The influence of plant species and water dynamics on nitrogen removal within stormwater biofilters

Stormwater biofiltration systems harness the processing capacity of plants, microbial communities and layered filter media to reduce excess stormwater runoff and pollutants generated by urban environments. These systems can satisfy multiple water quality treatment and flow regime restoration objectives within a concentrated space, but consistent treatment of nitrogen remains an elusive target. Variable performance stems from multiple nitrogen species and transformation processes, which can be stimulated or inhibited by a myriad of biogeochemical influences. These interactions have been extensively studied across natural and modified environments, but the complex inter-dependencies and feedbacks remain poorly understood. Even less is known of internal biofilter nitrogen processing, yet biofiltration systems are unique in their engineered design and highly ephemeral water and nutrient availability. Previous biofilter studies have demonstrated performance sensitivity to vegetation presence, plant species and extreme drying, and the benefits of a saturated zone and carbon source at depth. However, the lack of process understanding is restricting further design optimisation. In order to further characterise biofilter nitrogen performance, a laboratory-scale study using 245 experimental columns was undertaken with variation in plant species, inclusion or exclusion of a saturated zone and carbon source, wet and dry inflow frequencies and non-vegetated controls. This study aimed to identify and quantify processes for the first time using an isotope tracer and provide guidance on plant species selection and system design. Performance showed greater consistency and effectiveness across species than previous multiple-species biofilter studies, likely due to recent trends in design specifications towards lower media nutrient content. Under wet conditions, nitrate (NO3-) dominated the effluent and dictated performance variation. During these frequent inflows the saturated zone offered little additional treatment to stormwater immediately passing through the filter, but provided ongoing treatment to the volume stored between events. This only benefitted biofilters planted with less effective species and thereby acted to reduce variation between species. Oxygen availability fluctuated in the saturated zone; stormwater inflows delivered oxygen but anaerobic conditions re-established rapidly. The benefits of a saturated zone could be partially offset by greenhouse gas production, with instances of elevated nitrous oxide (N2O) and methane (CH4) concentrations, but concentrations were highly variable, typically low, and consumption processes may prevent emissions. An isotope tracer (15NO3-), applied on three occasions in the wet and dry, indicated the majority of stormwater NO3- was initially consumed by biotic assimilation (between 58 - 100%), while denitrification contributed only minor processing in the saturated zone (0 - 22%) in vegetated biofilters. Denitrifying bacteria appeared to receive only the NO3- remaining after assimilation and therefore the contribution by denitrification to nitrogen removal tended to be higher alongside less effective plant species. Hence, biofilter effectiveness correlated positively with plant assimilation. As a result, desirable species characteristics in the wet period reflected efficient uptake capacity, including high biomass and high root length, mass, surface area and length of fine roots, possessed by select sedges, reeds and trees (Leptospermum continentale, Juncus spp., Carex spp. and Melaleuca incana). However, prolonged dry periods reversed many of the relationships evident in wet conditions. Assimilation still effectively consumed most incoming NO3-, but inter-event desiccation strongly influenced performance and water conservation became critical. Effective performance correlated with a slower growth rate, low plant mass and photosynthetic capacity, short shoots and limited above ground mass. These traits embody the lawn grasses, which performed exceptionally well during dry conditions, possibly due to their low stature and high ground coverage, but also experimental artefacts. Consequently, their performance requires validation at the field scale. Non-vegetated controls with saturated zones were also effective across the dry period due to minimal water loss. Drying additionally introduced the challenge of poor organic nitrogen removal. Unlike NO3-, both particulate and dissolved organic nitrogen concentrations (PON and DON) showed little sensitivity to plant species or the presence of a saturated zone and carbon source. The findings indicate the need for designs to conserve moisture and incorporate a diversity of species characteristics and broad plant types in systems. The dominance of biotic assimilation underlines the critical need to understand the accumulation and turnover of organic matter in stormwater biofiltration systems, and determine if denitrification becomes a significant removal pathway within mature systems.