On a large coconut plantation planted on a deep peat soil in Sumatra, Indonesia, hybrid coconut yields are stabilizing at values well below their potential, even in plots where known production factors have been mastered since the outset: land preparation, water control, and mineral fertilization. To explain this situation, five hypotheses have been proposed. They were tested one by one in field trials. The authors have demonstrated that the main reason for stabilized yields was the Lepidoptera root pest Sufetula spp. and conclude by considering lines of research for controlling this pest in the context of rehabilitation and replanting.