Multi-site salt marsh restoration can recover key natural functions despite long-term structural deviations and site-level differences

Restoration studies often focus on short-term structural recovery, while achieving consistent, long-term functional outcomes across diverse sites remains a challenge, limiting scalability. We evaluated the long-term structural and functional effectiveness (average match to reference natural sites) and outcome consistency (dispersion around the average) of two salt marsh restoration methods in the Venice Lagoon: RC which favours tidal creek formation at the low shore, and RB, which does not. Sampling was conducted at mid and low shore levels in restored marshes over 10 years old. At the mid shore, neither method fully replicated the natural sediment characteristics, and restored sites often showed greater site-level differences than natural marshes. Nevertheless, average vegetation structure and function, including biomass and macrofaunal diversity, were comparable to natural sites. Soil carbon was also similar between restored and natural sites, although this likely reflects the carbon content of dredged sediments rather than new atmospheric CO2 sequestration. In contrast, at the low shore, neither method supported the native cordgrass Spartina maritima, despite sediment characteristics more closely resembling natural conditions. Only the non-native S. anglica colonised the low shore at RC sites, exhibiting biotic and functional attributes similar to those of the widespread S. anglica stands at natural sites, while RB sites remained unvegetated. These findings suggest that restoration methods promoting natural foreshore dynamics, such as RC, more effectively support vegetation establishment and replicate low-shore functions, even if these functions are delivered by a non-native species. We hypothesize that restoring a network of spatially diffuse sites may improve ecological outcomes by better replicating key natural structural and functional conditions, despite persistent structural deviations and elevated site-level differences.