Tailoring agroforestry technologies to the diversity of Rwandan smallholder agriculture

<strong>Keywords</strong>: food security, biophysical and socioeconomic conditions, farmer resource groups, productivity, economic evaluation, scenario analysis Smallholder livelihoods in sub-Saharan Africa(SSA) are constrained by a number of factors that limit food production and thereby threaten food security. Soil fertility is one of the major factors explaining the decrease in per capita food production in SSA. Nutrient deficiencies in particular N and P severely limit agricultural production in many regions in the tropics. Supply of adequate amounts of nutrients throughfertiliserapplication is therefore a prerequisite to balance soil fertility budgets and to boost food production. However, mineral fertilisers are not accessible to the large majority of smallholder farmers. Farmyard manure, an important source of organic fertiliser for smallholder farmers, is available at limited quantities due to low livestock densities in many regions, for example Rwanda. Agroforestry, a low-input technology, was shown to contribute to the enhancement of food production while ensuring sustainability in sub-Saharan Africa.Agroforestry may contribute to soil fertility by increasing nutrient availability and providing other various benefits and services. However, to be successful agroforestry technologies need to match the characteristics of different smallholder farming systems, like for example soil fertility status, socioeconomic status and farmer management. These factors are rarely studied in an integrated manner. This thesis aims to understand and characterise different farming systems, evaluate the potential for the most promising agroforestry practices and suggest the most suitable agroforestry recommendations for different farming systems in targeted agro-ecological zones of Rwanda. The approach combined characterization of farming systems, participatory tree testing, farmer’s evaluations of technologies, and scenario and trade-off analyses in two agro-ecological zones: Central Plateau (moderate altitude) and Buberuka (high altitude zone). Two locations, Simbi and Kageyo sectors were selected as representative study sites. Wealth ranking techniques allowed the identification of three farm resource groups (RGs). Though three farmer classes were identified in the two locations and referred as RG 1, RG 2 and RG 3 respectively, farmer classes were unique to each location. Averaged over sampled villages, 76% of all households belong to RG 1 class in Simbi versus 67% in Kageyo. This least resourced group with on average 0.20 ha of land and with 1 goat wasthe most vulnerable farmer groupin terms of food security (20 to 25% protein deficient).RG 2 (9 to 31%) was intermediate between RG 1 and RG 3. RG 3 (2 to 7 %) was the wealthiest (1 to 3 ha, 2 or more cattle) and food-secure for at least 10 months. Soil nutrient balances were negative in most farms due to small amounts of nutrients applied, which did not compensate for nutrient removal during harvest. From an agroforestry perspective, Simbi contrasted with Kageyo in tree diversity and density but tree niches and management were similar between the locations. The main agroforestry species may becategorisedinto three classes including timber, legume and fruit tree species based on the main functions. The results clearly indicated the need to improve soil fertility and food production using integrated soil fertility approaches that promote a combined use of agroforestry resources and other fertiliser sources to replenish the soil nutrients and improve the efficiency and cost effectiveness of inputs use at farm level. Experiments evaluated the potential effects of agroforestry species on production within different farming systems. <em>Tephrosia</em> species were tested as a source of mulch in coffee plantations in the Central Plateau agro-ecological zone. Application of <em>Tephrosia</em> mulch resulted in higher biomass and better economic returns when established in coffee fields, particularly when <em>Tephrosia</em>mulch was combined with NPK. Application of prunings of <em>Calliandra</em> increased maize productivity, net returns and the ratio between gross margin and costs of inputs on all farms except the richest farms. This positive effect of <em>Calliandra</em> was larger in Kageyo than in Simbi. The effect was even more pronounced with P application. The results indicated that fields responded differently within farms, and significant differences between locations were present. The assessment of fodder availability within different farming systems revealed that animal feeds are widely diversified, with <em>Pennisetum</em> being largely used in wealthier farms (RG 3), while RG 1 farmers use larger quantities of marshland-herbs and crop residues.There was a strong variation in seasonal feed availability. Napier and <em>Calliandra</em> were more available during the wet season, while banana pseudo-stems were used more in the dry seasons. Quantification of the year-round fodder availability showed that RG 1 farmers are unable to keep a cow, while RG 2 and RG 3 could keep local or improved cows under specific scenarios. Biophysical (rainfall, field type) and socio-economic conditions (wealth status) as well as farmer preferences were factors influencing the choice and performance of agroforestry technologies. The study recommends revisiting current agroforestry research policies and taking into account farmer’s preferences as priorities in the agroforestry research agenda.