ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2019

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1790

Biogeochemical Processes inDenitrifying Woodchip Bioreactorsand their Application in the MiningIndustry

ALBIN NORDSTRÖM

ISSN 1651-6214ISBN 978-91-513-0617-9urn:nbn:se:uu:diva-380386

Dissertation presented at Uppsala University to be publicly examined in Axel Hambergsalen,Geocentrum, Villavägen 16, Uppsala, Friday, 17 May 2019 at 09:00 for the degree ofDoctor of Philosophy. The examination will be conducted in English. Faculty examiner:Professor Gunno Renman (KTH Royal Institute of Technology, Department of SustainableDevelopment, Environmental Science and Engineering).

AbstractNordström, A. 2019. Biogeochemical Processes in Denitrifying Woodchip Bioreactors andtheir Application in the Mining Industry. (Biogeokemiska processer i kvävebarriärer medtillämpning inom gruvindustrin). Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1790. 65 pp. Uppsala: Acta UniversitatisUpsaliensis. ISBN 978-91-513-0617-9.

This thesis evaluates passive denitrifying woodchip bioreactors (DWBs) for the removal ofnitrate (NO3

-) in neutral pH mine drainage, where water is passed through a carbon-rich porousmatrix (e.g. woodchips) for the reduction of NO3

- to nitrogen gas. DWBs have been used forthe removal of NO3

- from water in various settings and are expected to operate with littlemaintenance for at least a decade; however, the processes controlling the emission of greenhousegases and other undesirable by-products, as well as the magnitude and variability in NO3

-

removal rates and how these develop over time, are not completely understood and were thefocus of this thesis.

Water treatment in DWBs was investigated in laboratory-scale column tests and in a pilot-scale bioreactor installed at the Kiruna iron ore mine, northern Sweden. Denitrification was themajor pathway for NO3

- removal in both experimental systems. Incoming NO3- concentrations

(up to 30.0 mg N L-1) were removed to below detection limits at temperatures and hydraulicresidence times (HRTs) between 5-22°C and ~1.9-2.6 days, respectively, without substantialproduction of nitrite or ammonium (NH4

+). NO3- removal was incomplete in both systems

when HRTs decreased to ~1 day, and/or as temperature decreased below 5°C in the pilot-scalebioreactor, under which conditions an increased production of nitrous oxide (N2O) and NH4

+

was observed (relative to the NO3- reduced).

In the column tests, non-ideal flow was detected and solute transport was described using adual-porosity model. Stagnant zones not transmitting flow did not participate in NO3

- removaland the fraction of immobile water increased with increases in the advection velocity, suggestingthat bioreactor performance could be enhanced by emphasizing design with low advectionvelocities.

The study demonstrated that dominating biogeochemical processes varied with time inthe pilot-scale bioreactor. There was a decline in organic carbon export and increase in pHand alkalinity that, based on a stoichiometric mass-balance, was suggested to be the resultof a change in fermentation end-products that provided a carbon source to the denitrifyingcommunity. The decline in NO3

- removal rates and biogeochemical process diversity, and thepreferential selection of denitrifiers with the genetic capacity for reduction to N2O, but notN2, are hypothesized to arise from the temporal development of syntrophic structures betweenfermenters and denitrifiers.

Keywords: water treatment, mine drainage, nitrate, nitrous oxide, DNRA, fermentation,syntrophy, temperature

Albin Nordström, Department of Earth Sciences, LUVAL, Villav. 16, Uppsala University,SE-75236 Uppsala, Sweden.

© Albin Nordström 2019

ISSN 1651-6214ISBN 978-91-513-0617-9urn:nbn:se:uu:diva-380386 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-380386)

Avhandling presenterad vid Uppsala universitet för att bli offentligt granskad i Axel Hambergs-alen, Geocentrum, Villavägen 16, Uppsala, fredagen den 17 maj 2019 klockan 09:00 för Tek-nologie doktorsexamen. Disputationen kommer att genomföras på engelska. Opponent: Profes-sor Gunno Renman (KTH Kungliga Tekniska Högskolan, Institutionen för hållbar utveckling, miljövetenskap och teknik). Referat Nordström, A. 2019. Biogeochemical Processes in Denitrifying Woodchip Bioreactors and their Application in the Mining Industry. (Biogeokemiska processer i kvävebarriärer med till-lämpning inom gruvindustrin). Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1790. 65 pp. Uppsala: Acta Universitatis Upsali-ensis. ISBN 978-91-513-0617-9. Denna avhandling utvärderar passiva kvävebarriärer (eng. denitrifying fixed-bed bioreactors) som en metod för att minska kvävehalter i gruvvatten (neutralt pH), där vattnet rinner genom en porös matris rik på kol (här träflis) för att omvandla nitrat (NO3

-) till kvävgas. Kvävebarriärer har tidigare använts i olika miljöer för att minska nitrathalter i vatten och förväntas fungera med minimalt underhåll i upp till tio år. Dock så är de processer som kontrollerar utsläpp av växt-husgaser och andra biprodukter, samt variabiliteten i nitratreduceringshastigheter och hur dessa förändras över tid, fortfarande relativt okända i dessa system och var fokus i denna avhandling.

Kvävereningen i kvävebarriärer utvärderades i kolonnstudier samt i en barriär i pilotskala som installerades vid Kirunagruvan i Kiruna. Denitrifikation var den huvudsakliga processen varmed NO3

- reducerades i båda systemen. Inkommande NO3- koncentrationer (upp till 30.0 mg N L-1)

reducerades till under detektionsgränsen vid temperaturer mellan 5-22°C och uppehållstider mel-lan ~1,9-2,6 dagar, utan betydande produktion av nitrit (NO2

-) eller ammonium (NH4+). Då up-

pehållstiden minskades till ~1 dag förblev inkommande NO3- ofullständigt reducerat i båda sy-

stemen; vid låga temperaturer (<5°C) och/eller uppehållstider om ~1 dag i barriären så ökade produktionen av lustgas (N2O) och NH4

+ (relativt den reducerade mängden NO3-).

Preferentiellt flöde i träflismatrisen påvisades i kolonnstudierna och ämnestransporten besk-revs med en ’dubbel-porositets’-modell. Zoner i träflismatrisen som inte deltog i flödet ökade tillsammans med flödeshastigheten i systemet och bidrog inte till att minska nitrathalter, vilket tydde på att reningskapaciteten i dessa system kan ökas genom att främja låga flödeshastigheter.

Den här studien demonstrerade att de dominerande biogeokemiska processerna i den pi-lotskaliga barriären förändrades med tid. Exporten av organisk kol minskade medan pH och alkalinitet ökade med tid vilket, baserat på en stökiometrisk mass-balans, kunde förklaras av en ändring i slutprodukten från fermentation. Minskningen i nitratreduceringshastigheter och mångfalden av biogeokemiska processer, samt främjandet av denitrifierare med N2O som slut-produkt, diskuterades utifrån en utveckling av syntrofi mellan fermenterare och denitrifierare. Nyckelord: kväverening, gruvvatten, nitrat, lustgas, DNRA, fermentation, syntrofi, temperatur Albin Nordström, Department of Earth Sciences, LUVAL, Villav. 16, Uppsala University, SE-75236 Uppsala, Sweden. © Albin Nordström 2019 ISSN 1651-6214 ISBN 978-91-513-0617-9 urn:nbn:se:uu:diva-380386 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-380386)

Till min familj

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Nordström, A., Herbert, R.B., 2017. Denitrification in a low-

temperature bioreactor system at two different hydraulic resi-dence times: laboratory column studies. Environmental Technol-ogy 38(11), 1362-1375.

II Nordström, A., Herbert, R.B., 2018. Determination of the major biogeochemical processes in a denitrifying woodchip bioreactor for treating mine drainage. Ecological Engineering 110, 54-66.

III Nordström, A., Herbert, R.B., 2019. Identification of the Tem-poral Control on Nitrate Removal Rate Variability in a Denitri-fying Woodchip Bioreactor. Ecological Engineering 127, 88-95.

IV Nordström, A., Hellman, M., Hallin, S., Herbert, R.B., 2019. Microbial Controls on Net Production of NOx Species in a Deni-trifying Woodchip Bioreactor. Manuscript.

Reprints were made with permission from the respective publishers.

In addition, the following papers related to the thesis but not appended to it have been published in conference proceedings:

Nordström, A., Herbert, R., 2017. Field-scale denitrifying woodchip bioreactor treating high nitrate mine water at low tem-peratures. In Wolkersdorfer C, Sartz L, Sillanpää A (Eds.) Mine water and circular economy, 1087-1094. Herbert, R.B., Nordström, A., 2017. Leachate generation and nitrogen release from small-scale rocks dumps at the Kiruna iron ore mine. In Wolkersdorfer C, Sartz L, Sillanpää A (Eds.) Mine water and circular economy, 140-146.

Co-authorship

Paper I I was responsible for the construction of the laboratory columns, conducted all the experimental work, performed part of the chemical analyses, performed all of the data analysis, and wrote the paper. Roger Herbert designed the ex-periment, helped in writing and editing of the paper, and framing the research question. Paper II I shared responsibility for the construction and design of the pilot-scale deni-trifying woodchip bioreactor with Roger Herbert, but had the main responsi-bility for planning and conducting the field work. I performed the profile sam-pling, data analysis, and wrote the paper. Roger Herbert helped in writing and editing of the paper. Paper III I performed all data analysis and wrote the paper. Roger Herbert helped in writing and editing of the paper. Paper IV I performed all sampling, extracted DNA from the Sterivex filters, performed all data analysis, and wrote the paper. Maria Hellman extracted DNA from the woodchips and performed the qPCR. Maria Hellman, Sara Hallin, and Roger Herbert helped in writing and editing of the paper.

Contents

1. Introduction ............................................................................................... 11 1.1 Nitrogen in the mining industry ......................................................... 11 1.2  The Nitrogen Cycle ......................................................................... 14 

1.2.1 Denitrification ............................................................................. 15 

2. Denitrifying Fixed-Bed Bioreactors .......................................................... 17 2.1 Reactive bioreactor substrates ............................................................ 18 

2.1.1 Carbon release and microbial competition ................................. 18 2.2 NO3

- removal rates ............................................................................. 19 2.2.1 NO3

- concentrations .................................................................... 19 2.2.2 Temperature ................................................................................ 20 2.2.3 Organic carbon availability and quality ...................................... 20 

2.3 Concerns with bioreactor performance .............................................. 21 2.3.1 Dissolved by-products ................................................................ 21 2.3.2 Gaseous by-products ................................................................... 22 2.3.3 Non-ideal flow ............................................................................ 22 

3. Research objectives ................................................................................... 24 

4. Materials and methods .............................................................................. 25 4.1 Study area ........................................................................................... 25 4.2 Experimental design ........................................................................... 25 

4.2.1 Column experiment .................................................................... 26 4.2.2 Pilot-scale denitrifying woodchip bioreactor .............................. 28 

4.4 Numerical modeling approaches ........................................................ 31 4.4.1 Paper I: Solute transport and hydraulic conditions ..................... 31 4.4.2 Paper II: Determination of major biogeochemical processes ..... 33 4.4.3 Paper III: Variability in NO3

- removal rate ................................. 34 4.4.5 Paper IV: Variability in the NOx reducing community .............. 35 

5. Summary of Papers ................................................................................... 37 5.1 Paper I: Column study ........................................................................ 37 5.2 Paper II: Pilot-scale denitrifying woodchip bioreactor ...................... 38 5.3 Paper III: Variability in NO3

- removal rates ....................................... 41 5.4 Paper IV: Variability in NOx-reducing community ............................ 43 

6. General discussion .................................................................................... 45 

6.1 Conditions for sustainable operation .................................................. 45 6.1.1 Potential design improvements ................................................... 46 6.1.2 Long-term sustainability and the importance of syntrophy ........ 47 

6.2 Potential full-scale application ........................................................... 48 

7. Conclusions and outlook ........................................................................... 50 

8. Acknowledgements ................................................................................... 52 

9. Sammanfatting på svenska ........................................................................ 54 

10. References ............................................................................................... 58 

11

1. Introduction

Since the development of the Haber-Bosch process in the early 20th century, an increased use of nitrogen (N) in industrial products has resulted in an in-creased export of N to aqueous ecosystems. This increase in N exports has been associated with acidification, eutrophication, hypoxia and the creation of ‘dead zones’, leading to habitat degradation and losses in biodiversity in the receiving waters (Galloway et al., 2003; Diaz and Rosenberg, 2008). The Bal-tic Sea and the northern Gulf of Mexico are but two present-day examples of water bodies where increased nutrient inputs (primarily from agriculture) have exacerbated such issues (Bianchi et al., 2010; Carstensen et al., 2014; Conley et al., 2009). Nutrient enrichment (N and phosphorus, P) is one of the main reasons for 60% and 87% of surface water bodies in the European Union (EU) and Sweden, respectively, not achieving a ‘good ecological status’, primarily due to the increased use of fertilizers during the last century (EEA, 2018; Naturvårdsverket, 2018).

Agriculture is not the only anthropogenic source of N exports to aquatic ecosystems. Mining activities, which commonly use N-based explosives, can result in an increased flux of N to nearby surface water and groundwater, with potential negative consequences (e.g. Chlot et al., 2013). Regardless of the source, there is a need to abate N release to aqueous systems from industrial activities. There are a number of techniques that may be used for the removal of N in water, and this thesis focuses on one of them: denitrifying fixed-bed bioreactors for the removal of N in mine drainage.

1.1 Nitrogen in the mining industry Ammonium nitrate (NH4NO3)-based explosives are the most commonly used blasting agents in most industries where explosives are used, as they are com-paratively inexpensive, powerful, and safe to use (Meyers and Shanley, 1990). However, NH4NO3 is highly soluble in water and although techniques are used to prevent contact with water (cf. Forsyth et al., 1995), studies have shown that up to 28% of the NH4NO3 based explosives remains undetonated in min-ing operations (Lindeström, 2012; Morin and Hutt, 2009); even with the most efficient operations, at least ~12% is lost. The undetonated explosives even-tually dissolve in rainfall/snowmelt that infiltrates into waste rock piles (Degnan et al., 2016; Nilsson and Widerlund, 2017; VVT, 2015), from which

12

it is exported to nearby surface water and groundwater as primarily nitrate (NO3

-) (Lindeström, 2012). On a mass basis, N-release is conceivably the most substantial export of

elements from mining activities with potentially detrimental effects on aquatic ecosystems (Boliden, 2017). For example, between approximately 0.2 tons N (Lovisagruvan) and 219.3 tons N (Malmberget) are, on average, annually re-leased per Swedish mine to receiving waters such as lakes, coastal waters, and watercourses, varying with the scale of production and mining technique (Ta-ble 1); underground mining releases more N than open pit mining (Lind-eström, 2012; VVT, 2015). As a reference, this is equivalent to the annual N-release from ~12 to ~11,542 hectares of Swedish farmland, using an average N-release of 19 kg N per hectare of Swedish farmland (SMED, 2016) (Table 1); similar figures were reported for N-release from a typical mine in Finland (VVT, 2015), which was equivalent to the N-release from 360-7860 hectares of Finnish farmland. With a total Swedish farmland area of 2,549,700 hectares (Jordbruksverket, 2018), the average annual N-release from Swedish mines is equivalent to a mere ~1.1 ± 0.4 % of the average annual N leached from Swe-dish farmland (cf. Table 1) and is hence not significant on a national scale. However, a mine is more often a point release rather than a diffuse release, and the effect may be significant at local scales.

There are no recommendations for best available technologies (BATs) to reduce N discharge from mining activities although there are a handful of po-tential alternatives (cf. VVT, 2015). Techniques that may be used for the re-moval of NO3

- from water are broadly classified as either physiochemical (e.g. ion-exchange, reverse osmosis, filtration, and chemical reduction) or biologi-cal. Physiochemical techniques have the disadvantage of being operationally expensive by producing secondary waste products (e.g. brines) which need further disposal (Mohseni-Bandpi et al., 2013; VTT, 2015). Biological tech-niques harness energy from N transformations within the ‘nitrogen cycle’, which are naturally carried out by organisms where available forms of N may be completely transformed into the comparatively unavailable N2(g). Produc-ing no secondary waste products, biological techniques for NO3

- removal in mine drainage would potentially be considered as BAT if high process effi-ciency would be secured at a reasonable cost (cf. Forsyth et al., 1995; VVT, 2015). For the purpose of removing NO3

- in mine drainage, the technique se-lected will also be conditioned by its volumetric capacity as the cumulative discharge of NO3

- laden waters from mines may be enormous. For example, ~9 Mm3 of water with elevated concentrations of NO3

- was discharged in 2016 from the Kiruna iron ore mine (Kiirunavaara, Table 1; LKAB 2017). With such a large volume, approaches where several techniques are integrated for full scale treatment of mine drainage may be needed.

13

Table 1. Average annual release of nitrogen (N)a from the active metal mines in Sweden (as of August 2018; SGU, 2019) based on emission data (total N) between 2007 and 2017 from Naturvårdsverket (2019), and calculated equivalent farmland area for each N release

Mine Annual N-release (tons)b Equivalent farmland (ha)c Aitik 40.9 ± 16.4 2154.2 ± 865.4 Björkdalsgruvan n.a. n.a. Garpenberg 18.3 ± 3.4 965.8 ± 176.7 Gruvberget n.a. n.a. Kankbergsgruvan 16.9 ± 8.3 889.0 ± 435.9 Kiirunavaara 153.0 ± 67.1 8052.6 ± 3530.0 Kristinebergsgruvan 8.9 ± 1.8 466.6 ± 95.4 Leveäniemi 25.0 ± 9.1 1317.2 ± 478.4 Lovisagruvan 0.2 ± 0.0 12.1 ± 1.1 Malmberget 219.3 ± 74.7 11542.0 ± 3932.9 Maurliden n.a. n.a. Mertainen n.a. n.a. Renströmsgruvan 14.9 ± 4.0 784.4 ± 212.9 Tapuligruvan n.a. n.a. Zinkgruvan 13.8 ± 3.4 725.6 ± 177.4 Sum 511.3 ± 188.2 26909.6 ± 9906.1

a Primarily NO3--N. Annual NH4

+-N releases from the Aitik, Garpenberg, and Malmberget mines were on average 4.6% (n = 3), 12.4% (n = 7), and 2.6% (n = 6) of the annual total N release from the respective mines when reported together with total N release; NH4

+ data was unavailable for the other mines (Naturvårdsverket, 2019). bAverage (± SD) annual load of total N to water between 2007 and 2017. c Equivalent farmland based on an average release of 19 kg N per hectare of Swedish farmland (SMED, 2016).

