de la transformation des gousses de huile de palme ou pour le semis et la récoltes des légumes. - Aucune variable décrivant l’état du sol sous les différents types d’écosystème et de jachère n’a pu être collectée. En effet, dans cette étude, nous avons prévu de collecter les variables : état hydrique du sol, notamment en fin de cycle du riz, et fertilité du sol au niveau des différents types d’écosystème et de jachère. Ces variables seraient pour plusieurs auteurs d’une grande importance afin d’expliquer les différents niveaux de performance et d’efficience observés sur une exploitation rizicole en Afrique de l’Ouest (Van Duivenbooden N. et al., 2000). Pour cet objectif, nous avons organisé une mission spécifique en Sierra Leone 123 où nous étions en contact avec des chercheurs du centre de recherche agronomique de Bombali (SLARI). Malheureusement, très peu d’expérimentations ont été menées en Sierra Leone depuis la fin de la guerre et il a été impossible de se procurer ce type de variable. Cependant, malgré l’absence de ce type de mesures, il nous semble que la quantité totale du travail, les doses de semis et la durée de la jachère (en partant de l’hypothèse que plus la jachère sur les plateaux est longue, plus la fertilité initiale du sol est élevée) expliquent, en grande partie, la faiblesse et la variabilité des rendements. Le stress hydrique est certainement un facteur assez déterminant notamment sur les plateaux, mais il nous paraît, vu le faible niveau de rendement du riz, moins contraignant que les autres facteurs précités. - Dans notre étude, l’information au niveau des enquêtes sur les préférences de consommation n’a concerné que les quantités de riz consommées par ménage. En revanche, les données de consommation des cultures secondaires (manioc, palmier à huile, légumes, patate douce) sont estimées à partir des tables du bilan alimentaire de l’OCDE. A cela, il faut ajouter que les quantités de riz consommées au niveau des enquêtes sont déclinées d’une façon annuelle et non par période (mensuelle par exemple) comme suggéré par Van Wijk M. T. et al. (2014). En effet, les niveaux de consommation globale par produit sont tributaires de la production au niveau de chaque saison et des moyens de stockages disponibles par ménage. Dans un milieu où les unités de stockage sont souvent insuffisantes et de mauvaises qualités (CORAF et al., 2008), il est souvent admis que les périodes de soudure (période creuse entre deux cycles de production) sont les plus critiques avec des niveaux de consommation les plus faibles (WAF, CILSS, FAO, CIRAD, Fewsnet, 2011). 2. Intensification des systèmes rizicoles Quelle place pour la culture du riz pour satisfaire des besoins alimentaires croissants? Retour sur l’utilisation du modèle de ménage : quels enseignements tirés? (Chapitre 4) Plusieurs études traitent des limites de l’agriculture de subsistance en Afrique de l’Ouest en termes de production et de capacité à fournir une alimentation durable pour les populations locales (Winter-Nelson A., 1997). Néanmoins, à l’échelle d’une exploitation/ménage, il est souvent question, dans ces études, d’évaluer la performance de ces systèmes et leur impact sur l’alimentation des ménages ruraux de façon qualitative aux dires d’expert (Winter-Nelson A., 124 1997). Plus le système est complexe, plus il serait difficile à l’expertise de prédire sa performance, à la fois technique et socio-économique (y compris la consommation), et son évolution dans un contexte changeant (Ewert F. et al., 2009). A l’échelle d’un territoire, d’un pays ou au niveau mondial, cette évaluation se fait dans la plupart des cas en utilisant des tableurs qui calculent, en s’appuyant sur des analyses statistiques, les flux de production, de consommation, d’importation et d’exportation sans réellement tenir compte de la diversité au niveau local (ménage, petite région) (Dossa L. H. et al., 2011) ; Bidogeza, Berentsen, De Graaff, & Oude Lansink, 2009; Joffre & Bosma, 2009; Kostov & McErlean, 2006; Maseda, Dıaz, & Alvarez, 2004; Somda, Kamuanga, & Tollens, 2005). Dans cette thèse, nous avons fait le choix de développer un modèle de ménage basé sur la programmation linéaire pour évaluer la performance socio-économique et technique des ménages rizicoles. En soi, cette approche n’est pas nouvelle, néanmoins, elle est innovante au moins sur quatre points essentiels. En effet, elle permet: - de simuler la complexité, pour la prise de décision du riziculteur au regard du choix des assolements (y compris la distribution des cultures par écosystème et des doses d’intrants par culture), des relations entre ressources disponibles, production et consommation au niveau d’un ménage ; - contrairement à plusieurs études antérieures qui maximisent de manière dissociée le revenu agricole et la consommation (Herrero M. et al., 2014; Husin L., 2012), de maximiser un revenu global qui tient compte conjointement du revenu agricole et de l’autoconsommation. Cela suppose la non-séparabilité entre les deux profits comme recommandé par plusieurs auteurs, lorsqu’il s’agit d’analyser le comportement des ménages ouest africains (Louhichi K. et al., 2014) ; - de tenir compte, lors de la simulation des comportements des ménages rizicoles, des quantités de riz stockées, vendues et autoconsommées. Cette spécificité des ménages rizicoles, non seulement en Sierra Leone, mais également en Afrique de l’Ouest, est un élément essentiel de la prise de décision des ménages en lien à la fois avec les niveaux de consommation et les doses de semis pour la saison suivante (Kargbo A. B., 2002) ; 125 - de considérer, lors de la simulation de la volatilité des prix des produits, l’effet des aléas climatiques (pluies tardives, stress hydrique tardif sur le riz plateau…), sur le rendement des principales cultures cultivées. Ces deux types de risque sont des éléments clés pour la prise de décision des ménages en ce qui concerne le choix des assolements et des niveaux de consommation. Cependant, certaines limites par rapport au modèle de ménage sont à mentionner. Elles concernent à la fois l’élaboration du modèle conceptuel et les bases de données mobilisées. - Le modèle conceptuel développé dans cette application pour représenter le lien entre ressources disponibles, production et consommation, présente deux principales limites par rapport au contexte de la zone d’études. La première concerne l’absence du facteur « temps ». En effet, la question des besoins alimentaires dans notre étude a été traitée d’une façon globale (à l’année) sans tenir compte des calendriers de production au cours de l’année des différentes cultures par exploitation (en relation avec le manque de données mensuelles comme expliqué dans la partie 2 de la discussion). Cela implique la non simulation des prises de décision pour les périodes de soudure où les disponibilités en riz sont notamment les plus faibles. Il serait donc souhaitable de considérer dans l’absolu un pas de temps mensuel pour mieux comprendre la compétition autours de la distribution des ressources, mais également les périodes de risques de pénurie en riz (figure 39). Néanmoins, la structure actuelle du modèle, en cas de disponibilité des données, est parfaitement adaptée pour simuler d’une façon mensuelle les prises de décision des riziculteurs. Ainsi, même si, globalement, la production rizicole semble être suffisante pour un ménage donné, cela pourrait cacher des disponibilités en riz insuffisantes à certaines périodes de l’année. Cette situation concerne l’ensemble des ménages étudiés mais surtout ceux dominés par un seul écosystème. Cette limite sur le facteur « temps » peut s’appliquer également à la disponibilité de la main d’œuvre qui est considérée, dans notre modèle, d’une façon globale sans tenir compte des besoins mensuels et des échanges possibles du facteur travail entre les ménages d’un seul village. En Afrique de l’ouest et plus particulièrement au niveau des zones rurales pauvres comme celle de Bombali, l’échange de biens (notamment alimentaires) et ressources (travail, semence…) est assez fréquent (Spencer D., Deen S., 2009). La non considération de ce type de comportement au niveau de notre travail constitue la deuxième 126 principale limite du modèle qui simule des comportements individuels par ménage sans tenir compte des échanges possibles entre ménages. - Les données mobilisées pour la modélisation : pour cet exercice de modélisation, nous avons mobilisé la base de données élaborée par le JRC. Elle est composée essentiellement de 3 parties : systèmes de culture, production et consommation (annexe 1). Même si cette base de données est assez exhaustive, elle reflète, dans sa conception, la façon avec laquelle les grands centres (type IFPRI, JRC) procèdent pour caractériser les comportements de production et de consommation. En effet, la partie production est renseignée par les entretiens avec les riziculteurs sans forcément créer un vrai lien avec la partie consommation. Par exemple, il serait difficile de connaître, à partir de cette base de données, les parcelles de riz qui ont servi pour la consommation, pour la vente ou encore pour les semences. En effet, il semble que, même si cette base de données a été conçue pour des études pluridisciplinaires, elle servira surtout à étudier de façon séparée les comportements de production de ceux de la consommation. Cela représente un frein important, lorsque nous souhaitons réaliser des études intégrées comme la nôtre. Dans ce registre, la notion d’ « activité » (Dusserre J. et al., 2012) telle qu’appliquée dans cet exercice pourrait être l’élément fédérateur pour une base de données où le lien entre production et consommation serait explicitement définie. Dans cas de figure, à chaque activité définie essentiellement par une culture, une rotation, un écosystème et un niveau d’intensification, est associé un vecteur de production, d’externalité, de consommation, de vente... D’autres types de limites moins structurelles sont également à mentionner. Dans cet exercice, pour déterminer les préférences alimentaires par ménage, nous nous sommes basés sur les données moyennes nationales déterminées par la FAO (OCDI, 2005). Il serait plus judicieux de recueillir ce type d’informations à partir d’enquêtes de terrain pour mieux cerner la diversité des préférences de consommation au niveau d’un territoire. Un autre point qui nous paraît important à mentionner concerne « la quantité considérable de résultats » obtenue par le modèle de ménage. En effet, comme l’atteste l’annexe 1, plusieurs résultats sont obtenus en utilisant ce modèle de ménage. Même si certains indicateurs/variables ont été choisis et calculés en fonction de l’objectif de l’étude, d’autres variables intermédiaires sont générées par le modèle et sont d’une importance capitale pour correctement comprendre les résultats. Néanmoins, même s’il apparaît frustrant de ne pas 127 pouvoir exploiter l’ensemble de ces variables, il serait quasiment impossible (au moins le temps d’une thèse) de toutes les utiliser. D’ailleurs, il serait plus prudent d’établir, avant même de commencer les simulations, une stratégie très claire (calée par rapport à l’objectif de l’étude) permettant de bien interpréter les résultats, sans être noyé sous la quantité considérable de résultats obtenus. Pluviometrie (mm) 600 500 400 300 200 100 0 Dec Dec Jan Jan Fev Fev Mar Mar Avr Mai Juin Jui Pluviometrie Avr Mai Défrichage/ Abattage Riz plateau Manioc Juin B Récolte (n+1) S Aout Jui D Préparation/ plant Sep Oct Aout Sep Oct O Nov Nov Récolte Récolte feuille Trans Palmier à huile Pépin. Riz bas-fond Rep Récolte L B: Brûlis D: Désherbage Figure 39 : Trans: transplantation Rep: Repiquage Pep: pépinière L: labour O: effrayer les oiseaux Répartition mensuelle des principales tâches agricoles pour un ménage rizicole au Nord de la Sierra Leone. La courbe de pluie est une moyenne calculée sur la période 19791990. (élaboration personnelle) 3. Retour sur les choix de production Alors que la production rizicole en Afrique subsaharienne a connu une croissance annuelle de 3,2 % entre 1961 et 2005, la croissance annuelle de la consommation de riz a été de 4,5 % durant cette même période. Selon le Centre du riz pour l’Afrique (ADRAO), le niveau 128 d’autosuffisance en riz en Afrique subsaharienne a connu une baisse, passant de 112 % en 1961 à 61 % en 2006; ce qui veut dire qu’aujourd’hui le continent s’approvisionne sur le marché international du riz pour satisfaire environ 39 % de ses besoins de consommation. Le coût de ces importations s’élève à presque 2 milliards de dollars par an. Ce résultat prouve l’incapacité des riziculteurs à intensifier leur production pour satisfaire des besoins alimentaires croissants. Notre étude confirme ce constat. En effet, les trois scénarios appliqués ne montrent qu’une amélioration partielle de la production et de la consommation en riz. Deux principales causes sont identifiables: - des moyens d’intensification faibles : faute d’un soutien de l’Etat, les riziculteurs en Afrique de l’Ouest, comme d’autres de par le monde (AFD et al., 2011), se sont orientés vers des cultures de rente pour générer du cash et intensifier leur agriculture. Néanmoins, il apparaît très clairement que ce type de culture (le palmier à huile dans notre exercice) vient concurrencer non seulement le riz, mais également d’autres cultures essentielles pour l’équilibre alimentaire des ménages. Cependant, l’instabilité des marchés (représenté par le risque autour des prix dans notre application) fait que, pour l’ensemble des ménages, la surface réservée au palmier à huile n’a