Adsorption of dissolved organic matter in aquatic ecosystems: Effects on composition and reactivity

(1)

UNIVERSITATISACTA

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1984

Adsorption of dissolved organic matter in aquatic ecosystems

Effects on composition and reactivity

MARLOES M. GROENEVELD

(2)

Dissertation presented at Uppsala University to be publicly examined in Zootissalen, Villavägen 9, Uppsala, Wednesday, 16 December 2020 at 13:15 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Associate Professor Suzanne Tank (University of Alberta, Department of Biological Sciences ).

Abstract

Groeneveld, M. M. 2020. Adsorption of dissolved organic matter in aquatic ecosystems.

Effects on composition and reactivity. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1984. 48 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-1055-8.

Inland waters receive organic matter from terrestrial ecosystems and in situ production. In transit from land to the ocean, dissolved organic carbon (DOC) may be mineralised to inorganic forms (CO2 and CH4) by microbial degradation and photodegradation. It may also transition from dissolved into particulate phase, and be transferred to the sediment and buried. One way in which this can happen is by adsorption of dissolved organic matter (DOM) to mineral particles.

This process is rarely studied in inland waters, since suspended particles are often in short supply. However, there are scenarios under which high particle concentrations occur, and in those cases, adsorption may have a substantial effect on DOM composition and reactivity. The overall aim of this thesis was to investigate the potential for DOM adsorption to inorganic particles and the resulting effect on DOM composition, as well as its biological reactivity. Three studies within this thesis focus on different types of surfaces waters in the boreal landscape of Sweden, and one study focuses on coastal moorland streams in the United Kingdom. Adsorption experiments were conducted under controlled laboratory conditions using batch experiments.

DOM quality was studied based on bulk optical properties, and composition was examined by high resolution mass spectrometry. Adsorption experiments using a commercially available reference clay (containing substantial amounts of aluminium and iron oxides) as the adsorbent show a widespread potential for DOM in inland water to adsorb to mineral particles. The extent of DOM adsorption in the experiments was regulated by two factors: 1) DOM composition, since compounds with a terrestrial signature were selectively adsorbed, and 2) water chemistry, as adsorption was impaired by pH>7 and higher concentrations of base cations. These general patterns were observed across surface waters with contrasting DOC concentrations, DOM composition and water chemistry parameters, and across spatial and temporal scales. In contrast, adsorption to suspended sediment derived from a glacial stream resulted in the removal of

‘protein-like’ DOM that is produced in situ, rather than terrestrially derived DOM. Hence, the mineralogy of particles may determine which DOM fraction is adsorbed. Experiments examining microbial degradation indicated that the effect of adsorption on the bioavailability of the remaining DOM depends on which DOM fraction is removed by the different adsorbents.

This thesis shows that adsorption to mineral particles in aquatic ecosystems is a highly relevant biogeochemical process that has the potential to alter DOM composition and thereby affect its biological reactivity.

Keywords: dissolved organic matter, DOM, carbon, aquatic, adsorption, mineral particles Marloes M. Groeneveld, Department of Ecology and Genetics, Limnology, Norbyv 18 D, Uppsala University, SE-75236 Uppsala, Sweden.

ISBN 978-91-513-1055-8

urn:nbn:se:uu:diva-423428 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-423428)

(3)

To my family

“If there is magic on this planet, it is contained in water”

Loren Eiseley

(4)

(5)

List of Papers

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

I Groeneveld, M., Catalán, N., Attermeyer, K., Hawkes, J., Einarsdóttir, K., Kothawala, D., Bergquist, J., & Tranvik, L.J.

(2020). Selective Adsorption of Terrestrial Dissolved Organic Matter to Inorganic Surfaces Along a Boreal Inland Water Con- tinuum. Journal of Geophysical Research: Biogeosciences, 125(3), e2019JG005236.

II Snöälv, J.T.C., Groeneveld, M., Hawkes, J., Kothawala, D., Quine, T.A. & Tranvik, L.J. (2020). Flocculation boundaries in the landscape: transformation processes of dissolved organic matter in coastal moorland streams, UK. Manuscript.

III Groeneveld, M., Kothawala D. & Tranvik, L.J. (2020). Season- ally variable controls on the interactions between dissolved organic matter and mineral particles in a Swedish agricultural river.

Manuscript.

