ES recovery
Our meta-analysis showed that restoration enhanced ES supply in degraded wetlands. The results also showed that it is more difficult to recover ES supply than to recover biodiversity; an alternative or complementary interpretation is that full recovery of ES supply takes longer than full recovery of biodiversity. Either interpretation is consistent with the meta-analysis by Rey Benayas et al. [9], but inconsistent with the analysis of North American wetlands by Dodds et al. [8].
Restoration did not enhance ES uniformly across all individual ES examined. We observed that restored wetlands provided, on average, 36% higher levels of provisioning, regulating and supporting ES than did degraded wetlands, but similar levels of cultural services. To be sure, we did not expect uniform recovery of all individual ES, given the heterogeneity of ES and wetland types included in the meta-analysis; wetlands types are known to differ in ecological dynamics, recovery rates and extents of recovery [7].
Our finding that restoration increased supply of provisioning services more than the supply of other ES may reflect the fact that, among the included studies, the desired outcomes when restoring provisioning services (e.g. abundance of target species) were generally better defined and more homogeneous than were objectives for regulating, supporting, and cultural services. Effect sizes for these last three services showed wide confidence intervals in our study, suggesting higher intra-class heterogeneity than effect sizes for provisioning services [12]. Small sample size may explain our finding that restoration did not significantly affect cultural services. Compared to natural wetlands, restored wetlands showed similar supply of provisioning and cultural services but lower supply of regulating services (mainly climate regulation, soil fertility and erosion) and supporting services (mainly biogeochemical cycles and provision of terrestrial habitat). The lower levels of climate and soil regulation, biological structure and biogeochemical cycles may reflect the intrinsically slow recovery rates reported for these surrogate variables [7]. In contrast, faster recovery rates have been reported for the water regulation variables in our study, such as hydrological dynamics and water quality, and these latter variables indeed showed full recovery.