Nitrate removal and secondary effects of a woodchip bioreactor for the treatment of subsurface drainage with dynamic flows under pastoral agriculture

Highlights

• Hydraulic residence time, available organic carbon, and inflow nitrate load affected nitrate removal.

• Long hydraulic residence time (>5 days) resulted in hydrogen sulphide production.

• Short hydraulic residence time (<4 days) resulted in net nitrous oxide production.

• Enhancing nitrate removal from dynamic drainage requires additional readily available carbon during high flows.

• High retention (89%) of dissolved reactive phosphorus in 2nd year merits further investigation.

Abstract

While enabling economically viable use of poorly drained soils, artificial subsurface drainage has also been found to be a significant pathway for nutrient transfers from agricultural land to surface waters. Thus, mitigating the impacts of agriculture on surface water quality needs to address nutrient transfers via subsurface drainage. Woodchip bioreactors are a promising mitigation option as demonstrated under arable agriculture in the mid-west of the USA. However, research is needed to ascertain their efficiency in removing nutrients from very flashy drainage flows common in New Zealand (NZ) pastoral agriculture and any possible pollution swapping (e.g. reduction of leaching losses vs. greenhouse gas emissions). Accordingly, a lined 78-m³ woodchip bioreactor was constructed on a dairy farm in the Hauraki Plains (Waikato, NZ) with a drainage area of 0.65 ha. Rainfall, flow, hydrochemistry and dissolved gases in the inflow and outflow were monitored for two drainage seasons (part of 2017, 2018). Based on the nitrate-N fluxes, the estimated nitrate removal efficiency of the bioreactor was 99 and 48% in 2017 and 2018, respectively. The higher removal efficiency in 2017 could be attributed to two reasons. Firstly, the substantially longer hydraulic residence time (HRT) of the water in the bioreactor (mean = 21.1 days vs 4.7 days in 2018) provided more opportunity for microorganisms to reduce the nitrate. A strong positive relationship between HRT and removal efficiency was also observed within the 2018 drainage season. Secondly, denitrification was supported in 2017 by greater electron donor availability. Evidence of this was the higher mass of DOC discharge from the bioreactor (318 mg C L⁻¹ of bioreactor volume vs 165 mg C L⁻¹ in 2018). Removal rates in the bioreactor varied from 0.67–1.60 g N m⁻³ day⁻¹ and were positively correlated with inflow nitrate loads. Pollution swapping was observed during the start-up phase of the bioreactor in both years (DOC, and DRP only in 2017) and during periods with very long HRTs (hydrogen sulphide (H₂S) and methane (CH₄) production). Substantially elevated discharges of DOC and DRP, as compared to inlet conditions, occurred during the initial start-up phase of the bioreactor in 2017 (3 to 3.5 pore volumes of the bioreactor), but only slightly elevated DOC and decreased DRP discharges were observed when drainage flow resumed at the start of the 2018 drainage season. Unexpectedly, cumulative DRP removal during the 2018 drainage season amounted to 89% of the DRP inflow into the bioreactor. Long HRTs (>5 days) enabled high nitrate removal efficiency (≥59%) and promoted complete reduction of nitrate to harmless dinitrogen gas but also promoted strongly reduced conditions, resulting in the production of H₂S and CH₄. On the other hand, short HRTs (<4 days) only allowed for moderate nitrate removal efficiency (≤43%) and constrained complete reduction of nitrate resulting in higher nitrous oxide concentrations in the outflow as compared to the inflow. Thus, nitrate removals above 50% were not able to be achieved without inducing H₂S and CH₄ generation. However, it may be achievable when the microbial community is provided with an additional source of readily available carbon during the critical periods when hydraulic flow and concomitant N load peaks occur.

Introduction

Artificial drainage has been instrumental in the viable use of poorly drained soils for agriculture. However, artificial drains can also provide a pathway for the fast transfer of unattenuated nutrients to streams and rivers (Algoazany et al., 2007; King et al., 2015a; Arenas Amado et al., 2017). This is a concern as the contribution of tile drainage to streamflow has been found to range from 42 to 86% in catchment studies conducted in agricultural lands in Ohio and Illinois in the USA and in Ontario, Canada (King et al., 2015b). In an effort to mitigate the impacts of artificial drainage on surface water quality, several measures have been proposed, including controlled drainage (Tan et al., 1999; Ballantine and Tanner, 2013) and denitrifying bioreactors (Schipper et al., 2010; Christianson et al., 2012b; Addy et al., 2016). A denitrifying bioreactor is fundamentally a pit filled with a source of carbon (e.g. woodchips, corn husks), which microorganisms use to transform nitrate through the process of denitrification into gaseous forms of nitrogen, mostly dinitrogen gas (N₂). Denitrifying bioreactors have been used to treat shallow contaminated groundwater in the form of a permeable reactive barrier (sometimes referred to as ‘denitrification wall’) intercepting shallow lateral flow (Schipper and Vojvodic-Vukovic, 1998; Robertson et al., 2000; Schipper et al., 2005; Barkle et al., 2008). The technology has also been used to treat discharge from artificial drainage systems. This can occur at the edge of the field by intercepting tile drainage before it discharges into open drains or in the open drains themselves (as an in-ditch bioreactor) (Robertson and Merkley, 2009; Schipper et al., 2010; Addy et al., 2016; Pfannerstill et al., 2016; Christianson et al., 2017a; Sarris and Burbery, 2018). The earliest known field-scale application of a bioreactor for treating artificial drainage using woodchips was in the 1990s in Canada (Blowes et al., 1994; Robertson et al., 2000). Since then, the suitability of field-scale bioreactors as a mitigation option for the impacts of agricultural drainage has been investigated worldwide, such as in Canada (van Driel et al., 2006; Robertson et al., 2009; Husk et al., 2017), Denmark (Bruun et al., 2016b; Bruun et al., 2017; Carstensen et al., 2019), Germany (Pfannerstill et al., 2016), Ireland (Fenton et al., 2016), New Zealand (Hudson et al., 2018; Goeller et al., 2019), and the USA (Chun et al., 2010; Christianson et al., 2012a; Ghane et al., 2015; Hassanpour et al., 2017). In the US, bioreactors have already been accepted as one of the US Department of Agriculture's Conservation Practices (Standard No. 605) and are being adopted increasingly in cropped lands (Christianson et al., 2012a; Hartz et al., 2017). However, a different bioreactor design and operation than in the US is required for the shallower subsurface drainage systems in lowland areas with accompanying highly variable flows and nitrate concentrations common in many agricultural lands, including pastoral lands in New Zealand (NZ). These conditions contrast with the generally deeper drainage systems in the US, which are often fed by snow meltwater and therefore treat a much steadier inflow and N loadings (Hassanpour et al., 2017).

