Cost-effectiveness of restoration approaches for above-ground and soil carbon accumulation
The amount of carbon accumulated per invested US$ in restored forests increased with time when only implementation costs were considered (F1,22 = 12.8, P = 0.001), because of the gradual increase of carbon stocks and the absence of additional costs after the first years of implementation (Fig. 4A). However, the increase of land opportunity costs was higher than the increase of carbon accumulation, which resulted in a decrease of the opportunity cost-effectiveness for carbon accumulation along the chronosequence (F1,22 = 10.5, P = 0.003; Fig. 4B). These opposing trajectories of implementation and accumulated land opportunity cost-effectiveness along the chronosequence nullified changes in total cost-effectiveness with age in restored forests (F1,22 = 2.8, P = 0.1; Fig. 4C). Implementation cost-effectiveness was significantly higher for second-growth forests than for plantations throughout the chronosequence (F1,22 = 9.4, P = 0.005; Fig. 4A). Opportunity cost-effectiveness showed a similar trend but the average difference between plantations and second-growth forests was not significant (F1,22 = 2.7, P = 0.1; Fig. 4B). Total cost-effectiveness, i.e. the amount of accumulated carbon standardized by the total cost of restoration (implementation + opportunity costs), was significantly higher for second-growth forests (F1,22 = 9.2, P = 0.006; Fig. 4C). On average, second-growth forests displayed a total cost-effectiveness that was 60% higher than plantations (15.1 kgC.US$-1 and 9.4 kgC.US$-1, respectively). Average differences between second-growth forests and plantations were higher for total cost-effectiveness than for its components (i.e. implementation and accumulated land opportunity cost-effectiveness). This indicates that differences in both implementation and opportunity costs contributed to the resulting differences in total cost-effectiveness between restoration practices. We note that one tree planting site was significantly older (61 years old) than the rest of the restoration dataset. Removing this site from the analyses did not change any results or conclusions of this work.