augmenté que partiellement au niveau de l’exploitation, et ce malgré la subvention donnée à l’installation de cette culture. En effet, l’instabilité des prix rend très risqué pour le ménage d’avoir une stratégie basée uniquement sur la production de palmier à huile. Pour réduire ce risque lié au marché, les ménages préfèrent toujours produire par eux-mêmes l’ensemble des produits de consommation au lieu de se les procurer au marché. Ce résultat se manifeste pour l’ensemble des scénarios, lorsque la consommation augmente, par l’accroissement de l’autoconsommation. Ce résultat explique pourquoi, aujourd’hui, cette culture (palmier à huile) reste peu présente au niveau des villages ouest africains. Ceci explique également pourquoi cette stratégie d’intensification basée sur les cultures de rente n’a pu se généraliser, bien qu’elle soit proposée et introduite depuis les années 60 (depuis l’époque coloniale) au niveau des pays de l’Afrique de l’Ouest. - des facteurs de production limités : l’histoire de l’agriculture moderne montre que l’intensification des systèmes de production passe essentiellement par l’application conjointe de plusieurs leviers techniques tels que la fertilisation, l’irrigation, de nouvelles variétés, la mécanisation… (Family farming & research; 2014). Cependant, en Sierra Leone (comme dans plusieurs régions ouest africaines) il est souvent question, dans plusieurs initiatives, de réduire 129 l’effet d’un seul facteur limitant (fertilisation, semence améliorés). Cela n’entraînera qu’une amélioration partielle du rendement. Le scénario encourageant l’installation du riz sur les bas-fonds prouve ce type de résultat. En effet, certes, les bas-fonds sont plus fertiles que les plateaux, mais en l’absence d’une main d’œuvre suffisante et de doses de semence faibles (cette liste peut être élargie à d’autres facteurs qui ne sont pas traités dans cette modélisation comme l’irrigation, les traitements phytosanitaires…) les productions restent faibles, et, par conséquent, insuffisantes pour générer assez de cash pour l’achat de légumes au marché (étant donné que les légumes ne peuvent être cultivés que sur les plateaux). Le même type de conclusion pourrait être appliqué au scénario qui prévoit la subvention des semences. Ce genre de résultat est d’autant plus « frustrant » qu’aujourd’hui les efforts de plusieurs initiatives (en Sierra Leone et plus largement en Afrique de l’Ouest) considèrent que le problème majeur de la production rizicole est essentiellement lié à la variété. Les défenseurs de cette idée prennent l’exemple de la production de riz en Asie au cours de la révolution verte des années 1960 et 1970 (Alpert E. et al., 2009). Pour plusieurs de ces auteurs (Fofana M. et al., 2011; Rodenburg J. et al., 2009), si les précédentes tentatives de faire évoluer la production de riz ont échoué, cela ne tient pas à l’approche, ou au mauvais choix des technologies, mais à l’absence de variétés améliorées adaptées aux conditions de culture en Afrique. Toutefois, de nos jours, de nombreuses expériences (y compris en Sierra Leone), (Bah A. A., 2013; Katic P. G., Namara R. E., Hope L., Owusu E., Fujii H., 2013) prouvent que l’utilisation de ces variétés par manque de moyens, d’investissement adéquats, d’un savoirfaire et plus largement sans une vraie politique d’accompagnement et de soutien, ne permet qu’une amélioration partielle des productions. « Malheureusement », aujourd’hui, la diffusion de ces techniques (nouvelles variétés, introduction de l’irrigation…) est liée à l’accroissement des investissements privés dans de nouvelles filières de riz étroitement contrôlées par de grandes entreprises qui s’intéressent uniquement à l’agriculture industrielle et qui s’installent aux dépens des ménages vivriers et, par conséquent, constituent une vraie menace pour l’agriculture paysanne (Lynch K. et al., 2013). Ce constat, tiré de l’expérience des ménages rizicoles asiatiques, est en train de 130 s’installer progressivement au niveau de la Sierra Leone d’après-guerre(Alpert E., Smale M., Hauser K., Bientema N., Pérez J., 2009). En terme de perspective, il serait intéressant de construire des scénarios qui combinent plusieurs mesures, à la fois techniques (nouvelle variété, irrigation, fertilisation…) et socioéconomiques (comme celles testées dans cette étude), et d’évaluer leurs effets sur la production et la consommation. Malgré les limites de notre approche de modélisation, elle présente l’avantage d’être assez représentative des stratégies de production des ménages rizicoles mais également simple pour être facilement extrapolable à d’autres régions en Afrique de l’Ouest. D’ailleurs, il serait très intéressant de reprendre, en partant de travaux existants comme ceux de Tittonell P., Muriuki A., Shepherd K. D., Mugendi D., Kaizzi K. C., Okeyo J., Verchot L., Coe R., Vanlauwe B. (2010), ce type de modélisation pour comprendre et comparer la logique de production des ménages (pas seulement rizicoles), leur résilience face au changement de contexte et les leviers nécessaires pour améliorer la production et la consommation de ces ménages. 131 Références bibliographiques Abdelrasoul A., Childs A., Clerisme J., Dzakuma E., Pittman B., Sogut E. (2013). Pionneering high-yield rice production in Sierra Leone: Recommendations for an outgrower model. The West African Rice Company. 56 p. Acet. (2010). The oil palm value capture opportunity in Africa. African center for economic transformation. 47 p. AFD, CIRAD, FIDA. (2011). Les cultures vivrières pluviales en Afrique de l’Ouest et du Centre. Akabundo I.O. (1991). Les principes de la théorie de l'écologie de la production. Université Agricole de Wageningen (TPE-LUW). Aker C., Block S., Ramachandran V., Timmer C.P. (2010). West African Experience with the world Rice Crisis, 2007–2008. Washington: Center for Global Development. Akintayo I., Cissé B., Zadji D. (2008). Guide pratique de la culture des NERICA de plateau. Contonou, Bénin Centre du riz pour l’Afrique (ADRAO). 39 p. Aldenderfer M.S., Blashfield R.K. (1984). Cluster analysis. Beverly Hills, USA: International Professional. Alpert E., Smale M., Hauser K., Bientema N., Pérez J. (2009). Investir sur les agriculteurs pauvres: une stratégie porteuse. Repenser la manière d’investir dans l’agriculture. Oxfam international. 64 p. Babu S.C., Gajanan S.N., Sanyal P. (2014). Chapter 2 - Implications of Technological Change, Post-Harvest Technology, and Technology Adoption for Improved Food Security— Application of t-Statistic. In: Babu Suresh C., Gajanan Shailendra N., Sanyal Prabuddha (eds). Food Security, Poverty and Nutrition Policy Analysis (Second Edition). San Diego: Academic Press. p. 29-61. BAD. (2010). Résumé du cadre général de la politique de réinstallation et plan d'action de réinstallation de la phase pilote. Banque Africaine de Développement. 29 p. Bah A.A. (2013). Strategies for Promoting Rice Self-sufficiency in Sierra Leone. Journal of Northeast Agricultural University (English Edition), vol. 20, n. 4, p. 78-86. Bahan F., Kéli J., Yao-Kouamé A., Gbakatchétché H., Mahyao A., Bouet A., Camara M. (2012). Caractérisation des associations culturales à base de riz (Oryza sp): cas du CentreOuest forestier de la Côte d’Ivoire. Journal of Applied Biosciences, vol. 56, p. 4118- 4132. Bal A. B. (ed). (2002). Compte rendu de la seconde revue régionale de la recherche rizicole. ADRAO-Le centre du riz pour l’Afrique, Réseau Ouest et Centre Africain du Riz (ROCARIZ). 263 p. 132 Bangura Z.H., Louis M. (2012). Country strategy paper and national indicative programme for the period 2008-2013. European Community. 128 p. Becker L., N'guessan Y. (2004). Le riz dans l'ancienne «Boucle du cacao» de Côte d'Ivoire. Autrepart, n. 31, p. 133-150. Becker M., Johnson D. (1999). Rice yield and productivity gaps in irrigated systems of the forest zone of Côte d'Ivoire. Field Crops Research, vol. 60, n. 3, p. 201-208. Becker M., Johnson D.E. (2001). Improved water control and crop management effects on lowland rice productivity in West Africa. Nutr. Cycl. Agroecosyst, vol. 59, n. 2, p. 119-127. Bidogeza J.C., Berentsen P.B.M. (2009). A typology of farms households for the umutara provence in Rwanda. Food Security, vol. 1, p. 321-335. Bossa A.Y., Diekkrüger B., Giertz S., Steup G., Sintondji L.O., Agbossou E.K., Hiepe C. (2012). Modeling the effects of crop patterns and management scenarios on N and P loads to surface water and groundwater in a semi-humid catchment (West Africa). Agricultural Water Management, vol. 115, p. 20-37. Braima J., Atcha-Ahowé. (2010). Gestion intégrée des nuisibles en production maraîchère: Guide pour les agents de vulgarisation en Afrique de l’Ouest. IITA. 125 p. Brouwer J., Fussell L.K., Herrmann L. (1993). Soil and crop growth micro-variability in the West African semi-arid tropics: a possible risk-reducing factor for subsistence farmers. Agriculture, Ecosystems & Environment, vol. 45, n. 3–4, p. 229-238. Bumb B.L., Johnson M.E., Porfirio A.F. (2011). Policy Options for Improving Regional Fertilizer Markets in West Africa. IFPRI, IFDC. 84 p. Carberry P.S. (2004). Management of rice production systems to increase productivity in the Gambia, West Africa. Ph.D: Faculty of the Graduate School of Cornell University, New York. 178 p. CARD. (2009). National rice development stratégy Sierra Leone (NRDS). 31 p. Carrere R. (2010). Le palmier à huile en Afrique, le passé, le présent et le future. World Rainforest Movement. Ceesay M.M. (2004). Management of rice production systems to oncrease producrtivity in the Gambia, West Africa. Doctor: Cornelle University Faculty of the Graduate School of Cornell University, 178 p. Chen J., Huang Y., Tang Y. (2011). Quantifying economically and ecologically optimum nitrogen rates for rice production in south-eastern China. Agriculture, Ecosystems & Environment, vol. 142, n. 3–4, p. 195-204. CORAF, WECARD. (2008). Strategy for improving agro-processing systems in rice, sorghum/ millet and cassava to enhance marketability in West Africa.Global Food Security Response Initiative. 133 CORAF, WECARD. (2014). Etude de chaine de valeur riz en Sierra Leone. Sierra Leone: Conseil ouest africain pour la recherche et le developpement agricoles. 71 p. CORAF, WECARD, GTZ, IITA. (2003). Le secteur semencier formel en Sierra Leone. 39 p. CRDI. (2013). Fait et chiffres sur l’alimentation et la biodiversité. Centre de Recherches pour le Développement international. CRISTO. (2010). Etude sur la main d'oeuvre agricole au Togo et au Bénin. Rapport 2010. Centre de Recherche et d’Ingénieries Sociales du Togo. 47 p. Deen A.S., Ngaujah. (2010). Characteristic of rice varieties in northern Sierra Leone. Freetown: Entreprise development services. Delarue J. (2007). Mise au point d'une méthode d'évaluation systémique d'impact des projets de développement agricole sur le revenu des producteurs: Etude de cas en région Kpele (République de Guinée). Annexes. Thèse de doctorat: Agro Paris Tech. 58 p. Diagne A., Midingoyi A., Wopereis M., Akintayo I. (2010). Increasing Rice Productivity and Strengthening Food Security through New Rice for Africa (NERICA). In: Chuhan-Pole Punam, Angwafo M (eds). Yes Africa Can: Success Stories from a Dynamic Continent. Word Bank. p. 253-268. Diallo B., Dembélé N., Staatz J. (2012). Compétitivité du riz local en Afrique de l’Ouest depuis la hausse des prix alimentaires mondiaux. PRESAO (Programme de Renforcement et de Recherche sur la Sécurité Alimentaire en Afrique de l'Ouest). 16 p. DIC. (2007). Sierra Leone an community profile. Department of Immigration and Citizenship,Commonwealth of Australia. 29 p. Dingkuhn M., Singh B.B., Clerget B., Chantereau J., Sultan B. (2006). Past, present and future criteria to breed crops for water-limited environments in West Africa. Agricultural Water Management, vol. 80, n. 1-3, p. 241-261. Dixon C.A., Johnson S.D., Fomba S. (2001). Management of iron toxicity in rice in the inland valley swamps in Sierra Leone. In: Audebert A., Narteh L.T., Kiepe P. , Millar D., Beks B. (eds). Management of iron toxicity in rice in the inland valley swamps in Sierra Leone. Africa Rice Center (WARDA). p. 82-95. Dossa L.H., Abdulkadir A., Amadou H., Sangare S., Schlecht E. (2011). Exploring the diversity of urban and peri-urban agricultural systems in Sudano-Sahelian West Africa: An attempt towards a regional typology. Landscape and Urban Planning, vol. 102, n. 3, p. 197206. Dusserre J., Chopart J.-L., Douzet J.-M., Rakotoarisoa J., Scopel E. (2012). Upland rice production under conservation agriculture cropping systems in cold conditions of tropical highlands. Field Crops Research, vol. 138, p. 33-41. Eklou A.S. (2013). Rice in West Africa: challenges and opportunities. Gro Intelligence. 134 Erenstein O., Sumberg J., Oswald A., Levasseur V., Koré H. (2006). What future for integrated rice–vegetable production systems in West African lowlands? Agricultural Systems, vol. 88, n. 2–3, p. 376-394. Everitt B. (1993). Claster Analysis. London: Edward Arnold. In Zhang, N. L. (2002). Hierarchical Latent Class Models for Cluster Analysis. Department of Computer Science Hong Kong, University of Science and Technology. Ewert F., van Ittersum M.K., Bezlepkina I., Therond O., Andersen E., Belhouchette H., Bockstaller C., Brouwer F., Heckelei T., Janssen S., Knapen R., Kuiper M., Louhichi K., Olsson J.A., Turpin N., Wery J., Wien J.E., Wolf J. (2009). A methodology for enhanced flexibility of integrated assessment in agriculture. Environmental Science & Policy, vol. 12, n. 5, p. 546-561. Fall A.A. (2000). Coûts de production et compétitivité du riz au niveau de l'exploitation en afrique de l'ouest et du centre. ISRA Sénégal. 