Evaluating the effectiveness of spontaneous vegetation for stormwater mitigation on green roofs

Green roofs can reduce stormwater runoff in urban areas by capturing rainfall. The extent of this capture is partially influenced by vegetation type and cover, which can be manipulated to optimise run-off reduction. However, in the absence of routine maintenance, planted green roof vegetation is often replaced by 'weedy' spontaneous species with unknown rainfall retention qualities. To better understand the role of spontaneous vegetation in green roof stormwater mitigation, we undertook a 100-day rainfall simulation involving 14 plant species that occur spontaneously on green roofs in Mediterranean-type climates. Green roof modules were filled with either 7 cm (shallow) or 14 cm (deep) substrate. The substrate was either left bare or sown with the spontaneous species community, which established approximately 100 % cover prior to the beginning of the rainfall simulation. During the simulation, modules were subjected to a "dry" and then a "wet" rainfall phase, each based on historical climate records from Melbourne, Australia. The "dry" treatment replicated the timing and depth of the driest rainfall period on record, while the "wet" treatment applied rainfall depths randomly selected from the 90th, 95th, and 99th percentiles of recorded rainfall. Rainfall retention, evapotranspiration, time to initiation of runoff and soil water content was measured for 17 rainfall events. Spontaneous vegetation cover and both species and functional diversity were measured at the end of each rainfall phase, and biomass was measured at the end of the wet phase. During the dry phase, modules with spontaneous vegetation cover retained 88 % of applied rainfall regardless of substrate depth and had 6 % greater retention than bare substrate. During the wet phase, deep substrate modules with spontaneous vegetation cover had 30 % greater retention than other treatment combinations. At the end of the wet phase, spontaneous vegetation in deep substrate had 42 % greater biomass, 19 % greater coverage and more than twofold greater functional richness than in shallow substrate. These findings demonstrate that spontaneous vegetation can increase stormwater retention on green roofs relative to bare substrate and have similar retention performance to commonly utilised species. However, the extent to which stormwater mitigation on green roofs is enhanced by spontaneous vegetation is dependent on factors that are more important for rainfall retention, such as substrate depth and rainfall patterns.