Analysing seed systems performance: the case of oil palm in Bénin

Key words: Bénin, genetic quality, growth dynamics, innovation, institutions, multistakeholders process, oil palm, perennials, physiological quality, pot size, seed quality, seed systems, smallholders, social learning. The seed supply system used by smallholder farmers is characterised by many dysfunctions. For perennials, including oil palm, there is a knowledge gap regarding these dysfunctions. In this thesis, we used oil palm as a case to analyse the performance of seed systems. We conducted social surveys, sampled farmers’ oil palm plots, ran field experiments and documented the social learning process of stakeholders involved. To identify major constraints in the seed system, we conducted a diagnostic study using farmers’ perspective. Jointly with stakeholders the following major constraints were identified (in order of importance): poor geographic distribution of official nurseries, poor genetic quality of palms in smallholder plots, high costs of hybrid seedlings, and poor seedling care in nurseries leading to poor physiological quality. The poor care was specifically mentioned in the eastern part of the study area. We investigated the reliability of genetic quality of seedlings supplied to smallholder farmers as one of the constraints that emerged from the diagnostic study. Main drivers of reliability in genetic quality over the past decades were analysed. Using event ecology approach, we document the historical events that may have affected the oil palm seed system. Proportions of hybrid palms varied with seedling supply source, farmers’ geographic position, seedling purchase price and year of planting. Socio-institutional mechanisms associated with observed variation in smallholder plantations were national policy change, local arrangements for seedling supply to smallholder farmers, and farmers’ personal characteristics. Local arrangements improved genetic quality in villages located far away from official supply sources. Villages where local seedling supply initiatives withdrew showed reduced genetic quality with farmers having fewer tenera in their fields than before. Membership of farmers’ organisation correlated positively with proportion of tenera. Farmer’s use of informal intermediaries showed negative effects on genetic quality they received. To evaluate the efficacy of on-going nursery management practices on the seedling phenotype at planting we conducted joint learning experiments. In two full 3 × 4 × 3 factorial experiments, bag size (small, medium, large), type of soil substrate ("forest" soil, household waste substrate, arable soil, and arable soil with animal manure) and fertiliser supply (no fertilisation, split dose every 15 days, and full dose every 30 days) were tested for their effects on seedling phenotype. Bag size proved the main factor determining oil palm seedling phenotype in both the 2011 and 2012 experiment. Although large-sized bags produced largest seedlings, medium-sized bags filled with a mixture of arable soil and animal manure without any fertiliser supply sustained seedling growth well and seemed the best balance between physiological quality and production cost. Growth variables were highly correlated. Height and root-collar diameter constituted good proxies to estimate seedling biomass differences between objects in a non-destructive way. To gain insight into temporal patterns of effects of bag size, substrate, fertiliser supply, and their interactions on seedling growth, we analysed the dynamics of oil palm seedling growth using monthly observations. Overall, in both experiments, bag size explained the largest proportion of experimental error and started to deviate earlier than substrate, fertiliser supply or their two and three way interactions. Curve fitting showed different growth models for height, root-collar diameter and number of leaves. The analysis of growth rates showed that (relative and absolute) growth rates were mainly affected by bag size in both years with larger F-values than for substrate, fertiliser supply, and interactions. Experimental findings indicated that pot size matters and cannot be compensated by fertiliser addition and should therefore be considered carefully for tree seedling production in nurseries. We analysed the joint experiment as a multi-stakeholder process and contributed to understand how the way of organising social learning affects stakeholders’ ownership of process outcomes. Stakeholders’ perceptions of seedling quality and their appreciation of treatment performance varied with the use they make of planting material. While farmers, as end-users, put forward seedling vigour when describing quality seedlings, nursery holders underlined production costs and reported that seedling quality is a compromise with production costs. Field observations further to the joint experiment indicated changes in practices among nursery holders, research, and farmers. The level of stakeholders’ involvement increased their participation, ownership of the learning process, and could lead to sustainable practices. The research approach developed in this thesis to analyse seed systems performance contributed to the methodology for seed systems analysis. It also contributed to knowledge of dysfunctions of seed systems for perennials, tree nursery seedling production and social learning processes.