Selecting tree species to restore forest under climate change conditions: Complementing species distribution models with field experimentation.

The long-term success of forest restoration programs can be improved using climate-based species distribution models (SDMs) to predict which tree species will tolerate climate change. However, as SDMs cannot estimate if species will recruit at these habitats, determining whether their predictions apply to early life-cycle stages of trees is critical to support such a usage. For this, we propose sowing seeds of the focal tree species under the current climate and simulated climate change conditions in target restoration sites. Thus, using of SDMs to design climate-adaptive forest restoration programs would be supported if the differences in habitat occupancy probabilities of species they predict between the current and future climate concurs with the observed differences in recruitment rates of species when sowed under the current climate and simulated climate change conditions. To test this hypothesis, we calibrated SDMs for Vachellia pennatula and Prosopis laevigata, two pioneer tree species widely recommended to restore human-degraded drylands in Mexico, and transferred them to climate change scenarios. After that, we applied the experimental approach proposed above to validate the predictions of SDMs. These models predicted that V. pennatula will decrease its habitat occupancy probabilities across Mexico, while P. laevigata was predicted to keep out their current habitat occupancy probabilities, or even increase them, in climate change scenarios. The results of the field experiment supported these predictions, as recruitment rates of V. pennatula were lower under simulated climate change than under the current climate, while no differences were found for the recruitment rates of P. laevigata between these environmental conditions. These findings demonstrate that SDMs provide meaningful insights for designing climate-adaptive forest restoration programs but, before applying this methodology, predictions of these models must be validated with field experiments to determine whether the focal tree species will recruit under climate change conditions. Moreover, as the pioneer trees used to test our proposal seem to be differentially sensitive to climate change, this approach also allows establishing what species must be prescribed to restore forests with a view to the future and what species must be avoided in these practices.

Modelling and validation of the spatial distribution of suitable habitats for the recruitment of invasive plants on climate change scenarios: An approach from the regeneration niche.

The regeneration niche concept states that plant species only occur in habitats where the environmental conditions allow their recruitment. This study focuses on this concept and proposes a novel approach for modelling and experimentally validating the distribution of suitable habitats for the recruitment of invasive plants under the current and future climate. The biological invasion of the Peruvian peppertree (Schinus molle) in Mexico is used as practical example. The values of eight bioclimatic variables associated to sites in which young, naturally established seedlings and saplings were detected were used to model the current distribution of recruitment habitats. A machine-learning algorithm of maximum entropy (MaxEnt) was used to calibrate the model and its output indicated the distribution of occurrence probabilities of young peppertrees in Mexico under the current climate. This model was projected on climate change scenarios predicted for the middle of this century, which indicated that the cover of suitable recruitment habitats for this invasive species will shrink. To validate these predictions, field experiments were performed at three sites where the model predicted reduced occurrence probabilities of young peppertrees. In these experiments, emergence and survival rates of peppertree seedlings were assessed under the current climate and under simulated climate change conditions. As seedling emergence and survival rates were lower under simulated climate change conditions, the experiments validated the model predictions. These results supported our proposal, which combines modelling and experimental approaches to make accurate and valid predictions about the distribution of suitable recruitment habitats for invasive plants in a warmer and drier world.