Plants seeded during ecological restoration often perish en masse, and researchers are searching for traits benefiting survival. However, it remains unclear if any particular traits uniformly benefit survival or if different traits are beneficial depending on environmental conditions following seeding. To determine if different traits are advantageous under different conditions, we conducted 30 seeding experiments (15 sites × 2 seeding years) in a big sagebrush (Artemisia tridentataNutt.) ecosystem. In each experiment, we estimated the probability each of four grass species would germinate and survive to the second growing season after planting.
We found survival probability rankings of the species varied across experiments, providing conclusive evidence different traits maximized survival under different conditions. This led us to develop a framework to account for trait fitness variation during ecological restoration. Specifically, based on survival probabilities in our experiments, we developed functions defining explicit restoration goals. Then we used optimization algorithms to assign species identities to seeds of a seed mix in a manner that maximized the probability of achieving the goals. Our native grass density goal was most consistently met by assigning seeds to two (of a possible three) species, with the presence of >1 species denoting a biodiversity effect. Our findings indicate it will be of value to seek and strategically combine ensembles of traits that increase chances at least some seed mix members (i.e. species, varieties) will be adapted to whatever plant establishment conditions arise. The framework we illustrate for selecting restoration germplasm should inspire a productive new area of research focused on using biodiversity as a tool for more consistently meeting plant establishment goals.