14

1.2 The Nitrogen Cycle In order to treat N-laden waters, a complete understanding of the nitrogen cy-cle is required, which refers to a set of reduction-oxidation (redox) reactions where the oxidation state of N changes. The nitrogen cycle is commonly re-stricted to seven pathways: nitrogen fixation, assimilation, mineralization, ni-trification (nitritation + nitratation), dissimilatory reduction of nitrate (nitrite) to ammonia/ammonium (DNRA), denitrification, and anaerobic ammonium oxidation (anammox). These pathways for N transformations are all facilitated by specialized microbes who harness energy from the N transformations.

Figure 1. Seven pathways for nitrogen (N) transformation within the nitrogen cycle. Based on Stein and Klotz (2016).

Nitrogen fixation refers to the process during which nitrogen gas (N2) is re-duced to ammonia/ammonium (NH3/NH4

+) (Canfield et al., 2010; Galloway et al., 2003; Figure 1). Most organisms depend on the supply of NH4

+ from microbes fixating nitrogen, or through the process of assimilatory reduction of NO3

- to NH4+, for the synthesis of biomass (R~NH2; Figure 1). NH4

+ is returned to the environment through mineralization of biomass when organ-isms die (Canfield et al., 2010).

In oxygenized (aerobic) environments, NH4+ is transformed into NO3

- through nitrification including the intermediate production of NO2

-, and po-tential production of nitrous oxide (N2O) (Canfield et al., 2010). In environ-ments where there is a limited availability of oxygen (anoxic/anaerobic envi-ronments), NO3

- is reduced to NO2- and further into NH3/NH4

+ or N2, with the prevailing end-product of NO2

- reduction largely being determined by the rel-ative availability of electron donors in the system (Burgin and Hamilton, 2007; Kox and Jetten, 2015). In anoxic/anaerobic environments with a large surplus of electron donors (reducing agents) relative to the availability of NO3

-/NO2-,

15

DNRA is the preferential pathway resulting in the production of NH3/NH4+.

In contrast, in environments where there is a moderate availability of electron donors, NO2

- is preferentially reduced to N2, through the intermediate produc-tion of nitric oxide (NO) and N2O, in the denitrifying pathway (Canfield et al., 2010; Burgin and Hamilton, 2007). Lastly, in environments where there is a low availability of organic carbon, NO2

- and NH4+ may simultaneously be

transformed into N2 through anammox (Burgin and Hamilton, 2007; Gonzá-lez-Cabaleiro et al., 2015).

Biological techniques using the anammox or denitrification pathways are the only two sustainable options for the removal of NO3

- in water as they com-pletely transform bioavailable N (NO3

-, NO2-, NH4

+) into N2(g), resulting in the permanent removal of the excessive N from the aqueous system (N2(g) is unavailable as an N-source for most organisms). DNRA is undesired as NH4

+ is oxygen consuming and may transform into NH3 which is toxic to aquatic biota. Moreover, NH4

+ is more easily assimilated into biomass (Kox and Jet-ten, 2015; Rittman and McCarthy, 2001) and is released back into the system when the organism dies. Whereas anammox technologies have been applied at full scale (e.g. Lotti et al., 2014), anammox requires equivalent amounts of NO2

- and NH4+ (reaction 1), and in mining settings, the initial partial transfor-

mation of some NO3- to NH4

+ in order to promote the anammox pathway is required (cf. VVT, 2015; Figure 1).

→ 2 (1)

Because of the general dominance of NO3- in mine drainage (Lindeström,

2012), denitrification is the most direct pathway for the removal of NO3-.

1.2.1 Denitrification In denitrification, five electrons (e-) are needed to reduce one mole of NO3

- into N2 along with the consumption of hydrogen ions (H+), leading to an in-crease in pH following (net) denitrification (reaction 2).

5 6 → 3 (2)

The net reduction of NO3- in denitrification is accomplished in five separate

steps, with each intermediate NOx reduction being performed by a specialized reductase (enzyme performing reduction). Strictly speaking, however, the de-nitrifying pathway starts with the reduction of NO2

- [to NO] by one of two distinct nitrite reductases (Zumft, 1997) that are encoded by the (generally mutually exclusive) functional genes nirS and nirK (Graf et al., 2014; Jones and Hallin, 2010; Zumft, 1997). NO is further reduced into N2O by nitric oxide reductases (encoded by functional gene group nor; Zumft, 1997), and lastly

16

N2O is reduced to N2 by nitrous oxide reductases encoded by the two phylo-genetically distinct functional genes nosZI and nosZII (Jones et al., 2013). Mi-croorganisms able to perform all of the above intermediate reaction steps, in-cluding NO3

- reduction to NO2-, are sometimes referred to as ‘canonical deni-

trifiers’ (Stein and Klotz, 2016). Technologies using denitrification for the removal of NO3

- have the poten-tial for systematic accumulation of NO2

-, NO, and/or N2O because of an un-balanced production and reduction of NOx species in the denitrifying pathway (Betlach and Tiedje, 1981). This is undesired because of the toxicity of NO2

- and NO, contribution to global warming by NO and N2O, and ozone depleting effect of NO. Causes for unbalanced rates of production and reduction of in-termediate NOx species may be physiochemical, biological, or a combination of these. Examples of physiochemical factors observed to affect intermediate NOx accumulation are temperature, pH, concentration of inhibiting com-pounds, and the type and availability of electron donor (e.g. Warneke et al., 2011a; Pan et al., 2012; Morley et al., 2014).

Biological causes for intermediate NOx accumulation are, for example, the relative increased genetic potential for specific NOx reductions in the denitri-fying pathway. For example, an increased abundance of functional genes en-coding for nitrite reductase (nirS + nirK) relative to functional genes coding for nitrous oxide reductase (nosZ) have been associated with emissions of N2O from biological treatment systems (e.g. Warneke et al., 2011a). Many bacteria only possess partial gene inventories to perform denitrification and denitrifi-cation is generally catalyzed by several co-cultures of denitrifiers in nature (Kox and Jetten, 2015; Stein and Klotz, 2016). nirK-type denitrifiers are more frequently associated with denitrifying pathways truncated to N2O relative to nirS-type denitrifiers due to their less common association with nor and nosZ genes (Graf et al., 2014), whereas nirS-type denitrifiers are generally thought to perform complete denitrification (Graf et al., 2014; Hallin et al., 2018). The physiochemical and biological factors affecting intermediate NOx accumula-tion are, however, mutually dependent as nirS- and nirK-type denitrifiers, and nosZI and nosZII are thought to occupy different ecological niches (Hallin et al., 2018; Jones and Hallin, 2010), implying that the biological potential for intermediate NOx reduction and production are contingent on specific envi-ronmental conditions.

17

2. Denitrifying Fixed-Bed Bioreactors

Denitrifying fixed-bed bioreactors are low-maintenance treatment systems where water with high concentrations of nitrate (NO3

-) is passed through a porous material supplying electrons for the reduction of NOx through denitri-fication. These systems are commonly designed as large trenches or containers filled with a carbon-rich porous material, where the NO3

- laden water is passed horizontally or upwards through the system for the reduction of NO3

- to N2(g) (cf. Schipper et al., 2010). Denitrifying fixed-bed bioreactors have previously been used to remove NO3

- in e.g. agricultural run-off (Ghane et al., 2015), aquaculture effluents (Christianson et al., 2017), hydroponic effluent (Warn-eke et al., 2011b), and mine drainage (Herbert et al., 2014).

Research on denitrifying fixed-bed bioreactors has mainly focused on iden-tifying a suitable electron donor that supports complete denitrification and high NO3

- removal rates while suppressing undesirable secondary reactions such as DNRA and methanogenesis (producing methane, CH4) (e.g. Cameron and Schipper, 2010; Feyereisen et al., 2016; Gibert et al., 2008; Greenan et al., 2006; Volokita et al., 1996; Warneke et al., 2011a). Effort has also been put into understanding the environmental controls on NO3

- removal rates (e.g. Addy et al., 2016; Ghane et al., 2015; Halaburka et al., 2017; Robertson, 2010; Schmidt and Clark, 2013; Warneke et al., 2011a), N2O emissions (e.g. Warn-eke et al., 2011a), and defining and improving the hydraulic conditions of these systems (e.g. Cameron and Schipper, 2011; Christianson et al., 2013; Herbert, 2011; Jaynes et al., 2016). Despite recent advances in our understand-ing of the above, there is still considerable uncertainty in terms of variability in and between systems, and in the underlying processes that create this vari-ability. This thesis focuses on the environmental parameters that control the emission of undesired by-products (NO2

-, N2O, H2S, CH4, organic carbon), NO3

- removal rates, and hydraulic conditions in denitrifying bioreactors. Hence, current research of electron donor availability, carbon release, NO3

- removal rates, and hydraulic conditions in denitrifying fixed-bed bioreactors is summarized in the sections below.

18

2.1 Reactive bioreactor substrates The porous material supplying electrons for denitrification is commonly re-stricted to carbon-rich media due to its abundance in nature, low cost (in com-parison to e.g. zerovalent iron), and for being a strong reducing agent (cf. Ap-pelo and Postma, 2005, p. 438). Woodchips are the preferred carbon source in denitrifying fixed-bed bioreactors because of their high permeability, moder-ate reactivity, and preferential promotion of denitrifying bacteria, allowing for sustained NO3

- removal up to decades following installation (Cameron and Schipper, 2010; Gibert et al., 2008; Robertson, 2010; Warneke et al., 2011a). Several alternative carbon-rich materials may also be used, e.g. maize cobs, cornstalks, wheat straw, cardboard, newspaper, leaves, compost, sawdust, and more (e.g. Cameron and Schipper, 2010; Feyereisen et al., 2016; Gibert et al., 2008; Greenan et al., 2006; Volokita et al., 1996). Although carbon sources other than woodchips may promote higher NO3

- removal rates, they are gen-erally not preferred because a higher reactivity may lead to the faster depletion of carbon (decreased longevity of treatment) and an increased export of or-ganic carbon, while an increased export of NH4

+ may occur due to the promo-tion of environmental conditions favorable for the DNRA pathway (e.g. Cam-eron and Schipper, 2010; Feyereisen et al., 2016; Gibert et al., 2008; Greenan et al., 2006; Schipper et al., 2010; Volokita et al., 1996). In addition, alterna-tive carbon sources have been demonstrated to be prone to higher exports of N2O and CH4 (Feyereisen et al., 2016; Warneke et al., 2011a), some lack sup-port structures for the denitrifying bacteria (newspaper; Volokita et al., 1996), and carbon sources giving rise to low permeabilities have been noted to pro-mote non-ideal flow which reduces their efficiencies (e.g. sawdust, Herbert et al., 2014; Schipper et al., 2004; see below). Henceforth, the focus lies on de-nitrifying woodchip bioreactors.

2.1.1 Carbon release and microbial competition Water-saturated woodchips are noted for creating an environment where de-nitrifying bacteria are preferentially promoted over other bacteria (Warneke et al., 2011a), likely due to the moderate reactivity of woodchips that maintain a moderate availability of electron donors. The supply of usable carbon sub-strate for the denitrifiers (and other non-cellulolytic bacteria) is dependent on the rate at which cellulose, the primary carbon substrate in ‘woody’ biomass (Sun and Cheng, 2002), is hydrolyzed into carbon substrates that are accessi-ble to denitrifiers (primarily glucose, C6H12O6; cf. Palmqvist and Hahn-Häger-dahl, 2000; Schipper and Vojvodić-Vuković, 2001). The solubilized organic carbon substrates are the target of competition by multiple terminal electron accepting processes (TEAPs) which oxidize the organic carbon for the reduc-tion of various electron acceptors (e.g. NOx, SO4

2-) to inorganic carbon. In non-equilibrium systems, reactions that quickly dissipate energy gradients are

19

preferentially promoted (Vallino and Algar, 2016), and available carbon will be preferentially directed to denitrification once oxygen levels are low. Com-petition for the available carbon may also take place between intermediate processes within the denitrification pathway and may control emissions of NO2

- (Lilja and Johnson, 2016) or N2O (Pan et al., 2013) from denitrifying systems.

In systems where there is low availability or competition for a shared sub-strate, less exergonic reactions may occur in synergism with more exergonic reactions in syntrophic metabolic structures by increasing the overall rate of energy dissipation in the system (cf. González-Cabaleiro et al., 2015). Fer-mentation, the transformation of soluble carbohydrates into organic acids and alcohols, is classically considered as a low exergonic reaction. Since the prod-ucts of fermentation may be used as (secondary) carbon substrates for TEAPs, fermenters and e.g. denitrifiers may co-operate in syntrophic metabolic struc-tures where fermenters control the supply of organic carbon to denitrifiers (cross-feeding, e.g. Hanke et al., 2016).

2.2 NO3- removal rates

NO3- removal rates in denitrifying woodchip bioreactors generally vary be-

tween 2 and 22 g N m-3 d-1 although NO3- removal rates as high as 73 g N m-3

d-1 have been observed (Hassanpour et al., 2017; Schipper et al., 2010; m-3 pertains to the bioreactor volume). The variability in the observed NO3

- re-moval rates may be ascribed to a variability in the environmental conditions wherein the NO3

- removal rates were determined. In terms of optimizing NO3-

removal in these treatment systems, high NO3- removal rates are desired as

this would increase the cost-efficiency by decreasing the size of the required reactor for the removal of a given mass of NO3

-. NO3- concentrations, temper-

ature, organic carbon availability, and the time since start-up (‘maturity’) are regarded as the significant controls on the NO3

- removal rate (Addy et al., 2016), although the interdependencies of these factors and the importance of other controls are uncertain.

2.2.1 NO3- concentrations

Paradoxically, NO3- removal rates in denitrifying woodchip bioreactors are

generally regarded as being operationally independent of the magnitude of NO3

- concentrations (‘zeroth-order kinetics’, cf. Halaburka et al., 2017; Rob-ertson, 2010; Schipper et al., 2010; Schmidt and Clark, 2013), although NO3

- removal rates that are proportional to the incoming NO3

- concentration have also been described (‘first-order kinetics’, e.g. Jaynes et al., 2016). For deni-trifying woodchip bioreactors in general, NO3

- removal rates significantly de-crease when NO3

- concentrations decrease below ~10 mg N L-1 (Addy et al.,

20

2016), indicating non-zeroth concentration dependency of the NO3- removal

rate at low NO3- concentrations. The pattern of increasing concentration de-

pendence as concentrations decrease is indicative of Monod kinetics (Hoover et al., 2016).

2.2.2 Temperature Temperature is cited as the most substantial control on NO3

- removal rates in denitrifying woodchip bioreactors (Halaburka et al., 2019; Schmidt and Clark, 2013) and NO3

- removal rates in denitrifying woodchip bioreactors are ob-served to increase with a factor of ~2-8 with a 10°C increase (Cameron and Schipper, 2010; Elgood et al., 2010; Feyereisen et al., 2016; Hoover et al., 2016; Jaynes et al., 2016; Schmidt and Clark, 2013; Warneke et al., 2011a). The range in observed temperature dependence suggests a considerable vari-ability between studies and a low precision in our understanding of the tem-perature dependence of NO3

- removal rates in these systems (also evident in the compilation by Addy et al., 2016). This variable temperature dependency may be related to the variation in the natural composition of the NO3

- reducing communities within and between denitrifying woodchip bioreactors. The NO3

- reducing community in any given denitrifying woodchip bioreactor will be represented by a consortium of microbial species that are adapted to margin-ally different temperature ranges (cf. Schipper et al., 2014) and, reflecting this, the composition of the microbial community in woodchip bioreactors is ob-served to have a high degree of seasonality (e.g. Porter et al., 2015). This sug-gests that the kinetics of NO3

- removal may change with changing environ-mental conditions due to the preferential promotion of different microbial spe-cies.

2.2.3 Organic carbon availability and quality Nitrate removal rates in denitrifying woodchip bioreactors are significantly higher, and less sensitive to temperature changes, for the general duration of one year following bioreactor start-up that is associated with an increased ex-port of organic carbon from the system (known as ‘flush-outs’) (Addy et al., 2016; Cameron and Schipper, 2010; David et al., 2016; Hassanpour et al., 2017; Hoover et al., 2016). NO3

- removal rates have been observed to increase with the availability of organic carbon (Warneke et al. 2011a), and the initial high (and relatively temperature-insensitive) NO3

- removal rates following bi-oreactor start-up are suggested to relate to an initially higher fraction of solu-ble, easily degradable, carbon in the pore water (e.g. David et al., 2016; Has-sanpour et al., 2017), which is often described in terms of organic carbon ‘quality’. Indeed, the organic carbon ‘quality’ has been observed to decrease over time in denitrifying woodchip bioreactors (increased lignocellulosic in-

21

dex, cf. Ghane et al., 2018; Schmidt and Clark, 2013); biogeochemical reac-tion rates of processes feeding on carbon substrate of ‘high quality’ (easily available, labile) will be comparatively high and insensitive to changes in tem-perature, whereas processes feeding on carbon substrates of ‘low quality’ (less available, recalcitrant) will exhibit relatively low reaction rates and be rela-tively sensitive to temperature changes (Sierra, 2012). This suggests that the characteristic temporal variability in NO3

- removal rates in denitrifying wood-chip bioreactors are largely related to the ‘quality’ of organic carbon, and may reflect changes in the microbial community as this have been linked to the availability of organic carbon in these systems (Porter et al. 2015).