20 p. FAO. (2002). La sécurité alimentaire en Afrique subsaharienne. 105 p. FAO. (2002). Pertes après récolte: un concept mal défini ou mal utilisé. FAO. (2003). Gestion de la fertilité des sols pour la sécurité alimentaires en Afrique subsaharienne. 63 p. FAO. (2011). Stratégie de gestion des risques de catastrophe en Afrique de l’Ouest et au Sahel - FAO (2011 -2013). 52 p. FAOSTAT. (2013). Indices économiques. 9 p. Farooq M., Basra S.M.A., Wahid A. (2006). Priming of field-sown rice seed enhances germination, seedling establishment, allometry and yield. Plant Growth Regulation, vol. 49, n. 2-3, p. 285-294. Farooq M., Wahid A., Lee D.J., Ito O., Siddique K.H.M. (2009). Advances in drought resistance of rice. Critical Reviews in Plant Sciences, vol. 28, n. 4, p. 199-217. Fofana M., Futakuchi K., Manful J.T., Yaou I.B., Dossou J., Bleoussi R.T.M. (2011). Rice grain quality: A comparison of imported varieties, local varieties with new varieties adopted in Benin. Food Control, vol. 22, n. 12, p. 1821-1825. Franke A.C.v.d.B., G.J. Giller, K.E. (2014). Which farmers benefit most from sustainable intensification ? An ex-ante impact assessment of expanding grain legume production in Malawi. European Journal of Agronomy, vol. 58, p. 28-38. Gaiser T., Judex M., Mouinou Igué A., Paeth H., Hiepe C. (2011). Future productivity of fallow systems in Sub-Saharan Africa: Is the effect of demographic pressure and fallow reduction more significant than climate change? Agricultural and Forest Meteorology, vol. 151, n. 8, p. 1120-1130. 135 Gollin D., Rogerson R. (2014). Productivity, transport costs and subsistence agriculture. Journal of Development Economics, vol. 107, p. 38-48. Gomez y Paloma S.A., S Saravia Matus, S Lakoh, A Baudouin, M Hites, G Sammeth, F. (2012). Rural poverty reduction and food security: the case of smallholders in Sierra Leone. Luxembourg: Publications Office of the European Union. 208 p. Graef F., Sieber S., Mutabazi K., Asch F., Biesalski H.K., Bitegeko J., Bokelmann W., Bruentrup M., Dietrich O., Elly N., Fasse A., Germer J.U., Grote U., Herrmann L., Herrmann R., Hoffmann H., Kahimba F.C., Kaufmann B., Kersebaum K.C., Kilembe C., Kimaro A., Kinabo J., König B., König H., Lana M., Levy C., Lyimo-Macha J., Makoko B., Mazoko G., Mbaga S.H., Mbogoro W., Milling H., Mtambo K., Mueller J., Mueller C., Mueller K., Nkonja E., Reif C., Ringler C., Ruvuga S., Schaefer M., Sikira A., Silayo V., Stahr K., Swai E., Tumbo S., Uckert G. (2014). Framework for participatory food security research in rural food value chains. Global Food Security, vol. 3, n. 1, p. 8-15. GRAIN. (2012). Projets agricoles en Afrique de l'Ouest. Haefele S.M., Banayo N.P.M., Amarante S.T., Siopongco J.D.L.C., Mabesa R.L. (2013). Characteristics and management options for rice–maize systems in the Philippines. Field Crops Research, vol. 144, p. 52-61. Hardiman R.T. (1990). Use of cluster analysis for identification and classification of farming systems in Qingyang County, Central Northern China. Agricultural Systems, vol. 33, p. 115125. Harding S.S., Taylor D.R., Jalloh A.B., Mahmood N., Dixon C.A., Johnson S.D. (2012). Evaluation of the Efficacy of Different Rates of Herbicides on Weed Growth and Grain Yield of Two Rice Varieties in Two Rice Ecologies in Sierra Leone. American Journal of Experimental Agriculture, vol. 2, n. 4, p. 607-615. Herrero M., Thornton P.K., Bernués A., Baltenweck I., Vervoort J., van de Steeg J., Makokha S., van Wijk M.T., Karanja S., Rufino M.C., Staal S.J. (2014). Exploring future changes in smallholder farming systems by linking socio-economic scenarios with regional and household models. Global Environmental Change, vol. 24, p. 165-182. Holt C.A., Laury S.K. (2014). Assessment and Estimation of Risk Preferences. In: Marchina M, Viscusi K (eds). Handbook of the economics of Economics of Risk and Uncertainty. Vol.1. Elsevier. p. 135-201. Hoogh I.D., Osiro A., Wiegers E., Fornah D., Sawi M., Rhodes E. (2011). Improving Nutrition through Agriculture: Challenges and Opportunities. Wageningen University and Research Centre. 80 p. Husin L. (2012). The rubber farm household's behavior on economic activities to achieve family food security. Biological & Environmental Engineering Society, p. 169-174. Ignacio A.C., Vyn T.J. (2012). Physiological perspectives of changes over time in maize yield dependency on nitrogen uptake and associated nitrogen efficiencies: A review. Field Crops Research, vol. 133, p. 48–67. 136 IRRI. (2009). Quality seeds. International rice research institute. Iversen V. (2002). Idiosyncrasies of child labour in peasant households in sub-Saharan Africa: anthropological observations and the economics of labour obligations and exchange. Understanding Children's Word. 43 p. Jalloh A. (2006). Farming systems/agricultural land use in Sierra Leone: Land-use Planning for Optimizing Agricultural Production Project. FAO. 53 p. Jolliffe I.T. (1986). Principal Component Analysis. Spinger. 153 p. Kargbo A.B. (2002). Le secteur semencier en Sierra Leone. Semence et Matériel de Plantation en Afrique Occidentale : Bulletin du réseau sur les semences en Afrique occidentale (WASNET), n. 10, p. 21-28. Katic P.G., Namara R.E., Hope L., Owusu E., Fujii H. (2013). Rice and irrigation in West Africa: Achieving food security with agricultural water management strategies. Water Resources and Economics, vol. 1, p. 75-92. Kayombo B., Lal R. (1993). Tillage systems and soil compaction in Africa. Soil and Tillage Research, vol. 27, n. 1-4, p. 35-72. Keating B., Carberry P., Bindraban P., Asseng S., Meinke H., J. D. (2010). Eco-efficient agriculture: Concepts, challenges and opportunities. Crop Science, vol. 109, p. 109-119. Kent R., Johnson D.E., Becker M. (2001). The influences of cropping system on weed communities of rice in Côte d’Ivoire, West Africa. Agriculture, Ecosystems and Environment, n. 87, p. 299-307. Kostrowicki J. (1983). Land-use survey, agricultural typology and land-use systems: introductory remarks. In: Jansen, L. J. M (ed.) Rural Systems: Harmonization of land-use class sets to facilitate compatibility and comparability of data across space and time Chiba (2005). Japan. (n. 1). Kouassi B. (ed.) (2008). Les déterminants de la production agricole en Afrique de l'Ouest. Burkina Faso, Côte d’ivoire, Ghana et Togo. Paris: Karthala. 119 p. Langyintuo A.S., Wilson D. (2005). Characterizing the constraints for the adoption of a Callopogonium mucunoides improved fallow in rice production systems in northern Ghana. Agriculture, Ecosystems and Environment, vol. 110, p. 78-90. Larbi A. (2011). Country Pasture: Forage Resource Profiles. Sierra Leone FAO. Louhichi K., Gomez y Paloma S. (2014). A farm household model for agri-food policy analysis in developing countries: Application to smallholder farmers in Sierra Leone. Food Policy, vol. 45, p. 1-13. Louhichi K., Sergio Gomez y P., Belhouchette H., Allen T., Fabre J., Blanco M., Chenoune R., Acs S., Flichman G. (2013). Modelling Agri-Food Policy Impact at Farmhousehold Level in Developing Countries (FSSIM-Dev). 147 p. 137 Lynch K., Maconachie R., Binns T., Tengbe P., Bangura K. (2013). Meeting the urban challenge? Urban agriculture and food security in post-conflict Freetown, Sierra Leone. Applied Geography, vol. 36, p. 31-39. Maclean J.L., Dawe D.C., Hardy B., Hettel G.P., (eds). (2002). Rice almanac: Source book for the most important economic activity on earth. Third Edition. International Rice Research Institute. 256 p. Madry W., Gozdowski D., Roszkowska-Mądra B., Dąbrowski M., Lupa W. (2005). Diversity and typology of farms accordingto farming system: a case study for a dairy region of podlasie province, Poland. Electronic journal of polish agricultural universities, vol. 12, n. 2, p. 521-531. MAFFS, NARCC, IAR, RRS. (2005). Crop production Gidelines for Sierra Leone Sierra Leone Ministry of agriculture forestry and food security, National agricultural research coordinating council, Institute of agricultural research, Rice research station, Njala university college. 101 p. Mbetid-Bessanee E. (2004). Faiblesse de la main-d’œuvre familiale et diversification des activités dans les exploitations agricoles de la zone cotonnière en Centrafrique. Tropicultura, vol. 22, n. 2, p. 88-92. Melendez C.J. (2006). Maintaining the Yield of Upland Rice under Intensified Land Use in Slash and Burn Systems of West Africa. Thesis: University of Bonn, 70 p. MFMR M. (2004). Agricultural sector review and agricultural development strategy. Main Report. Freetown, Sierra Leone: Ministry of agricultural Forestry and Food Security, Ministry of Fisheries and Marine Resources. Mokuwa A., Nuijten, E. (2014). Processes Underpinning Development and Maintenance of Diversity in Rice in West Africa: Evidence from Combining Morphological and Molecular Markers. PLoS ONE, vol. 9, n. 1, p. e85953. NASDP. (2009). Sierra Leone’s comprehensive African agriculture development programme: Plan 2010-2030. Freetown: National Sustainable Agriculture Development. 46 p. NEPAD, FAO. (2009). Support to NEPAD-CAADP implementation. New Partnership for Africa’s Development (NEPAD), Comprehensive Africa Agriculture Development Programme (CAADP), Food and Agriculture Organization of the United Nations. 51 p. Niangado O. (2010). Varietal development and seed system in West Africa: Challenges and opportunities. Second Africa Rice Congress: Innovation and Partnerships to Realize Africa’s Rice Potential. Bamako, Mali, 22-26. Norman D.W., Worman F.D., J.D S., Modiakgotla E. (1995). The farming systems approach to development and appropriate technology generation. Farm System Management. OCDE. (2011). Enjeux Ouest- africains : Crise rizicole de 2008: chocs et nouveaux enjeux. 12 p. 138 Perrault F., Baldeh Y., Wahome J., Zayid J. (2013). Sierra Leone: Document de strategie pays 2013-2017. Groupe de la banque africaine de developpement, Departement ORWB/SLFO/OSFU. 57 p. Rafflegeau S. (2008). Dynamique d'implantation et conduite technique des plantations villageoise de palmier à huile au Cameroun : facteurs limitants et raisons des pratiques. AgroParisTech. Paris. 148 p. RARC. (2010). Multi-national development of complementary technologies for upland Nerica-Rocuper Sierra Leone Agricultural Research Institute (SLARI). Agricultural research centre. Resimao. (2007). Réseau des systèmes d’information des marchés en Afrique de l’Ouest. RESIMAO. Rodenburg J., Saito K., Kakaï R.G., Touré A., Mariko M., Kiepe P. (2009). Weed competitiveness of the lowland rice varieties of NERICA in the southern Guinea Savanna. Field Crops Research, vol. 114, n. 3, p. 411-418. Rodenburg J.J.Z., S Kiepe, P Narteh , L T Dogbe, W Wopereis, M.C.S. (2013). Sustainable rice production in African inland valleys: Seizing regional potentials through local approaches. Agricultural Systems, vol. 123, p. 1-11. Saito K., Azoma K., Sokei Y. (2010). Genotypic adaptation of rice to lowland hydrology in West Africa. Field Crops Research, vol. 119, n. 2–3, p. 290-298. Saito K., Futakuchi K. (2009). Performance of diverse upland rice cultivars in low and high soil fertility conditions in West Africa. Field Crops Research, vol. 111, n. 3, p. 243-250. Saito K., Futakuchi K. (2014). Improving estimation of weed suppressive ability of upland rice varieties using substitute weeds. Field Crops Research, vol. 162, p. 1-5. Sakane N., Van Wijk M.T., Langensiepen M., Becker M. (2014). A quantitative model for understanding and exploring land use decisions by smallholder agrowetland households in rural areas of East Africa. Agriculture. Ecosystems & Environment, vol. 197, n. 1, p. 59-17. Sakurai S. (2006). Intensification of rainfed lowland rice production in West Africa: present status and potential green revolution. The Developing Economies, vol. 2, p. 232–51. Sammeth F., Lakoh A., Baudouin M., Hites G., Gomez y Paloma S. (2010). Impact of rural poverty reduction strategies. The case of smallholders in Sierra Leone. p. 23. Cape Town, South Africa Conference: African Association of Agricultural Economists (AAAE). Sanoh N. (2013). Migrants en période de crise dans l'espace CEDEAO. 13 p. Sékou D., Condé A. (2011). Etude sur la compétitivité du riz local en Guinée. 24 p. 139 Sene M. (1996). Influence of fallow on sorghum production (Sorghum bicolor (L) Moench, variety CE 14566) in farmer’s fields in Senegal. (Institut Sénégalais des Recherches Agricoles (ISRA) BP 3 120 Dakar - Sénégal). Serpantié G.B., O. (2001). Fertilité et jachère en Afrique de l’ouest. p. 21-83. SLIEPA. (2012). Sierra Leone investment outreach campaign: Opportunities for investors in the oil palm sector. Sierra Leone Investment and Export Promotion Agency. 40 p. Smaling E.M.A., Dixon J. (2006). Adding a soil fertility dimension to the global farming systems approach, with cases from Africa. Agriculture, Ecosystems and Environment, vol. 116, p. 15-26. Spencer D., Deen S. (2009). Economics of rice production in Sierra Leone- Report of a Survey in Three Northern Districts. 19 p. Sultan B., Baron C., Dingkuhn M., Sarr B., Janicot S. (2005). Agricultural impacts of large-scale variability of the West African monsoon. Agricultural and Forest Meteorology, vol. 128, n. 1-2, p. 93-110. Tanaka A., Saito K., Azoma K., Kobayashi K. (2013). Factors affecting variation in farm yields of irrigated lowland rice in southern-central Benin. European Journal of Agronomy, vol. 44, p. 46-53. Therond O., Belhouchette H., Janssen S., Louhichi K., Ewert F., Bergez J.