IV Groeneveld, M., Jakobsson, E., Hawkes, J., Tittel, J., Kothawala, D. & Tranvik, L.J. (2020). Effects of adsorption to glacial suspended sediment and phototransformations on the bioavailability of dissolved organic matter from a thawing palsa mire. Manu- script.

The reprint of Paper I was made with permission from the publisher.

(6)

Additional Papers

In addition to the papers included in this thesis, I have contributed to the following papers:

I Groeneveld, M., Tranvik, L., Natchimuthu, S., & Koehler, B.

(2016). Photochemical mineralisation in a boreal brown water lake: Considerable temporal variability and minor contribution to carbon dioxide production. Biogeosciences, 13(13), 3931–3943.

II Attermeyer, K., Catalán, N., Einarsdottir, K., Freixa, A., Groeneveld, M., Hawkes, J.A., Bergquist, J., & Tranvik, L.J.

(2018). Organic Carbon Processing During Transport Through Boreal Inland Waters: Particles as Important Sites. Journal of Geophysical Research: Biogeosciences, 123(8), 2412–2428.

III Attermeyer, K., Andersson, S., Catalán, N., Einarsdottir, K., Groeneveld, M., Székely, A.J., & Tranvik, L.J. (2019). Potential terrestrial influence on transparent exopolymer particle concentrations in boreal freshwaters. Limnology and Oceanography, 64(4): 2455-2466.

(7)

Abbreviations

A254

DOC DOM EEM ESI-MS H/C O/C PARAFAC PLS PCoA SPE SUVA

absorbance at 254 nm dissolved organic carbon dissolved organic matter excitation-emission matrix

electrospray ionisation mass spectrometry hydrogen to carbon ratio

oxygen to carbon ratio parallel factor analysis partial least squares

principal coordinate analysis solid phase extraction specific UV absorbance

(9)

Introduction

Everything you see exists together in a delicate balance. … [We] need to understand that balance and respect all the crea-

tures, from the crawling ant to the leaping antelope. … When we die, our bodies become the grass. And the antelope eat the grass. And so, we are all connected in the great circle of life.

‘The Lion King’, 1994

The origin of life and the onset of organic matter cycling

All living things consist of organic compounds: molecules made from atoms that are continuously recycled. This exchange of material happens during the life time of an organism, and between the end of one organism and the start of another. Since the buildings blocks of life are present in limited amount, all elements have been recycled over and over again, ever since life first came into existence. As such, the atoms that now make up our body, may once have been part of an antelope, a blade of grass, a single-celled alga, a bacterium, etc…

The main elements of life are carbon, hydrogen, oxygen, nitrogen, phos- phorus and sulphur. Since carbon easily forms bonds with other carbon atoms, it can be seen as the ‘elemental backbone’ from which more complex organic molecules are made. Natural organic matter is a pool of carbon-based compounds that are essentially the waste products of past life, and the building blocks for new life. Degradation encompasses the chemical, physical processes and biological processes by which dead biomass becomes available for uptake again. The community of decomposers that is responsible for the biological degradation, consists of animals, such as worms, but mostly smaller organisms such as fungi and bacteria, all of which gain energy and biomass from the process.

Life originated in water, which has always remained an important site of the cycling of organic matter. This has been illustrated in Darwin’s notion of the first molecules forming ‘in some warm little pond’, through the ideas of Oparin and Haldane of a pre-biotic soup and the Miller-Urey experiment where ammonia, methane and hydrogen were combined into amino acids under the influence of heat and electricity, to more recent research that point to hydrothermal vents at the boundaries of tectonic plates in the deep sea and

(10)

inland geothermal springs as the possible birth places of life (Martin et al.

2008, Mulkidjanian et al. 2012, McCollom 2013). Water is also important for organic matter decomposition. Although organic matter is degraded in both terrestrial and aquatic ecosystems, molecules become available for microbial uptake when they are in solution. Also in soils, degradation depends on the occurrence of water, e.g. in pores (Marschner and Kalbitz 2003).

At the geological timescale, even rocks pas through a cycle. Seemingly permanent entities like mountains are slowly converted into smaller particles by landslides and erosion, transported in rivers and by wind to the ocean where they are buried in the sediments as sand, silt and clay, and returned to the Earth’s crust from where they at some point may surface again as new mountains. We will return to these mineral particles and how they may interact with organic matter in aquatic ecosystems later on.