The expected longevity of these systems is often based on the total carbon content in the woodchip material which yields values ~40 years (e.g. Long et al., 2011). These estimates assume that the carbon delivery to the denitrifiers remain constant; if the carbon delivery chain breaks down, it is likely that the bioreactors will suffer a ‘premature’ termination in comparison to the stoichi-ometric longevity estimate (Schipper and Vojvodić-Vuković, 2001).

2.3 Concerns with bioreactor performance As discussed in the above section, there are several parameters that influence NO3

- removal rates (e.g. temperature, organic matter quality) and thus must be considered in the optimization of bioreactor treatment. While the degree of NO3

- removal is the most common criteria for evaluating bioreactor perfor-mance (Addy et al., 2016; Schipper et al., 2010), performance is also evaluated in terms of the release of undesirable by-products and hydraulic efficiency. These performance indicators are discussed in the following sections.

2.3.1 Dissolved by-products Dissolved by-products include NH4

+ and NO2-. NH4

+ may be produced from DNRA provided favorable environmental conditions which is generally char-acterized by the high availability of electron donors (organic carbon) relative to electron acceptors (NO3

-, NO2-) (Burgin and Hamilton, 2007). Alterna-

tively, NH4+ may be produced from the mineralization of organic matter under

aerobic conditions (section 1.2). NO2- may be produced from the unbalanced

rate of NO2- production (NO3

- reduction) and NO2- reduction (cf. Betlach and

Tiedje, 1981), and may be the result of competition for the available organic carbon substrate in denitrifiers able to perform complete denitrification (Lilja and Johnson, 2016). The environmental conditions that promote the relative production and accumulation of NH4

+ and NO2-, and how these changes over

time, is not well established in denitrifying woodchip bioreactor systems.

22

2.3.2 Gaseous by-products The potential production of greenhouse gases from the denitrifying woodchip bioreactors (NO, N2O, and CH4; and CO2) negatively affects bioreactor per-formance from a perspective of sustainability as we do not want to solve one environmental problem by creating another (‘pollution swapping’, see Healy et al., 2014).

CH4 emissions from denitrifying woodchip bioreactors have been observed to increase with temperature (Elgood et al., 2010; Warneke et al., 2011a), with CH4 being produced from the comparatively less energetic methanogenesis once all NO3

- was depleted by denitrification. Methanogenesis represents the final decay of the organic carbon source in the absence of other electron ac-ceptors (Schlesinger and Bernhardt, 2013). CH4 emissions from these systems are therefore likely not significant provided sufficient availability of NO3

- or other electron acceptors.

Previous studies have shown that up to ~10% of the reduced NO3- may be

exported from denitrifying woodchip bioreactors as N2O (David et al., 2016; Elgood et al.,, 2010; Feyereisen et al., 2016; Ghane et al., 2015; Greenan et al., 2009; Warneke et al., 2011), with most of the N2O emissions from the bioreactor being dissolved in water (e.g. Warneke et al., 2011). However, there are no previous studies (to the author’s knowledge) where NO emissions from denitrifying woodchip bioreactors have been determined. Increased N2O production has been observed in environments not promoting the growth of N2O reducers, or to preferentially promote NO2

- reducers relative to N2O re-ducers, with the relative promotion of N2O reducers being related to tempera-ture and the type of carbon source used in denitrifying fixed-bed bioreactors (Feyereisen et al., 2016; Warneke et al., 2011a). Differences in N2O emissions from denitrifying fixed-bed bioreactors using different sources of carbon may thus be related to the relative availability of the various carbon sources, and the dynamics of N2O emissions may be related to changes in the competition for the available carbon.

2.3.3 Non-ideal flow In a perfectly mixed bioreactor, water will come in contact with the entire pore space as it flows through the reactive material. For a flow Q and total pore volume V, the hydraulic residence time (HRT) is defined as V/Q. However, for most bioreactors, not all pores participate and flow may be concentrated along certain pathways. Preferential flow, alternatively short circuiting or non-ideal flow, is a non-generic descriptive term referring to the channeling of flow along preferred pathways which offer less resistance to flow, as for ex-ample in porous media where the permeability matrix is anisotropic (Hen-drickx and Flury, 2001; Gerke, 2006). In denitrifying woodchip bioreactors, this may result in an overall decrease in NO3

- removal efficiency because of

23

an inefficient volume utilization (Cameron and Schipper, 2011), and may trig-ger less energetic reactions such as methanogenesis in zones not transmitting flow (Grieβmeier et al., 2017).

The occurrence of non-ideal flow will depend on bioreactor hydraulics. Woodchips used in denitrifying bioreactors have grain sizes of typically a few centimeters, with the individual woodchip grains being separated by relatively large voids, in comparison to the pore space internal to each woodchip grain (Jaynes et al., 2016). This is conceptually equivalent to a dual-porosity system in which solute transport primarily takes place within the larger void space between the woodchip particles (‘mobile zone’), while no flow takes place within the woodchip fragments (‘immobile zone’) (Jaynes et al., 2016). How-ever, studies have shown that immobile zones are not only restricted to the intra-grain porosity, but may also incorporate inter-grain porosity (e.g. Jaynes et al., 2016). Immobile water zones in denitrifying woodchip bioreactors may be substantial (e.g. Cameron and Schipper, 2011; Christianson et al., 2013; Jaynes et al., 2016). For example, Jaynes et al., (2016) estimated that ~78% of the pore space in the bioreactor studied could be considered as immobile, pre-sumably due to the presence of ‘dead end’ pores and centralized flow. Whereas solutes may be exchanged between the intra-woodchip pore space and the void space between the individual woodchip through diffusion (cf. Herbert, 2011; Jaynes et al., 2016), the availability of NO3

- within immobile zones will be limited if the rate of solute exchange is low.

There are several techniques that have been suggested as a way to minimize the occurrence of non-ideal flow behavior in denitrifying bioreactors, such as the use of ‘baffles’ and/or changing the placement of the inlet and outlet (Christianson et al., 2013). Furthermore, it has been suggested that short-cir-cuiting increases with hydraulic loading to denitrifying woodchip bioreactors (David et al., 2016), and the volume of immobile water zones may be affected by the HRT applied in the system.

24

3. Research objectives

The work presented within this thesis was conducted as part of the VINNOVA project ‘miNing’ wherein three different techniques for the removal of NO3

- in mine site drainage were evaluated (denitrifying fixed-bed bioreactors, wet-lands, and pond denitrification)1. In this thesis, the denitrifying woodchip bi-oreactor technique is evaluated and implemented as a potential treatment method for the removal of NO3

- in mine site drainage with the following spe-cific research objectives:

Determine the hydraulic conditions and solute transport character-istics in woodchip media, and how these affect NO3

- removal rates (paper I)

Determine the capacity of a denitrifying woodchip bioreactor for the removal of NO3

- from mine drainage in a low temperature en-vironment (papers I and II).

Considering that denitrifying woodchip bioreactors are designed to operate passively over at least a decade, we were also interested in understanding how these systems behave and develop over time in order to determine their long-term sustainability, with the following specific research objectives:

Determine the major biogeochemical processes in a denitrifying woodchip bioreactor, their interactions and how these develop over time, and the processes that control this development (papers II, III, IV).

Identify the parameters that controls the variability in NO3- removal

rates in denitrifying woodchip bioreactors (paper III) Identify the parameters that control the relative abundances of func-

tional genes involved in denitrification, DNRA, and anammox (the ‘NOx-reducing community’) and their association with the produc-tion of NO2

-, N2O, and NH4+ (paper IV).

1 See Choudhury (2018), Hellman (2018), and Nilsson (2018) for additional information.

25

4. Materials and methods

4.1 Study area The Kiruna iron ore mine (Kiirunavaara, Table 1) is operated by the mining company Luossavaara-Kiirunavaara Aktiebolag (LKAB) and is located above the Arctic Circle in the subarctic climate of northern Sweden. The mean an-nual air temperature in the Kiruna region was -3°C, with mean monthly air temperatures varying between -13.9°C in January and 12°C in July for the period 1961-1990 (SMHI, 2018a). For the same period, snow covered the ground for 200-225 days per year (SMHI, 2018b).

The Kiruna iron ore (apatite-iron oxide) is mined below ground through sublevel caving and the unprocessed rock is hoisted to the surface where it is screened (separation of ore-grade and waste rock). The ore-grade rock is then further processed, in the presence of water (so-called process water), in order to increase the concentration of valuable elements. The waste rock is depos-ited in piles on the ground surface or is used for secondary purposes (LKAB, 2018); between 2015 and 2017, ~9.2 M tons of waste rock were yearly depos-ited in piles (LKAB, 2016; 2017; 2018).

Process water that is not recirculated within the mine facilities is released into the recipient lake Mettä Rakkurijärvi, following sedimentation and clari-fication steps in a series of pond systems. In 2015 and 2016, ~9 Mm3 of pro-cess water (neutral pH) with an average NO3

- concentration of ~28 mg N L-1 was released into Mettä Rakkurijärvi (LKAB, 2016; 2017; 2018), with an es-timated annual export of ~231 tons of N as NO3

- (primary N-species) (LKAB, 2018); more than 99% of the NO3

- released was retained within the Rakku-risystem (LKAB, 2016; 2017). In 2015, the ecological status of the Mettä Rak-kurijärvi lake was classified as ‘moderate’ on the basis of nutrient status, and elevated concentrations of NO3

- were noted (VISS, 2019).

4.2 Experimental design This thesis is based on the results of two woodchip bioreactor systems: a la-boratory-sized denitrifying woodchip bioreactor (laboratory columns, paper I), and a pilot-scale denitrifying woodchip bioreactor installed at the Kiruna iron ore mine, northern Sweden (papers II-IV). The details on the construction

26

of these systems, and procedures for sampling and chemical analysis, are laid out in papers I and II, and only broadly summarized here.

4.2.1 Column experiment The column experiment was performed in order to test the NO3

- removal ca-pacity of a reactive mixture of decorticated pine woodchips and sewage sludge at the low temperatures expected at the study site, and to define the hydraulic conditions and solute transport in the woodchip media and their effect on NO3

- removal rates. Three Plexiglas columns (Figure 2) were filled with a mixture of decorcitated pine woodchips and activated sewage sludge (5:1 v/v) from Kungsängen waste water plant (Uppsala, central Sweden). Baffles were in-serted in one of the columns as a potential technique to reduce preferential flow behavior by preventing flow along the edges of the columns (Figure 2). The columns were named following their differences in characteristics, with the size of woodchip fragments varying between the different columns: large-chip (LC) column, small-chip (SC) column, and small-chip column with baf-fles (SCB).

Figure 2. (a) Experimental set-up for the column experiment and conceptual view of flow lines (arrows in columns) in columns without baffles (LC and SC), and in col-umn with baffles (SCB). Note that the SC and SCB columns were placed in an incu-bator for temperature adjustments. (b) Top-view of bottom plate of the columns showing the radial grooving distributing inlet flow at the center of the plate to the edges (same principle for top plate: collecting effluent from the entire cross-section of the column).

Columns were operated in upward flow mode and the inlet and outlet of the columns were designed as to distribute and collect influent and effluent from the entire cross-section of the column: bottom and top cover plates of the col-umns were radially grooved and a 1 cm thick sand layer (medium-grained) were placed at the bottom of each column (Figure 2).

27

Sewage sludge diluted in distilled water was recirculated in the columns using a peristaltic pump for an initial 10 days to establish the denitrifying com-munity within the woodchip media; 110 mg L-1 sodium acetate (NaCH3COO) was added as an external carbon source during the recirculation stage. The columns were then flushed with distilled water for the duration of two days to remove residual acetate, and thereafter continuously fed with an aqueous so-lution containing 30 mg NO3

--N L-1 and 400 mg SO42- L-1 using a peristaltic

pump for 92 days (see Table 2 for operating conditions). Influent SO42- con-

centration was decreased to 100 mg L-1 toward the end of the experiment due to interferences between peak areas during analysis using ion chromatog-raphy.

Table 2. Calculated theoretical hydraulic residence time (HRT) and operational tem-peratures in the large chip column (LC), short chip column (SC), and short chip col-umn with baffles (SCB) for the duration of the column experiment.

Days Theoretical HRT (days) Temperature (°C) LC SC SCB LC SC SCB

a-12 to -2 (Recirculation) 1.1 1.2 1.1 22.0 22.0 22.0 -2 to 0 (Acetate flushing) 1.1 1.2 1.1 22.0 22.0 22.0 0 to 21 2.5 2.7 2.4 22.0 22.0 22.0 21 to 33 2.5 2.7 2.4 22.0 12.0 12.0 33 to 59 2.5 2.7 2.4 22.0 5.0 5.0 59 to 80 0.8 0.9 0.8 22.0 5.0 5.0 80 to 92 0.8 0.9 0.8 22.0 12.0 12.0

To determine the extent of non-ideal flow behavior, two tracer tests using chloride (Cl-, 100 mg L-1) as a conservative tracer were performed in the SCB at the ‘high’ and ‘low’ HRTs (2.4 days and 0.8 days, respectively; Table 2) following the termination of the normal sampling scheme (day 92).

Pore water profiles were sampled on days 47, 80, and 92 of the column experiment (at different temperatures and HRTs, see Table 2) whereas effluent was grab sampled approximately three times a week. Samples were filtered (0.2µm polyethersulfone membrane) and frozen until analysis; samples in-tended for the analysis of cations were acidified through the addition of 0.005 M sulfuric acid (H2SO4). For the tracer tests, a CF-2 fraction collector (SPEC-TRUM®) was used to collect samples. Effluent and pore water concentrations of NO3

-, NO2-, SO4

2-, Cl-, and NH4+ were determined using ion chromatog-

raphy (Metrohm Professional Ion Chromatograph 850, Department of Earth Sciences, Uppsala University). The pH, alkalinity, and total organic carbon (TOC) in the effluent water were sampled once a week during the column experiment; pH was determined using a pH meter (model 713, Metrohm Ltd) whereas alkalinity was determined by titration to end-point with 5 mM H2SO4. TOC was determined on a Shimadzu TOC-L analyzer.

28

4.2.2 Pilot-scale denitrifying woodchip bioreactor The pilot-scale denitrifying woodchip bioreactor was constructed close to the clarification pond of the Kiruna iron ore mine site (Kiirunavaara, Table 1), Kiruna (northern Sweden), in June 2015, with the overall purpose to evaluate the potential of the denitrifying woodchip bioreactor technology for the re-moval of NO3

- in mine drainage. Pore water, gas and woodchip samples were collected and analyzed for the determination of solute and gas concentrations, functional gene abundances, and environmental parameters. Analytical results were interpreted in order to obtain a better understanding of dominating bio-geochemical processes in the bioreactor, and to evaluate the removal of NO3

- from mine water in the pilot-scale denitrifying woodchip bioreactor. Time-series of the parameters allowed the study of the development of the biogeo-chemical environment.

Designed as a sub-surface system, a trench with a trapezoidal cross-section was excavated, and the bottom of the trench was lined with a 1.5 mm thick high-density polyethylene (HDPE) geomembrane in order to eliminate the hy-draulic contact with the surrounding soil (Figure 3). Five ‘inner walls’ of ply-wood were placed in the bioreactor to force the water to flow along the bottom of the bioreactor, and prevent water from flowing on top of the bioreactor (‘Wall 1’ to ‘Wall 5’, Figure 3). Polyvinyl chloride (PVC) pipes (n = 18) were attached to the inner walls at various positions and depths in the bioreactor and used to sample pore water (Figure 3); five of the PVC pipes (center center-line) were equipped with the BAT technology for pore water sampling (see Torstensson, 1984)2. The non-BAT PVC pipes ended with a 10 cm screen placed in a polyester plastic net filled with medium grained sand intended to filter small particles from the sample. Six static gas chambers were placed on top of the woodchip surface of the pilot-scale denitrifying woodchip bioreac-tor to determine surface gas fluxes (three close to the inlet, and three close to the outlet). In addition, 17 temperature probes (T107, Campbell Scientific®) were embedded within the mixture of woodchips sewage sludge at various positions.

2 Only pore water samples from the bottom center-line non-BAT PVC pipes were presented and used for data analysis in papers II, III, and IV.

29

Figure 3. Design of pilot-scale denitrifying woodchip bioreactor: subsurface con-struction with trapezoidal cross-section, trench lined with a 1.5 mm high-density polyethylene membrane, impermeable geomembrane

The bioreactor was filled with a mixture of decorticated pine woodchips (~210 m3) and digested sewage sludge (2.5 m3). Contrary to the sewage sludge used in the column study (paper I), a digested form of sewage sludge was used as the loss of microbial activity was expected to be less during the transport be-tween Luleå and Kiruna. The digested sewage sludge was mixed with water (~ 1:10 v/v) and distributed manually on top of the woodchips in two layers at 0.5 m and 1 m above the bottom of the bioreactor when filling the bioreactor; ~10 L of sewage slurry was added per square meter of woodchip surface, yielding a woodchip to sewage sludge ratio of 84:1 (v/v). In addition, sewage sludge (not slurry) was scattered on top of the woodchip surfaces, and the woodchips were then covered with an impermeable geotextile, and the biore-actor was partly covered with a layer 0.9 m thick layer of glacial till.

Inlet water was pumped from the clarification pond and distributed across the width of the bioreactor using a perforated drainage pipe (Figure 3) to de-crease the potential for preferential flow (cf. “horizontal-diffuse flow” in Cameron and Schipper, 2011). Pore water was initially recirculated in the bi-oreactor for ten days, where after the denitrifying woodchip bioreactor was operated for two consecutive field-seasons extending between the 22nd of June 2015 to the 20th of November 2015 (days 0-151, first operational year) and the 9th of May to the 21st of October 2016 (days 322-490, second operational year), respectively. The HRT of the denitrifying woodchip bioreactor was adjusted several times during the two operational years using a ball valve controlling the pump discharge (see Table 3 for operating conditions).

30

Table 3. Variation in calculated theoretical hydraulic residence time (HRT, pore vol-ume/pump discharge, pore volume = 112.3 m3) for the duration of the bioreactor op-eration. Reprinted with permission from Elsevier, originally published in Nordström and Herbert (2018).