-E., Wery J., Heckelei T., Olsson J.A., Leenhardt D., Van Ittersum M. (2009). Methodology to translate policy assessment problems into scenarios: the example of the SEAMLESS integrated framework. Environmental Science & Policy, vol. 12, n. 5, p. 619-630. Timmer C.P. (2010). Reflections on food crises past. Food Policy, vol. 35, n. 1, p. 1-11. Tittonel P., Giller K. (2013). When yield gaps are poverty traps: The paradigm of ecological intensification in African smallholder agriculture. Field Crops Research, vol. 143, p. 76-90. Tittonell P. (2013). Farming Systems Ecology Towards ecological intensification of world agriculture. Wageningen university. 42 p. Tittonell P. (2014). Livelihood strategies, resilience and transformability in African agroecosystems. Agricultural Systems, vol. 126, p. 3-14. Tittonell P., Muriuki A., Shepherd K.D., Mugendi D., Kaizzi K.C., Okeyo J., Verchot L., Coe R., Vanlauwe B. (2010). The diversity of rural livelihoods and their influence on soil fertility in agricultural systems of East Africa - A typology of smallholder farms. Agricultural Systems, vol. 103, n. 2, p. 83-97. Toulmin C., Guèye B. (2003). Transformation de l'agriculture ouest africaine et rôle des exploitations familiales. 106 p. 140 Tsujimoto Y., Muranaka S., Saito K., Asai H. (2013). Limited Si-nutrient status of rice plants in relation to plant-availableSi of soils, nitrogen fertilizer application, and ricegrowingenvironments across Sub-Saharan AfricaYasuhiro. Field Crops Research, vol. 155, p. 1-9. Usai M.G., Casu S., Molle G., Decandia M., Ligios S., Carta A. (2006). Using cluster analysis to characterize the goat farming system in Sardinia. Livestock Science, vol. 104, n. 1– 2, p. 63-76. Van Duivenbooden N., Pala M., Studer C., Bielders C.L., Beukes D.J. (2000). Cropping systems and crop complementarity in dryland agriculture to increase soil water use efficiency: a review. NJAS - Wageningen Journal of Life Sciences, vol. 48, n. 3, p. 213-236. Van Ittersum M.K., Cassman G.K., Grassini P., Wolf J., Tittonell P., Hochman Z. (2013). Yield gap analysis with local to global relevance-A review. Field Crops Research, n. 143, p. 4-17. Van Wijk M.T., Rufino M.C., Enahoro D., Parsons D., Silvestri S., Valdivia R.O., Herrero M. (2014). Farm household models to analyze food security in a changing climate: A review. Global Food Security, vol. 3, n. 2, p. 77-84. Von braun J. (1988). Effects of Technological Change in Agriculture. Von Food Consumption and Nutrition: Rice in a West African Setting. World Development, vol. 16, n. 9, p. 108-10. WAF, CILSS, FAO, CIRAD, Fewsnet. (2011). Crise rizicole, évolution des marchés et sécurité alimentaire en Afrique de l'Ouest. 61 p. Wakatsuki T.N., J.C Igwe, C.A. (2008). Evaluation of sawah rice management system in an inland valley in southeastern Nigeria: Soil chemical properties and rice yield. Paddy Water Environment, vol. 6, n. 3, p. 299-307. Ward J.H. (1963). Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association, vol. 58, p. 236-244. Webber H., Gaiser T., Ewert F. (2014). What role can crop models play in supporting climate change adaptation decisions to enhance food security in Sub-Saharan Africa? Agricultural Systems, vol. 127, p. 161-177. Whitbread A.M., Robertson M.J., Carberry P.S., Dimes J.P. (2010). How farming systems simulation can aid the development of more sustainable smallholder farming systems in southern Africa. European Journal of Agronomy, vol. 32, n. 1, p. 51-58. Winter-Nelson A. (1997). Rural taxation in Ethiopia, 1981–1989: a policy analysis matrix assessment for net producers and net consumers. Food Policy, vol. 22, n. 5, p. 419-431. Worou O.N., Gaiser T., Saito K., Goldbach H., Ewert F. (2012). Simulation of soil water dynamics and rice crop growth as affected by bunding and fertilizer application in inland valley systems of West Africa. Agriculture, Ecosystems & Environment, vol. 162, p. 24-35. 141 Yamah A. (2002). The practice of the system of rice intensification in Sierra Leone. In: Uphoff N. Fernandes E.C.M., Longping Y., Jiming P., Rafaralahy S., Rabenandrasana J. (eds) (ed.) Assessments of the system of rice intensification (SRI).Proceedings of an international conference held in Sanya, China, April 1-4, 2002. Ithaca, NY: CIIFAD-SRI. (n. 103-105). Yemadje R.H., Crane T.A., Visso P.V., Mongb R.L., Richards P., Kossou D.K., Kuyper T.W. (2012). The political ecology of land management in the oil palm based cropping system on the Adja plateau in Benin. Wageningen Journal of Life Science, vol. 60, n. 63, p. 91-99. 142 Annexe_1 : Structure de la base de données Questionnaires _A : 197 lignes Questionnaires A Number Name of the farmer Name of the enumerator Region/Province District Chiefdom Section Village Household number Gender composition (M=Male; F=Female) Age composition Educational level Occupation of members (F= Fulltime; P=Part-time) Major economic activities (A=Agri; PT= Petty trading; PE= Paid Employment; ASE= Artisan self employment) Sources of Income (Y=Yes; N=No) Most important source of income (1= most important; 4=least important) Access to staple food Debt (1=the most important; 2=the second most important) Local Governance Social/Group Leadership Membership of Farmer Association Or Groups (Y=Yes; N=No) Perceived Benefits (G=Great; M=Medium; L=Little; N=None) Organizations you are working with (Y=Yes; =No) Kind of support received (Y=Yes; =No) Adequacy of kind of support (A=Adequate; N=Not Adequate; =Not Provided) Impact of support on food security (GI= Great Improvement; LI= Little Improvement; NI= No Improvement) Level of Improvement in production (GI= Great Improvement; LI= Little Improvement; NI= No Improvement) Level of Improvement in Income (GI= Great Improvement; LI= Little Improvement; NI= No Improvement) Sustainability (Y=Yes; N=No; D=Don't know) Improvement during last 2 years? (B=Better; S=Same; Worse) Type of Crops Grown and area cultivated last year (in number of acres) The Most Important crop (1= the most important) Livestock Type (and number of each type) Rainfall (N=Normal;R=Rainy;Y=Yes;N=No) Soil condition Use of the following in the field? Seeds (Y=Yes; N=No; B=Bad; A=Average; G=Good) Maize Storage Change in planting area Condition of the physical facilities Condition of communications opportunities (G=good; P=Poor; NE=None Exist; Y=Yes; N=No) 143 Questionnaires_B : 197 lignes - Labour input Activity 1. Nursery establishment Hired labour (Man days) Family Labour (Man days) 2. Bush Clearing 3. Land preparation 4. Planting 5. Weeding Hired labour (Man days) Family Labour (Man days) Hired labour (Man days) Family Labour (Man days) Hired labour (Man days) Family Labour (Man days) Hired labour (Man days) Family Labour (Man days) 6. Pruning Hired labour (Man days) Family Labour (Man days) 7. Spraying Hired labour (Man days) 8. Staking/stoking Hired labour (Man days) Family Labour (Man days) 9. Harvesting Hired labour (Man days) Family Labour (Man days) 10. Grading Hired labour (Man days) Family Labour (Man days) 11. Fermenting Hired labour (Man days) Family Labour (Man days) 12. Drying Hired labour (Man days) Family Labour (Man days) 13. Fencing Hired labour (Man days) Family Labour (Man days) 14. Miling Hired labour (Man days) Family Labour (Man days) 15. Marketing Hired labour (Man days) Family Labour (Man days) 16. Other Hired labour (Man days) Family Labour (Man days) 17. Under Brushing Hired labour (Man days) Family Labour (Man days) 18. Processing Hired labour (Man days) Family Labour (Man days) Household Number Name of the farmer Village Name of tree crop 1 Field size (acres) Units Male Female Total Male Female Children Total days Area Labor cost per day (Le) Male Male Male Male Male Male Male Female Total Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) Male Male Area Labor cost per day (Le) 144 - Labour Input - Food crops Activity 1. Brushing/Felling/Clearing Hired labour (Man days) Family Labour (Man days) 2. Brushing and Mounding Hired labour (Man days) Family Labour (Man days) 3. Plowing and seeding Hired labour (Man days) Family Labour (Man days) 4. Harrowing Hired labour (Man days) Family Labour (Man days) 5. Planting of minor crops in the mix Hired labour (Man days) Family Labour (Man days) 6. First bird scaring Hired labour (Man days) Family Labour (Man days) 7. Puddling Hired labour (Man days) Family Labour (Man days) 8. Transplanting Hired labour (Man days) Family Labour (Man days) Male Female Total Male Female Children Total days Area Labor cost per day (Le) Male Female Total Male Female Children Total days Area Labor cost per day (Le) Male Female Total Male Female Children Total days Area Labor cost per day (Le) Specifications Male Female Total Male Female Children Total days Area Labor cost per day (Le) Specifications Male Female Total Male Female Children Total days Area Labor cost per day (Le) Specifications Male Female Total Male Female Children Total days Area Labor cost per day (Le) Specifications Male Female Total Male Female Children Total days Area Labor cost per day (Le) Specifications Male Female Total Male Female Children Total days Area Labor cost per day (Le) Specifications 145 - Physical Inputs 7.5.1 Individual Items 1.Hoes 2. Cutlasses 3. Slathers 4. Harvesting Knives 5. Tractor Quant Unit Cost Age Life Expectancy Total Cost Quant Unit Cost Age Life Expectancy Total Cost Quant Unit Cost Age Life Expectancy Total Cost Specifications Quant Unit Cost Age Life Expectancy Total Cost Quant Unit Cost Age Life Expectancy Total Cost Specifications 146 - Other inputs 1. Seed Rice Quant Unit Unit Cost Age Total Cost Quant Unit Unit Cost Age Total Cost Quant Unit Unit Cost Age Total Cost Quant Unit Unit Cost Age Total Cost 2. Rice Seedlings 3. Cassava Cuttings Other Inputs 4. Beans/Vegetable Seeds 5. Others Quant Unit Unit Cost Age Total Cost - Yield/Output Food crops 8.2. Yield/Output Food crops Crop Type 1. Upland Rice 2. IVS/Swamp Rice 3. Cassava 4. Ground Nuts 5. Beans/vegetables 1. Rice - upland Size/acres Size composition Unit Quantity Quantity Consumed Quantity Sold Quantity stocked Unit Unit cost Total Income Notifications Size/acres Size/acres Size/acres Size/acres Estimated Loss in percent 147 - Yield Losses 1. Rice - upland 2. Rice - lowland 3. Cassava 4. Ground Nuts 5. Beans/vegetable Estimated Loss in percent Pre-Harvest if unit Post Harvest if unit Principal cause Other Don't know Estimated Loss in percent Estimated Loss in percent Estimated Loss in percent Estimated Loss in percent 148 Annexe_2 : les grandes étapes de l’itinéraire technique à Bombali. Cette description s’est basée sur une revue de bibliographie mais aussi essentiellement sur des entretiens effectués avec les riziculteurs lors de mon séjour dans la zone d’étude. Le défrichage et l’abattage d’une parcelle représentent la première activité, cette tâche est réalisée par les hommes en fin de saison des pluies (à partir de novembre- décembre) et peut se prolonger jusqu’au mois de mai, soit un mois avant les premières pluies de mars. L’agriculteur commence par couper les arbustes et débroussaille la parcelle souvent laissée au préalable en jachère. Cette opération est faite par des outils rudimentaires avec une machette et une hache. Ces arbres sont laissés sur la parcelle puis éventuellement éliminés en les brûlant. La parcelle de jachère longue est souvent mal nettoyée laissant les troncs d’arbres sur place et non calcinés. La date de cette opération est variable selon le type de relief et les variétés sélectionnées. Le défrichage du riz plateau commence souvent au mois de novembre et peut se prolonger jusqu’au mois de mars, tandis que, sur les bas-fonds, le défrichage est réalisé vers la fin mai. En terme d’organisation, l’agriculteur commence par défricher à partir de mois de novembre sur les plateaux. Sur les bas-fonds, souvent inondés pendant une longue période de l’année, le défrichage est moins pénible et plus rapide que sur les plateaux. Ainsi, il est souvent réalisé au mois de mai afin de bien préparer les sols au repiquage du riz au mois de juillet. Le brûlis est la deuxième étape de l’itinéraire technique. Cette action a pour objectif de mettre à la disposition de la culture de riz des résidus de la jachère « la flore de la friche ». Cette étape est réalisée uniquement pour le riz plateaux car,après une jachère, une accumulation importante de biomasse est souvent observée, contrairement au sol des bas-fonds où la jachère n’est pas appliquée. Cette tâche est réalisée avant la période des pluies qui commence en mars, pour que les cendres des végétaux ne soient pas lessivées. D’après les agriculteurs du village, en cultivant sur des friches de longue durée (plus de 10 ans), les rendements sont faibles par rapport aux friches de durée moyenne (6 à 7ans). Ils estiment que la biomasse produite sur des friches de longue durée n’arrive pas à être brûlée correctement. Par conséquent, des éléments grossiers (tronc) restent sur le sol et pourrissent provoquant ainsi de multiples maladies. Par contre, lorsque la friche arborée ou herbacée est plus jeune (moins de 7ans), les éléments minéraux sont vite brûlés et stockés au sol. 149 Le semis est assuré par les femmes, et éventuellement par les enfants. Cette étape a lieu entre le mois de juin pour les sols plateaux et peut être prolongée jusqu’en juillet sur les bas-fonds et les Bolilands. Ainsi, la date de semis du riz plateau doit coïncider avec les premières pluies pour que le sol soit suffisamment humide et la levée du riz soit assurée. Sur les bas-fonds, le semis se fait souvent en juillet en passant par la transplantation des plants qui sont multipliés en pépinière au mois d’avril. Avant la transplantation, un labour