Several investigations on bioreactors have been conducted in NZ, but these were mainly under laboratory settings with controlled environments (Cameron and Schipper, 2010; Warneke et al., 2011c; Cameron and Schipper, 2012). Limited field-scale investigations dealt with treating wastewater effluents from domestic, glasshouse, dairy shed, or laboratories (Warneke et al., 2011a; Warneke et al., 2011b; Tanner et al., 2012; Rambags et al., 2016) or intercepting shallow groundwater as a ‘denitrification wall’ (Schipper et al., 2004; Schipper et al., 2005; Barkle et al., 2008; Long et al., 2011). Several field-scale studies targeted treating artificial drainage water (Hudson et al., 2018; Goeller et al., 2019), but these were either not on an area with highly variable and seasonal flows and/or not considering potential secondary effects.

Bioreactors have been found to be effective in removing nitrate from drainage water but detrimental side effects (i.e., pollution swapping) have also been reported. Removal efficiency (RE) of nitrate from artificial drainage by bioreactors has been reported to range from 12 to 76% of the nitrate load (Jaynes et al., 2008; Christianson et al., 2012b; Hassanpour et al., 2017). Removal rates (RR) have been found to vary between 0.01 and 15 g nitrate-N m⁻³ day⁻¹ (Schipper et al., 2010; Christianson et al., 2012a; Addy et al., 2016; Hassanpour et al., 2017; Griessmeier et al., 2019). On the other hand, negative side effects include; high concentrations of dissolved organic matter and/or phosphorus in the outflow especially soon after installation, emission of greenhouse gases such as nitrous oxide (N₂O) and methane (CH₄), and production of odorous hydrogen sulphide gas (H₂S) (Schipper et al., 2010; Herbstritt, 2014; Healy et al., 2015; Weigelhofer and Hein, 2015). An incomplete denitrification process results in excess N₂O emitted through the surface of the bioreactor or discharged as dissolved gas in the outflow. Moreover, denitrification – being a microbially-mediated process – generally follows the succession of electron-accepting processes based on the energy generated (O₂ > NO₃⁻ > Mn (IV) > Fe (III) > SO₄²⁻ > CO₂) (McMahon and Chapelle, 2008). Thus, if the bioreactor has become strongly reduced with the absence or very low concentrations of dissolved oxygen (O₂) and nitrate, sulphate (SO₄²⁻) and carbon dioxide (CO₂) reduction (methanogenesis) may occur, resulting in the production of H₂S and CH₄, respectively.

Several studies have looked into the relationships between environmental conditions (e.g. flow, temperature, etc.) and nitrate removal and/or greenhouse gas production, but these were mostly done in controlled settings in laboratories (Greenan et al., 2009; Christianson et al., 2011; Healy et al., 2012; Christianson et al., 2013a; Healy et al., 2015; Weigelhofer and Hein, 2015; Bruun et al., 2016a; Hoover et al., 2016; Lepine et al., 2016; Bock et al., 2018a; Soupir et al., 2018). There are also limited field-scale or in situ studies dealing with agricultural drainage water (Elgood et al., 2010; Christianson et al., 2013b; David et al., 2016; Hassanpour et al., 2017; Davis et al., 2019; Goeller et al., 2019; Martin et al., 2019). Moreover, some of these studies also applied constraints deviating from natural flow conditions (Davis et al., 2019; Martin et al., 2019). Thus the main objective of this research was to assess the applicability and performance of denitrifying bioreactor technology to reduce nitrate loads from subsurface drains with very flashy drainage flows and variable nitrate concentrations as common in New Zealand pastoral agriculture. We aimed to identify the factors affecting the performance and pollution-swapping side-effects of a bioreactor receiving highly variable pastoral drainage flows and concentrations, and to identify potential modifications to optimise its efficiency.