Recruitment, a multi-stage process with unpredictable result: the case of a Sapotaceasae in French Guianan forest

The net result of the multi-stage recruitment from seed production to seedling establishment was assessed along 3 consecutive years in Chrysophyllum lucentifolium (Sapotaceae) in a mature rain forest of French Guiana. According to year, between 37 and 42 % of seed production was primarily dispersed away from fruiting trees without damage, especially by primates. Seeds dispersed in howler monkeys' (Alouatta seniculus) defecation were serendipitously buried by dung beetles (14%), eaten by terrestrial vertebrates such as rodents, or destroyed by insects or pathogens. The proportion of seeds secondarily dispersed by scatterhoarding rodents Varied from 3 to 17 % according to year. The survival rate of unremoved and uneaten seeds after 20 days, varied from 2 to 56 % according to year and pattern of spatial distribution generated by primary dispersal (clump or scattered). The empirical model based on the seed and seedling fate diagram showed that, according to years, from 6 to 15 % of produced seeds lead to seedling establishment away from parent tree. Variations in patterns of seed removal and predation by rodents were the most crucial factors governing the effectiveness of primary and secondary dispersers and the spatial distribution of seedlings. Between-year variations of recruitment were important and stochastic because the fluctuations of seed production and seed dispersal/predation varied independently. The proposed empirical model has thus no predictive value. However it can be useful for instance to test the effects of seed size and of some faunal perturbations.