Organic matter in inland waters

Since carbon is the most abundant element in organic matter, organic carbon and organic matter are often used in an almost synonymous way. While we study organic matter transformation processes, we focus on carbon as the main component that we can measure and therefore ‘track’ throughout the environ- ment (Prairie and Cole, 2009). In addition, the carbon cycle has in recent decades become even more relevant through the link between increasing carbon dioxide (CO2) and methane (CH4) concentrations in the atmosphere and rising global temperatures. A main research objective is therefore to calculate the global carbon budget: in order to anticipate and mitigate global change, we need to know how much carbon moves around between the geosphere, bio- sphere and atmosphere, and we need to understand the processes underlying the carbon cycle (Ciais et al. 2013). This includes understanding of the natural carbon cycle as well as anthropogenic alterations or disruptions. The following section will give a brief overview of organic matter, or organic carbon, cycling in inland waters.

Surface waters such as lakes, rivers and wetlands contain only 0.3% of all the worlds freshwater (Shiklomanov 1993), but they are important sites for organic matter processing. Organic matter can be produced in aquatic ecosystems through primary production (autochthonous matter), but it can also be exported from terrestrial systems (allochthonous matter). The pool of dissolved organic matter consists of many thousands of individual compounds that differ in their molecular composition and reactivity biogeochemical processes. Along the aquatic continuum, that ranges from streams and lakes to rivers that eventually flow into the oceans, organic matter is not only transported but also transformed (Cole et al. 2007), which is summarised in Figure 1. These transformations occur in degradation processes that may alter the structure of the organic matter or mineralise it to an inorganic form. Mineral- isation of organic matter is an important contributor to CO2 and CH4 in many aquatic systems, which leads to outgassing to the atmosphere in supersaturated

(11)

systems (Cole et al., 1994). Mineralisation can occur through microbial respiration of OM (Duarte and Prairie, 2005) and through photochemical degradation of the coloured, or chromophoric, dissolved organic matter (CDOM; Gra- néli et al., 1996). Another pathway by which organic matter is removed from solution is sedimentation. This can happen when DOM flocculates into particulate form or adsorbs to mineral surfaces. While substantial efforts have been made to study microbial (Amon and Benner 1996, Fasching et al. 2014, Catalán et al. 2016) and photochemical degradation (Koehler et al. 2014, Cory et al. 2014), and to some extent also DOM losses by flocculation (Droppo and Ongley 1997, Von Wachenfeldt and Tranvik 2008), the role of DOM adsorption to mineral particles for the losses along the continuum has received limited attention. Although adsorption of DOM has been suggested as another potential removal process for DOM (Seidel et al. 2015), studies that system- atically investigate the potential and qualitative effects of adsorption on the DOM pool across a variety of inland waters are currently lacking.

Figure 1. Organic matter processing and transport in inland waters.

Organic matter adsorption in soils

Within the field of soil science, adsorption is known as an important process that can stabilise organic matter to a solid surface. Mineral-associated organic matter potentially postpones or prevents microbial degradation by protecting it from enzymatic decay (Kaiser and Guggenberger 2000, Kalbitz et al. 2005, Hunter et al. 2016). How much organic matter and which compounds adsorb depends on the nature of the sorbent (i.e. the inorganic particles), the sorbate (i.e. the organic compounds), and water chemistry characteristics such as pH and ionic strength (Sollins et al. 1996). Specific characteristics of the sorbent

(12)

make it more or less prone to attract organic matter compounds. For example, aluminium and iron oxides and hydroxides often play a role in adsorption, since their hydroxyl ligand group can be traded for a negatively charged organic matter molecule in a process called ligand exchange (Tipping 1981a, McKnight et al. 1992, Kaiser et al. 1996 Yoon et al. 2004). As for the organic matter, aromatic compounds with many functional groups, such as carboxylic and hydroxylic groups that allow for ligand exchange with metals on mineral surfaces, are likely to adsorb to mineral surfaces (Gu et al. 1994, Kaiser et al.

1996, Kalbitz et al. 2005, Kleber et al. 2015). Conversely, aliphatic compounds with few of these functional groups adsorb less strongly to the mineral soil and thus turn over quicker (Kalbitz et al. 2005). Additionally, cation bridges can form between negatively charged clay and OM compounds (Theng 1976, Ahmed et al. 2002) and high ionic strength may neutralise the negative charges and thereby decrease repulsive forces between and within organic molecules, which allows for more compounds to be ‘packed’ on the same surface (Shen 1999).

