Search

Many tropical pioneer species depend on the presence of high seed densities in the soil for successful recruitment following canopy disturbance (Cheke et al. 1979, Dalling & Hubbell 2002, Guevara Sada & Gómez Pompa 1972, Whitmore 1983). However determinants of variation in the composition and abundance of soil seed banks remain poorly understood. Seed bank densities can be affected by rates of seed predation and pathogen infection on the surface and in the soil, by intrinsic rates of loss in viability following dispersal, and by variation in the timing and duration of fruit production (Dalling et al. 1997, Garwood 1983, Murray & Garcia 2002). Here we compare seasonal fluctuations in seed bank density in five Panamanian forests varying in elevation and seasonality of precipitation (Table 1). We predict that lowland forests should show stronger intra-annual fluctuation in seed bank densities than montane forests because seed production and loss rates should be higher under conditions of greater resource availability, and where consistent high temperatures support greater abundance or activity of seed predators and pathogens (Brühl et al. 1999). Secondly, among lowland sites, we predict greater fluctuations in seed bank densities at drier, more seasonal sites where seasonally favourable conditions for seedling recruitment may select for interspecific synchrony in fruit production (Daubenmire 1972, Garwood 1983).

Variation among forests in environmental and biotic conditions may strongly influence seed fate with important consequences for the abundance and distribution of plant species. Here we examine the post-dispersal seed removal rates of six pioneer species (Cecropia peltata, Miconia argentea, Luehea seemannii, Trema micrantha, Apeiba aspera and Jacaranda copaia) from the soil surface at five sites in Panama varying in elevation (0–1100 m) and seasonality (0–4 mo dry season). We compared removal rates of washed seeds placed in vertebrate exclosures, invertebrate exclosures, and unprotected controls in January and June. Overall, removal rates of unprotected seeds were similar among sites. Almost all seed removal could be attributed to litter ants in two subfamilies (Myrmicinae and Ponerinae). Little or no removal was recorded from invertebrate exclosures while vertebrate exclosures had no effect on removal either in lowland and montane forests. Seed removal rates were high for four animal-dispersed species (mean 45% removed over 2 d), whereas two wind-dispersed species were largely untouched (mean 2% removed). These results indicate that seed dispersal characteristics, rather than site characteristics, may be the strongest determinant of the post-dispersal seed fate of pioneers.