superficiel, à l’aide d’une houe (puddling), se fait sur les bas-fonds inondés en permanence. La quantité moyenne de semence de riz nécessaire pour un hectare est estimée à 50 kg/ha. Cette semence est souvent auto produite par l’agriculteur (multiplication traditionnelle) en conservant une partie de la production de l’année précédente. Dans certains cas, les agriculteurs se procurent des semences améliorées (exp Nerica) auprès des ONG et des associations locales. Dans la pratique, l’agriculteur cultive des variétés locales principalement Rok 14 qui a en moyenne un cycle de 140 jours. Par ailleurs, pendant la saison sèche, les agriculteurs cultivent pendant trois mois (janvier-mars) du maïs ou du piment sur les bas-fonds selon la disponibilité de la main d’œuvre (Rhodes, 2005). Sur les plateaux, les agriculteurs cultivent en association jusqu’à six cultures différentes (on parle souvent de parcelle « super marché») (riz, manioc, sorgho, maïs, sésame). Dans ce type de système, les cultures mixtes et associées sont appliquées. Les cultures mixtes représentent un mélange de graines avec le riz (principalement le sésame et le sorgho). Ce mélange est semé en même temps entre le mois d’avril et juin. Nous retrouvons également les cultures dites associées comme le manioc qui est planté en mai, deux semaines avant le semis du riz plateau. La distance entre deux plants de manioc est en moyenne d’un mètre. La rotation appliquée par les agriculteurs se fait pendant la période humide, le sol est occupé par le riz, puis en saison sèche, on trouve du manioc ou de l’arachide en monoculture. Etape 4 : Le désherbage est généralement réalisé une seule fois, deux à trois semaines après le semis. Cette opération réalisée manuellement est souvent assurée par les femmes et les enfants. C’est la tâche la plus consommatrice en temps et la plus coûteuse,quand elle est assurée par la main d’œuvre non familiale. En revanche, à partir du stade laiteux du grain (mi juillet-août), les enfants et les personnes âgées surveillent les parcelles contre les attaques d’oiseaux. 150 Etape 5 : La récolte commence au mois d’août et peut se prolonger jusqu’au mois de décembre selon les variétés qui peuvent durer de 90 à150 jours (précoce, tardive) et le type de sol. En effet, les variétés cultivées sur les plateaux sont souvent précoces (août-septembre) par rapport aux variétés tardives cultivées sur les IVS (novembre-décembre). La récolte est effectuée généralement par les femmes, qui coupent la tige à une vingtaine de centimètres sous l’épi à l’aide d’un couteau. Les épis sont mis en botte sur les troncs d’arbre défrichés auparavant à une distance de 1 m du sol pour sécher le grain. A la fin de la récolte, les femmes collectent les bottes de riz et procèdent au battage et vannage pour récolter les grains de riz. Les grains sont étalés sur un plastique ou sur une dalle en béton de 80*100 m pour le séchage. D’après les agriculteurs, cette technique de récolte génère une perte de grains estimée à 30% de la récolte totale au champ. L’agriculteur donne la priorité au riz plateau, il représente la première production de l’année pour satisfaire les besoins familiaux (Bahan F., Kéli J., Yao-Kouamé A., Gbakatchétché H., Mahyao A., Bouet A., Camara M., 2012). Par ailleurs, le riz IVS est également cultivé en parallèle avec le riz plateau dans le cas où les besoins en terres, en semences et en main d’œuvre sont disponibles, c’est pourquoiune concurrence de main d’œuvre s’observe au moment du désherbage du riz plateau et du repiquage du riz IVS. 151 Annexe_3 : Modèle bio économique (Household model) objectif .. uu-phi*ectype=E=U; utilmoy .. Z +(sum(prc,autoconso(prc)*prix(prc)))- sum((mo,p),trslr(mo,p))*slr -sum((c,sol,t,j),couts(c,sol,t)$ct(c,sol,t,j)*x(c,sol,t,j)) -smc*psmc*pss=E=UU; recette .. sum(pr,vente(pr)*prix2(pr))=E=Z; recette1(em) .. sum(pr,vente(pr)*PXAL(PR,EM))=E=Z2(em); prodd(pr).. sum((c,sol,t,j)$ct(c,sol,t,j),x(c,sol,t,j)*proo(pr,c,sol,j,t))=E= production(pr); prodd1(c,pr).. sum((sol,t,j)$ct(c,sol,t,j),x(c,sol,t,j)*proo(pr,c,sol,j,t))=E= production1(c,pr); proddal(pr,am).. sum((c,sol,t,j)$ct(c,sol,t,j),x(c,sol,t,j)*PM(pr,c,sol,j,t,am))=E= prodal(pr,am); utilal(EM,am).. Z2(em)+ (sum(prc,autoconsoal(prc,am)*prix(prc)))sum ((mo,p),trslr(mo,p))*slr -sum((c,sol,t,j),couts(c,sol,t)$ct(c,sol,t,j)*x(c,sol,t,j)) -smc*psmc*pss=E=Ual(am,em); risque1(am,em)..uu-ual(am,em)=e=dev(am,em); ecart..sqrt(sum((am,em),sqr(dev(am,em)))/(card(am)*card(em)))=e=ectype; distrib2(pr).. production(pr)-stock(pr)-vente(pr)=E=autoconso(pr); consomcont(prc) .. conso(prc) =E= autoconso(prc)+ achat(prc); terre(sol) .. sum((c,t,j)$ct(c,sol,t,j),x(c,sol,t,j))=L= dterre(sol); terre2 .. sum ((cr,sol,t),x(cr,sol,t,'j1'))=L= sum ((sol,t),x('jach',sol,t,'j1'))/6; terre3 .. sum ((cr,sol,t),x(cr,sol,t,'j2'))=L= sum ((sol,t),x('jach',sol,t,'j2'))/8; semences.. sum((cr,sol,t,j),semm(cr,sol,j,t)$ct(cr,sol,t,j)*x(cr,sol,t,j))*perte=E=stock('grain')+smc; contsem .. smc*psmc*pss =L= cashsem*CA ; cash .. Z-sum((c,sol,t,j),couts(c,sol,t)$ct(c,sol,t,j)*x(c,sol,t,j))-sum((mo,p),trslr(mo,p))*slr -smc*psmc*pss =E=CA; achatlim .. sum(pr,achat(pr)*prix2ini(pr))=L= cashachat*CA; prodcalconv .. sum(pr,conso(pr)*conversion(pr))*10000=E= calories*365*10.25; coef .. sum(prc,partini(prc)*lprix2(prc))-sum(prc,partini(prc)*lprix2ini(prc))=E= A; consom(prc) .. consoini(prc)+(uu/prix2(prc))*(beta(prc)*((log(uu)-luuini)+beta(prc)*A)) +gamma(prc)*(lprix2(prc)-lprix2ini(prc))=E= conso(prc); 152 travaill(mo,p) .. sum((c,sol,t,j),x(c,sol,t,j)*bt(c,sol,t,mo,j,p))=l= travail(mo,p)+trslr(mo,p); travaill1(c,sol,t,j).. sum((mo,p),bt(c,sol,t,mo,j,p))=e=bt1(c,sol,t,j) ; Annexe_4 : Politique agricole et rurale en Sierra Leone a. Effective Policy Framework and Capacity Building • Re-organisation and training of agriculture staff and other relevant service providers. • Farmer empowerment through local capacity building to organise themselves, effectively express their demands for the various support services they require and take steps towards a more commercially oriented approach to agriculture. • Promotion of matching grant – for small investment schemes. b. Increasing Food Production: • Rehabilitation of the infrastructure necessary for the production and distribution of good quality planting material. • Promotion of improved rice cultivars and alternative root crops to fill the hunger gap. • Development of a localised community based system of seed multiplication on a semicommercial basis, which could also encourage entry of the private sector into this market. • Mechanisation programme support in order to cultivate suitable large land areas in low land ecologies to bring about accelerated productivity using improved farming practices to achieve increased rice production, and to encourage youths back into the rural sector. It should be geared towards strengthening private investment in machine hire services. • Facilitate access to and management of credit from bank(s). • In the livestock sector, rehabilitation of laboratories and veterinary clinics as well as of abattoirs. • The fisheries sector needs to boost the Monitoring, Control and Surveillance (MCS) activities under the Artisanal Fisheries Development Project for Sierra Leone to protect its marine fisheries resources and bring about increased fish supplies to the domestic and export market and contribute to poverty reduction. c. Improving rural services: • Extension services: create a demand driven semiautonomous extension system to take over from the current dysfunctional government system. This will build on initiatives such as the present Farmer Field Schools, to involve a number of different stakeholders: the private sector, NGOs and research and training institutes. 153 • Research system: restructure by bringing all research programmes under a unified management structure, the Sierra Leone Agricultural Research Institute, to develop and implement a National Collaborative Agricultural Research Programme. Research will be primarily adaptive, and respond to the needs expressed by farmers in the areas of crop production, livestock, fisheries, post-harvest technology, natural resource management and agricultural policy. • Rural finance: restructure so as to ensure access by farmers, both men and women, to provide credit for input use and seasonal credit. Credit will also be required for equipment purchase, and credit lines should be organised for farmer’s association and small-scale farmer groups, and to encourage the participation of other private sector investors in various aspects of the livestock industry. These service delivery programmes should be developed in collaboration with farmers. d. Rural Infrastructure: • Roads, this programme will be undertaken by central and district roads, transportation and works departments as part of an expected wider, national intervention to bring the country up to a recognizable international standard of provision. Maintenance of all but the major roads will be the responsibility of district councils, and it is necessary that suitable funds are made available for this purpose. • Irrigation and drainage works rehabilitation, and handling and storage facilities; it is recommended that these should be implemented through districts, chiefdoms and farmer groups, to be financed mainly by loans, with a significant beneficiary contribution. Towards the end of 2009, the GoSL published The National Sustainable Agriculture Development Plan (NSADP) which follows from the second generation PRSP, the Agenda for Change and the Vision of commercializing agriculture, forestry, fisheries and livestock through linking small to large farmers to market economies (NSADP, 2009). The Plan therefore not only reorganises precedent efforts and policies in the agricultural sector but sets four specific major sub-programmes that have the specific objective of increasing agriculture sector growth from 2% to 6% per annum by 2015. As with past documents, the NSADP is in line with the first Millennium Development Goal and World Food Summit targets (NSADP, 2009). The four subprogrammes include: • Commercialisation of key commodities including a small-holder commercialization scheme and medium and large farm producer’s promotion scheme. 154 • Agriculture infrastructure with focus on rehabilitation and upgrading of feeder roads, development of irrigable swamps, rehabilitation and modernisation of storage and processing facilities and rehabilitation and construction of research centres and MAFFS/MFMR facilities. • Private sector promotion which will focus on policies and legislation to encourage sustainable domestic and international investments in the agricultural and fisheries sector. • Sector coordination and management to improve transparent, efficient and effective sector coordination and management. The latter is accompanied by a change in cultivation methods for the major agricultural areas of Sierra Leone: “A gradual shift will be encouraged from damaging and low yielding but diversified slash and burn upland rice systems towards more stable perennial and tree crops with inter-planting of rice and diverse crops including livestock” (NSADP, 2009). Simultaneously the government intends to promote in the uplands the cultivation of legumes that allow improving the quality of the soil while fixing nitrogen and enhancing its fertility. For the cultivation of tree crops, the government plans to address long-term land security issues and lease holding payments to communities. In the case of the inland valley rice system (IVS), the introduction of water control structures and cropping systems that have both rice and legumes are envisaged in the NSADP. For this purpose, non-photoperiod sensitive varieties will be emphasised for double cropping (rice-rice and rice-legume). In this area, land security issues will also be resolved mainly to foster the long term investment in water control structures. - Agricultural taxation and tariff policies According to Jalloh (2006) the key features in the Income Tax Act and Tariff Regime with relevance to agriculture for import duties may be summarised as follows: • Lower duty rate of 5% on raw materials and inputs, capital goods and social products including all basic educational materials, pharmaceutical products for primary health care and agricultural machinery; • Import duty rate of 20% for immediate and 30% for final goods as defined in the tariff; • Duty draw back system for imported inputs and all exports; • Elimination of export taxes for exportoriented industries; • Zero duty rate on imports of raw materials for industries with a market share of 60% or more for that product; • Sales tax rate of 20% on all imports, except capital goods; 155 • Domestic sales tax of 20% on domestic output. However, companies with turnover of less than. Le 200 million are exempt from paying domestic sales tax on outputs; these companies are instead required to pay sales tax on only imported inputs; • Import duty on rice is 15%. For income tax they are: • Reduced corporate tax of 35% is payable by all companies; • Income earned from rice farming is exempt from tax for a period of 10 years from the date of commencement of the activity for both incorporated and unincorporated businesses; • The threshold for income tax on employment income is Le 1 million, while the top marginal rate of tax for employees, the selfemployed and property owners is 35%, which applies to most small scale farmers; • Payment of payroll tax for foreigners currently ranges from Le 250,000 – Le 1 million; • The amount of investment allowance to be