Days Pump discharge (m-3 day-1) Theoretical HRT (days) a(-10 to 0) 14.0 8.0 b0 to 45 43.2 2.6 46 to 52 0 Pump malfunction 53 to 151 59.4 1.9 c152 to 321 0 Winter intermission 322 to 370 43.2 2.6 371 to 428 48.9 2.3 429 to 490 109.6 1.0

a Pore water recirculation, in practice infinite HRT. b Through-flow initiated by opening outlet. c Through-flow in the bioreactor was terminated during the winter as the low air temperatures risked freezing of the inlet tubing.

Influent and effluent water from the bioreactor were grab sampled approxi-mately two times a week during the operational years by LKAB personnel. Pore water from sampling tubes was sampled once a month during the opera-tional years (days 1, 22, 57, 85, 113, 330, 365, 400, 428, 456, and 477) using a peristaltic pump; ~2 L of pore water was discarded to assure a representative sample. Chemical analyses (NO3

-, NO2-, NH4

+, SO42-, TOC, pH, alkalinity) of

the water samples were carried out by LKAB’s accredited laboratory services following standardized procedures (see paper II for details). Dissolved gases (pore water) and surface gas fluxes were sampled monthly together with the pore water samples, transported to Uppsala via ground transport, and analyzed together for N2O and CH4 on a gas chromatograph equipped with an electron capture detector (Clarus 500 GC, Perkin Elmer, CT, USA) at the Swedish University of Agricultural Sciences (SLU, see paper II for details). Surface gas fluxes was calculated in R using the R-package ‘HMR’ (Pedersen, 2015) and reported as the average of six spatially separated static gas chambers. The temporal and spatial variation in the composition of the microbiologi-cal community in the woodchip bioreactor was sampled by filtering pore water in SterivexTM filter units (0.22 µm Millipore Express® polyetherosulfone membrane), during each pore water sampling event (excluding days 1 and 22, see above), and by collecting woodchips samples following the termination of bioreactor operations (day 490). DNA was extracted from the woodchips and SterivexTM filter units, and the abundance of specific marker genes coding for denitrification (nirK, nirS, nosZI, nosZII), DNRA (nrfA), and anammox (hdh) was determined using quantitative real-time PCR (qPCR) (see paper IV for additional details).

31

4.4 Numerical modeling approaches Mathematical models may be used to describe and study natural systems in simplified forms through the parameterization of comparatively complex pro-cesses. In the context of this thesis, numerical modeling was used as a tool for studying solute transport and hydraulic conditions in woodchip media (paper I) and for studying temporal changes in the biogeochemistry in denitrifying woodchip bioreactors (papers II-IV).

4.4.1 Paper I: Solute transport and hydraulic conditions Chloride breakthrough curves obtained from the column tracer tests (paper I, section 4.2.1) were used to define the hydraulic conditions, and to study solute transport, within woodchip media using a dual-porosity model. Dual-porosity models are based on the division of the pore space into mobile and immobile zones, with flow taking place in the mobile zone, and solutes being exchanged between the mobile and immobile zones through diffusion. Omitting adsorp-tion, as there is no evidence that supports the adsorption of anions to ‘woody’ media (cf. Jaynes et al., 2016), the governing equation (advection-dispersion equation, ADE) controlling the distribution of solutes in one-dimensional, steady-state, dual-porosity systems may be expressed according to equations (3) and (4) (adapted from van Genuchten and Wagenet, 1989).

(3)

(4)

Subscripts m and im refers to mobile and immobile pore waters; θ is the volu-metric water content (saturated porosity), D is the dispersion coefficient, c is the solute concentration, v is the pore water velocity, and α is the rate of mass transfer between the mobile and immobile water zones. Chloride breakthrough curves were fitted to an analytical solution of the ADE equation in the soft-ware CXTFIT (Toride et al., 1995) through non-linear least squares using a deterministic non-equilibrium approach (equations 3 and 4) where CXTFIT fits v (equation 5), Dm, the fraction of immobile volumetric water content to the total water content (β, equation 6), and a dimensionless expression of the mass transfer coefficient (ω, equation 7).

(5)

(6)

32

(7)

For β < 1.0, immobile water zones are present and preferential flow behavior is expected, and lower end values of the dimensionless mass transfer coeffi-cient (0≤ω≤100) indicate low to non-existent solute exchange between mobile and immobile water phases.

We determined the effect of the baffles on the NO3- removal efficiency in

the laboratory sized denitrifying woodchip bioreactor (paper I) by simulating NO3

- removal in the SC and SCB columns in the geochemical software PHREEQC (Parkhurst and Appelo, 1999); NO3

- removal in the LC column was not modeled (see paper I for details). Mathematical definitions of NO3

- removal rate kinetics in denitrifying woodchip bioreactors is generally aligned on forms resembling equation (8) (cf. Ghane et al., 2015; Jaynes et al., 2016).

(8)

Where r is the NO3- removal rate, k is the temperature dependent rate constant,

and [NO3-]n represents an exponential dependence on the NO3

- concentration of the order n (0, 1, …). For reactions with variable rate orders, a Monod ex-pression may be used to describe the NO3

- concentration dependency which defines a rate order that is dependent on the NO3

- concentration (equation 9): the rate order moves from approximately zeroth (n = 0) to first (n = 1) order as the NO3

- concentration approaches the half-saturation concentration (Ks).

(9)

Ks is defined as the NO3- concentration where the actual reaction rate (r) is half

of the maximum reaction rate (here k). The temperature dependency of chem-ical reaction rates is classically expressed through the rate constant k using the Arrhenius law (equation 10) or modifications thereof (e.g. Jaynes et al., 2016), and predicts an exponential increase in reaction rate with temperature.

(10) In the Arrhenius law (equation 10): k, R, T, A, and Ea is the rate constant, universal gas constant, the temperature [K], the pre-exponential factor, and the Arrhenius activation energy (minimum energy for reaction), respectively. NO3

- removal rates were calculated from the sampled NO3- concentration

profiles (see details in paper I). Solute transport in the columns was defined using a physical non-equilibrium (dual-porosity) model defined in

33

PHREEQC, which was coupled with NO3- removal rate kinetics defined us-

ing the Monod expression (equation 9) for the removal of NO3-. The temper-

ature dependency of the rate constant (k, equation 9) was defined using the Arrhenius law (equation 10); Ks was determined individually for each col-umn from the NO3

- concentration profiles (see paper I for details). NO3

- removal was modeled in the SC and SCB columns in two scenarios with Ea, A, and Ks retrieved from the SC column (SC scenario) or the SCB column (SCB scenario), respectively. In both scenarios, parameters for the physical non-equilibrium model were retrieved from the fitting of the chloride breakthrough curves in the SCB column. It was assumed that the SC (or SCB) scenario should be able predict NO3

- removal in the SCB (or SC) column if the baffles had no effect on the removal.

4.4.2 Paper II: Determination of major biogeochemical processes The major biogeochemical processes operating in the pilot-scale denitrifying woodchip bioreactor were determined using in a stoichiometric mass-balance constrained by pH and alkalinity (paper II). Denitrification, DNRA, sulfate reduction, and fermentation were defined stoichiometrically using glucose (C6H12O6) as the carbon substrate. When used as an electron donor for TEAPs, glucose is oxidized into inorganic carbon (H2CO3

*, HCO3-, CO3

2-), resulting in characteristic changes of pH and alkalinity in the pore water. Intermediate steps of NOx reduction in denitrification and DNRA were defined separately, allowing for the accumulation of NO2

- and N2O whereas NO accumulation was assumed to be negligible. During fermentation, a fraction of the glucose is also transformed to more ‘simple’, lower molecular weight, organic carbon substrates (e.g. butyrate and acetate), under the simultaneous consumption of alkalinity (HCO3

-). It was assumed that the organic carbon (TOC) exported from the pilot-scale denitrifying woodchip bioreactor was an end-product from anaerobic glucose fermentation, which based on stoichiometry could be either butyrate (C3H7COO-), acetate (CH3COO-), or a combination of these.

From the observed changes in the concentration of NO3-, NO2

-, N2O, NH4+,

SO42-, and TOC in the pilot-scale denitrifying woodchip bioreactor, pH and

alkalinity in the effluent from the pilot-scale denitrifying woodchip bioreactor were modeled under three scenarios incorporating different biogeochemical processes: denitrification and DNRA (scenario 1); denitrification, DNRA, and sulfate reduction (scenario 2); and denitrification, DNRA, sulfate reduction, and fermentation (scenario 3). Modeled effluent pH and alkalinity was com-pared to the observed values under the assumption that all alkalinity was bi-carbonate (HCO3

-) and that there was no degassing of CO2(g) (see details in paper II). The end-product(s) of glucose fermentation was determined through a stochastic simulation (see details in paper II) where the stoichiometry of

34

glucose fermentation was selected by minimizing the absolute residuals be-tween the observed and modeled pH and alkalinity.

4.4.3 Paper III: Variability in NO3- removal rate

The temporal variability in the NO3- removal rate in the pilot-scale denitrify-

ing woodchip bioreactor was studied in paper III with the objective of identi-fying the selective pressure that resulted in the general decline of NO3

- re-moval rates over time in denitrifying woodchip bioreactors. We hypothesized that a decrease in the carbon quality resulted in a decrease in the NO3

- removal rates over time and used the relative temperature sensitivity of the NO3

- re-moval rate as a proxy for the carbon quality (cf. Sierra et al., 2012). Carbon ‘quality’ was defined accordingly to Bosatta and Ågren (1999) as ‘the number of enzymatic steps required to release as carbon dioxide a carbon atom from an organic compound’ where low quality carbon is characterized by a larger number of steps required for the oxidation of organic carbon to inorganic CO2 (Bosatta and Ågren, 1999).

NO3- removal rates were calculated as the mass NO3

- removed between two points of pore water sampling points divided by elapsed time between them (assuming plug-flow conditions, details in paper III) and equation (8) was used to define the NO3

- removal kinetics of the NO3- removal rate. Recognizing that

the NO3- removal rate in the pilot-scale denitrifying woodchip bioreactor was

controlled by a community of NO3- reducers that were adapted to marginally

different temperatures (cf. Schipper et al., 2014), the temperature dependency of the NO3

- removal rate was defined using the Eyring equation (equation 11, from Garcia-Viloca et al., 2004).

∆ ‡

(11)

Where kb, h, and c0 is the Boltzmann’s constant, the Planck’s constant, and the standard state concentration (1 mol L-1), respectively. γ(t) is the generalized transmission coefficient and is generally close to unity (Feller, 2013). For en-zymes, the Gibb’s free energy of activation (∆G‡

, minimum energy for reac-tion) is non-linearly dependent on the temperature due to the difference in heat capacity (∆C‡

p) between the ground state enzyme substrate complex and tran-sition state enzyme substrate complex as defined by the macromolecular rate theory (MMRT; equation 12, Arcus et al., 2016; Hobbs et al., 2013; Schipper et al., 2014). This results in a temperature optimum (Topt) of enzymatic reac-tion rates above and below which the reaction rate declines (equation 13, Hobbs et al., 2013).

∆ ‡ ∆ ‡ ∆ ‡ ∆ ‡ ∆ ‡ln / (12)

35

∆ ‡ ∆ ‡

∆ ‡ (13)

∆H‡T0 and ∆S‡

T0 are the differences in enthalpy and entropy of activation be-tween the ground state enzyme substrate complex and the transition state en-zyme substrate complex, at a reference temperature (T0, here 25°C), respec-tively (Hobbs et al., 2013).

Since NO3- removal rates have been shown to be higher during the initial

13 months of operation in denitrifying woodchip bioreactors (Addy et al., 2016), NO3

- removal rates from the first and second operational years were analyzed separately. Firstly, equation (8) with k expressed using the Eyring equation was fitted to the determined NO3

- removal rates as a function of tem-perature and NO3

- concentration in two scenarios assuming zeroth (n=0) and first (n=1) order reaction rates, respectively. This was done to obtain general values for ∆G‡ and reaction orders for the two operational years, respectively. The NO3

- concentration dependency of the NO3- removal rate was studied us-

ing the Monod equation (equation 9) by fitting it to the variability in NO3-

removal rate with NO3- concentration (see details on fitting in paper III). Sec-

ondly, the variability in ∆G‡ with temperature was determined by selecting the ‘optimal’ ∆G‡ for NO3

- removal through a Monte Carlo procedure for each sampled NO3

- concentration profile separately (see details in paper III). The temperature dependency of ∆G‡ was then explained using MMRT by fitting equation (12) to the obtained variability in ∆G‡ with temperature through non-linear regression (NLS) in combination with a 1000 non-parametric bootstrap procedure (see details in paper III). Each bootstrapped set of ∆H‡

T0, ∆S‡T0,

∆C‡p, and Topt were regarded as representing an individual NO3

- reducer. Fi-nally, the relative temperature sensitivity was defined using Q10 (equation 14; Schipper et al., 2014) and used as a proxy for carbon quality where increases in Q10 are synonymous with increases in the relative temperature sensitivity, and suggests a decrease in carbon quality (cf. Sierra, 2012).

ln ∆ ‡ ∆ ‡

(14)

The enthalpy of reaction, ∆H‡, is linearly dependent on temperature [i.e. ∆H‡ = ∆H‡

T0 + ∆C‡p(T-T0)] (Schipper et al., 2014).

4.4.5 Paper IV: Variability in the NOx reducing community The temporal variability in the NOx reducing community in the pilot-scale de-nitrifying woodchip bioreactor and its association to changes in concentrations of NO2

-, N2O, and NH4+ was studied in paper IV. To assess whether the NOx-

36

reducing community in the pore water was representative of the NOx-reducing community in the woodchip media, we first evaluated the similarity between the NOx-reducing community in the four different sample sources: pore water, influent water, woodchips, and digested sewage sludge. Differences in gene abundances between the different sample sources were tested using a Kruskal-Wallis test, and a two-sided Wilcoxon test was used to test differences in gene abundances between specific sample sources individually. Non-metric Multi-dimensional Scaling (NMDS; R-package ‘vegan’ Oksanen et al., 2018) was used to illustrate structural differences in the concatenated NOx-reducing com-munities (i.e. based on all functional genes) between samples from the pore water (including outlet samples), influent water, woodchips, and digested sewage sludge (see paper IV for details). The abundances of the functional genes in denitrification (nirS, nirK, nosZI, nosZII), DNRA (nrfA), and anam-mox (hdh) from each individual sample (irrespective of source, time of sam-pling, or position in the pilot-scale denitrifying woodchip bioreactor) was as-sumed to represent a specific NOx-reducing community. The differences in the concatenated NOx-reducing communities were further characterized by us-ing an analysis of similarity (ANOSIM) and by comparing the functional gene diversity (based on the Shannon’s diversity index, H’) between the four dif-ferent sample sources (see paper IV for details).

To assess the temporal variability in the NOx-reducing community of the pilot-scale denitrifying woodchip bioreactor, we compared the similarity (ANOSIM, functional gene diversity) between the NOx-reducing community in pore water in all samples from the same day of sampling, with the NOx-reducing community in the woodchip samples. Lastly, changes in pore water concentrations of NO2

-, N2O, NH4+, TOC, pore water pH and bed temperature

was correlated with changes in the structure of the NOx-reducing community in the pore water by using the function envfit from the R-package ‘vegan’ (Oksanen et al., 2018) (see paper IV for details). In addition, a permuted Spearman’s rank analysis was performed to correlate changes in the abun-dances of functional genes in denitrification, DNRA, and anammox with changes in solute concentrations (see paper IV for details).

37

5. Summary of Papers

5.1 Paper I: Column study Influent NO3

- concentrations (30 mg N L-1) were reduced to below detection limit in all columns, at all temperatures (5°C, 12°C, and 22°C) when the HRT was high (~2.5 days), without substantial production of NO2

- or NH4+. Sulfate

reduction was noted in the LC column (22°C) during the high HRT when all NO3

- was removed to below detection limits. At the low HRT (~1 day), NO3-

removal in all columns at all temperatures was incomplete, and NO3- removal

occurred concurrently with the production of NH4+ in the LC column (22°C).

The low production of NH4+ relative to the NO3

- reduced, and an increase in pH and alkalinity across the columns, indicated denitrification as the main pathway for NO3

- removal. Immobile water zones were detected in the SCB column at both HRTs sug-

gesting preferential flow behavior, with the volumetric fraction of immobile water zones increasing from ~12% to ~25% at the high and low HRT, respec-tively. Solute exchange between the mobile and immobile water was close to non-existent (ω ≈ 0) at both HRTs, implying that immobile water zones were not participating in the NO3

- removal. NO3- removal rates increased with tem-

perature and decreased with HRT, and Ks was determined as 4.46 ± 0.3, 2.25 ± 1.16, and 10.61 ± 5.86 mg NO3

--N in the SC, SCB, and LC columns at the high HRT, respectively.

The modeling exercise showed that the baffles did not have any substantial effect on the NO3

- removal in the SCB column; equation (9) with parameters retrieved from either the SC column or SCB at the low HRT was able to pre-dict the observed NO3

- removal in both the SCB and SC column at the low HRT, with no substantial differences. Moreover, the combined Arrhenius-Monod equation with parameters from the low HRT were unable to predict the observed NO3

- removal at the high HRT, substantially underestimating the observed NO3

- removal. Nevertheless, the Arrhenius parameters obtained from the SC and SCB columns, respectively, were different, which indicated that the installation of the baffles may have had some influence on the kinetics of NO3

- removal. However, the apparent differences may also have been due to the natural variability of the woodchip-sewage sludge mixture.

38

5.2 Paper II: Pilot-scale denitrifying woodchip bioreactor The average NO3

- concentration in the influent and effluent waters from the pilot-scale denitrifying woodchip bioreactor was 22.0 and 5.4 mg N L-1, re-spectively, for the duration of the two operational years. NO3

- was removed to below detection limits (0.06 mg N L-1) whenever the temperatures were high (>5°C) and the HRT was between 1.9 and 2.6 days (Table 3, Figure 4a), whereas NO3

- removal was incomplete when temperatures were low (<5°C) and/or when the HRT decreased to 1 day (day 429) (Table 3, Figure 4a).