Organic matter adsorption in aquatic ecosystems

The role of adsorption is well-studied in estuaries (Hedges and Keil 1999, Uher et al. 2001, Tremblay and Gagné 2009), and in marine sediments (e.g.

Keil et al. 1994, Mayer 1994a) where the OM to inorganic surface ratio is low, and the available surface area controls the amount of adsorption (Mayer 1994b). In inland waters, the ratio of organic matter to available surfaces is generally high. Although concentrations of mineral particles are often low in surface waters, this is not always the case. Erosion of various origins, whether from natural causes or human disturbances, may lead to increased load of suspended solids (Walling and Fang 2003, Syvitski et al. 2005). Adsorption to suspended particles may intercept organic matter and prevent it from being transported to the ocean or from being transformed by other processes. In this way, adsorption may affect carbon cycling. The importance and effect of adsorption on DOM composition and reactivity has thus far not been studied in much detail across freshwater ecosystems. Earlier studies have often considered specific fractions of the DOM pool (e.g. Tipping 1981b, Day et al. 1994, Gu et al. 1995), and when the total DOM pool was studied (e.g. Meier et al.

1999, Luider et al. 2003) the spatial and temporal extent of the study was limited to a single sample taken for one or at most a few sites.

(13)

Aims of the Thesis

Adsorption plays an important role in the biogeochemistry of carbon, but there is limited understanding of this process in inland waters. In these systems, the focus has been on photochemical and microbial transformation and degradation processes of DOM. However, when high suspended particle concentrations occur in the water column and the adsorption capacity of mineral surfaces in the water column is high, the process of adsorption would effectively bury organic matter to the lake sediments where the rates of microbial DOM degradation tend to be very low. Since the process of adsorption tends to be selective towards specific chemical groups, with some compounds preferentially being adsorbed, the remaining dissolved pool of OM could be altered accordingly. This change in DOM composition may subsequently affect its reactivity to other biogeochemical processes such as photodegradation and microbial degradation. The aim of this thesis was therefore to investigate the effects of adsorption to mineral surfaces on DOM composition and reactivity in aquatic ecosystems.

The thesis addresses the following questions:

• Is there a widespread potential for DOM adsorption in aquatic ecosystems? (Paper I-III)

• What is the effect of adsorption on DOM composition? (Paper I-IV) Do different adsorbents remove different DOM fractions?

• Which factors regulate DOM adsorption potential? Do these factors differ on a spatial scale compared to a temporal scale? (Paper I and III)

• Do changes in DOM composition as a result of adsorption affect its reactivity to other processes? (Paper III and IV)

(14)

Methods

Materials

Water samples

Paper I includes a set of 30 surface water samples (three peats, seven streams and rivers and twenty lakes) from across Sweden (58-63° N, 12-17° E), which cover a wide range of DOC concentrations (3-36 mgL^-1) and water residence times (hours to decades), as well as diversity in DOM composition and water chemistry parameters. Paper II focuses on a coastal moorland in the UK (51°

N, 3° W) and includes eight second and third order streams characterised by low DOC concentrations (~1 mgL^-1) and one headwater pool (9 mgL^-1 DOC).

In Paper III, river Fyrisån (60° N, 17° E), which was also included in Paper I, is sampled on a monthly basis during the course of a year. During this time, DOC concentrations varied between 10 and 26 mgL^-1. For Paper IV water was collected from the palsa mire Storflaket (16 mgL^-1 DOC) in a discontinuous permafrost zone (68°N, 19°E). A schematic overview of the four papers included in the thesis, their spatial and temporal scale, and which adsorbents were used, is given in Figure 2.

Figure 2. Schematic overview of the spatial (horizontal axis) and temporal (vertical axis) scale of the individual papers of this thesis. The colours of the cells indicate which adsorbents and/or flocculant were used.

(15)

Adsorbents

In Paper I, II and III, a commercially available reference clay, IPT.32 Plastic Clay (Bureau of Analysed Standards LTD, Middlesbrough, UK), was used as the adsorbent. The main constituents of the clay are SiO2, Al2O3, Fe2O3 and TiO2. The clay originates from an alluvium consisting of fluvial, marine and aeolian deposits, and is composed of approximately 80% kaolinite, 10%

quartz, 5% illite, 5% feldspar and a trace of smectite (Hosterman et al. 1987).

