The expansion of tree species within a grassland-woodland matrix relies heavily on an abundant seed crop, which may be affected by many factors; among these, browsing by livestock has gone largely ignored. Here we explore how seed production is affected by tree size, grazing by large domestic herbivores and interannual climate fluctuations in a small-sized tree (Polylepis australis, Rosaceae), which dominates the remnant forests in the high mountains of central Argentina. Seed number and seed mass per tree were quantified in 100 different-sized trees located in areas with and without livestock. Using seed traps we also determined seed rain variation in 15 plots between 2005 and 2008. The estimated number of seeds per tree ranged from 0 to 224,000 and increased with tree height (power relation); for trees smaller than 354 cm in height, seed production was lower in areas with livestock and higher in areas without livestock. Mass per seed increased with tree height, with no significant livestock effect. The largest seed rain was recorded in 2008, probably due to growth of seed trees and not in response to climate. To promote seed production and forest expansion, livestock should be excluded from areas with small trees, such as forest edges.

Domestic livestock are widespread in seasonally dry forests, likely causing forest degradation and limiting tree seedling establishment. Shrubs can play an important role in facilitating tree regeneration, by protecting trees from livestock damage and ameliorating unfavorable abiotic conditions. We aimed at disentangling the relative contribution of grazing exclusion, long‐term forest conservation, and the potential facilitation effect of shrubs on the performance of saplings of the native tree Kageneckia lanceolata. We planted 400 saplings in grazed and ungrazed areas situated both in a preserved and a degraded forest. In each situation, we established planting plots in three accompanying vegetation treatments: herbs, a nonleguminous spiny shrub and a leguminous spiny shrub. Survival of 3‐year‐old saplings was 10‐fold higher in the preserved than in the degraded forest and 2‐fold higher in the ungrazed than in the grazed site. Differences in survival among accompanying vegetation treatments were much smaller than between grazing treatments. Survival significantly increased with increasing protection by shrubs only in the degraded site. Sapling growth patterns were fairly similar to survival patterns, with no growth in the degraded forest, except for limited growth under both shrubs in the ungrazed site. We conclude that, in selecting plantation sites for the study species, forest condition and grazing exclusion should be prioritized over microsite selection based on neighboring vegetation.

Forest restoration is a great challenge when saplings are repeatedly pruned by leaf-cutting ants and pesticides are not permitted. Based on four hypotheses (distance-dependent foraging, plant repellence, structural complexity, and nutrient content), we predicted that sapling performance increases and ant pruning decreases: (1) away from leaf-cutting ant nests; (2) in unpalatable or (3) tall vegetation patches; and (4) in non-fertilized soils. We set up a full factorial field experiment to test the hypotheses; we also monitored sapling performance without and with plastic tree shelters used by practitioners to deter ant pruning. Our results showed predicted sapling survival from 40 to 80% per year for distances to nests ranging from 4 to 223 m, respectively. Ant pruning decreased with increasing distance, with null pruning at distances >121 m. Variation in sapling survival, but not in ant pruning, was also explained by the interaction between vegetation palatability and nutrient condition. Vegetation height did not explain sapling survival or pruning. Plastic tree shelters increased survival by 3.5 times and reduced pruning by half. Height of surviving saplings was not affected by any treatment. Our results show the potential for active forest restoration without the use of tree shelters away from leaf-cutting ant nests, while tree shelters are a viable option near leaf-cutting ant nests.