Effluent NO2- concentrations were generally lower (except for days 1-7),

whereas effluent concentrations of NH4+ were generally higher, than their re-

spective influent concentrations (Figures 4b and 4c). NO2- and NH4

+ effluent concentrations were higher during the first operational year in comparison to the second operational year (Figures 4b and 4c). Sulfate reduction was most evident during the first operational year (~days 25-50), when all NO3

- was reduced to below detection limits relatively close to the bioreactor inlet (data not shown), and when the bioreactor temperature was relatively high (>10°C) (Figure 4d). Surface gas fluxes and dissolved concentrations of CH4 declined with time from bioreactor start-up (data not shown). Effluent TOC concentra-tions exponentially declined, whereas pH and alkalinity progressively in-creased, with time from start-up (Figures 4e-4g); initial biogeochemical con-ditions in the pilot-scale denitrifying woodchip bioreactor were acidic and al-kalinity-consuming (Figures 4e and 4f).

NO3- removal was associated with increases in pH and alkalinity: the gen-

eral low production of NO2-, N2O, and NH4

+ suggested that complete denitri-fication (full reduction to N2(g)) was the main pathway for NO3

- reduction. However, incomplete NO3

- removal resulted in an increased production of N2O and NH4

+ relative to the NO3- reduced, which suggested that the relative

importance of denitrification truncated to N2O and DNRA (as pathways for NO3

- removal) increased as NO3

- removal decreased. When less than ~10% of the incoming NO3

- was reduced in the bioreactor, ~100% of the reduced NO3-

were exported as N2O, whereas N2O emissions represented 0.10±0.05% of the reduced NO3

- when NO3- removal was complete. The production of NH4

+ rel-ative to the NO3

- reduced was on average low (3.8%), but increased to between ~8-50% when NO3

- removal was incomplete at temperatures (<5°C). The as-sociation between high levels of TOC and low concentrations of NO3

- (high C/NO3

- ratio) during the first operational year suggested that NH4+ was pro-

duced from DNRA. Furthermore, the initial acidic and alkalinity-consuming conditions, and the decline in effluent concentrations of TOC following start-up suggested a fermentative signature on the TOC ‘flush-out’.

39

Figure 4. Influent (●) and effluent (○) concentrations of NO3

- (a), NO2- (b), NH4

+ (c), SO4

2- (d), pH (e), alkalinity (f), and TOC (g) from the pilot-scale denitrifying woodchip bioreactor. Temperature of the pilot-scale denitrifying woodchip bioreac-tor is shown in (a). Reprinted with permission from Elsevier, originally published in Nordström and Herbert (2018).

40

Denitrification, DNRA, sulfate reduction, and fermentation were confirmed as major biogeochemical processes in the pilot-scale denitrifying woodchip bioreactor by the stoichiometric mass balance model (Figure 5): fermentation was required to account for the low effluent pH and alkalinity consumption observed in the bioreactor during the first 63 days of operation. The stochastic formulation of fermentation reaction stoichiometry required to predict effluent TOC concentrations suggested that the organic carbon in the TOC ‘flush-out’ was in the form of acetate and step-wise increased to butyrate as pH in the bioreactor increased. This further suggested that the organic carbon substrate used by the TEAPs changed from butyrate to acetate with time from bioreactor start-up.

Figure 5. Simulated and observed effluent pH (a) and alkalinity (b) for scenario 1 (denitrification and DNRA), scenario 2 (denitrification, DNRA, sulfate reduction), and scenario 3 (denitrification, DNRA, sulfate reduction, and fermentation). Re-printed with permission from Elsevier, originally published in Nordström and Her-bert (2018).

41

5.3 Paper III: Variability in NO3- removal rates

Determined NO3- removal rates were higher and more variable during the first

operational year (9.7 ± 10.7 g N m-3 d-1) in comparison to the second opera-tional year (5.5 ± 5.2 g N m-3 d-1), indicating a decline in NO3

- removal rates over time in the bioreactor. Despite overall poor fits, first and zeroth rate or-ders best described the NO3

- concentration dependency of the NO3- removal

rates from the first and second operational years, respectively. This was ac-companied by an increase in ∆G‡ and a decrease in Ks, between the two oper-ational years. The change in NO3

- removal kinetics suggested a process by which NO3

- reducers were selected based on the increase in ∆G‡, and that the decrease in Ks was also the result of an increase in ∆G‡ (decrease in NO3

- removal rate): Ks determined at the scale of the pilot-scale denitrifying wood-chip bioreactor also included mass-transport limitations and thereby scaled with the NO3

- removal rate (Shaw et al., 2013; Shaw et al., 2015). The ‘optimal’ ∆G‡ for NO3

- reduction was selected for each sampled NO3-

concentration profile, assuming first and zeroth order reaction rates for the first and second operational year, respectively; MMRT was used to explain the variability in ∆G‡ with temperature for each operational year separately. The mean Topt decreased from 24.2°C to 16.0°C between the two operational years (Table 4), which indicated an adaption/selection of the NO3

- reducing community to low temperature conditions. This was also seen when simulat-ing effluent NO3

- concentrations in the pilot-scale denitrifying woodchip bio-reactor, with NO3

- reduction clearly being dominated by NO3- reducers with

low temperature optimums during the second operational year (Figure 6). In addition, the range of fitted MMRT parameters (∆H‡

T0, ∆S‡T0, ∆C‡

p, Topt) de-creased between the two operational years (Table 4), which suggested homog-enization of the NO3

- reducing community. The relative temperature sensitivity (Q10) increased between the opera-

tional years and was negatively correlated with temperature. Using Q10 as a proxy for the carbon quality in the bioreactor, this implied that the carbon quality decreased with temperature and between the operational years. From the definition of carbon quality by Bosatta and Ågren (1999), this was inter-preted as an increased cycling of organic carbon over time and at lower tem-peratures by the fermentative community prior to terminal use by the NO3

- reducers in the bioreactor. With this interpretation, this indicated an increased dependence of the NO3

- reducing community on the supply of organic carbon from the fermenting community that may be related to increasingly special-ized NO3

- reducers that fed on a specific organic carbon substrate excreted by the fermenting community (increased cross-feeding, i.e. the development of syntrophic structures).

The interpretation of an increased dependency of the NO3- reducers on the

supply of organic carbon from the fermentative community challenged the predicted theoretical longevity of the woodchip bioreactor technology based

42

on the total carbon content of the woodchips. With a potential breakdown in the syntrophic relationship between the fermenters, which provide organic carbon for the NO3

- reducers, NO3- removal in denitrifying woodchip bioreac-

tors may face a premature decline despite carbon storage in the woodchips.

Table 4. Range of fitted macromolecular rate theory parameters for the first and sec-ond operational years (NSE = Nash-Sutcliffe Model Efficiency coefficient). Re-printed with permission from Elsevier, originally published in Nordström and Her-bert (2019).

Year Parameter Range of fitted parameters

NSE Minimum Mean Maximum

1

∆H‡T0 (kJ mol-1) -156.3 -33.5 67.8

0.54 ∆S‡

T0 (J mol-1 K-1) -862.9 -437.1 -86.5 ∆C‡

p (kJ mol-1 K-1) -17.9 -10.7 -4.9 Topt (°C) 16.4 24.2 38.2

2

∆H‡T0 (kJ mol-1) -584.2 -262.7 34.5

0.82 ∆S‡

T0 (J mol-1 K-1) -2392.8 -1290.4 -270.0 ∆C‡

p (kJ mol-1 K-1) -49.6 -27.9 -8.0 Topt (°C) 13.3 16.0 29.6

Figure 6. Simulated NO3

- effluent concentrations using the macromolecular rate the-ory (MMRT) parameters determined through the bootstrap procedure for the first and second operational year, respectively. Mean MMRT simulation using the mean MMRT parameters. NSE = Nash-Sutcliffe Model Efficiency coefficient. Reprinted with permission from Elsevier, originally published in Nordström and Herbert (2019).

43

5.4 Paper IV: Variability in NOx-reducing community The overall NOx-reducing community in the pore water was significantly dif-ferent (based on ANOSIM) from the NOx-reducing communities in the influ-ent water (R = 0.91, p= 0.001), woodchips (R = 0.88, p = 0.001), and digested sewage sludge (0.83, p = 0.001); the functional gene diversity was signifi-cantly different (Kruskal-Wallis: χ2 = 47.88, df = 3.0, p<0.01) in the pore wa-ter (H’ = 1.1 ± 0.2), influent water (H’ = 1.3 ± 0.2), woodchips (H’= 0.3 ± 0.1), and sewage sludge (H’ = 0.6 ± 0.0). However, functional gene abun-dances relative to each other in the pore water were comparable to the func-tional gene abundances in the woodchip media (nirS > nrfA > nirK > nosZII > nosZI; data not shown). In addition, both the NOx-reducing community and the functional gene diversity in the pore water became increasingly similar to that in the woodchip media, for pore water sampled closer to the sampling date of the woodchip media (data not shown).

The NOx-reducing community in the pilot-scale denitrifying woodchip bi-oreactor became increasingly enriched in nirS, nosZII, and nrfA, and relatively depleted in nosZI, between the operational years (Figure 7). These structural changes were significantly associated with pore water concentrations of NO3

- (R2 = 0.34, p = 0.001), NO2

- (R2 = 0.30, p = 0.003), N2O (R2 = 0.17, p = 0.022), NH4

+ (R2 = 0.14, p = 0.040), TOC (R2 = 0.28, p = 0.003), and bioreactor tem-perature (R2 = 0.46, p = 0.001) (Figure 7).

The abundances of nirS, nirK, nosZII, and nrfA significantly increased be-tween the operational years, whereas the abundance of nosZI significantly de-creased. As indicated in paper II, pore water concentrations of NO3

- and N2O were significantly higher during the second operational year, whereas pore water concentrations of NO2

- were significantly lower. Pore water concentra-tions of NO3

- was most strongly correlated (negatively) with bioreactor tem-perature, whereas pore water concentrations of N2O was most strongly corre-lated with the ratio of nirS to nirK (nirS:nirK) and Σnir to ΣnosZ [(nirS + nirK) / (nosZI + nosZII)], both ratios significantly increasing between the opera-tional years. Σnir:ΣnosZ was most strongly associated with the abundance of nirS. The abundance of the nrfA gene was not associated with pore water con-centrations of NH4

+ but was correlated with pore water concentrations of TOC and (negatively) NO3

-, which suggested that the DNRA pathway was sup-pressed by unfavorable environmental conditions (low C/NO3

- ratios). Denitrifying pathways truncated to N2O are preferentially promoted when

there is competition for the electrons released from oxidation of the organic carbon electron donor (cf. Pan et al., 2013). It was argued that the development of temperature dependent syntrophic structures led to a change in the relative availability of different carbon substrates (cf. papers II and III), an increased competition for the available carbon, and the enrichment in nirS-type denitri-fiers representing denitrifying pathways truncated to N2O. This led to a loss

44

of denitrifiers able to perform complete denitrification (represented by the de-pletion in nosZI). As nosZII is frequently associated with bacteria that do not have the genetic make-up needed for N2O production (Graf et al., 2014; Jones et al., 2014), the enrichment in nosZII was interpreted as the decoupling of intermediate steps in the denitrification pathway onto several organisms which eliminated intra-organism competition for the available carbon (Pan et al., 2013). The lack of correlation between the abundance of nrfA and pore water NH4

+ was argued to be attributed to the enrichment of nrfA in the NOx-reduc-ing community over time as the result of a developing fermenting community.

The observed temporal pressure for the development of denitrifying path-ways truncated to N2O increases the potential for N2O release from denitrify-ing woodchip bioreactors over time.

Figure 7. Non-metric multidimensional scaling ordination based on Bray-Curtis dis-similarities for the abundance of the functional genes nirS, nirK, nosZI, nosZII, and nrfA in pore water. Arrows show significant correlations (p<0.05) between func-tional gene abundances and environmental variables: pore water concentrations (NO3

-, NO2-, N2O, NH4

+, and TOC). Circles (cyan) and diamonds (magenta) shows samples from the first and second operational years, respectively. Labels nirS, nirK, nosZI, nosZII, and nrfA denote species scores.

45

6. General discussion

Nitrogen-laden discharge from mining activities to surface- and groundwater may have potentially negative consequences on aquatic ecosystems. The re-sults presented in this thesis show that denitrifying woodchip bioreactors can be used for the removal NO3

- from neutral pH mine drainage. However, the results also point toward limitations of the denitrifying woodchip bioreactor technology that need to be considered, and call for further research. In this section, the use of denitrifying woodchip bioreactors for the removal of NO3

- in mine drainage is discussed on the basis of sustainability, operational condi-tions, and application within the context of removing NO3

- from neutral pH mine drainage.

6.1 Conditions for sustainable operation From the results of paper II, two partly conflicting criteria must be met for a sustainable operation of denitrifying woodchip bioreactors: (1) non-limiting NO3

- concentrations throughout the bioreactor, and (2) complete NO3- re-

moval must be achieved in the denitrifying woodchip bioreactor. These two criteria converge into an operational optimum wherein complete NO3

- re-moval is achieved at the outlet position in the denitrifying woodchip bioreac-tor. If complete NO3

- removal is achieved far from the outlet in the bioreactor system, secondary, less energetic reactions (sulfate reduction, methanogene-sis) will be activated. In addition, DNRA may become increasingly favorable at relatively limited availabilities of NO3

-, which may increase the net produc-tion of NH4

+ in the bioreactor (cf. paper IV). For example, this was seen fol-lowing the start-up of the pilot-scale denitrifying woodchip bioreactor when NO3

- concentrations were low in large parts of the bioreactor and effluent NH4

+ concentrations were relatively high (cf. paper II). On the other hand, if NO3

- is incompletely removed in the bioreactor, the emission of N2O and NH4+

relative to the NO3- removed will increase (cf. paper II).

To meet the criteria for a sustainable operation of denitrifying woodchip bioreactors, detailed a-priori knowledge on the environmental dependencies of the NO3

- removal rates is required in order to facilitate the design (dimen-sions and appropriate HRT) of the denitrifying woodchip bioreactor at the spe-cific environmental conditions at the site of implementation. However, the site-specific environmental conditions are likely to change over the course of

46

an operational year with repercussive effects on the NO3- removal rate. In or-

der to meet the criteria for sustainable operation over the course of one oper-ational year, the HRT (~ mass flux of NO3

-) will have to be controlled and varied as the NO3

- removal rate varies: decreasing the HRT as the NO3- re-

moval rate increases, or increasing the HRT as the NO3- removal rate de-

creases. On the other hand, adjustments in the HRT may affect the volume of immobile zones (cf. paper I), thereby further affecting the NO3

- capacity of the system and increasing the risk for CH4 emissions from the denitrifying woodchip bioreactor (cf. Grieβmeier et al., 2017). Meeting the criteria for sus-tainable operation at all times during an operational year of a hypothetical de-nitrifying woodchip bioreactor is therefore made complicated by the variabil-ity in environmental conditions and NO3

- removal rates, and adjusting the HRT over the course of an operational year may not be the best approach. An alternative approach would be to stabilize the variability in NO3

- removal rates in a given system, thereby decreasing the need for adjusting the HRT over the course of an operational year. Our results suggest that it is the development of temperature dependent syntrophic structures between denitrifiers and ferment-ers that creates the variability in NO3

- removal rates in a given system (paper III), and that an adaption to environmental conditions (i.e. temperatures) may be responsible for the variability between systems. Numerical prediction of the adaption of the NO3

- removal rates in denitrifying woodchip bioreactors would improve the design of these systems with respect to sustainability. In addition, stabilizing temperature in the denitrifying woodchip bioreactor will most likely also stabilize NO3

- removal rates, which also could help with meet-ing the criteria of sustainable operation.

6.1.1 Potential design improvements The cost-efficiency of denitrifying woodchip bioreactors could be increased by either extending the duration of an operational year or increasing NO3

- re-moval rates. NO3

- removal rates could be increased and the length of an oper-ational year could be extended by increasing the temperature in denitrifying woodchip bioreactors. In New Zealand, passive solar heating of a denitrifying woodchip bioreactor has been shown to increase bed temperatures by ~1-6°C at air-temperatures between ~8-22°C (Cameron and Schipper, 2011). Passive solar heating can be maximized by considering inlet and outlet designs: ‘downward’ flow regimes in which influent water is distributed across the sur-face and exits along the base of the bioreactor promotes the largest increases in treatment temperatures (Cameron and Schipper, 2011). Note, however, that increasing the temperature of the influent water may exacerbate non-ideal flow regimes in the bioreactor by increasing the buoyancy of the warm influ-ent water whereby flow is increasingly concentrated along the surface of the bioreactor (Cameron and Schipper, 2011). In addition, increasing tempera-tures may increase the net production of N2O (see above).

47

Decreasing preferential flow behavior in denitrifying woodchip bioreactors may be used to increase NO3

- removal rates by increasing the ‘active’ volume participating in NO3

- removal (i.e. decreasing immobile water zones, cf. paper I). From the discussion in paper I, the volume of immobile zones decreased as HRT increased (i.e. the advection velocity decreased), and the NO3

- removal rate was inversely proportional to the advection velocity. This could be be-cause of an increased influx of oxygen as advection velocity in the system increases (Halaburka et al., 2019). Promoting low advection velocities in de-nitrifying woodchip bioreactors may thus increase NO3

- removal rates which may be achieved by increasing the cross-sectional area (A) of the bioreactor: given an influent flux of water (Q), the advection velocity (v) in the bioreactor decreases with increasing cross-sectional area (v=Q/A), and wider designs of denitrifying woodchip bioreactors should be considered, e.g. ‘denitrification walls’ (see Schipper et al., 2010). Low(er) advection velocities may also be achieved by distributing the influent water across the cross-sectional area of the bioreactor rather than at a single point, which is conceptually equivalent to the ‘horizontal-diffuse’ design by Cameron and Schipper (2011) which was shown to decrease short-circuiting.