The clay was selected because of its affinity for DOM compared to two other mineral standards (granite and kaolinite), which was determined in a pilot experiment. Although the reference clay is not a material that naturally occurs in any of the studied sites, it has the advantage of facilitating highly standard- ised experiments and comparisons between different sites as well as between different studies. Paper II compared the effect of the reference clay with that of an artificial seawater standard, Tropic Marin^® Pro-Reef. This standard is technically not an adsorbent but a flocculant, since DOM does not adsorb to it but flocculates under the influence of the charged ions. The major ions in this standard are Cl^-, Na⁺, Mg²⁺, SO42-, Ca²⁺ and K⁺ (Atkinson & Bingman, 2008). Paper IV used suspended sediment collected from a glacial stream (68°N, 18°E) as the adsorbent. The suspended particles were pre-concentrated in the field by repeatedly allowing the particles to settle in canisters, siphoning off the top half of the water and refilling the canisters with additional stream water. The suspended sediment was afterwards converted into a powder by centrifugation and freeze-drying.

Experimental setup

At the heart of all papers included in this thesis is a set of analyses focused on DOM composition (optical and mass spectrometry), and the central question of how DOM composition is affected by certain processes. The experimental setup therefore follows the same general pattern for all studies: the initial DOM composition is determined, the DOM is subjected to a specific process, the DOM composition is determined again and compared to the initial composition (Fig. 3). In the experiments, DOM is operationally defined as the organic matter fraction that passes through a filter with a pore size (usually 0.7 µm).

Briefly, batch adsorption experiments were performed by adding the adsorbent (or flocculant) to a sample of filtered water. The particles were kept in suspension on a home-built magnetic stirrer that allowed for up to twelve sample to be stirred simultaneously. The samples were stirred for at least 20 hours so that an equilibrium could be reached between the DOM in solution and the organic matter adsorbed to the particles. Afterwards, the samples were centri- fuged to remove the bulk of the particles, and re-filtered to make sure that no

(16)

particles remained in solution, and the composition of the remaining DOM could be analysed.

In Paper III and IV, microbial degradation was assessed in bio-incubation assays. After filtered water samples were pre-processed (e.g. part of the DOM was removed by adsorption), they were inoculated with a small volume (5 or 10%) of the initial sample, which was passed through a filter with pore size 1.2 µm (Paper III) or a 20 µm plankton net (Paper IV). The samples were then left at room temperature in the dark for a number of weeks (8 weeks in Paper III and 5 weeks to 3 months in Paper IV), after which the difference in DOC concentration was determined. This was compared to the DOC loss in the control, where the DOM had not been pre-processed. In Paper IV, inorganic nu- trients were added to ensure that microbial degradation was carbon-limited, not nutrient-limited, and the DOC measurements were complemented with measurements of oxygen consumption.

Figure 3. Illustration of the experimental setup, examining the change in DOM com- position in response to experimental treatments, including adsorption to inorganic particles, microbial degradation and photochemical degradation. The van Krevelen diagrams showing individual molecules as determined by mass spectrometry (above) and fluorescence spectra (below) illustrate data from lake Svarttjärn (one of the sites in Paper I), before (left) and after (right) adsorption to the reference clay.

(17)

In Paper IV, the effect of adsorption to particles on DOM composition was compared to that of photodegradation. Here, water samples were exposed to UV-light for 22 hours in autoclavable polypropylene bags. During irradiation, the samples were placed in a water bath at room temperature. Dark controls, where the bags were covered with aluminium foil were included next to the light-exposed samples. The amount of organic carbon that was photomineral- ised (0.8 mgL^-1) was similar to the amount adsorbed (1.1 mgL^-1).

DOM characterisation: methods and data analysis

DOM composition is commonly analysed by means of its bulk optical properties and via high resolution mass spectrometry. Correlations are often found between the results of the two methods, but it is good to keep in mind that we look at different DOM sub-pools with either of the methods. This happens as a consequence of selectivity during sample preparation and the principles of detection for each of the methods. For example, when samples are analysed using high pressure size exclusion chromatography coupled to a charged aerosol detector (CAD), a diode array detector and electrospray ionisation mass spectrometry (ESI-MS, Orbitrap), the CAD is assumed to capture the entire DOM pool. Figure 4a shows that, relative to the CAD signal, the UV signal and the MS signal are to some extent separated on the chromatogram. As such, the two methods target different, though partly overlapping, DOM fractions and thus provide complementary insights, while some DOM is not detected by either of the methods (Fig. 4b). The following sections briefly describe the general methods for optical properties and mass spectrometry as well as how the data were analysed.

