Rooftop plants: Community and nutrient dynamics of New York City green roofs

Green, vegetated, roofs mitigate urban environmental degradation, thereby improving human health, and increasing available habitat space. Compared to succulent species, non-succulent plant species increase the productivity and ecological function of green roofs, but survivorship is generally low because of stressful environmental conditions. Two different non-succulent eight-species assemblages that were derived from native plant communities (Hempstead Plains: HP, Rocky Summit: RS) were established on ten New York City rooftops. While each assemblage initially had equal proportions of grasses and forbs, grasses dominated the HP assemblage and forbs dominated the RS assemblage after 23 months. Within-roof factors explained 71% of variation in survivorship, with biotic factors accounting for 54% of the overall variation, and abiotic factors explaining 17% of the variation. Among roofs, greater solar radiation and colonist cover correlated with lower survivorship. Sixty-three colonist species were recorded and colonist composition differed among roofs, but 12 species accounted for 90% of colonist dry mass. Eighty percent of the dry mass from the HP assemblage were forb species, while 31% of the dry mass collected from the RS assemblage were forb species. In June and July, non-grass species accounted for over 95% of dry mass, while grass species accounted for over 95% of dry mass in August. Prior to planting the HP and RS resident assemblages, plant composition of green roof plots was assessed after a colonization period of May to August. Species composition differed among roofs, but functional diversity indices, comprised of predominantly annual (84%) and grass (72%) species, did not. Portulaca oleracea (succulent) and Amaranthus spp. (forb) were identified during colonist surveys. In a rooftop microcosm experiment, three species of Amaranthus spp. and P. oleracea reduced runoff volume (22–69%) compared to the control, yet P. oleracea reduced runoff volume (29–61%) and total dissolved phosphorous (21–39%) less than the Amaranthus spp. Overall, these results suggest that long-term plant composition and function of green roofs depend on both local and regional abiotic and biotic factors. Accounting for the effects of regional variation on local green roof vegetation will be critical for effective citywide implementation of ecologically beneficial green roofs