6.1.2 Long-term sustainability and the importance of syntrophy Denitrifying woodchip bioreactors are passive systems designed to operate without maintenance for at least a decade. For a sustainable NO3

- removal, it is imperative to understand how denitrifying woodchip bioreactors naturally evolve with respect to NO3

- removal efficiency and emission of undesired by-products (NH4

+, NO2-, N2O, H2S, and CH4). It has been shown that denitrifying

woodchip bioreactors maintain ~50% of their initial NO3- removal capacity

for up to a decade after installation (Moorman et al., 2010; Robertson, 2010), and possibly up to 15 years if compared with similar systems (Robertson et al., 2008), however, little is known of long-term emissions of NO2

-, N2O, NH4

+, H2S, and CH4. The results of papers II, III, and IV enable us to discern temporal trends in the development of processes controlling NO3

- removal rates and the emission of NH4

+, NO2-, N2O, H2S, and CH4 from the pilot-scale

denitrifying woodchip bioreactor. Our results suggest that these temporal trends may be controlled by the development of syntrophic metabolic struc-tures in the pilot-scale denitrifying woodchip bioreactor, which is elaborated upon below.

In paper II it was suggested that the ‘waste’ carbon exported from the pilot-scale denitrifying bioreactor (i.e. TOC) moved over time from primarily con-taining acetate to containing butyrate, which is in agreement with two recent studies on denitrifying woodchip bioreactors where a decline in the export of acetate over time was observed (Grieβmeier et al., 2017; Grieβmeier and Gescher, 2018). It was suggested that acetate produced from fermentation was increasingly used as the carbon substrate for the TEAPs (cf. Grieβmeier et al.,

48

2017), implying the development of syntrophic structures between these pro-cesses over time. As further discussed in paper IV, the impetus for this devel-opment may have been the competition for a limited shared resource (carbon) that triggered the development of an energy-efficient microbial structure (cf. González-Cabaleiro et al., 2015). This could have transpired as the initial ‘wasteful’ utilization of the primary carbon substrate (glucose), leading to the secondary excretion and export of ‘waste’ carbon (acetate), promoting mi-crobes that were efficient scavengers of the secondary carbon substrate (cf. Gudelj et al., 2016). This would lead, with time, to the preferential promotion of denitrification over DNRA, sulfate reduction, and methanogenesis (paper II), NO2

- reducers over N2O reducers (paper IV), and the decrease in NO3-

removal rates (paper III). These results are important as they point toward a temporal pressure for reduced emissions of NH4

+, NO2-, H2S, and CH4, de-

creases in NO3- removal efficiencies (~ NO3

- removal rates), and an increase in the production of N2O, in the pilot-scale denitrifying woodchip bioreactor; these effects have all been observed to some degree (papers II-IV).

There is a clear need for continuous observations of long-term N2O emis-sions from denitrifying woodchip bioreactors. Nevertheless, our hypothesis concerning the development of syntrophic structures as the underlying control to the temporal trends in the pilot-scale denitrifying woodchip bioreactors al-lows for some speculative predictions. Hanke et al. (2016) observed that the syntrophic exchange between fermenters and denitrifiers was temperature-de-pendent. Denitrifiers were outcompeted by fermenters at lower temperatures (10°C), whereas the two engaged in successful syntrophic structures (or the fermenters were outcompeted by the denitrifiers) at higher temperatures (25°C; Hanke et al., 2016). With the development of syntrophic structures over time, this suggests that NO3

- removal efficiencies may suffer ‘premature’ declines at relatively cold temperatures ( 10°C) (paper III), whereas denitri-fying woodchip bioreactors operating at relatively high temperatures (~25°C) may suffer from higher net productions of N2O (paper IV, cf. Hanke et al., 2016; Warneke et al., 2011a, Warneke et al., 2011b).

6.2 Potential full-scale application While the bioreactor in this study obtained water for treatment directly from the Kiruna clarification pond, which is not considered a viable treatment al-ternative over the long term, a possible application of the denitrifying wood-chip bioreactor technology would be to collect drainage directly from waste rock piles. This is currently under investigation (NITREM, 2019). However, based on the timing and character of leachate generation from waste rock de-posits, this approach faces several challenges (see Nordström and Herbert, 2017a). One of these challenges is the intermittent production of leachate from waste rock piles that would result in fluctuating HRTs in the denitrifying

49

woodchip bioreactor, moving from stagnant conditions when no leachate is produced to increasingly lower HRTs as the generated volume of leachate in-creases, together with associated changes in the biogeochemistry of the deni-trifying woodchip bioreactor. For example, stagnant conditions may lead to the complete depletion of NO3

- in the bioreactor, triggering sulfidic to meth-anogenic conditions in the bioreactor resulting in the undesired release of H2S and CH4 (paper II). On the other hand, during periods of high and fast leachate generation, the HRT in the bioreactor would be lower which may result in the incomplete reduction of NO3

- and an increased relative export of N2O from the bioreactor (paper II). In addition, the results of paper I suggests that de-creases in HRT may increase the preferential flow in denitrifying woodchip bioreactors; periods of high leachate generation from waste rock deposits may therefore be related a relatively low volume utilization of the bioreactor.

For a sustainable operation of the denitrifying woodchip bioreactor inter-cepting waste rock leachate, the diversion of leachate collected from waste rock piles for intermediate storage in an accumulator tank or reservoir prior releasing it to the bioreactor would alleviate problems associated with variable HRTs in the bioreactor. This would allow for an increased control on the NO3

- availability in the denitrifying woodchip bioreactor, and an increased control on the HRT of the system, by distributing the collected waste rock leachate evenly throughout the operational year of the system; there would also be the potential to engineer this system to include passive solar heating. Neverthe-less, this would increase the cost of the method, particularly at larger scales, but this should be weighed against the reduced release of greenhouse gases (CH4, N2O) and bioavailable N (NH4

+).

50

7. Conclusions and outlook

The export of nitrogen (N) from mines to local waterways such as lakes, wa-tercourses, and coastal areas is significant and may lead to acidification, eu-trophication, and hypoxia in the receiving waters, resulting in habitat degra-dation and losses in biodiversity. This thesis evaluated and implemented the passive denitrifying woodchip bioreactor technology for the removal of nitrate (NO3

-) in neutral pH mine drainage. Some specific conclusions from the re-sults of this thesis are the following:

The denitrifying woodchip bioreactor technology is a potential al-ternative for the removal of NO3

- in neutral pH mine drainage. Our results show that incoming NO3

- concentrations (~22.0-30.0 mg N L-1) were reduced to below detection limits (≤ 0.06 mg N L-1) as temperatures and hydraulic residence times (HRTs) were relatively high (≥5°C and ~1.9-2.6 days, respectively) without any substantial production of nitrite (NO2

-), nitrous oxide (N2O), or ammonium (NH4

+). However, at the onset of low temperatures (≤5°C), and/or when the HRT was low (~1 day), NO3

- removal was incomplete and the production of NO2

-, N2O, and NH4+ relative to the NO3

- reduced increased. In addition, sulfate (SO4

2-) reduction and methanogene-sis commenced once all NO3

- had been depleted in the system. For a sustainable operation, the method requires continuous supply

of NO3- in order to prevent the activation of secondary undesirable

reactions such as sulfate reduction and methanogenesis. Moreover, operations should strive for complete NO3

- removal in order to min-imize the production of N2O and NH4

+ relative to the NO3- reduced.

However, complete NO3- removal should only occur at (or close to)

the outlet of the bioreactor in order to avoid prevailing NO3- limit-

ing conditions in the system which could increase the net produc-tion of NH4

+ by triggering environmental conditions favorable for DNRA.

Potential design improvements of these systems include engineered solutions for the reduction of immobile water zones in the wood-chip media that do not participate in NO3

- removal, and engineered solutions for increasing bed temperatures or controlling the HRT.

51

NO3- removal rates decreased over time in the pilot-scale denitrify-

ing woodchip bioreactor. The diversity of biogeochemical processes decreased over time in

the pilot-scale denitrifying woodchip bioreactor, with denitrifica-tion outcompeting the less energetic reactions SO4

2- reduction and methanogenesis over time.

Denitrifiers with denitrifying pathways truncated to N2O were pref-erentially promoted, and pore water concentrations of N2O in-creased, over time in the pilot-scale denitrifying woodchip bioreac-tor.

The final three conclusions suggest that the temporal development of the bio-geochemical environment and several key characteristics of the denitrifying woodchip bioreactor technology may be explained within the context of de-veloping syntrophic structures between the fermenting and denitrifying com-munities. The temporal decline in organic carbon export from and NO3

- re-moval rates in the pilot-scale denitrifying woodchip may have been the result of this, as well as the decline in the diversity of biogeochemical processes and preferential promotion of denitrifying pathways truncated to N2O.

There is a need for an increased understanding of the syntrophic metabo-lism between fermenters and denitrifiers in denitrifying woodchip bioreactors. This should start with focused research on the carbon cycling processes in denitrifying woodchip bioreactors as our understanding of the terminal elec-tron accepting processes in these systems, particularly the NOx-reducing sys-tem, is comparatively advanced. Nevertheless, studying syntrophic metabo-lisms is difficult as this involves the study of transient chemical species that are produced and then quickly consumed.

Comparative studies of ‘young’ and ‘mature’ denitrifying woodchip biore-actors using metagenomics could be used for studying the full community structure and its development in denitrifying woodchip bioreactors. Likewise, comparative studies between N2O emissions between ‘young’ and ‘mature’ denitrifying woodchip bioreactors could provide a general idea on the tem-poral development of these systems.

On a more applied level, if the denitrifying woodchip bioreactor technology is to be used on a commercial basis, there is a need to facilitate the design of these systems. Selecting the appropriate residence time at the specific envi-ronmental conditions (NO3

- concentration, temperature) for the design of these system currently requires in-depth expert knowledge. Moreover, our results have shown that these reactors adapt to the environment conditions given, wherefore selecting appropriate design criteria may be difficult. An increased understanding of how these systems adapt and develop, as partly identified in this thesis, and an increased knowledge on how this can be formulated in nu-merical models, will help with this objective.

52

8. Acknowledgements

I am very grateful for having been given the opportunity to pursue a doctoral degree and there are several people and institutions that I would like to acknowledge for this. This thesis would not have been written without the aid of talented teachers as I never thought I would find chemistry interesting. Roger Herbert (main supervisor), you have been my teacher for these past ten years (!); in 2009 you introduced me to hydrology and later showed me the pragmatism of chemistry which lit my interest in biogeochemistry. Despite having spent the past five years thinking (and dreaming) of biogeochemistry - currently, the nrfA gene quite often materializes in my dreams - my interest in the subject has only grown stronger. You have been a constant pillar of support throughout this time and it has been a privilege to have had you as my super-visor. I can’t thank you enough. Fritjof Fagerlund (assistant supervisor), our contact was sporadic but highly valuable to me. Thank you for talking some sense into me when I was under the illusion that my research had gone down the drain.

My time as a doctoral student would not have been possible without the financial support of LKAB, Boliden, and VINNOVA, the Swedish Innovation Agency, under grant 2014-01134, who are gratefully acknowledged. I have had the fantastic opportunity of visiting LKABs facilities in Kiruna numerous times during my studies. Although weather was sometimes discouraging for hydrological field work, my motivation was kept high by the interest taken in my project by LKAB employees, the dedication of everyone affiliated with my project, and from the food at Magnetiten. I am especially grateful to Johan Jonasson for all the assistance provided when constructing the bioreactor, and to Malin Suup for putting up with my endless inquiries about chemical results from the bioreactor.

I also want to extend my gratitude to all the people affiliated with the ‘miN-ing’ project. It has been a true privilege to share the knowledge of such tal-ented people. Special thanks to Sara Hallin and Maria Hellman for the inval-uable comments provided on paper IV and who throughout this project have been answering my many, many, many inquiries on microbiological techni-calities with patience and a kind heart. Thanks are also extended to Christo-pher Jones who provided unconditional support on matters of gas sampling and analysis, and the use of Sterivex filters, and to Robert Almstrand who took interest in ‘my’ bioreactor and helped with the DNA extraction from the Sterivex filters.

53

To past and present colleagues at and out of the Department of Earth Sci-ences (UU), thank you for creating an inspirational and friendly environment at the department. Maryeh Hedayati and Kenechukwo (Kaycee) Okoli, my long(est)-term office mates, thank you for personalizing the PhD experience; Agnes Soto Gomez, Diana Fuentes-Andino, Elisa Savelli, and Nicole Schaf-fer, my short(er)-term office mates; Jean-Marc Mayotte, Reinert Huseby Karlsen, Beatriz Quesada Montano, Marc Girons Lopez, Adam Dingwell, Lino Nilsson, Carmen Vega Riquelme, Mattias Winterdahl, Liang Tian, Nino Amvrosiadi, Saba Joodaki, Kristina Conrady, Farzad Basirat, and many more; thank you all for many interesting discussions not always concerning science. Special thanks to Rickard Pettersson who provided help related to the use of Trimble GPS, and to Joachim Audet who provided unconditional support on the use of static gas chambers to measure surface gas fluxes.

Mamma, Pappa, Roger (svärfar), Margot och Sivert, tack för att ni alltid har ställt upp när tiden inte har räckt till. Ronja, tack för att du har ställt upp under dessa år och tålmodigt lyssnat då jag pratat om regressionsanalyser, vat-tenföringsmätningar, stökiometri, med mera. Utan dig så hade jag aldrig klarat detta. Freja, att du svarar ’dator’ på frågan om vad du ska arbeta med när du blir stor får mig att reflektera över den tid som jag har lagt ned på detta arbete. Att få bryta av och titta på bilder på ’Bamse’ och lejon, leka apa, läsa Alfons, och ibland bara springa runt, har varit ovärderliga inslag under denna tid.

54

9. Sammanfatting på svenska

Biogeokemiska processer i kvävebarriärer med tillämpning inom gruvin-dustrin

Den ökade användningen av kvävebaserade produkter har i takt med industri-aliseringen ökat spridningen av kväveföreningar till vattendrag, sjöar, kust-områden, och hav. Denna spridning har potentiella konsekvenser så som för-surning, övergödning, och syrebrist i recipienterna, vilket kan leda till en för-sämring av livsviktiga förutsättningar i akvatiska miljöer och därmed en minskning av den biologiska mångfalden. De sprängmedel som idag används inom industriella verksamheter är främst baserade på ammoniumnitrat (NH4NO3) då de är förhållandevis effektiva och säkra att använda till en för-hållandevis låg kostnad. Studier inom gruvindustrin har emellanåt visat att upp mot 28 % av det använda sprängmedlet inte detonerar och läcker ut till närlig-gande vattendrag, sjöar, och kustområden, främst i form av nitrat (NO3

-), och även med de mest effektiva brytningsmetoderna så uppgår denna del till minst 12% av det använda sprängmedlet. Det övergripande målet med den här av-handlingen har varit att utvärdera tillämpningen av kvävebarriärer (eng. deni-trifying bioreactors) för att minska kvävexporten från gruvindustrin, som en del i det större VINNOVA projektet ’miNing’.

En kvävebarriär är en passiv reningsmetod som använder sig av denitrifi-kation varmed NO3

- reduceras till kvävgas (N2(g)) för att minska NO3- halter

i vatten. Vatten med höga halter av NO3- passeras genom en matris av träflis

där organiskt kol frigörs och skapar en miljö som främjar denitrifierande bak-terier. Studier har dock visat att reningskapaciteten i dessa system minskar med ~50% över tid, men stabiliseras någorlunda efter deras första år i bruk; att en del NO3

- lämnar systemet som lustgas (N2O) eller ammonium (NH4+);

att vissa delar av kvävebarriären inte är aktiva p.g.a. preferentiella flöden i träflismatrisen; och att oönskade processer så som metanproduktion kan före-komma i dessa system. Dynamiken varmed ovanstående sker är inte fastlagd och studier har visat en variabilitet mellan olika system, vilka båda är av in-tresse då de har en potentiell negativ påverkan på kvävebarriärer som renings-metoder ur ett kostnads- och hållbarhetsperspektiv. Av särskilt intresse är hur denna dynamik förändras över tid och vad som driver detta, då reningssyste-met förväntas fungera passivt utan underhåll under flera år (~10-40 år) efter dess installation. Därför var ett övergripande forskningsmål i denna avhand-ling att förstå denna dynamik och hur den förändras över tid.

55

Denna avhandling är baserad på resultat från en kvävebarriär i laboratorie-skala (artikel I), och en kvävebarriär i pilotskala som installerades på LKABs gruvområde i Kiruna (artikel II-IV). I dessa två system så studerades tillämp-ningen av kvävebarriärer för att minska NO3

- i gruvvatten med neutralt pH (artikel I och II), preferentiellt flöde i träflismatrisen (artikel I), de biogeoke-miska processerna aktiva i kvävebarriären och deras förändring över tid (arti-kel II-IV).

Laboratorieförsöket bestod av tre kolonner (10 cm i diameter, 40 cm långa) som fylldes med en blandning av avbarkad träflis (tall) och aktivt slam. I en av kolonnerna så installerades ring-formade skiljeväggar som ett försök att minska tendensen för preferentiellt flöde genom att förhindra flöde längst si-dorna i kolonnen. Destillerat vatten innehållande 30 mg NO3

--N L-1 och 400 mg SO4

2- L-1 (100 mg SO42- L-1 mot slutet av experimentet) pumpades genom

kolonnerna under 92 dagar. Temperaturen och uppehållstiden i kolonnerna ändrades under experimentets gång och två spårämnesförsök (100 mg Cl- L-1) utfördes i kolonnen med skiljeväggarna för att studera ämnestransport i träflis mediet. Vid en uppehållstid om ~2.5 dygn så reducerades 30 mg NO3

--N L-1 till under detektionsnivåer vid alla temperaturer (5°C, 12°C, 22°C) i alla ko-lonner utan någon betydande produktion av nitrit (NO2

-) och ammonium (NH4

+). När uppehållstiden minskades till ~1 dygn så förblev inkommande NO3

- ofullständigt reducerat och NO3- koncentrationen i utloppet från alla ko-

lonner översteg 15 mg NO3--N; i en av kolonnerna (22°C) så ökade utlopps-

koncentrationer av NH4+ kontinuerligt från dess att uppehållstiden minskat.