Optical properties

Optical measurements are based on the attribute of certain DOM compounds to absorb and, in a subset of these cases, to emit light at certain wavelengths.

Absorbance spectra were determined between 250 and 600 nm. From these spectra, a range of variables that serve as approximations for certain DOM characteristics can be calculated: integrated absorbance between 250 and 600 nm as a measure of the total amount of chromophoric DOM, A420 as a proxy for water colour, SUVA254 (DOC-specific UV absorbance at 254 nm)) as a proxy for aromaticity (Weishaar et al. 2003), the 250/365 absorbance ratio (Peuravuori and Pihlaja 1997) and spectral slope (Helms et al. 2008) as prox- ies for non-aromatic and low molecular weight compounds.

Fluorescence excitation-emission matrices (EEMs, see Fig. 3 for examples) were collected with excitation wavelength between 250 and 445 nm (5 nm in-

(18)

Figure 4. a) Normalised signals for the charged aerosol detector (CAD), UV at 254 nm (diode array detector) and mass spectrometry (MS) for the Storflaket palsa mire sample (Paper IV). b) Conceptual representation of the different DOM sub-pools from the perspective of different analytical techniques. The figure shows that there is a fraction of the total DOM pool that is not detected by optical methods or mass spectrometry. The area of the circles is arbitrary and does not reflect actual relative size of the sub-pools.

crements) and emission wavelengths between 300 and 600 nm (4 nm incre- ments). Three commonly used indices were used as approximations for DOM character and origin: fluorescence index (Cory and McKnight 2005), freshness index (Parlanti et al. 2000) and humification index (Ohno et al. 2002). Parallel factor analysis (PARAFAC; Bro 1997, Murphy et al. 2013) was used to extract the underlying, independently varying components in the EEMs across a set of samples in Papers I, II and IV.

Analysis of the ‘topography’ of the EEM showed that the location of the maximum fluorescence intensities (i.e. the ‘peaks’) was different when look- ing at different samples representing a gradient of water residence times (Pa- per I) and changed in response to the adsorption treatments (Papers I-III).

Since these peak shifts themselves provide information about DOM composition, they were investigated in more detail in Papers I and III. A useful way of showing which parts of an EEM have been removed or added as a result of certain process is to subtract the EEM obtained after the process from the initial EEM. The EEM of the fluorescent DOM that has been added or removed is called the ∆EEM.

Mass spectrometry

DOM composition also was analysed by mass spectrometry (MS), which is a technique that generates ions from individual organic molecules. These ions are then separated by their mass-to-charge ratio (m/z) and the abundance of each of the m/z ratios, or compounds, is quantified. In Paper I, II and IV, DOM

(19)

samples were first concentrated and stripped from salts by solid-phase extraction (SPE), after which the samples were eluted with methanol. The sample were subsequently loaded by an autosampler to the Orbitrap mass spectrome- ter with the ESI source operated in negative mode. Once the m/z spectra were obtained, noise was removed and formulas assigned according to in-house routines (Hawkes et al. 2016).

The resulting data were visualised by plotting the H/C versus the O/C ratio for each identified compound in van Krevelen diagrams (for examples, see Fig. 3). The data were often simplified by reducing the list of the abundances of a few thousand compounds to the average H/C and O/C ratio for each sample, weighted by the relative abundance of each compound in the sample.

Compositional changes as a result of adsorption were analysed across the entire data set. Bray-Curtis dissimilarities were calculated between all the samples in the data set. This dissimilarity matrix was then used in a principal coordinate analysis (PCoA), which is a multidimensional scaling method that can be used for exploring and visualising similarities/differences between samples by projecting the multidimensional dataset in a low-dimensional space. In Paper I, II and IV, only the first two coordinates are considered.

Other variables pertaining to the samples were added to a PCoA plot as arrows representing the correlations of these variables with the principal coordinates in the plot. In Paper I and II, the first principal coordinate, which explains most of the variation, correlated with the treatment (i.e. samples before and after treatment were separated on the first axis). Compositional changes in the samples were therefore visualised by plotting the correlations of the individual compounds with the principal coordinate representing the treatment in van Krevelen diagrams. Here, the colour of the individual compounds indicates whether it was generally lost or preserved with the treatment across the dataset.

(20)

Results and Discussion

Is there a widespread potential for DOM adsorption in aquatic ecosystems?

