There was fruiting of the dipterocarps of Pantai Aceh Forest Reserve, Penang, Peninsular Malaysia in September 1986. The opportunity was taken to investigate the establishment and seedling survival of Shorea curtisii, Shorea multiflora and Shorea pauciflora over the subsequent year. The germination and seedling survival of Shorea curtisii was compared between canopy gaps and forest understorey sites.

Under the shaded conditions none of the species grew beyond the two leaf stage over their first year. Shorea multiflora seedlings suffered a 72% mortality, significantly lower than that of Shorea pauciflora (89%) and Shorea curtisii (93%). The major cause of mortality appeared to be drought, though Shorea curtisii was also prone to predation by small mammals.

In a gap, Shorea curtisii seeds probably germinated less well but seedling survival (28% versus 7%) and growth was better than canopy-shaded conspecifics. Seedlings in gaps escaped small mammal predation.

Spatial and temporal variation in seedling dynamics was assessed using records of community-wide seedling demography collected with identical monitoring methods at four tropical lowland forests in Panama, Malaysia, Ecuador and French Guiana for periods of between 3 and 10 y. At each site, the fates of between 8617 and 391 777 seedlings were followed through annual censuses of the 370–1008 1-m2 seedling plots. Within-site spatial and inter-annual variation in density, recruitment, growth and mortality was compared with among-site variability using Bayesian hierarchical modelling to determine the generality of each site's patterns and potential for meaningful comparisons among sites. The Malaysian forest, which experiences community-wide masting, was the most variable in both seedling density and recruitment. However, density varied year-to-year at all sites (CVamong years at site = 8–43%), driven largely by high variability in recruitment rates (CV = 40–117%). At all sites, recruitment was more variable than mortality (CV = 5–64%) or growth (CV = 12–51%). Increases in mortality rates lagged 1 y behind large recruitment events. Within-site spatial variation and inter-annual differences were greater than differences among site averages in all rates, emphasizing the value of long-term comparative studies when generalizing how spatial and temporal variation drive patterns of recruitment in tropical forests.