deducted from business income is 5% of the cost of the relevant asset;57 • Repatriation of after tax profits or dividends is subject to the payment of withholding tax of 10%; • Repatriation of original loan or interest payment thereon, known-how fees and other services at the exchange rate prevailing at the time of repatriation; • Capital allowance deduction is allowed for depreciation of a taxpayer’s depreciable assets. The most significant effect of the tariff regime is an excise tax on imported fuel of 50 %, the landed cost of petrol and 41 % for diesel. This affects tradable costs of all inputs, as well as marketing and processing costs. Materials directly related to the production in all the subsectors (crops, livestock, fisheries and forestry), however, face reduced import duties. Specific machinery (tractors, and appliances, harvesters, veterinary drugs and implements) can be imported at a lower duty rate of 5 %, compared to 50 – 100 % for luxury cars. Concerning protectionist measures, as stated above, rice is subject to 15 % import duty. Nonetheless, (at the moment of writing) with the ongoing negotiations for a common external tariff harmonisation for the ECOWAS26 and UEMOA27 regions, the government would no longer be able to influence tariffs on both agricultural inputs and outputs. - Rice: Government support and intervention The government through the Ministry of Trade, the Rice Department, the Rice Corporation, and the Sierra Leone Produce Marketing Board (SLPMB) has been deeply involved in rice trading. According to the report issued by IFPRI (2009) the local rice operation of the SLPMB was rather unsatisfactory as less and less local rice was sold. The Economic 156 Community of West African States (ECOWAS) is a regional group of fifteen West African countries The West African Economic and Monetary Union (also known as UEMOA from its name in French, Union économique et monétaire ouest-africaine) to it because of the unattractive floor prices set by the government for local rice. Starting in 1987, the monopoly of SLPMB in the rice trade declined considerably when its foreign reserves from cocoa and coffee trade plummeted due to low world prices; it thus ceased operation. Consequently, by the late 1980s, the private sector had assumed a dominant role in the marketing of both local and imported rice in Sierra Leone. However, the estimated proportion of local rice that has been marketed since the 1980s has averaged about 20 percent of annual production, implying that most of the rice produced locally is consumed by farm households (IFPRI, 2009). Since the war, the marketing of rice in Sierra Leone has not fundamentally changed, except that the volumes of local rice production are muchlower and of reduced quality (given the absence of sufficient milling facilities28) now than before the war. In fact, although the rate of self-sufficiency in rice increased in Sierra Leone between 2002 and 2007, the country still imports substantial proportion of rice, which has increased from 120 000 tons per annum during the pre-war period to 230 000 tons post war, a quantity short of the national requirement of 530 000 tons of milled rice per annum (MAFFS, 2009 & FAOSTAT). The market structure for rice and other major agricultural commodities sold by farmers in Sierra Leone (i.e. maize, cassava, groundnuts, and vegetables) generally follows a producerwholesaler- retailer-consumer pattern. - Sector Constraints and Challenges Although Sierra Leone is naturally endowed with adequate land, water and climatic conditions (to enable the agricultural sector to contribute to high economic growth and food security) the national context is one of the most 28 In 2004, a total of 53 small scale rice mills existed in Sierra Leone. 60% of these mills were located in the Northern region (National Rice Development Strategy. Sierra Leone, 2009) severely deprived in Sub-Saharan Africa in terms of institutional facilities (FAO, 2005). The majority of the farmers in Sierra Leone operate as basic subsistence food production units, who, for the most part, do not use improved techniques and inputs. Crop production is thus characterised by low yields and productivity insufficient to supply the food requirements of their farm households. The slow growth of the sector (all crops included) may therefore be attributed to the interplay of several factors influencing farmers’ behaviour and farm productivity. 157 Annexe_5: Résultats des scénarios - Scénario palmier à huile_classe_1 Classe_1 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION TECHN TECHN TECHN ACHAT ACHAT TOT SEMENCES TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 autoCONSO autoCONSO autoCONSO autoCONSO autoCONSO TOTAL grain leg huile pd maniocc TOTAL TOTAL CONSO VENTES leg leg huile huile pd pd pd maniocc maniocc TOTAL grain leg huile pd maniocc grain leg huile pd maniocc dens1 dens2 dens3 grain SEM SEM UTILSATION JOURS J t t t t t t LEONES LEONES LEONES t LEONES t LEONES t t LEONES t LEONES t t t t t t t t t t t ha ha ha t t t t j ha J JOUR rizleg manioc rizpat palm legume papate douce jach ha ha ha ha ha avers rizleg rizleg manioc rizpat rizpat palm grain leg huile pd maniocc ha PERSONNE haut haut basfond haut haut rizleg manioc rizpat palm jach phi grain leg maniocc grain pd huile t t t t t 0% it1 60 0,76 0,09 0,17 0,08 0,54 1,64 3586406 2198079 3651363 0,83 131119 0,13 202439 0,00 2,39 3754762 1,93 429882 5,28 0,98 0,05 0,10 0,04 0,34 0,64 0,92 0,30 0,00 2,47 0,65 1,03 0,12 0,10 0,10 0,07 319 1,03 6,16 0,17 1653 1,03 0,82 0,65 1,16 0,00 0,00 6,16 1,03 0,82 0,65 1,16 6,16 0,15 0,35 0,92 2,47 0,29 0,00 0,30 0,64 0,09 0,17 0,08 0,54 20% it2 60 0,76 0,09 0,17 0,08 0,54 1,64 3589125 2200961 3654949 0,83 130187 0,17 268880 0,00 2,39 3754762 1,65 367533 5,04 0,98 0,05 0,10 0,04 0,34 0,64 0,92 0,34 0,00 2,19 0,65 0,98 0,03 0,12 0,10 0,10 0,08 318 1,02 6,10 0,17 1654 1,02 0,73 0,65 1,32 0,00 0,00 6,10 1,02 0,73 0,65 1,32 6,10 0,15 0,35 0,92 2,19 0,29 0,00 0,34 0,64 0,09 0,17 0,08 0,54 40% it3 60 0,76 0,09 0,17 0,08 0,54 1,64 3592236 2205148 3660167 0,81 127398 0,22 344501 0,00 2,39 3754762 1,69 376703 5,11 0,98 0,05 0,10 0,04 0,34 0,64 0,90 0,39 0,00 2,23 0,65 0,87 0,11 0,12 0,11 0,11 0,08 326 0,99 5,93 0,17 1656 0,99 0,74 0,65 1,51 0,00 0,00 5,93 0,99 0,74 0,65 1,51 5,93 0,15 0,35 0,90 2,23 0,29 0,00 0,39 0,64 0,09 0,17 0,08 0,54 60% it4 60 0,76 0,09 0,17 0,08 0,54 1,64 3595784 2210388 3666716 0,79 123866 0,28 429206 0,00 2,39 3754762 1,82 406543 5,28 0,98 0,05 0,10 0,04 0,34 0,65 0,88 0,45 0,00 2,36 0,65 0,74 0,22 0,12 0,12 0,12 0,09 337 0,95 5,71 0,17 1658 0,95 0,79 0,65 1,72 0,00 0,00 5,71 0,95 0,79 0,65 1,72 5,71 0,15 0,35 0,88 2,36 0,29 0,00 0,45 0,65 0,09 0,17 0,08 0,54 80% it5 60 0,77 0,09 0,17 0,08 0,54 1,65 3599833 2217006 3675014 0,76 119262 0,34 528004 0,00 2,39 3754762 2,08 464311 5,57 0,98 0,05 0,10 0,04 0,34 0,65 0,85 0,51 0,00 2,62 0,65 0,55 0,35 0,12 0,13 0,13 0,10 353 0,90 5,43 0,17 1662 0,90 0,87 0,65 1,96 0,00 0,00 5,43 0,90 0,87 0,65 1,96 5,43 0,15 0,36 0,85 2,62 0,29 0,00 0,51 0,65 0,09 0,17 0,08 0,54 100% it6 60 0,77 0,09 0,17 0,08 0,54 1,65 3604484 2225591 3685821 0,72 112985 0,42 649635 0,00 2,39 3754762 2,53 563963 6,06 0,98 0,05 0,10 0,04 0,34 0,65 0,81 0,59 0,00 3,07 0,65 0,31 0,53 0,12 0,15 0,15 0,11 376 0,84 5,04 0,17 1666 0,84 1,02 0,65 2,27 0,00 0,00 5,04 0,84 1,02 0,65 2,27 5,04 0,15 0,36 0,81 3,07 0,29 0,00 0,59 0,65 0,09 0,17 0,08 0,54 158 - Scénario palmier à huile_classe_2 Classe_2 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION TECHN TECHN TECHN ACHAT ACHAT ACHAT ACHAT TOT SEMENCES TRSAL CR CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 autoCONSO autoCONSO autoCONSO autoCONSO 0% it1 30,00 0,43 0,04 0,04 0,08 0,62 1,21 1934948 1491820 1978995 0,55 85737 TOTAL J grain t leg t huile t pd t maniocc t TOTAL t TOTAL LEONES CONSO LEONES VENTES LEONES leg t leg LEONES huile t huile LEONES pd t 0,00 pd t 1,29 pd LEONES 2023487 TOTAL t 1,83 grain t 0,47 leg t 0,05 huile t 0,05 pd t 0,04 maniocc t 0,36 grain t 0,43 leg t 0,59 huile t 0,04 pd t 0,00 maniocc t 0,62 dens1 ha 0,36 dens2 ha 0,78 dens3 ha grain t leg t maniocc t SEM t 0,04 SEM t 0,04 UTILSATION t 0,03 JOURS j 169 j1 j2 j 0,78 ha ha 6,27 J 0,12 JOUR PERSONNE 889 rizleg haut 0,78 manioc haut 0,21 rizpat basfond 0,36 palm haut 0,15 Legumes 0,00 P douce 0,00 jach haut 6,27 ha rizleg 0,78 ha manioc 0,21 ha rizpat 0,36 ha palm 0,15 ha jach 6,27 avers phi 0,15 rizleg grain 0,27 rizleg leg 0,59 manioc maniocc 0,62 rizpat grain 0,16 rizpat pd 0,00 palm huile 0,04 grain t 0,43 leg t 0,04 huile t 0,04 pd t 0,08 20% it2 30,00 0,42 0,08 0,04 0,08 0,62 1,23 1957353 1468701 1991469 0,49 76609 0,08 266625 0,00 1,29 2023487 1,85 0,47 0,05 0,05 0,04 0,36 0,42 0,56 0,12 0,00 0,62 0,36 0,75 0,04 0,04 0,03 176 0,75 6,01 0,12 896 0,75 0,21 0,36 0,44 0,00 0,00 6,01 0,75 0,21 0,36 0,44 6,01 0,15 0,26 0,56 0,62 0,16 0,00 0,12 0,42 0,08 0,04 0,08 40% it3 30,00 0,32 0,04 0,04 0,08 0,10 0,58 2315977 690312 2737576 60% it4 30,00 0,37 0,02 0,04 0,08 0,62 1,13 3250329 732207 3666820 1,89 6698137 0,00 1,29 2023487 3,17 0,47 0,05 0,05 0,04 0,36 0,16 1,86 6618382 0,00 1,29 2023487 3,15 0,47 0,05 0,05 0,04 0,36 0,16 1,93 0,00 0,36 1,90 0,00 0,26 0,36 0,15 0,04 0,10 0,04 0,04 0,03 332 0,21 0,02 0,36 0,04 0,04 0,03 336 573 812 0,36 7,41 0,00 0,00 0,09 0,36 7,32 0,00 0,00 0,36 7,41 0,09 0,36 7,32 0,15 0,15 0,16 0,00 1,93 0,16 0,26 0,16 0,00 1,90 0,16 0,04 0,08 0,04 0,08 80% it5 30,00 0,50 0,09 0,04 0,01 0,62 1,26 3784774 893787 4127966 0,07 10840 1,54 5479966 0,00 1,36 2133459 2,97 0,47 0,05 0,05 0,04 0,36 0,24 0,16 1,58 0,00 0,62 0,36 0,16 0,26 0,06 0,06 0,04 324 0,16 0,95 0,16 975 0,16 0,21 0,36 6,09 0,00 0,00 0,95 0,16 0,21 0,36 6,09 0,95 0,15 0,08 0,16 0,62 0,16 0,00 1,58 0,24 0,09 0,04 0,01 100% it6 30,00 0,55 0,09 0,04 0,08 0,62 1,38 4066596 1145567 4338492 0,19 29312 1,23 4360114 0,00 1,29 2023487 2,70 0,47 0,05 0,05 0,04 0,36 0,29 0,28 1,27 0,00 0,62 0,36 0,10 0,20 0,25 0,06 0,06 0,05 297 0,20 0,10 2,02 0,15 1057 0,30 0,21 0,36 4,88 0,00 0,00 2,02 0,30 0,21 0,36 4,88 2,02 0,15 0,13 0,28 0,62 0,16 0,00 1,27 0,29 0,09 0,04 0,08 159 - Scénario palmier à huile_classe_3 classe_3 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION TECHN TECHN ACHAT ACHAT ACHAT ACHAT ACHAT TOT SEMENCES TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 autoCONSO autoCONSO autoCONSO autoCONSO autoCONSO TOTAL grain leg huile pd maniocc TOTAL TOTAL CONSO VENTES huile huile pd pd pd TOTAL grain leg huile pd maniocc grain leg huile pd maniocc dens1 dens3 grain leg huile maniocc SEM SEM UTILSATION JOURS j1 ha J JOUR rizleg manioc rizpat palm legume papate douce jach ha ha ha ha ha avers rizleg rizleg manioc rizpat rizpat palm grain leg huile pd maniocc 0% it1 26 0,45 0,03 0,01 0,30 0,41 1,20 2749361 1325327 3097559 0,01 26724 0,00 2,21 3468834 2,22 0,47 0,05 0,05 0,10 0,36 0,31 0,03 0,01 0,00 0,41 0,66 0,03 0,14 0,00 0,01 J t t t t t t LEONES LEONES LEONES t LEONES t t LEONES t t t t t t t t t t t ha ha t t t t t 0,07 t 0,07 t 0,05 j 138,41 j 0,03 ha 0,20 j 0,16 PERSONNE 994,64 haut 0,03 haut 0,14 basfond 0,66 haut 0,04 0,00 0,00 haut 0,20 rizleg 0,03 manioc 0,14 rizpat 0,66 palm 0,04 jach 0,20 phi 0,72 grain 0,02 leg 0,03 maniocc 0,41 grain 0,30 pd 0,00 huile 0,01 t 0,31 t 0,03 t t 0,30 t 0,41 20% it2 26 0,44 0,03 0,01 0,30 0,22 1,01 2770013 1270432 3053496 0,06 150856 0,00 2,21 3468834 2,27 0,47 0,05 0,05 0,10 0,36 0,30 0,02 0,06 0,00 0,22 0,66 0,02 0,14 0,02 0,01 40% it3 26 0,44 0,03 0,01 0,30 0,22 1,01 2799143 1270432 3082626 0,06 150856 0,00 2,21 3468834 2,27 0,47 0,05 0,05 0,10 0,36 0,30 0,02 0,06 0,00 0,22 0,66 0,02 0,14 0,02 0,01 0,07 0,07 0,05 138,24 0,02 0,09 0,15 899,50 0,02 0,07 0,66 0,23 0,00 0,00 0,09 0,02 0,07 0,66 0,23 0,09 0,72 0,01 0,02 0,22 0,30 0,00 0,06 0,30 0,02 0,07 0,07 0,05 138,24 0,02 0,09 0,15 902,97 0,02 0,07 0,66 0,23 0,00 0,00 0,09 0,02 0,07 0,66 0,23 0,09 0,72 0,01 0,02 0,22 0,30 0,00 0,06 0,30 0,02 0,30 0,22 0,30 0,22 60% it4 26 0,44 0,03 0,04 0,30 0,10 0,92 2847621 1327332 3131104 0,06 150856 0,00 2,21 3468834 2,27 0,47 0,05 0,05 0,10 0,36 0,30 0,00 0,10 0,00 0,06 0,66 0,00 0,15 0,03 80% it5 26 0,45 0,03 0,04 0,30 0,10 0,92 2896449 1327332 3179932 0,06 150856 0,00 2,21 3468834 2,27 0,47 0,05 0,05 0,10 0,36 0,30 0,00 0,10 0,00 0,06 0,66 0,00 0,15 0,03 100% it6 26 0,45 0,03 0,04 0,30 0,10 0,92 2948150 1315449 3235885 0,07 169855 0,00 2,21 3468834 2,27 0,47 0,05 0,05 0,10 0,36 0,30 0,04 0,07 0,07 0,05 138,10 0,00 0,01 0,06 956,35 0,00 0,02 0,66 0,38 0,00 0,00 0,01 0,00 0,02 0,66 0,38 0,01 0,72 0,00 0,00 0,06 0,30 0,00 0,10 0,30 0,00 0,04 0,30 0,06 0,04 0,07 0,07 0,05 138,10 0,00 0,01 0,06 962,17 0,00 0,02 0,66 0,38 0,00 0,00 0,01 0,00 0,02 0,66 0,38 0,01 0,72 0,00 0,00 0,06 0,30 0,00 0,10 0,30 0,00 0,04 0,30 0,06 0,10 0,07 0,07 0,05 137,48 0,11 0,00 0,66 0,15 0,03 963,46 0,66 0,41 0,00 0,00 0,66 0,41 0,72 0,30 0,00 0,11 0,30 0,04 0,30 160 - Scénario palmier à huile_classe_4 Classe_4 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION TECHN TECHN ACHAT ACHAT ACHAT TOT SEMENCES TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 autoCONSO autoCONSO autoCONSO autoCONSO TOTAL J grain t leg t huile t pd t maniocc t TOTAL t TOTAL LEONES CONSO LEONES VENTES LEONES leg t leg LEONES huile t huile LEONES pd t pd t pd LEONES TOTAL t grain t leg t huile t pd t maniocc t grain t leg t huile t pd t maniocc t dens1 ha dens3 ha grain t huile t SEM t SEM t UTILSATION t JOURS j j1 j J ha J j JOUR PERSONNE rizleg haut manioc haut rizpat basfond palm haut legume p douce jach haut ha rizleg ha manioc ha rizpat ha palm ha jach avers phi rizleg grain rizleg leg manioc maniocc