Direct seeding is potentially a more cost-effective alternative to conventional tree planting for restoring tropical forest ecosystems. However, seed loss, due to removal and damage by animals, can substantially reduce seedling establishment. Therefore, this study examined the impact of seed predation on seedling establishment of five tree species, native to upland evergreen forests of northern Thailand: Hovenia dulcis, Alangium kurzii, Prunus cerasoides, Choerospondias axillaris and Horsfieldia amygdalina. We tested the hypothesis that excluding animals would significantly reduce seed removal, and increase both germination and seedling survival. The objective was to calculate a composite index of the relative suitability of the species studied for direct seeding.

Seeds were placed on the ground in a deforested site and subjected to five predator-exclusion treatments: wire cage, insecticide, cage ​+ ​insecticide, open cage and no exclusion (control).

Seed loss was highest for H. amygdalina (the largest seed tested). Across species, wire cages significantly reduced seed loss by 12.4% compared with controls (P ​< ​0.001) suggesting that vertebrates were the major seed predators. Seed germination ranged from 0 to 77% among the species tested. Based on relative species-performance scores (combining measures of survival and seedling growth), P. cerasoides was the most suitable species for direct seeding, followed by A. kurzii and C. axillaris, whilst H. dulcis and H. amygdalina were unsuitable. H. dulcis had small seeds with low seed germination, whereas H. amygdalina was subjected to high seed removal.

Exclusion of seed predators and the selection of suitable species may substantially increase the success of direct seeding, as a technique for restoring upland evergreen forest ecosystems. Testing more species for their suitability is needed, to provide more diverse options for forest restoration.