The potential for DOM adsorption was assessed in Papers I-III using batch experiments with the commercially available reference clay as an absorbent at a fixed particle concentration. Paper I shows that at least 22 to 75% of the DOC across Swedish surface waters can be adsorbed to these clay particles.

In Paper II, we were unable to accurately determine the low DOC concentrations in the moorland streams. However, we observed changes in DOM composition, which indicates that there is also a potential for adsorption in these waters. In Paper III, where we followed the potential for adsorption in a river over time, the adsorption potential varied between 22 and 43%. The different papers demonstrate that a substantial fraction of the DOM can be adsorbed onto inorganic particles in a variety of inland waters, and that this potential exists across spatial and temporal scales. Since the particle concentration in the batch experiments was much higher than most naturally occurring concentrations, this suggests that the adsorption potential in the batch experiments greatly exceeds the actual adsorption taking place in inland waters. Hence, in addition to previously established processes that modify and degrade DOM (photochemical degradation, biodegradation and flocculation), these results illustrate that adsorption has the potential to be an important geochemical process in situations where high particle concentrations occur, for example as a result of mineral erosion though agriculture, mining or forestry practices.

What is the effect of adsorption on DOM composition?

The decrease in DOC concentrations as a result of adsorption was accompa- nied by a change in the DOM composition. That is to say, adsorption did not remove all compounds evenly from the DOM pool, but targeted specific molecules. In the experiments where the reference clay was used as an absorbent, we found that coloured, aromatic compounds were preferentially removed.

This manifested itself in changes in the bulk optical properties. Absorbance variables, such as A420 and SUVA254, and the humification index decreased as a result of selective adsorption, while the fluorescence index and freshness index increased. Fluorescence intensities decreased predominantly for the

(21)

Figure 5. A comparison of the distribution of individual mass to charge peaks derived from mass spectrometry and assigned a molecular formula are plotted here as a function of elemental ratios (H/C and O/C) for different samples and treatments. The colour of the individual peaks reflects the strength and direction of the spearman rank correlation coefficients, with red dots being selectively preserved during the treatment and blue dots being selectively lost during treatment.

high emission wavelengths, as shown by the PARAFAC components in Paper I and Paper II (an example of a ∆EEM for adsorption to the reference clay is given below in Fig. 7a). The molecular data shows a preferential loss of DOM compounds with a low H/C ratio and a high O/C ratio. This corresponds to compounds classified as unsaturated and aromatic, with oxygen-containing functional groups that are known to engage in adsorption through ligand exchange, such as carboxyl and hydroxyl groups. This pattern was observed for both Swedish surface waters in Paper I (Fig. 5a) and Exmoor streams in Paper

(22)

II (Fig. 5b). The observed changes in DOM composition after adsorption to the reference clay resemble changes observed after exposure to other iron- and aluminium-containing mineral surfaces (e.g. Banaitis et al. 2006, Chassé et al.

2015, Fleury et al. 2017, Galindo and Del Nero 2014, Lv et al. 2016).

Salinity-induced flocculation in the Exmoor stream samples in Paper II showed a tendency to remove DOM compounds with a high O/C ratio (>0.5) and intermediate H/C ratio (0.5<H/C<1.5) from solution (Fig. 5c). This again suggests that oxygen containing functional groups, such as carboxyl and hydroxyl groups, are involved. Saline mixing had a weaker effect on DOM composition than adsorption to the reference clay (fewer compounds showed a correlation with treatment), but the quantitative effect is difficult to compare since clay and salt concentrations were arbitrarily chosen. The effect of combined clay and saline mixing in the Exmoor stream samples (Fig. 5d) greatly resembled the effect of adsorption to clay alone (Fig. 5b).

In Paper IV, adsorption of DOM from a palsa mire to suspended sediment derived from a glacial stream in northern Sweden selectively removed ‘protein-like’ fluorescent DOM. This could be observed from the bulk optical properties (the ∆EEM for adsorption to the glacial suspended sediment is given below in Fig. 7b), but not with our mass spectrometry method.

Different adsorbents thus appear to have a different effect on DOM composition. Although none of the studies directly compared the effects of all adsorbents on DOM quality (i.e. by taking a water sample and exposing it to all different adsorbents), this result emerges from the different papers. That mineral surfaces themselves influence which fraction of the DOM pool of a given sample adsorbs is well established in the soil literature (Kaiser et al. 1996, Kothawala et al. 2009, Fleury et al. 2017) and to some extent also in aquatic settings (Meier et al. 1999). If different adsorbents remove different DOM fractions, then it is important to note that the results presented here are limited to the adsorbents used. The use of different clay minerals or suspended sediment eroded by a glacier in an area with a different geology may lead to adsorption of different DOM fractions.

