Tropical forest restoration is a critical component of global strategies to conserve biodiversity and sequester carbon, particularly in highly deforested areas where lands are being removed from agriculture but resources to restore these lands are limited.  Establishing tree plantations on degraded lands is an important restoration tool but is costly and labor intensive.  Planting tree “islands” is less expensive and simulates the nucleation process of succession but has rarely been tested as a restoration strategy.  There is little information on the role of island size in attracting seed dispersers, the potential for island expansion, or how different planting arrangements affect seed dispersal by birds and bats.  The amount of forest cover in the landscape surrounding a restoration site may also be a key driver of seed rain patterns.  We tested how three restoration approaches affected arrival of forest seeds at 11 experimental sites spread over a 100 km² area in southern Costa Rica.  Each site had three 50 × 50 m treatments: (1) control - natural regeneration, (2) island - planting tree patches of three sizes (4 × 4, 8 × 8, and 12 × 12 m), and (3) plantation - planting an area uniformly.  Four tree species were used (Terminalia amazonia [Combretaceae], Vochysia guatemalensis [Vochysiaceae], Erythrina poeppigiana, and Inga edulis [both N-fixing Fabaceae]).  Seed rain was measured for 18 months beginning ~2 yr after planting.

Results/Conclusions

Plantations received the most zoochorous tree seeds (266.1 ± 64.5), islands were intermediate (210.4 ± 52.7), and controls were lowest (87.1 ± 13.9) (seeds/m²/yr ± SE).  Greater tree seed deposition in the planted areas was due to birds (0.51 ± 0.18 seeds/m²/day) rather than bats (0.07 ± 0.03).  The seed rain was primarily small-seeded early-successional species and dispersal of large-seeded and late-successional forest species was extremely rare.  Large and medium islands received twice as many zoochorous tree seeds as small islands and locations at 2, 4, and 12-m outside island edges, suggesting there is a minimum size threshold for islands necessary to increase dispersal of zoochorous tree seeds and that seed rain outside of planted areas is strongly reduced.  Local restoration planting design was more important for enhancing seed dispersal than the amount of remnant forest cover within the surrounding 100- and 500-m areas.  Planting trees in plantations and island ‘nuclei’ substantially increased arrival of early-successional tree seeds, and represents a broadly applicable strategy for facilitating forest succession on abandoned agricultural lands.