Is there more than meets the eye? Seed bank analysis of a typical novel ecosystem, the extensive green roof

Abstract Questions (a) Will seed banks develop on extensive green roofs (EGRs) – prime examples of novel ecosystems – as in (semi‐)natural systems? If so, which type of seed bank (transient or persistent) can be expected? (b) Do EGR characteristics, such as roof age and substrate depth, influence seed bank seed density and species richness? (c) Can differences in successional stages be noticed in the seed bank and vegetation on EGRs? Location Flanders and Brussels regions of Belgium. Methods We conducted a field sampling campaign in which the vegetation and seed bank of 109 EGRs were sampled. Information on roof characteristics (e.g. age, substrate depth) was also collected. Seedling emergence was monitored under standardized greenhouse conditions. Species traits were collected from several existing databases. Data analysis consisted of (generalized) linear mixed modelling. Results A total of 18,466 seedlings from 139 different species emerged in the seed bank analysis. Average similarity between vegetation and the seed bank was low, and the majority of species could be considered spontaneous colonizers. Most species were classified as having a transient seed bank. Life forms were shown to be significantly different between vegetation and seed bank. Older EGRs had higher seed bank species richness and seed density, both of which were also positively influenced by sampling date and the vegetation species richness. A discrepancy between seed bank and vegetation successional state, characterized by functional signatures, was present for EGRs. Conclusions EGRs do develop seed banks and show parallels with seed banks in other novel ecosystems. Seed bank species richness and seed density are influenced by temporal and vegetation related factors. We can also confirm EGR seed bank succession and the discrepancy between seed bank and vegetation successional state, as often observed in natural and novel systems. Furthermore, we argue that seed banks can function as biodiversity reservoirs, which can be considered in EGR vegetation restoration if a set of several conditions is fulfilled. Overall, this study explores seed bank analysis on EGR as a novel ecosystem model. It thereby provides multiple insights in processes impacting EGR plant species diversity, both on and in the soil.