Drivers of planted seedling survival and growth in a Mediterranean‐climate woodland restoration experiment

Abstract Introduction The restoration of degraded ecosystems is an important means of promoting and conserving biodiversity. However, restoration projects on average fail to reach target biodiversity levels within the time frames they are studied. Competition from non‐native groundlayer grasses and forbs, potential facilitative effects of plant–plant interactions, and species‐specific responses to these processes are all thought to be important in determining outcomes of tree and shrub‐based restoration. Objective To quantify the effects of initial seedling height, herbicide application, nitrogen addition, starting soil conditions, soil depth (using distance to exposed bedrock as a proxy), and plot species richness on the survival and growth of seedlings in an experimental restoration project. Methods The survival and growth of over 11,000 individual trees and shrubs consisting of eight different species were measured over a 12‐year period. Results We found that herbicide application had a lasting negative signature on the survival of plants and a positive effect on growth. In contrast, nitrogen addition had no significant effect on plant survival and growth. Larger seedling size was correlated with increased survival and growth more than 10 years after planting; however, these responses were species‐specific. Importantly, approximately 50% of seedlings died over the 12‐year monitoring period, and this mortality was concentrated in Hakea lissocarpha , Banskia sessilis , Eucalyptus astringens , and Acacia microbotrya . [Correction added on 21 April 2026, after first online publication: Callistemon phoeniceus was replaced with Acacia microbotrya in the preceding sentence.] Conclusion Standard management practices such as weed control can have unintended consequences for plant survival. Initial seedling height strongly predicts survival and growth more than a decade after planting, meaning future restoration outcomes may be predicted based on early measurements.