rizpat grain rizpat pd palm huile grain t leg t pd t maniocc t 0% it1 65 0,65 0,09 0,04 0,08 0,80 1,66 3930855 1458732 4166804 0,06 8783 0,13 131627 0,00 3,11 4889045 3,29 0,61 0,05 0,06 0,04 0,45 0,45 0,15 0,13 0,00 0,80 0,84 0,15 0,20 0,04 0,09 0,09 0,07 182 0,15 0,88 0,16 1831,99 0,15 0,27 0,84 0,48 0,00 0,00 0,88 0,15 0,27 0,84 0,48 0,88 0,39 0,07 0,15 0,80 0,38 0,00 0,13 0,45 0,09 0,08 0,80 20% it2 65 0,65 0,09 0,04 0,08 0,80 1,66 3931495 1452357 4164756 0,05 7983 0,13 141360 0,00 3,11 4889045 3,30 0,61 0,05 0,06 0,04 0,45 0,45 0,14 0,13 0,00 0,80 0,84 0,14 0,20 0,04 0,09 0,09 0,07 183 0,14 0,84 0,16 1840,56 0,14 0,27 0,84 0,52 0,00 0,00 0,84 0,14 0,27 0,84 0,52 0,84 0,39 0,07 0,14 0,80 0,38 0,00 0,13 0,45 0,09 0,08 0,80 40% it3 65 0,64 0,09 0,05 0,08 0,80 1,66 3932180 1446193 4162887 0,05 7210 0,14 150770 0,00 3,11 4889045 3,30 0,61 0,05 0,06 0,04 0,45 0,44 0,14 0,14 0,00 0,80 0,84 0,14 0,20 0,05 0,09 0,09 0,07 183 0,14 0,82 0,16 1849,03 0,14 0,27 0,84 0,55 0,00 0,00 0,82 0,14 0,27 0,84 0,55 0,82 0,39 0,07 0,14 0,80 0,38 0,00 0,14 0,44 0,09 0,08 0,80 60% it4 66 0,64 0,09 0,05 0,08 0,80 1,66 3932909 1440226 4161185 0,04 6461 0,15 159880 0,00 3,11 4889045 3,31 0,61 0,05 0,06 0,04 0,45 0,44 0,13 0,15 0,00 0,80 0,84 0,13 0,20 0,05 0,09 0,09 0,07 184 0,13 0,79 0,16 1857,42 0,13 0,27 0,84 0,59 0,00 0,00 0,79 0,13 0,27 0,84 0,59 0,79 0,39 0,06 0,13 0,80 0,38 0,00 0,15 0,44 0,09 0,08 0,80 80% it5 66 0,64 0,09 0,05 0,08 0,80 1,66 3933679 1434442 4159641 0,04 5736 0,16 168711 0,00 3,11 4889045 3,31 0,61 0,05 0,06 0,04 0,45 0,44 0,13 0,16 0,00 0,80 0,84 0,13 0,20 0,05 0,09 0,09 0,07 184 0,13 0,76 0,16 1865,73 0,13 0,27 0,84 0,62 0,00 0,00 0,76 0,13 0,27 0,84 0,62 0,76 0,39 0,06 0,13 0,80 0,38 0,00 0,16 0,44 0,09 0,08 0,80 100% it6 66 0,63 0,09 0,06 0,08 0,80 1,66 3934490 1428828 4158247 0,03 5031 0,17 177281 0,00 3,11 4889045 3,31 0,61 0,05 0,06 0,04 0,45 0,44 0,12 0,17 0,00 0,80 0,84 0,12 0,20 0,06 0,09 0,09 0,07 185 0,12 0,73 0,16 1873,98 0,12 0,27 0,84 0,65 0,00 0,00 0,73 0,12 0,27 0,84 0,65 0,73 0,39 0,06 0,12 0,80 0,38 0,00 0,17 0,44 0,09 0,08 0,80 161 - Scénario bas-fond_classe_1 Classe_1 test PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE autoCONSO autoCONSO autoCONSO autoCONSO autoCONSO VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES ACHAT ACHAT ACHAT ACHAT INIT INIT INIT INIT INIT STOCK TOT SEMENCES TECHN TECHN TECHN TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 X1bf Xbf travtot travtot travtot travtot X1bf X1bf Xbf Xbf TOTAL grain leg huile pd maniocc grain leg huile pd maniocc quantiteconso revenutotal revenutotalrisque autoCONSO grain leg huile pd maniocc grain grain leg leg huile huile pd pd pd maniocc maniocc quantite grain leg huile SEM grain leg huile pd maniocc grain SEM UTILSATION dens1 dens2 dens3 JOURS J t t t t t t t t t t t LEONES LEONES LEONES t t t t t t LEONES t LEONES t LEONES t t LEONES t LEONES t t t t t t t t t t t t t ha ha ha j ha J JOUR rizleg manioc rizpat palm Legumes Patate douce jach ha ha ha ha ha avers rizleg rizleg manioc rizpat rizpat palm basfond basfond rizleg manioc rizpat palm basfond basfond basfond basfond ha j PERSONNE haut haut basfond haut haut haut haut rizleg manioc rizpat palm jach phi grain leg maniocc grain pd huile ha ha j j j j dens1 dens3 dens1 dens3 0% it1 20% it2 40% it3 60% it4 80% it5 100% it6 60 0,64 0,93 0,37 0,00 1,50 0,95 0,09 0,17 0,08 0,54 1,83 3660323 3572817 2204241 0,64 0,09 0,17 0,08 0,54 60 0,64 0,93 0,37 0,00 1,50 0,95 0,09 0,17 0,08 0,54 1,83 3660323 3572817 2204241 0,64 0,09 0,17 0,08 0,54 60 1,24 0,93 0,37 0,00 1,50 1,18 0,09 0,01 0,08 0,54 1,90 4658948 4371043 3331704 1,18 0,09 60 1,79 48 1,69 0,24 0,00 1,50 1,18 0,02 0,01 0,08 0,54 1,83 4658948 4580246 3334796 1,18 0,08 0,00 1,42 1,18 0,02 0,01 0,08 0,54 1,83 4658948 4633894 3334796 1,18 0,08 0,54 60 0,88 0,93 0,37 0,00 1,50 1,13 0,09 0,06 0,08 0,54 1,90 4658948 4502872 2647783 0,88 0,09 0,06 0,08 0,54 0,01 0,08 0,54 0,51 1336707 0,01 0,08 0,54 0,47 1215177 0,84 132101 0,20 312866 0,84 132101 0,20 312866 0,84 132101 0,37 576429 0,84 132101 0,31 479915 0,23 359078 0,07 112498 0 2,39 3754762 0,96 214080 4,39 0,31 0 2,39 3754762 0,96 214080 4,39 0,31 0,10 0,98 0,05 0,10 0,04 0,34 0,10 0,98 0,05 0,10 0,04 0,34 0,10 0,07 0,65 1,03 0,10 0,07 0,65 1,03 305 1,03 6,21 0,17 1532 1,03 0,50 0,65 1,43 0,00 0,00 6,21 1,03 0,50 0,65 1,43 6,21 0,20 0,35 0,93 1,50 0,29 0,00 0,37 305 1,03 6,21 0,17 1532 1,03 0,50 0,65 1,43 0,00 0,00 6,21 1,03 0,50 0,65 1,43 6,21 0,20 0,35 0,93 1,50 0,29 0,00 0,37 0,65 127,22 38,06 145,60 54,34 0,65 127,22 38,06 145,60 54,34 0,65 0,65 0 7,45 11708446 0,96 214080 9,63 0,01 0,17 0,98 0,05 0,10 0,04 0,34 0,07 0,24 0,18 1,98 1,03 0 4,37 6859866 0,96 214080 6,48 0,25 0,15 0,98 0,05 0,10 0,04 0,34 0,15 0,12 1,17 1,03 604 1,03 6,21 0,17 1365 1,03 0,50 1,98 1,43 0,00 0,00 6,21 1,03 0,50 1,98 1,43 6,21 0,20 0,35 0,93 1,50 0,89 0,00 0,37 1,33 0,65 127,22 38,06 444,03 54,34 1,33 422 1,03 6,21 0,17 1447 1,03 0,50 1,17 1,43 0,00 0,00 6,21 1,03 0,50 1,17 1,43 6,21 0,20 0,35 0,93 1,50 0,53 0,00 0,37 0,52 0,65 127,22 38,06 262,11 54,34 0,52 0,65 0,65 0 0 2,01 3165357 0,96 214080 3,72 0,15 238039 0,88 195853 1,57 0,02 0,02 0,98 0,05 0,10 0,04 0,34 0,10 0,10 0,07 0,59 0,04 1,74 137 0,98 0,05 0,10 0,04 0,34 0,05 0,05 0,04 0,10 0,04 1,88 14 4,93 5,57 1333 0,11 0,71 2,26 1,14 0,00 0,00 4,93 0,11 0,71 2,26 1,14 4,93 0,20 1333 0,11 0,69 1,91 0,53 0,00 0,00 5,57 0,11 0,69 1,91 0,53 5,57 0,20 1,50 1,79 0,00 0,24 1,61 0,65 1,42 1,69 0,00 0,08 1,26 0,65 38,06 123,48 35,31 0,07 1,54 0,48 0,17 35,99 13,65 12,07 0,05 1,21 0,02 0,63 162 - Scénario bas-fond_classe_2 Classe_2 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE autoCONSO autoCONSO autoCONSO autoCONSO autoCONSO VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION STOCK TECHN TECHN TECHN ACHAT ACHAT ACHAT TOT SEMENCES TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 X1bf Xbf travtot travtot travtot travtot X1bf X1bf Xbf Xbf TOTAL grain leg huile pd maniocc quantiteconso revenutotal revenutotalrisuqe autoCONSO grain leg huile pd maniocc grain grain leg leg huile huile pd pd pd maniocc maniocc quantite grain leg huile pd maniocc grain leg huile pd maniocc grain dens1 dens2 dens3 leg huile SEM SEM UTILSATION JOURS J t t t t t t LEONES LEONES LEONES t t t t t t LEONES t LEONES t LEONES t t LEONES t LEONES t t t t t t t t t t t t ha ha ha t t t t t j ha ha JOUR rizleg manioc rizpat palm Legumes Patate douce jach ha ha ha ha ha avers rizleg rizleg manioc rizpat rizpat palm basfond basfond rizleg manioc rizpat palm basfond basfond basfond basfond PERSONNE haut haut basfond haut haut haut haut rizleg manioc rizpat palm jach phi grain leg maniocc grain pd huile ha ha j j j j dens1 dens3 dens1 dens3 0% it1 20% it2 40% it3 60% it4 80% it5 100% it6 30,00 0,42 0,05 0,04 0,00 0,62 1,21 2033932 1991513 1427953 0,42 0,05 0,04 0,08 0,62 30,00 0,42 0,05 0,04 0,00 0,62 1,21 2033932 1991513 1427953 0,42 0,05 0,04 0,08 0,62 30,00 0,63 0,09 0,01 0,00 0,62 1,43 4717195 4134842 1926846 0,63 0,09 30,00 0,63 0,09 0,01 0,00 0,62 1,43 4717195 4494540 1926846 0,63 0,09 30,00 0,63 0,02 0,01 0,00 0,62 1,36 4717195 4596603 1914434 0,63 24,95 0,63 0,02 0,01 0,00 0,62 1,36 4717195 4671877 1914434 0,63 0,08 0,62 1,27 3317127 0,54 84500 0,08 0,62 0,25 647057 0,48 75504 0,04 60464,63 0,08 0,62 1,43 3719920 0,54 84500 0,08 0,62 1,20 3118173 0,44 69274 0,04 60464,63 0,04 60464,63 0,04 60464,63 0 1,29 2023487 0,07 15415 1,89 0,47 0,05 0,05 0,04 0,36 0,43 0,59 0,04 1,37 0,69 0,01 0,36 0,78 0 1,29 2023487 0,07 15415 1,89 0,47 0,05 0,05 0,04 0,36 0,43 0,59 0,04 1,37 0,69 0,01 0,36 0,78 0,03 0,04 0,03 170 0,78 6,25 0,12 880 0,78 0,23 0,36 0,15 0,00 0,00 6,25 0,78 0,23 0,36 0,15 6,25 0,15 0,27 0,59 0,69 0,16 0,00 0,04 0,03 0,04 0,03 170 0,78 6,25 0,12 880 0,78 0,23 0,36 0,15 0,00 0,00 6,25 0,78 0,23 0,36 0,15 6,25 0,15 0,27 0,59 0,69 0,16 0,00 0,04 0,36 96,08 17,49 80,64 5,70 0,36 96,08 17,49 80,64 5,70 0,36 0,36 0 17,35 27249673 2,05 456218 21,07 0,47 0,05 0,05 0,04 0,36 2,30 0,53 0,04 17,43 2,67 0,48 4,59 0,71 0 6,74 10595652 0,62 137905 8,13 0,47 0,05 0,05 0,04 0,36 1,07 0,57 0,04 6,82 1,24 0,19 1,80 0,76 0,01 0,01 0,48 0,37 1158 0,71 5,66 0,12 1068 0,71 0,89 4,59 0,15 0,00 0,00 5,66 0,71 0,89 4,59 0,15 5,66 0,15 0,24 0,53 2,67 2,06 0,00 0,04 4,23 0,36 87,08 67,64 1027,16 5,70 4,23 0,19 0,14 503 0,76 6,09 0,12 1068 0,76 0,41 1,80 0,15 0,00 0,00 6,09 0,76 0,41 1,80 0,15 6,09 0,15 0,26 0,57 1,24 0,81 0,00 0,04 1,44 0,36 93,58 31,42 402,28 5,70 1,44 0,36 0,36 0 0 2,94 4624995 0,84 186458 5,25 0,47 0,05 0,05 0,04 0,36 2,19 0,12 188764 0,01 2710 1,44 0,47 0,05 0,05 0,04 0,36 1,96 0,04 3,02 1,46 0,13 0,80 0,04 0,20 0,63 0,05 0,05 2,03 0,02 0,01 2,15 0,02 0,01 0,13 0,10 191 0,05 0,04 9 6,54 6,81 1029 1029 0,49 2,83 0,15 0,00 0,00 6,54 0,21 2,20 0,15 0,00 0,00 6,81 0,49 2,83 0,15 6,54 0,15 0,21 2,20 0,15 6,81 0,15 1,46 2,19 0,00 0,04 2,47 0,36 0,63 1,96 0,00 0,04 1,84 0,36 36,95 178,25 5,70 0,80 1,67 16,04 11,80 5,70 0,05 1,79 0,36 0,36 163 - Scénario bas- fond _classe_3 Classe_3 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE autoCONSO autoCONSO autoCONSO autoCONSO VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION STOCK TECHN TECHN ACHAT ACHAT ACHAT ACHAT ACHAT TOT SEMENCES TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 X1bf Xbf travtot travtot travtot travtot X1bf X1bf Xbf Xbf TOTAL J grain t leg t huile t pd t maniocc t quantiteconso t revenutotal LEONES revenutotalrisuqeLEONES autoCONSO LEONES grain t leg t pd t maniocc t huile t huile LEONES pd t pd t pd LEONES maniocc t maniocc LEONES quantite t grain t leg t huile t pd t maniocc t grain t leg t huile t pd t maniocc t grain t dens1 ha dens3 ha grain t leg t huile t maniocc t SEM t SEM t UTILSATION t JOURS j ha ha JOUR rizleg manioc rizpat palm Legumes Patate douce jach ha ha ha ha ha avers rizleg rizleg manioc rizpat rizpat palm basfond basfond rizleg manioc rizpat palm basfond basfond basfond basfond PERSONNE haut haut basfond haut haut haut haut rizleg manioc rizpat palm jach phi grain leg maniocc grain pd huile ha ha j j j j dens1 dens3 dens1 dens3 0% it1 20% it2 40% it3 60% it4 80% it5 100% it6 26 0,44 0,03 0,01 0,00 0,61 1,40 3101587 2749340 1328348 0,31 0,03 0,30 0,42 0,01 19893 26 0,44 0,03 0,01 0,00 0,61 1,40 3101587 2749340 1328348 0,31 0,03 0,30 0,42 0,01 19893 26 0,54 0,05 0,01 0,00 0,62 1,52 3709278 2840994 1823493 0,54 0,05 0,30 26 0,58 0,05 0,01 0,00 0,62 1,57 4350282 3593956 1933019 0,58 0,05 0,30 26 0,58 0,02 0,01 0,00 0,62 1,53 4350282 3804596 1925301 0,58 26 0,58 0,02 0,01 0,00 0,62 1,53 4350282 3934795 1925301 0,58 0,30 0,30 0,01 19893 0,01 19893 0,01 19893 0,01 19893 0 0 0 0 2,21 3468834 2,21 3468834 5,34 8383985 4,66 7317712 2,22 0,47 0,05 0,05 0,10 0,36 0,31 0,03 0,01 2,51 0,42 2,22 0,47 0,05 0,05 0,10 0,36 0,31 0,03 0,01 2,51 0,42 5,34 0,47 0,05 0,05 0,10 0,36 0,69 0,05 0,01 5,64 4,67 0,47 0,05 0,05 0,10 0,36 0,61 0,05 0,01 4,96 0,66 0,03 0,12 0,66 0,03 0,12 0,16 1,48 0,05 0,03 1,30 0,05 0,01 0,20 0,07 0,07 0,05 138 0,03 0,21 0,15 1068 0,03 0,14 0,66 0,03 0,00 0,00 0,21 0,03 0,14 0,66 0,03 0,21 0,72 0,02 0,03 0,42 0,30 0,00 0,01 0,01 0,20 0,07 0,07 0,05 138 0,03 0,21 0,15 1068 0,03 0,14 0,66 0,03 0,00 0,00 0,21 0,03 0,14 0,66 0,03 0,21 0,72 0,02 0,03 0,42 0,30 0,00 0,01 0,01 0,62 0,16 0,12 315 0,05 0,33 0,16 1226 0,05 0,01 0,62 0,11 0,14 0,11 275 0,05 0,33 0,16 1289 0,05 1,48 0,03 0,00 0,00 0,33 0,05 1,30 0,03 0,00 0,00 0,33 0,05 1,48 0,03 0,33 0,72 0,03 0,05 1,30 0,03 0,33 0,72 0,03 0,05 0,66 4,85 10,59 147,84 1,14 0,66 4,85 10,59 147,84 1,14 0,67 0,00 0,01 0,82 0,66 7,65 0,59 0,00 0,01 0,64 0,66 7,65 332,26 1,14 0,82 292,26 1,14 0,64 0,66 0,66 0,66 0,66 0 0 3,46 5433027 1,14 186390 4,61 0,47 0,05 0,05 0,10 0,36 0,69 2,67 4191652 1,14 186390 3,82 0,47 0,05 0,05 0,10 0,36 0,67 0,01 3,76 1,14 0,11 0,99 0,27 0,01 2,97 1,14 0,09 0,78 0,35 0,02 0,01 0,62 0,02 0,01 0,62 0,11 0,08 226 0,09 0,07 179 1266 1266 0,38 1,26 0,03 0,00 0,00 0,38 1,14 0,03 0,00 0,00 0,38 1,26 0,03 0,38 1,14 0,03 0,72 0,72 1,14 0,69 0,00 0,01 0,60 0,66 1,14 0,67 0,00 0,01 0,48 0,66 28,93 221,54 1,14 0,48 0,12 0,51 0,15 28,93 174,96 1,14 0,35 0,12 0,43 0,23 164 - Scénario bas-fond_classe_4 Classe_4 test CONSO CONSO CONSO CONSO CONSO CONSO UTILITE UTILITE UTILITE autoCONSO autoCONSO autoCONSO autoCONSO autoCONSO VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES INIT INIT INIT INIT INIT PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION STOCK TECHN TECHN ACHAT ACHAT ACHAT ACHAT TOT SEMENCES TRSAL CR JACH2 CONTROL CALORIES ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 X1bf Xbf travtot travtot travtot travtot X1bf X1bf Xbf Xbf TOTAL grain leg huile pd maniocc quantiteconso revenutotal revenutotalrisuqe autoCONSO grain leg huile pd maniocc grain grain leg leg huile huile pd pd pd maniocc maniocc quantite grain leg huile pd maniocc grain leg huile pd maniocc grain dens1 dens3 grain leg huile SEM SEM UTILSATION JOURS J t t t t t t LEONES LEONES LEONES t t t t t t LEONES t LEONES t LEONES t t LEONES t LEONES t t t t t t t t t t t t ha ha t t t t t t j ha ha JOUR rizleg manioc rizpat palm Legumes Patate douce jach ha ha ha ha ha avers rizleg rizleg manioc rizpat rizpat palm basfond basfond rizleg manioc rizpat palm basfond basfond basfond basfond PERSONNE haut haut basfond haut haut haut haut rizleg manioc rizpat palm jach phi grain leg maniocc grain pd huile ha ha j j j j dens1 dens3 dens1 dens3 0% it1 20% it2 40% it3 60% it4 80% it5 100% it6 65 0,63 0,09 0,08 0,00 0,80 1,68 4166954 3930855 1459087 0,45 0,09 65 0,63 0,09 0,08 0,00 0,80 1,68 4166954 3930855 1459087 0,45 0,09 72 0,76 0,09 0,03 0,00 0,80 1,76 4576271 4185271 2001915 0,66 0,09 69 0,74 