Den övergripande låga produktionen av NH4+ i förhållande till den reducerade

mängden NO3-, och ökningen i pH och alkalinitet över kolonnerna, påvisade

denitrifikation som den huvudsakliga processen varmed NO3- reducerades.

Spårämnesförsöken påvisade preferentiellt flöde där flödet var koncentrerat till en begränsad del av porvolymen (mobila zonen) och att porvolymen som inte deltog i flödet (immobila zonen) ökade från ~12% till ~25% då uppehålls-tiden minskade. Ämnestransport mellan de mobila och immobila zonerna (dif-fusion) var närapå obefintlig, vilket innebar att de immobila zonerna inte hel-ler bidrog till att minska NO3

- halter i det inkommande vattnet. Det kunde dock inte visas att de ringformade skiljeväggarna bidrog till att öka renings-kapaciteten i kolonnerna.

Kvävebarriären i pilotskala konstruerades på LKABs gruvområde i Kiruna i juni 2015: ett dike fylldes med en blandning av avbarkad träflis och rötat avsloppsslam, varpå delar av träflis-slam blandningen täcktes med moränjord (Figur 4). Gruv- och processvatten (neutralt pH) från klarningsdammen på gruvområdet med en genomsnittligt NO3

- koncentration på 22,0 mg N L-1 pumpades genom kvävebarriären under två operativa år och flera olika uppe-hållstider användes under de operativa åren. Det första operativa året sträckte sig mellan 22:a juni och 20:de november 2015, och det andra operativa året mellan 9:de maj och 21:a oktober 2016. Inflödet till kvävebarriären stoppades

56

mellan de operativa åren p.g.a. låg lufttemperatur (< -20°C). Porvatten analy-serades en gång i månaden för NO3

-, NO2-, NH4

+, SO42-, TOC, pH, funktionella

gener som kodar för olika processer för omvandling av NOx, m.m. (se ’paper II’ och ’paper IV’) inklusive gaser (N2O, CH4) lösta i porvatten och som av-gasades från kvävebarriärens yta. Inlopps- och utloppsvatten från kvävebar-riären analyserades två gånger i veckan, och i samband med provtagningen av porvattnet. Efter det att kvävebarriären tagits ur bruk så analyserades även trä-flisen för olika funktionella gener som kodar för olika processer som omvand-ling NOx.

Inkommande NO3- halter reducerades till under detektionsnivåer (0,06 mg

N L-1) vid uppehållstider om ~1,9-2,5 dagar och vid temperaturer över 5°C, utan betydande produktion av NO2

-, N2O, eller NH4+; produktionen av N2O

var genomsnittligen 0,1% av den reducerade mängden NO3- under dessa för-

hållanden. Vid en uppehållstid om ~1 dag och/eller vid temperaturer under 5°C så var NO3

- reduceringen i kvävebarriären ofullkomlig och en högre andel av det reducerade NO3

- övergick till N2O och/eller NH4+; då ⪍ 10% av det

inkommande NO3- reducerades så övergick ~100% till N2O.

Den biogeokemiska miljön i kvävebarriären utvecklades mellan de två ope-rativa åren. En mass-balans visade att mångfalden av biogeokemiska proces-ser minskade över tid, där denitrifikation blev den dominerande processen i kvävebarriären under det andra operativa året; DNRA (eng. ’dissimilatory ni-trate reduction to ammonium’), sulfatreduktion, och metangenes minskade i omfattning under det andra operativa året. Exporten av organiskt kol från kvä-vebarriären minskade exponentiellt från uppstart vilket var kopplat till en ök-ning i pH och alkalinitet i kvävebarriären. Mass-balansen visade att denna koppling kontrollerades av fermentation och att slutprodukten av fermentat-ionen övergick från huvudsakligen butyrat till acetat över tid. Det föreslogs att denna process var katalysatorn till minskningen i mångfalden av biogeo-kemiska processer över tid i systemet.

Reduceringshastigheter av NO3- minskade och kinetiken för NO3

- reduce-ring förändrades mellan de två operativa åren. NO3

- reduceringshastigheten blev oberoende av NO3

- koncentration med tid från uppstart av barriären och temperaturoptimum för NO3

- reduceringen minskade från 24.2°C till 16.0°C. Nitratreduceringen blev mer känslig för temperaturförändringar mellan de två operativa åren, vilket användes som en indikation på att ’kvalitén’ på kolsub-stratet som användes av de NO3

- reducerande mikroberna försämrades och att tillgängligheten av organisk kol alltmer kontrollerades av fermentation i kvä-vebarriären (artikel III). Detta tolkades som att syntrofi mellan fermenterare och NO3

- reducerare utvecklades över tid i kvävebarriären, och NO3- reduce-

rare blev mer beroende av produktionen av kolsubstrat från fermenterare un-der det andra operativa året.

Mängden av funktionella gener som kodar för olika NOx-reduceringssteg förändrades mellan de två operativa åren. Övergripande så ökade mängden av

57

funktionella gener som kodar för NO2--reducerare (Σnir = nirK + nirS) i för-

hållande till mängden av funktionella gener som kodar för N2O-reducerare (ΣnosZ = nosZI + nosZII); mängden nirS, nirK, och nosZII ökade, medan mängden nosZI minskade, mellan de operativa åren. Även mängden av den funktionella genen som kodar för DNRA (nrfA; eng. ’Dissimilatory nitrate reduction to ammonium’; NH4

+ produktion) ökade över tid, men kontrollerade inte NH4

+ produktionen, vilken istället begränsades av ogynnsamma förhål-landen (låg C/NO3

- kvot). Den funktionella genen hdh som kodar för anammox (eng. ’Anaerobic ammonium oxidation’) påvisades inte i kvävebar-riären. Halter av N2O var starkt kopplade till mängden nirS i porvattnet, vilken i sin tur hade den starkaste inverkan på den ökade mängden Σnir relativt mäng-den ΣnosZ i porvattnet mellan de operativa åren. Detta medförde att potentia-len för NO2

- reduktion över N2O reduktion ökade över tid och innebar en för-höjd risk för ökad produktion av N2O; förhöjda halter av N2O, medan mins-kade halter av NO2

-, i porvattnet påvisades mellan de två operativa åren. För-ändringen i strukturen på det NOx-reducerande mikrobiella samhället mellan de operativa åren föreslogs vara ett resultat av en ökad konkurrens för det till-gängliga organiska kolet i systemet, och diskuterades utifrån en utveckling av syntrofi mellan fermenterare och denitrifierare i systemet.

Övergripande så visade resultaten från de två kvävebarriärsystemen att me-toden är ett möjligt alternativ för minskning av kvävehalter i gruvvatten med neutralt pH. För ett hållbart nyttjande av dessa reningssystem krävs dock att tillgängligheten av NO3

- alltid förblir god för att förhindra aktiveringen av oönskade reaktioner såsom sulfatreduktion eller metan produktion. Vidare så förespråkas det att all inkommande NO3

- ska reduceras i kvävebarriären för att minska produktionen av N2O och NH4

+ i förhållande till den reducerade mängden NO3

-.

58

10. References

Addy, K., Gold, A.J., Christianson, L.E., David, M.B., Schipper, L.A., Ratigan, N.A., 2016. Denitrifying Bioreactors for Nitrate Removal: A Meta-Analysis. Journal of Environmental Quality 45, 873-888. doi:10.2134/jeq2015.07.0399

Appelo, C.A.J., Postma, D., 2005. Geochemistry, Groundwater, and Pollution (Sec-ond Edition), A.A. Balkema, Leiden.

Arcus, V.L., Prentice, E.J., Hobbs, J.K., Mulholland, A.J., Van der Kamp, M.W., Pud-ney, C.R., Parker, E.J., Schipper, L.A., 2016. On the temperature dependence of enzyme-catalyzed rates. Biochemistry 55, 1681-1688. doi:10.1021/acs.bio-chem.5b01094

Betlach, M.R., Tiedje, J.M., 1981. Kinetic Explanation for Accumulation of Nitrite, Nitric Oxide, and Nitrous Oxide During Bacterial Denitrification. Applied Envi-ronmental Microbiology 42(6), 1074-1084.

Bianchi, T.S., DiMarco, S.F., Cowan Jr, J.H., Hetland, R.D., Chapman, P., Day, J.W., Allison, M.A., 2010. The science of hypoxia in the Northern Gulf of Mexico: A review. Science of the Total Environment 408, 1471-1484. doi:10.1016/j.sci-totenv.2009.11.047

Boliden, 2017. Metaller för långsiktigt värdeskapande - Årsredovisning 2017 [Metals for long term value - annual 2017]. New Boliden. (in Swedish).

Bosatta, E., Ågren, G.I., 1999. Soil organic matter quality interpreted thermodynami-cally. Soil Biology and Biochemistry 31(13), 1889-1891. doi:10.1016/S0038-0717(99)00105-4

Burgin, A.J., Hamilton, S.K., 2007. Have we overemphasized the role of denitrifica-tion in aquatic ecosystems? A review of nitrate removal pathways. Frontiers in Ecology and the Environment 5, 89-96. doi:10.1890/1540-9295(2007)5[89:HWOTRO]2.0.CO;2

Cameron, S.G., Schipper, L.A., 2011. Evaluation of passive solar heating and alterna-tive flow regimes on nitrate removal in denitrification beds. Ecological Engineer-ing 37, 1195-1204. doi:10.1016/j.ecoleng.2011.02.020

Cameron, S.G., Schipper, L.A., 2010. Nitrate removal and hydraulic performance of organic carbon for use in denitrification beds. Ecological Engineering 36, 1588-1595. doi:10.1016/j.ecoleng.2010.03.010

Canfield, D.E., Glazer, A.N., Falkowski, P.G., 2010. The Evolution and Future of Earth’s Nitrogen Cycle. Science 330, 192-196. doi:10.1126/science.1186120

Carstensen, J., Andersen, J.H., Gustafsson, B.G., Conley, D.J., 2014. Deoxygenation of the Baltic Sea during the last century. Proceedings of the National Academy of Sciences of the United States of America 111(15), 5628-5633. doi:10.1073%2Fpnas.1323156111

Chlot, S., Widerlund, A., Husson, E., Öhlander, B., Ecke, F., 2013. Effects on nutrient regime in two recipients of nitrogen-rich mine effluents in northern Sweden. Ap-plied Geochemistry 31, 12-24. doi:10.1016/j.apgeochem.2012.11.016

59

Choudhury, M.I., 2018. Nitrogen removal by wetlands in a cold climate: understand-ing interactions between macrophytes and microorganisms (Doctoral disserta-tion). Swedish University of Agricultural Sciences, Uppsala, Sweden.

Christianson, L., Helmers, M., Bhandari, A., Moorman, T. 2013. Internal hydraulics of an agricultural drainage denitrification bioreactor. Ecological Engineering 52, 298-307. doi:10.1016/j.ecoleng.2012.11.001

Christianson, L.E., Lepine, C., Sibrell, P.L., Penn, C., Summerfelt, S.T., 2017. Deni-trifying woodchip bioreactor and phosphorus filter pairing to minimize pollution swapping. Water Research 121, 129-139. doi:10.1016/j.waterres.2017.05.026

Conley, D.J., Björck, S., Bonsdorff, E., Carstensen, J., Destouni, G., Gustafsson, B.G., Kuosa, H., Meier, H.E.M., Müller-Karulis, B., Nordberg, K., Norkko, A., Nürn-berg, G., Pitkänen, H., Rabalais, N.N., Rosenberg, R., Savchuk, O.P., Slomp, C.P., Voss, M., Wulff, F., Zillén, L., 2009. Hypoxia-Related Processes in the Bal-tic Sea. Environmental Science & Technology 43, 3412-3420.

doi:10.1021/es802762a David, M.B., Gentry, L.E., Cooke, R.A., Herbstritt, S.M., 2016. Temperature and

Substrate Control Woodchip Bioreactor Performance in Reducing Tile Nitrate Loads in East-Central Illinois. Journal of Environmental Quality 45, 822-829. doi:10.2134/jeq2015.06.0296

Degnan, J.R., Böhlke, J.K., Pelham, K., Langlais, D.M., Walsh, G.J., 2016. Identifi-cation of Groundwater Nitrate Contamination from Explosives Used in Road Construction: Isotopic, Chemical, and Hydrologic Evidence. Environmental Sci-ence & Technology 50, 593-603. doi:10.1021/acs.est.5b03671

Diaz, R.J., Rosenberg, R., 2008. Spreading Dead Zones and Consequences for Marine Ecosystems. Science 321, 926-929. doi:10.1126/science.1156401

EEA, 2018. European waters - Assessment of status and pressures 2018. European Environmental Agency, report no 7/2018. doi:10.2800/303664

Elgood, Z., Robertson, W.D., Schiff, S.L., Elgood, R., 2010. Nitrate removal and greenhouse gas production in a stream-bed denitrifying bioreactor. Ecological En-gineering 36, 1575-1580. doi:10.1016/j.ecoleng.2010.03.011

Feyereisen, G.W., Moorman, T.B., Christianson, L.E., Venterea, R.T., Coulter, J.A., Tschirner, U.W., 2016. Performance of Agricultural Residue Media in Laboratory Denitrifying Bioreactors at Low Temperatures. Journal of Environmental Quality 45, 779-787. doi:10.2134/jeq2015.07.0407

Forsyth, B., Cameron, A., Miller, S., 1995. Explosives and Water Quality. Vancouver, B.C.; Sudbury, ON; Denver, CO: Golder Associates Ltd.

Galloway, J.N., Aber, J.D., Erisman, J.W., Seitzinger, S.P., Howarth, R.W., Cowling, E.B., Cosby, B.J., 2003. The Nitrogen Cascade. BioScience 53(4), 341-356. doi:10.1641/0006-3568(2003)053[0341:TNC]2.0.CO;2

Garcia-Viloca, M., Gao, J., Karplus, M., Truhlar, D.G., 2004. How enzymes work: analysis by modern rate theory and computer simulations. Science 303, 186-195. doi:10.1126/science.1088172

Gerke, H.H., 2006. Preferential flow descriptions for structured soils. Journal of Plant Nutrition and Soil Science, 169:382-400. doi:10.1002/jpln.200521955

Ghane, E., Fausey, N.R., Brown, L.C., 2015. Modeling nitrate removal in a denitrifi-cation bed. Water Research 71, 294-305. doi:10.1016/j.watres.2014.10.039

Ghane, E., Feyereisen, G.W., Rosen, C.J., Tschirner, U.W., 2018. Carbon Quality of Four-Year-Old Woodchips in a Denitrification Bed Treating Agricultural Drain-age. Transactions of the ASABE. 61(3), 995-1000. doi:10.13031/trans.12642

60

Gibert, O., Pomierny, S., Rowe, I., Kalin, R.M., 2008. Selection of organic substrates as potential reactive materials for use in denitrification permeable reactive barrier (PRB). Bioresource Technology 99, 7587-7596.

doi:10.1016/j.biortech.2008.02.012 Graf, D.R.H., Jones, C.M., Hallin, S., 2014. Intergenomic Comparisons Highlight

Modularity of the Denitrification Pathway and Underpin the Importance of Com-munity Structure for N2O Emissions. PLOS one 9, e114118. doi:10.1371/jour-nal.pone.0114118

Greenan, C.M., Moorman, T.B., Kaspar, T.C., Parkin, T.B., Jaynes, D.B., 2006. Com-paring Carbon Substrates for Denitrification of Subsurface Drainage Water. Jour-nal of Environmental Quality 35, 824-829. doi:10.2134/jeq2005.0247

Greenan, C.M., Moorman, T.B., Parkin, T.B., Kaspar, T.C., Jaynes, D.B., 2009. De-nitrification in wood chip bioreactors at different water flows. Journal of Envi-ronmental Quality 38(4), 1664-1671. doi:10.2134/jeq2008.0413

Grieβmeier, V., Bremges, A., McHardy, A.C., Gescher, J., 2017. Investigation of dif-ferent nitrogen reduction routes and their key microbial players in wood chip-driven denitrification beds. Scientific Reports 7, 17028. doi:10.1038/s41598-017-17312-2

Grieβmeier, V., Gescher, J., 2018. Influence of the Potential Carbon Sources for Field Denitrification Beds on Their Microbial Diversity and the Fate of Carbon and Nitrate. Frontiers in Microbiology 9, 1313. doi:10.3389/fmicb.2018.01313

González-Cabaleiro, R., Ofiţeru, I.D., Rodríguez, J., 2015. Microbial catabolic activ-ities are naturally selected by metabolic energy harvest rate. The ISME Journal 9, 2630-2641. Doi:10.1038/ismej.2015.69

Gudelj, I., Kinnersley, M., Rashkov, P., Schmidt, K., Rosenzweig, F., 2016. Stability of Cross-Feeding Polymorphism in Microbial Communities. PLoS Computa-tional Biology 12(12), e1005269. doi:10.1371/journal.pcbi.1005269

Halaburka, B.J., LeFevre, G.H., Luthy, R.G., 2017. Evaluation of Mechanistic Models for Nitrate Removal in Woodchip Bioreactors. Environmental Science & Tech-nology, 51, 5156-5164. doi:10.1021/acs.est.7b01025

Halaburka, B.J., LeFevre, G.H., Luthy, R.G., 2019. Quantifying the temperature de-pendence of nitrate reduction in woodchip bioreactors: experimental and modeled results with applied case-study. Environmental Science: Water Research & Tech-nology. doi:10.1039/c8ew00848e

Hallin, S., Philippot, L., Löffler, F.E., Sanford, R.A., Jones, C.M., 2018. Genomics and Ecology of Novel N2O-Reducing Microorganisms. Trends in Microbiology 26, 43-55. doi:10.1016/j.tim.2017.07.003

Hanke, A., Jasmine, B., Hargesheimer, T., Tegetmeyer, H.E., Sharp, C.E., Strous, M., 2016. Selective pressure of temperature on competition and cross-feeding within denitrifying and fermentative microbial communities. Frontiers in Microbiology 6, 1461. doi:10.3389/fmicb.2015.01461.

