Vegetated (green) roofs have become common in many cities and are projected to continue to increase in coverage, but little is known about the ecological properties of these engineered ecosystems. An important question is what the key mechanisms for nitrogen (N) and phosphorus (P) cycling are, and how tightly these nutrients are retained within the system. Ideally, green roofs will tightly recycle nutrients so as to maintain roof green-ness while minimizing downstream nutrient pollution. However, current understanding of green roof nutrient dynamics comes mainly from short-term studies of nutrients in runoff from newly-constructed systems, giving little insight into likely changes in nutrient storage over the roof’s 50+ year life span. To address this knowledge gap, we examined long-term variation in nutrient stocks and fluxes and the vegetation community, using a chronosequence study with 17 similar extensive (shallow substrate) green roofs installed over 22 years in Malmö, Sweden. We anticipated that the roofs, initially nutrient-poor, would accumulate N and P in substrate until reaching steady-state at an intermediate age. We also expected that changes in the substrate would be accompanied by changes in the vegetation community and nutrient retention. To test these hypotheses, stocks of C, N and P in vegetation and soil were measured along the chronosequence, as were the vegetation community composition and concentrations of nutrients in roof runoff.
We found that over the course of the 22-year chronosequence, green roof substrate depth approximately doubled, substrate C and N increased by approximately 7-fold, and substrate P increased 4-fold, indicating a buildup of organic nutrient-rich soil. There was no indication of a leveling off of substrate nutrient accumulation over time. Plant community composition and diversity were expected to vary with age as a function of nutrient availability and successional development, but in an initial October sampling, there were no significant differences in Shannon diversity or total plant coverage with age. However, vegetation communities did shift over time: lichens, for instance, were only observed on roofs that were 15+ years old, with coverage up to ~20% on some of the older roofs. Vegetation sampling in other seasons is ongoing, as are nutrient runoff measurements. Many gaps remain in the understanding of nutrient dynamics in green roof ecosystems. We encourage the next generation of studies to explore the role in nutrient cycling of microbial communities relative to plant communities, and how these communities interact and change over seasonal and decadal time scales.