0,09 0,05 0,00 0,80 1,76 4576271 4274895 1672232 0,53 0,09 64 0,76 0,02 0,03 0,00 0,80 1,69 4576271 4381769 2259680 0,76 58 0,76 0,02 0,03 0,00 0,80 1,69 4576271 4484455 2291191 0,76 0,08 0,80 0,08 0,80 0,08 0,80 0,08 0,80 0,08 0,80 1,36 3539809 0,03 0,08 0,80 0,91 2374415 0,06 8828 0,12 131086 0,06 8828 0,12 131086 0,06 8828 0,12 131086 0,04 6308 0,12 131086 0,12 131086 0,09 98664 0 0 0 0 3,11 4889045 3,11 4889045 4,87 7649548 3,29 0,61 0,05 0,06 0,04 0,45 0,45 0,15 0,12 3,19 0,80 3,29 0,61 0,05 0,06 0,04 0,45 0,45 0,15 0,12 3,19 0,80 5,05 0,61 0,05 0,06 0,04 0,45 0,66 0,15 0,12 4,95 0,80 3,87 6075116 0,34 63203 4,37 0,61 0,05 0,06 0,04 0,45 0,53 0,13 0,12 3,95 1,14 0,84 0,15 0,18 0,84 0,15 0,18 1,30 0,15 0,10 1,04 0,13 0,21 0,08 0,09 0,09 0,07 182 0,15 0,88 0,16 1982 0,15 0,27 0,84 0,48 0,00 0,00 0,88 0,15 0,27 0,84 0,48 0,88 0,39 0,07 0,15 0,80 0,38 0,00 0,12 0,08 0,09 0,09 0,07 182 0,15 0,88 0,16 1982 0,15 0,27 0,84 0,48 0,00 0,00 0,88 0,15 0,27 0,84 0,48 0,88 0,39 0,07 0,15 0,80 0,38 0,00 0,12 0,84 20,61 20,30 188,16 18,24 0,84 20,61 20,30 188,16 18,24 0,03 0,14 0,14 0,11 279 0,15 0,88 0,16 1987 0,15 0,27 1,30 0,48 0,00 0,00 0,88 0,15 0,27 1,30 0,48 0,88 0,39 0,07 0,15 0,80 0,59 0,00 0,12 0,46 0,84 20,61 20,30 291,74 18,24 0,46 0,05 0,11 0,11 0,09 229 0,13 0,78 0,16 2051 0,13 0,38 1,04 0,48 0,00 0,00 0,78 0,13 0,38 1,04 0,48 0,78 0,39 0,06 0,13 1,14 0,47 0,00 0,12 0,20 0,84 18,35 28,85 232,66 18,24 0,20 0,84 0,84 0,84 0,84 0 0 1,60 2518772 3,07 576027 6,15 0,61 0,05 0,06 0,04 0,45 2,21 0,25 389882 3,02 567358 4,28 0,61 0,05 0,06 0,04 0,45 1,72 0,12 1,68 3,87 0,10 0,44 2,24 0,12 0,33 3,82 0,05 0,09 1,87 0,02 0,03 0,02 0,10 0,07 152 0,05 0,04 76 0,01 1929 1929 1,29 2,68 0,48 0,00 0,00 1,27 1,96 0,48 0,00 0,00 0,01 1,29 2,68 0,48 0,39 1,27 1,96 0,48 0,01 0,39 3,87 2,21 0,00 0,12 1,84 0,84 3,82 1,72 0,00 0,12 1,12 0,84 98,19 99,22 18,24 0,44 1,40 97,02 19,34 18,11 0,09 1,03 0,84 0,84 165 - Scénario semence _classe_1 Classe_1 test CONSO CONSO CONSO CONSO CONSO CONSO INIT INIT INIT INIT INIT CALORIES UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES ACHAT ACHAT TOT SEMENCES PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 TECHN TECHN TECHN TRSAL CR JACH2 CONTROL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL TOTAL grain leg huile pd maniocc TOTAL grain leg huile pd maniocc JOUR TOTAL CONSO VENTES leg leg huile huile pd pd pd maniocc maniocc TOTAL grain SEM SEM UTILSATION grain leg huile pd maniocc rizleg rizleg manioc rizpat rizpat palm dens1 dens2 dens3 JOURS J t t t t t t t t t t t PERSONNE LEONES LEONES LEONES t LEONES t LEONES t t LEONES t LEONES t t t t t t t t t t grain leg maniocc grain pd huile ha ha ha j ha ha rizleg manioc rizpat palm legumes papate jach ha ha ha ha haut haut basfond haut ha ha haut rizleg manioc rizpat palm 0% 20% 40% 60% 80% 100% it1 60 0,76 0,09 0,17 0,08 0,54 1,64 0,98 0,05 0,10 0,04 0,34 1347 3572553 2205135 3660152 0,84 132244 0,21 320785 0,00 2,39 3754762 0,88 196213 4,32 0,12 0,10 0,10 0,07 0,64 0,93 0,38 0,08 1,42 0,35 0,93 1,42 0,29 0,08 0,38 0,65 1,04 it2 60 0,79 0,09 0,17 0,08 0,54 1,67 0,98 0,05 0,10 0,04 0,34 1371 3631950 2265639 3736909 0,69 108661 0,27 419324 0,00 2,39 3754762 5,47 1220481 8,83 0,12 0,17 0,17 0,13 0,67 0,78 0,44 0,08 6,01 0,38 0,78 6,01 0,29 0,08 0,44 0,65 it3 60 0,81 0,09 0,17 0,08 0,54 1,69 0,98 0,05 0,10 0,04 0,34 1391 3701832 2316819 3803846 0,73 115082 0,27 411298 0,00 2,39 3754762 4,67 1042399 8,06 0,12 0,17 0,17 0,13 0,69 0,82 0,44 0,08 5,21 0,39 0,82 5,21 0,29 0,08 0,44 0,65 it4 60 0,83 0,09 0,17 0,08 0,54 1,71 0,98 0,05 0,10 0,04 0,34 1412 3776379 2369994 3875429 0,77 121753 0,26 402606 0,00 2,39 3754762 3,85 857958 7,27 0,12 0,18 0,18 0,14 0,71 0,86 0,43 0,08 4,39 0,42 0,86 4,39 0,29 0,08 0,43 0,65 it5 60 0,85 0,09 0,17 0,08 0,54 1,73 0,98 0,05 0,10 0,04 0,34 1433 3855871 2425148 3951960 0,82 128673 0,25 393249 0,00 2,39 3754762 2,99 667220 6,45 0,12 0,19 0,19 0,14 0,73 0,91 0,42 0,08 3,53 0,44 0,91 3,53 0,29 0,08 0,42 0,65 it6 60 0,87 0,09 0,17 0,08 0,54 1,75 0,98 0,05 0,10 0,04 0,34 1456 3940584 2482242 4033744 0,86 135836 0,25 383244 0,00 2,39 3754762 2,11 470303 5,61 0,12 0,19 0,19 0,15 0,75 0,95 0,42 0,08 2,65 0,46 0,95 2,65 0,29 0,08 0,42 0,65 0,78 432 0,78 4,69 0,17 0,78 2,00 0,65 1,69 0,16 0,02 4,69 0,78 2,00 0,65 1,69 0,82 418 0,82 4,94 0,17 0,82 1,74 0,65 1,67 0,17 0,02 4,94 0,82 1,74 0,65 1,67 0,86 403 0,86 5,19 0,17 0,86 1,46 0,65 1,65 0,17 0,02 5,19 0,86 1,46 0,65 1,65 0,91 388 0,91 5,45 0,17 0,91 1,18 0,65 1,63 0,18 0,02 5,45 0,91 1,18 0,65 1,63 0,95 372 0,95 5,73 0,17 0,95 0,88 0,65 1,60 0,19 0,02 5,73 0,95 0,88 0,65 1,60 304 1,04 6,21 0,17 1,04 0,47 0,65 1,45 0,21 0,02 6,21 1,04 0,47 0,65 1,45 166 - Scénario semence _classe _2 Classe_2 test CONSO CONSO CONSO CONSO CONSO CONSO INIT INIT INIT INIT INIT CALORIES UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES ACHAT TOT SEMENCES PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 TECHN TECHN TECHN TRSAL CR CR JACH2 CONTROL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff autoCONSO autoCONSO autoCONSO autoCONSO autoCONSO coeif legume coeif patate TOTAL grain leg huile pd maniocc TOTAL grain leg huile pd maniocc JOUR TOTAL CONSO VENTES grain grain leg leg pd pd pd TOTAL SEM SEM UTILSATION grain leg huile pd maniocc rizleg rizleg manioc rizpat rizpat palm dens1 dens2 dens3 JOURS J t t t t t t t t t t t PERSONNE LEONES LEONES LEONES t LEONES t LEONES t t LEONES t t t t t t t t t grain leg maniocc grain pd huile ha ha ha j ha ha rizleg manioc rizpat palm legume papate jach ha ha ha ha ha avers grain leg huile pd maniocc haut haut basfond haut haut rizleg manioc rizpat palm jach phi t t t t t 0% 20% 40% 60% 80% 100% it1 30 0,43 0,04 0,04 0,08 0,62 1,21 0,47 0,05 0,05 0,04 0,36 889 1934948 1491820 1978995 it2 30 0,43 0,05 0,04 0,08 0,62 1,22 0,47 0,05 0,05 0,04 0,36 895 1953326 1493413 1997383 it3 30 0,43 0,06 0,04 0,08 0,62 1,23 0,47 0,05 0,05 0,04 0,36 901 1971711 1495039 2015777 it4 30 0,46 0,05 0,01 0,08 0,62 1,22 0,47 0,05 0,05 0,04 0,36 849 1995982 1479205 2039970 0,55 85737 0,00 1,29 2023487 1,83 0,04 0,04 0,03 0,43 0,59 0,04 0,00 0,62 0,27 0,59 0,62 0,16 0,00 0,04 0,36 0,78 0,54 84144 0,00 1,29 2023487 1,82 0,04 0,04 0,03 0,43 0,59 0,04 0,00 0,62 0,27 0,59 0,62 0,16 0,00 0,04 0,36 0,78 0,53 82518 0,00 1,29 2023487 1,81 0,04 0,04 0,03 0,43 0,59 0,04 0,00 0,62 0,27 0,59 0,62 0,16 0,00 0,04 0,36 0,78 it6 30 0,47 0,09 0,01 0,08 0,62 1,27 0,47 0,05 0,05 0,04 0,36 875 2118798 1493706 2167651 0,19 486131 0,93 146664 0,00 1,29 2023487 2,41 0,17 0,17 0,13 0,65 1,02 0,01 0,00 0,62 0,49 1,02 0,62 0,16 0,00 0,01 0,36 169 169 169 1,02 238 1,02 1,02 238 1,02 0,78 6,27 0,12 0,78 0,21 0,36 0,15 0,07 0,00 6,27 0,78 0,21 0,36 0,15 6,27 0,15 0,43 0,04 0,04 0,08 0,62 0,09 0,00 0,78 6,27 0,12 0,78 0,21 0,36 0,15 0,07 0,00 6,27 0,78 0,21 0,36 0,15 6,27 0,15 0,43 0,05 0,04 0,08 0,62 0,09 0,00 0,78 6,27 0,12 0,78 0,21 0,36 0,15 0,07 0,00 6,27 0,78 0,21 0,36 0,15 6,27 0,15 0,43 0,06 0,04 0,08 0,62 0,09 0,00 0,61 95211 0,00 1,29 2023487 1,89 0,06 0,06 0,04 0,46 0,66 0,01 0,00 0,62 0,30 0,66 0,62 0,16 0,00 0,01 0,36 0,68 0,14 176 0,14 0,68 6,34 0,13 0,83 0,21 0,36 0,04 0,08 0,00 6,34 0,83 0,21 0,36 0,04 6,34 0,15 0,46 0,05 0,01 0,08 0,62 0,09 0,00 it5 30 0,46 0,07 0,01 0,08 0,62 1,25 0,47 0,05 0,05 0,04 0,36 863 2034784 1485322 2083686 0,19 491997 0,95 149182 0,00 1,29 2023487 2,43 0,17 0,17 0,13 0,65 1,02 0,01 0,00 0,62 0,49 1,02 0,62 0,16 0,00 0,01 0,36 6,14 0,17 1,02 0,21 0,36 0,04 0,09 0,00 6,14 1,02 0,21 0,36 0,04 6,14 0,15 0,46 0,07 0,01 0,08 0,62 0,09 0,00 6,14 0,17 1,02 0,21 0,36 0,04 0,09 0,00 6,14 1,02 0,21 0,36 0,04 6,14 0,15 0,47 0,09 0,01 0,08 0,62 0,09 0,00 167 - Scénario semence_classe_3 Classe_3 test CONSO CONSO CONSO CONSO CONSO CONSO INIT INIT INIT INIT INIT CALORIES UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES ACHAT ACHAT ACHAT ACHAT ACHAT TOT SEMENCES PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 TECHN TECHN TRSAL CR JACH2 CONTROL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff autoCONSO autoCONSO autoCONSO autoCONSO coeif legume coeif patate TOTAL grain leg huile pd maniocc TOTAL grain leg huile pd maniocc JOUR TOTAL CONSO VENTES huile huile pd pd pd TOTAL grain leg huile maniocc SEM SEM UTILSATION grain leg huile pd maniocc rizleg rizleg manioc rizpat rizpat palm dens1 dens3 JOURS J t t t t t t t t t t t PERSONNE LEONES LEONES LEONES t LEONES t t LEONES t t t t t t t t t t t t t grain leg maniocc grain pd huile ha ha j ha ha rizleg manioc rizpat palm legume papate jach ha ha ha ha ha avers grain leg pd maniocc haut haut basfond haut haut rizleg manioc rizpat palm jach phi t t t t 0% 20% 40% 60% 80% 100% it1 26 0,45 0,09 0,01 0,30 0,41 1,26 0,47 0,05 0,05 0,10 0,36 1032 2751806 1327433 3100367 0,01 21961 0,20 2,21 3468834 2,22 0,13 0,06 0,01 it2 25 0,44 0,09 0,01 0,30 0,31 1,16 0,47 0,05 0,05 0,10 0,36 983 2785486 1287249 3135625 0,01 19291 0,20 2,21 3468834 2,22 0,12 0,04 0,01 0,31 0,07 0,07 0,05 0,32 0,05 0,01 0,23 it3 25 0,45 0,09 0,01 0,30 0,31 1,16 0,47 0,05 0,05 0,10 0,36 988 2819369 1288288 3171618 0,01 15735 0,20 2,21 3468834 2,21 0,12 0,03 0,01 0,31 0,07 0,07 0,05 0,32 0,06 0,01 0,23 it4 25 0,45 0,09 0,01 0,30 0,31 1,16 0,47 0,05 0,05 0,10 0,36 992 2853263 1289346 3207672 0,00 12114 0,20 2,21 3468834 2,21 0,13 0,03 0,01 0,31 0,07 0,07 0,05 0,32 0,06 0,00 0,23 it5 25 0,45 0,09 0,01 0,30 0,31 1,17 0,47 0,05 0,05 0,10 0,36 997 2887168 1290424 3243788 0,00 8425 0,20 2,21 3468834 2,21 0,13 0,03 0,01 0,31 0,07 0,07 0,05 0,32 0,06 0,00 0,23 it6 25 0,46 0,09 0,01 0,30 0,31 1,17 0,47 0,05 0,05 0,10 0,36 1002 2921085 1291523 3279968 0,00 4666 0,20 2,21 3468834 2,21 0,13 0,03 0,01 0,31 0,07 0,07 0,05 0,32 0,06 0,00 0,23 0,03 0,05 0,03 0,06 0,03 0,06 0,03 0,06 0,03 0,06 0,30 0,23 0,01 0,66 0,05 131 0,05 0,33 0,16 0,05 0,30 0,23 0,01 0,66 0,06 131 0,06 0,33 0,16 0,06 0,30 0,23 0,00 0,66 0,06 131 0,06 0,34 0,16 0,06 0,30 0,23 0,00 0,66 0,06 131 0,06 0,34 0,16 0,06 0,30 0,23 0,00 0,66 0,06 131 0,06 0,35 0,16 0,06 0,66 0,03 0,00 0,06 0,33 0,05 0,66 0,02 0,00 0,06 0,33 0,06 0,66 0,02 0,00 0,06 0,34 0,06 0,66 0,01 0,00 0,06 0,34 0,06 0,66 0,01 0,00 0,06 0,35 0,06 0,66 0,03 0,33 0,71 0,32 0,05 0,30 0,66 0,02 0,33 0,71 0,32 0,06 0,30 0,66 0,02 0,34 0,71 0,32 0,06 0,30 0,66 0,01 0,34 0,71 0,32 0,06 0,30 0,66 0,01 0,35 0,71 0,32 0,06 0,30 0,00 0,09 0,00 0,09 0,00 0,09 0,00 0,09 0,00 0,09 0,07 0,07 0,05 0,31 0,03 0,01 0,23 0,41 0,02 0,03 0,41 0,30 0,23 0,01 0,66 0,03 138 0,03 0,20 0,16 0,03 0,14 0,66 0,03 0,00 0,06 0,20 0,03 0,14 0,66 0,03 0,20 0,71 0,31 0,03 0,30 0,41 0,00 0,09 168 - Scénario semence_ classe_4 Classe_4 test CONSO CONSO CONSO CONSO CONSO CONSO INIT INIT INIT INIT INIT CALORIES UTILITE UTILITE UTILITE VENTES VENTES VENTES VENTES VENTES VENTES VENTES VENTES ACHAT ACHAT ACHAT TOT SEMENCES PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 PRODUCTION1 TECHN TECHN TRSAL CR JACH2 CONTROL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL ASSOL coeff autoCONSO autoCONSO autoCONSO autoCONSO coeif legume coeif patate TOTAL grain leg huile pd maniocc TOTAL grain leg huile pd maniocc JOUR TOTAL CONSO VENTES leg leg huile huile pd pd pd TOTAL grain huile SEM SEM UTILSATION grain leg huile pd maniocc rizleg rizleg manioc rizpat rizpat palm dens1 dens3 JOURS J t t t t t t t t t t t PERSONNE LEONES LEONES LEONES t LEONES t LEONES t t LEONES t t t t t t t t t t t grain leg maniocc grain pd huile ha ha j ha ha rizleg manioc rizpat palm legumes patate douce jach ha ha ha ha ha avers grain leg pd maniocc haut haut basfond haut haut rizleg manioc rizpat palm jach phi t t t t 0% 20% 40% 60% 80% 100% it1 65 0,64 0,09 0,05 0,08 0,80 1,67 0,61 0,05 0,06 0,04 0,45 1882 3930855 1458733 4166804 0,06 8783 0,13 131627 0,00 3,11 4889045 3,29 0,20 0,05 0,09 0,09 0,07 0,45 0,15 0,13 0,30 0,80 0,07 0,15 0,80 0,38 0,30 0,13 0,84 0,15 182 0,15 0,88 0,16 0,15 0,27 0,84 0,48 0,09 0,08 0,88 0,15 0,27 0,84 0,48 0,88 0,39 0,45 0,09 0,08 0,80 0,65 0,10 it2 65 0,65 0,09 0,04 0,08 0,80 1,67 0,61 0,05 0,06 0,04 0,45 1857 3975087 1469002 4215458 0,06 10072 0,11 115950 0,00 3,11 4889045 3,29 0,20 0,04 0,09 0,09 0,07 0,45 0,15 0,11 0,30 0,80 0,07 0,15 0,80 0,38 0,30 0,11 0,84 0,15 181 0,15 0,92 0,16 0,15 0,27 0,84 0,42 0,10 0,08 0,92 0,15 0,27 0,84 0,42 0,92 0,39 0,45 0,09 0,08 0,80 0,65 0,10 it3 65 0,66 0,09 0,06 0,08 0,80 1,68 0,61 0,05 0,06 0,04 0,45 1917 4019435 1479879 4264611 0,07 11436 0,09 99344 0,00 3,11 4889045 3,28 0,20 0,06 0,09 0,09 0,07 0,46 0,16 0,09 0,30 0,80 0,08 0,16 0,80 0,38 0,30 0,09 0,84 0,16 180 0,16 0,98 0,16 0,16 0,27 0,84 0,36 0,11 0,08 0,98 0,16 0,27 0,84 0,36 0,98 0,39 0,46 0,09 0,08 0,80 0,65 0,10 it4 65 0,66 0,09 0,07 0,08 0,80 1,70 0,61 0,05 0,06 0,04 0,45 1976 4063907 1491445 4314319 0,08 12887 0,08 81687 0,00 3,11 4889045 3,27 0,20 0,07 0,09 0,09 0,07 0,46 0,17 0,08 0,30 0,80 0,08 0,17 0,80 0,38 0,30 0,08 0,84 0,17 179 0,17 1,03 0,17 0,17 0,27 0,84 0,30 0,11 0,08 1,03 0,17 0,27 0,84 0,30 1,03 0,39 0,46 0,09 0,08 0,80 0,65 0,10 it5 65 0,66 0,09 0,08 0,08 0,80 1,71 0,61 0,05 0,06 0,04 0,45 2033 4108511 1504356 4364914 0,09 14507 0,06 61975 0,00 3,11 4889045 3,26 0,20 0,08 0,09 0,09 0,07 0,47 0,18 0,06 0,30 0,80 0,09 0,18 0,80 0,38 0,30 0,06 0,84 0,18 178 0,18 1,09 0,17 0,18 0,27 0,84 0,23 0,12 0,08 1,09 0,18 0,27 0,84 0,23 1,09 0,39 0,47 0,09 0,08 0,80 0,65 0,10 it6 65 0,68 0,09 0,08 0,08 0,80 1,73 0,61 0,05 0,06 0,04 0,45 2062 4152992 1543282 4428286 0,12 19391 0,00 2550 0,00 3,11 4889045 3,24 0,20 0,08 0,09 0,09 0,07 0,48 0,21 0,00 0,30 0,80 0,10 0,21 0,80 0,38 0,30 0,00 0,84 0,21 174 0,21 1,28 0,17 0,21 0,27 0,84 0,01 0,14 0,08 1,28 0,21 0,27 0,84 0,01 1,28 0,39 0,48 0,09 0,08 0,80 0,65 0,10 169