Hassanpour, B., Giri, S., Pluer, W.T., Steenhuis, T.S., Geohring, L.D., 2017. Seasonal performance of denitrifying bioreactors in the Northeastern United States: Field trials. Journal of Environmental Management 202, 242-253. doi:10.1016/j.jen-vman.2017.06.054

Healy, M.G., Barrett, M., Lanigan, G.J., Serrenho, A.J., Ibrahim, T.G., Thornton, S.F., Rolfe, S.A., Huang, W.E., Fenton, O., 2014. Optimizing nitrate removal and eval-uating pollution swapping trade-offs from laboratory denitrification bioreactors. Ecological Engineering. 74, 290-301. doi:10.1016/j.ecoleng.2014.10.005

Hellman, M., 2018. Microbial nitrate removal from mining waters (Licentiate thesis). Swedish University of Agricultural Sciences, Uppsala, Sweden.

61

Hendrickx, J.M.H., Flury, M., 2001. Uniform and Preferential Flow Mechanisms in the Vadose Zone. In: Conceptual Models of Flow and Transport in the Fractured Vadose Zone (2001) U.S. National Committee for Rock Mechanics, Panel on Conceptual Models of Flow and Transport in the Fractured Vadose Zone. Na-tional Academy Press, Washington D.C.

Herbert, R.B. Jr., 2011. Implications of non-equilibrium transport in heterogeneous reactive barrier systems: Evidence from laboratory denitrification experiments. Journal of Contaminant Hydrology 123, 30-39.

doi:10.1016/j.jconhyd.2010.12.004 Herbert, R., Nordström, A., 2017. Leachate generation and nitrogen release from

small-scale rock dumps at the Kiruna iron ore mine, in: Wolkersdorfer, C., Sartz, L., Sillanpää, M., Häkkinen, A. (Eds.), 13th International Mine Water Association Congress- Mine Water & Circular Economy. Lappeenranta University of Tech-nology, Finland.

Herbert, R.B., Winbjörk, H., Hellman, M., Hallin, S., 2014. Nitrogen removal and spatial distribution of denitrifier and anammox communities in a bioreactor for mine drainage treatment. Water Research 66, 350-360. doi:10.1016/j.wa-tres.2014.08.038

Hobbs, J.K., Jiao, W., Easter, A.D., Parker, E.J., Schipper, L.A., Arcus, V.L., 2013. Change in Heat Capacity for Enzyme Catalysis Determines Temperature Depend-ence of Enzyme Catalyzed Rates. ACS Chemical Biology 8(11), 2388.2393. doi:10.1021/cb4005029

Hoover, N.L., Bhandari, A., Soupir, M.L., Moorman, T.B., 2016. Woodchip Denitri-fication Bioreactors: Impact of Temperature and hydraulic Retention Time on Ni-trate Removal. Journal of Environmental Quality 45, 803-812. doi:10.2134/jeq2015.03.0161

Jaynes, D.B., Moorman, T.B., Parkin, T.B., Kaspar, T.C., 2016. Simulating Woodchip Bioreactor Performance Using a Dual-Porosity Model. Journal of Environmental Quality 45, 830-838. doi:10.2134/jeq2015.07.0342

Jones, C.M., Graf, D.R.H., Bru, D., Philippot, L., Hallin, S., 2013. The unaccounted yet abundant nitrous oxide-reducing microbial community: a potential nitrous ox-ide sink. The ISME Journal 7, 417-426. doi:10.1038/ismej.2012.125

Jones, C.M., Hallin, S., 2010. Ecological and evolutionary factors underlying global and local assembly of denitrifier communities. The ISME Journal, 4, 633-641. doi:10.1038/ismej.2009.152

Jones, C.M., Spor, A., Brennan, F.P., Breuil, M.C., Bru, D., Lemanceau, P., Griffiths, B., Hallin, S., Philippot, L., 2014. Recently identified microbial guild mediates soil N2O sink capacity. Nature Climate Change 4, 801-805. doi:10.1038/NCLI-MATE2301

Jordbruksverket, 2018. Jordbruksmarkens använding 2018, preliminär statistik [Use of agricultural land in 2018, preliminary statistics]. Jordbruksverket, JO 10 SM 1801. (in Swedish).

Kox, M.A.R., Jetten, M.S.M., 2015. The Nitrogen Cycle, in: Lugtenberg, B. (Ed.), Principles of Plant-Microbe Interactions, Microbes for Sustainable Agriculture, Springer, Switzerland.

Lilja, E.E., Johnson, D.R., 2016. Segregating metabolic processes into different mi-crobial cells accelerates the consumption of inhibitory substrates. The ISME Jounal 10, 1568-1578. doi:10.1038/ismej.2015.243

Lindeström, L., 2012. Kväveutsläpp från gruvindustrin [Nitrogen releases from the mining industry]. Stockholm: SveMin. (in Swedish).

LKAB, 2016. Miljörapport LKAB Kiruna 2015 [Environmental Report LKAB Kiruna 2015]. Luossavaara-Kiirunavaara AB (in Swedish).

62

LKAB, 2017. Miljörapport LKAB Kiruna 2016 [Environmental Report LKAB Kiruna 2016]. Luossavaara-Kiirunavaara AB (in Swedish).

LKAB, 2018. Miljörapport LKAB Kiruna 2017 [Environmental Report LKAB Kiruna 2017]. Luossavaara-Kiirunavaara AB (in Swedish).

Long, L.M., Schipper, L.A., Bruesewitz, D.A., 2011. Long-term nitrate removal in a denitrification wall. Agriculture, Ecosystems, and Environment 140, 514-520. doi:10.1016/j.agee.2011.02.005

Lotti, T., Kleerebezem, R., Hu, Z., Kartal, B., de Kreuk, M.K., van Erp Taalman Kip, C., Kruit, J., Hendrickx, T.L.G., van Loosdrecht, M.C.M., 2014. Pilot-scale eval-uation of anammox-based mainstream nitrogen removal from municipal wastewater. Environmental Technology 36(9), 1167-1177. doi:10.1080/09593330.2014.982722

Meyers, S., Shanley, E.S., 1990. Industrial explosives - a brief history of their devel-opment and use. Journal of Hazardous Materials 23, 183-201.

Mohseni-Bandpi, A., Elliot, D.J., Zazouli, M.A., 2013. Biological nitrate removal pro-cesses from drinking water supply-a review. Journal of Environmental Health Science & Engineering 11, 35. doi:10.1186/2052-336X-11-35

Moorman, T.B., Parkin, T.B., Kaspar, T.C., Jaynes, D.B., 2010. Denitrification activ-ity, wood loss, and N2O emissions over 9 years from a wood chip bioreactor. Ecological Engineering 36, 1567-1574. doi:10.1016/j.ecoleng.2010.03.012

Morin, K.A., Hutt, N.N., 2009. Mine-Water Leaching of Nitrogen Species from Ex-plosive Residues. Proceedings, GeoHalifax2009. 1549-1553.

Morley, N.J., Richardson, D.J., Baggs, E.M., 2014. Substrate Induced Dentirification over or under Estimates Shifts in Soil N2/N2O Ratios. PLoS One 9(9), e108144. doi:10.1371%2Fjournal.pone.0108144

Naturvårdsverket, 2018. Miljömålen: Årlig uppföljning av Sveriges nationella miljö-mål 2018 [Environmental objectives: Annual report on Swedens national envi-ronmental objectives]. Rapport 6833.

Naturvårdsverket, 2019. Swedish Environmental Protection Agency. Swedish Pollu-tant Release and Transfer Register. https://utslappisiffror.naturvardsver-ket.se/?epslanguage=en (Accessed 2019-01-07).

Nilsson, L., 2018. Nitrogen-cycling tracing methods: case studies at cold-climate mine sites in northern Sweden (Doctoral dissertation). Luleå University of Tech-nology, Luleå, Sweden.

Nilsson, L., Widerlund, A., 2017. Tracing nitrogen cycling in mining waters using stable nitrogen isotope analysis. Applied Geochemistry 84, 41-51. doi:10.1016/j.apgeochem.2017.05.025

NITREM, 2019. NITREM: NITrogen REMoval from waste rock. https://www.ni-trem.eu/ (Accessed 2019-02-15).

Nordström, A., Herbert, R.B., 2017. Denitrification in a low-temperature bioreactor system at two different hydraulic residence times: laboratory column studies. En-vironmental Technology 38(11), 1362-1375.

doi:10.1080/09593330.2016.1228699 Nordström, A., Herbert, R.B., 2018. Determination of the major biogeochemical pro-

cesses in a denitrifying woodchip bioreactor for treating mine drainage. Ecologi-cal Engineering 110, 54-66. doi:10.1016/j.ecoleng.2017.09.018

Nordström, A., Herbert, R., 2017a. Field-scale denitrifying woodchip bioreactor treat-ing high nitrate mine water at low temperatures, in: Wolkersdorfer, C., Sartz, L., Sillanpää, M., Häkkinen, A. (Eds.), 13th International Mine Water Association Congress- Mine Water & Circular Economy. Lappeenranta University of Tech-nology, Finland.

63

Nordström, A., Herbert, R.B., 2019. Identification of the Temporal Control on Nitrate Removal Rate Variability in a Denitrifying Woodchip Bioreactor. Ecological En-gineering 127, 88-95. doi:10.1016/j.ecoleng.2018.11.015

Oksanen, J., Guillaume Blanchet, F., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P.R., O’Hara, R.B., Simpson, G.L., Solynos, P., Stevens, M.H.H., Szoecs, E., Wagner, H., 2018. Vegan: Community Ecology Package. R package version 2.5-2. https://CRAN.R-project.org/package=vegan

Palmqvist, E., Hahn-Hägerdal, B., 2000. Fermentation of lignocellulosic hydroly-sates. II: inhibitors and mechanisms of inhibition. Bioresource Technology 74(1), 25.33. doi:10.1016/S0960-8524(99)00161-3

Pan, Y., Ni, B.J., Bond, P.L., Ye, L., Yuan, Z., 2013. Electron competition among nitrogen oxides reduction during methanol-utilizing denitrification in wastewater treatment. Water Research 47(10), 3273-3281. doi:10.1016/j.watres.2013.02.054

Pan, Y., Ye, L., Ni, B.J., Yuan, Z., 2012. Effect of pH on N2O reduction and accumu-lation duringdenitrification by methanol utilizing denitrifiers. Water Research 46(15), 4832-4840. doi:10.1016/j.watres.2012.06.003

Parkhurst, D.L., Appelo, C.A.J., 1999. User’s Guide to PHREEQC (version 2): a com-puter program for speciation, batch-reaction, one-dimensional transport, and in-verse geochemical calculations. Denver (CO), USGS (Water-Resources Investi-gations; no 99-4259).

Pedersen, A.R., 2015. HMR: Flux Estimation with Static Chamber Data. R Package Version 0.4.1. http://cran.r-project.org/package=HMR

Porter, M.D., Andrus, J.M., Bartoleiro, N.A., Rodriguez, L.F., Zhang, Y., Zilles, J.L., Kent, A.D., 2015. Seasonal Patterns in Microbial Community Composition in De-nitrifying Bioreactors Treating Subsurface Agriculture Drainage. Microbial Ecol-ogy 70(3), 710-723. doi:10.1007/s00248-015-0605-8

Rittman, B.E., McCarty, P., 2001. Biotechnology: Principles and Applications. McGraw-Hill Book Co., New York.

Robertson, W.D., 2010. Nitrate removal rates in woodchip media of varying age. Eco-logical Engineering 36, 1581-1587. doi:10.1016/j.ecoleng.2010.01.008

Robertson, W.D., Vogan, J.L., Lombardo, P.S., 2008. Nitrate Removal Rates in a 15-Year-Old Permeable Reactive Barrier Treating Septic System Nitrate. Ground-water Monitoring & Remediation 28, 65-72. doi:10.1111/j.1745-6592.2008.00205.x

Schipper, L.A., Barkle, G.F., Hadfield, J.C., Vojvodić-Vuković, M., Burgess, C.P., 2004. Hydraulic constraints on the performance of a groundwater denitrification wall for nitrate removal from shallow groundwater. Journal of Contaminant Hy-drology 69, 263-279. doi:10.1016/S0169-7722(03)00157-8

Schipper, L.A., Hobbs, J.K., Rutledge, S., Arcus, V.L., 2014. Thermodynamic theory explains the temperature optima of soil microbial processes and high Q10 values at low temperatures. Global Change Biology 20(11), 3578-3586.

doi:10.1111/gcb.12596 Schipper, L.A., Robertson, W.D., Gold, A.J., Jaynes, D.B., Cameron, S.C. 2010. De-

nitrifying bioreactors - An approach for reducing nitrate loads to receiving waters. Ecological Engineering 36, 1532-1543. doi:10.1016/j.ecoleng.2010.04.008

Schipper, L.A., Vojvodić-Vuković, M., 2001. Five years of nitrate removal, denitrifi-cation and carbon dynamics in a denitrification wall. Water Research 35, 3473-3477. doi:10.1016/S0043-1354(01)00052-5

Schlesinger, W.H., Bernhardt, E.S., 2013. Biogeochemistry 3rd Edition - An Analysis of Global Change. Academic Press, doi:10.1016/B978-0-12-385874-0.00007-8

64

Schmidt, C.A., Clark, M.W., 2013. Deciphering and modeling the physiochemical drivers of denitrification rates in bioreactors. Ecological Engineering 60, 276-288. doi:10.1016/j.ecoleng.2013.07.041

SGU, 2019. Swedish Geological Survey. https://www.sgu.se/bergsstaten/gruvor/gru-vor-i-sverige-2/ (Accessed 07.01.2019) (in Swedish)

Shaw, A., Takács, I., Pagilla, K.R., Murthy, S., 2013. A new approach to assess the dependency of extant half-saturation coefficients on maximum process rates and estimate intrinsic coefficients. Water Research 47, 5986-5994. doi:10.1016/j.wa-tres.2013.07.003.

Shaw, A., Takacs, I., Pagilla, K., Riffat, R., DeClippeleir, H., Wilson, C., Murthy, S., 2015. Toward universal half-saturation coefficients: describing extant Ks as a function of diffusion. Water Environment Research 87(5), 387-391. doi:10.2175/106143015X14212658614072

Sierra, C.A., 2012. Temperature sensitivity of organic matter decomposition in the Arrhenius equation: some theoretical considerations. Biogeochemistry 108, 1-15. doi:10.1007/s10533-011-9596-9

SMHI, 2018a. Swedish Meteorological and Hydrological Institute. https://www.smhi.se/klimatdata/meteorologi/temperatur (Accessed 18.10.09) (in Swedish)

SMHI, 2018b. Swedish Meteorological and Hydrological Institute. https://www.smhi.se/klimatdata/meteorologi/sno (Accessed 18.10.09) (in Swe-dish)

SMED, 2016. Läckage av näringsämnen från svensk åkermark [Nutrient leaching from Swedish farmland]. Svensk MiljöEmissionsData, report Nr 189 (in Swe-dish).

Stein, L.Y., Klotz, M.G., 2016. The nitrogen cycle. Current Biology 26(3), R94-R98. doi:10.1016/j.cub.2015.12.021

Sun, Y., Cheng, J., 2002. Hydrolysis of lignocellulosic materials for ethanol produc-tion: a review. Bioresource Technology 83(1), 1-11. doi:10.1016/S0960-8524(01)00212-7

Toride, N., Leij, F.J., van Genuchten, M.T., 1995. The CXTFIT code for estimating transport parameters from laboratory or field tracer experiments: version 2.0. Riv-erside (CA): U.S. Salinity Laboratory, Agricultural Research Service, U.S. De-partment of Agriculture, Research report; no. 137.

Torstensson, B.A., 1984. A New System for Ground Water Monitoring. Groundwater Monitoring & Remediation 4, 131-138. doi:10.1111/j.1745-6592.1984.tb00904.x

Vallino, J.J., Algar, C.K., 2016. The Thermodynamics of Marine Biogeochemical Cy-cles: Lotka Revisited. Annual Review of Marine Science 8, 333-356. doi:10.1146/annurev-marine-010814-015843

VISS, 2019. Vatteninformationssystem Sverige. https://viss.lansstyrelsen.se/Wa-ters.aspx?waterMSCD=WA68517355&managementCycleName=Cykel_2. (Ac-cessed 2019-02-20).

van Genuchten, M.Th., Wagenet, R.J., 1989. Two-site/two-region models for pesti-cide transport and degradation: theoretical development and analytical solutions. Soil Science Society of America Journal 53(5), 1303-1310.

Volokita, M., Belkin, S., Abeliovich, A., Soares, M.I.M., 1996. Biological denitrifi-cation of drinking water using newspaper. Water Research 30, 965-971. doi:10.1016/0043-1354(95)00242-1

VVT, 2015. Nitrogen compounds at mines and quarries: Sources, behavior and re-moval from mine and quarry water - Literature study. VVT Technical Research Centre of Finland Ltd, VVT technology 226, ISSN 2242-1211.

65

Warneke, S., Schipper, L.A., Bruesewitz, D.A., McDonald, I., Cameron, S., 2011b. Rates, controls and potential adverse effects of nitrate removal in a denitrification bed. Ecological Engineering 37, 511-522. doi:10.1016/j.ecoleng.2010.12.006

Warneke, S., Schipper, L.A., Matiasek, M.G., Scow, K.M., Cameron, S., Bruesewitz, D.A., McDonald, I.R., 2011a. Nitrate removal, communities of denitrifiers and adverse effects in different carbon substrates for use in denitrification beds. Water Research 45, 5463-5475. doi:10.1016/j.watres.2011.08.007

Zumft, W.G. 1997. Cell Biology and Molecular Basis of Denitrification. Microbiol-ogy and Molecular Biology Reviews, 61, 533-616.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1790

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-380386

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2019