Tree seedling vitality improves with functional diversity in a Mediterranean common garden experiment

Reforestation with multiple tree species is a promoted strategy to mitigate global change and to improve forest resistance against natural hazards. Dryland reforestation often fails because seedlings suffer from harsh conditions in degraded areas. Positive species interactions can overcome recruitment drawbacks by ameliorating environmental stress, but there is a strong need to advance functional insights from well-designed experiments. We studied the vitality of 19,712 tree seedlings from 12 species in a Mediterranean common garden experiment (Sardinia). Vitality was assessed as an integrated index of foliage discoloration and defoliation measures, which are in dry areas potential indicators of early plant performance. The experimental design properly replicated all monocultures and a selection of mixed communities with different levels of species richness (SR) and functional diversity (FD). From the second year onwards, a water availability treatment (irrigated versus non-irrigated) was added to the design. In the second year, seedling vitality was strongly determined by species identity and irrigation, but ecological interactions between trees were not relevant. In the third year, however, broad-leaved species were significantly more vigorous in mixed assemblages. Importantly, FD was identified as a seven times stronger predictor compared to SR. This suggests that a certain degree of trait diversification is essential to benefit from facilitative interactions. The positive FD effects were principally mediated by the presence of pines (P. pinea, P. pinaster and P. halepensis) in the neighborhood of broad-leaved trees. The latter had, on average, a 23% greater likelihood to have the highest vitality score in mixture with pines. The creation of a favorable physical and biotic neighborhood by pines is likely caused by their fast juvenile growth and adequate crown light transmission. FD effects on seedling vitality were positive, but contrary to the stress-gradient hypothesis, they were of similar magnitude in both irrigated and non-irrigated blocks. We conclude that local neighborhood facilitation provides essential assistance for broad-leaved trees passing a critical seedling stage in semi-arid regions. This knowledge can contribute to increased success rates in forest rehabilitation in these regions