Agroecological approaches require a complete understanding of the agroecosystems by considering complex and countless interactions. Agroforestry systems, that combine at least one ligneous perennial with at least one crop or cattle species, often aim at optimizing ecological and economical interactions among their components. They encompass highly contrasted agroecosystems, from mechanized input-intensive plantations intercropping only two species to family grown, highly diverse and ecologically intensive agroforests. Cocoa-based agroforestry systems have been widely described in the literature for the high taxonomic and functional diversity of the soil biota, especially of larger-sized organisms such as earthworms and macroinvertebrates. However, the interactions between farmer's practices and soil macrofauna are poorly documented. A common practice all over cocoa producing countries consists in piling the harvested pods on a determined area of the plantation floor before opening them. The cocoa beans are extracted and carried out of the plantations, but pod husks remain on the floor. In this study, we compared pod harvesting sites and sites free of pod husks for the diversity of soil macrofauna in the leaf-litter, the 0-10 cm and the 10 – 20 cm soil layers. Based on 60 soil and leaf litter samples, we compared mature (aged 11-25 years) with old (aged > 70 years) cocoa-based agroforestry plantations in the Dominican Republic. We found that under the cocoa pods, macrofauna density and taxa richness were significantly higher and bulk density was lower. This result is enhanced by the age of the cocoa plantation, as the accumulation of pod husks on a determined harvesting arena tends to be higher over time. The trade-off between a recommendation to spread pod husks over the plantation floor when harvesting cocoa and the current sanitary recommendations for harvesting is discussed.