Banque de graines du sol et déterminants de la germination du tali, Erythrophleum suaveolens (Guill. & Perr.) Brenan

This study assesses the abundance of Erythrophleum suaveolens seeds in the soil bank of central African humid forests. The work was carried out in two types of forests in northern Congo: the Celtis forest, growing on clay-sandy to sandy-clay soils and the Manilkara forest, growing on sandy soils. E . suaveolens stems (dbh ≥ 10 cm) were counted in two 400 ha blocks, and the diametric structures of the stands were compared. To assess seed abundance in the soil bank, 80 pits (2 x 40 pits per forest type) were then dug at the foot of 20 trees (10 per forest), removing three contiguous layers of 10 cm each, i.e. to a total depth of 30 cm. Dormancy of the seeds collected was tested by germination trials after treatment with H 2 SO 4 . Five seeds taken in the Celtis forest down to a depth of 20 cm were used to estimate their age by Accelerator Mass Spectroscopy (AMS). In both forest types, the diametric structures are characterized by a Gaussian distribution, illustrating a regeneration deficit. While stem densities (dbh ≥ 10 cm) are similar in the Celtis and Manilkara forests at 0.85 and 1.05 stems/ha respectively, seed densities are significantly higher in the Celtis forest (8.55 seeds/m 2 ) than in the Manilkara forest (0.15 seed/m 2 ). The maximum percentage of germination obtained was 19.1% for seeds having undergone no treatment. Germination rates in the acid-treated batches were lower. These seeds could remain intact in the soil bank for at least ten years. The factors that could influence seed density variations are discussed and silvicultural recommendations are formulated.