Which factors regulate DOM adsorption potential?

The factors that regulate adsorption to the reference clay in Paper I and III were investigated by means of partial least squares regression with percentage DOC removed as the dependent variable. Both factors relating to DOM composition and environmental variables were considered in the analyses. In Pa- per I, this was done on a spatial scale (30 different sites) and in Paper III on a temporal scale (monthly samples from one site during the course of a year).

On both a temporal and a spatial scale, the determinants of adsorption potential fall into two categories: DOM composition and ionic composition. Both studies identify quality indicators of coloured, allochthonous DOM, such as

(23)

SUVA254, A420 and HIX, to be positively related to percentage DOC adsorbed.

This is because it is the DOM with a more terrestrial character that gets adsorbed to the clay particles. Indicators of lower molecular weight, microbial origin and autochthonous production, such as the 250/365 absorbance ratio, fluorescence index and freshness index, were negatively related to percentage DOC adsorbed.

For the ionic composition, pH and concentrations of cations, such as calcium and magnesium, were found to impair DOM adsorption. The pH depend- ence of DOM adsorption onto clay surfaces has been well established in the literature (e.g. Tipping 1981a, Davis 1982, Gu et al. 1994). Protonation at low pH decrease negative electrostatic interactions between the DOM and the mineral particles which enables adsorption. At higher pH (>7 in Paper I), depro- tonation hinder adsorption. Although additions of cations are known in stim- ulate adsorption by neutralizing negative surfaces charges and forming cation bridges (Day et al. 1984, Preston and Riley 1982, Tipping 1981a), in Paper I and III, samples with naturally high divalent cation concentrations show a lower adsorption potential, which is somewhat of a paradox. In Paper I, we speculated that a fraction of the DOM that could have adsorbed to particles in the batch experiments had already reacted with cations in the natural systems and flocculated out of solution. In Paper III, we studied the river Fyrisån, which is characterised by relatively high base cation concentrations (particularly calcium ions) originating from glacial and sedimentary deposits in the catchment. Again, we identified the concentration of cations to have a negative relationship with adsorption potential. Furthermore, we identified a possible connection between the input of terrestrial DOM, the input of cations and discharge. At high discharge, especially in autumn, surface runoff and export from the upper soil horizons brings in substantial amounts of terrestrial DOM, resulting in a higher adsorption potential (Fig. 6a). At low discharge, this terrestrial supply of DOM is diminished, and we suspect that groundwater from the subsoils and substratum has a proportionally larger influence on water chemistry by supplying base cations, resulting in a lower adsorption potential.

We tested the predictive power of calcium concentration and SUVA420 (as a proxy for terrestrial DOM) in adsorption potential with a multiple linear regression model. The model explained 92% of the variation in adsorption potential in the river Fyrisån. The significant interaction term implies that at low concentrations of terrestrial DOM (i.e. at low SUVA420), calcium concentration has a strong negative effect on adsorption potential (Fig. 6b). With an increasing terrestrial influence (i.e. higher SUVA420), adsorption potential is less affected by calcium concentrations: the negative effect of calcium becomes saturated by the input of terrestrial DOM. However, it is important to note that using correlations to understand mechanisms, as we have done here, has limitations. In order to resolve the calcium paradox, further laboratory studies that e.g. experimentally manipulate calcium concentrations would be required. Also, the multiple linear regression model described here is based

(24)

on data obtained for one river, which has a relatively high conductivity due to the geology of the catchment. If similar relationships exist between adsorption potential, DOM quality and cation concentrations in other systems (e.g. with lower cation concentrations) is unclear.

Figure 6. a) Conceptual sketch of the hypothesised connections between discharge, input of terrestrial DOM and base cations. The left side of the river indicates low discharge conditions (in black), and the right side depicts high discharge conditions (in white). b) Visualisation of the interaction term in the multiple linear regression model (∆DOC = Ca * SUVA420). Different regression lines for the relationship between calcium concentration and adsorption potential are plotted for different levels of SUVA420. The data points on which the model is based are represented by the black dots.

(25)

Do changes in DOM composition as a result of adsorption affect its reactivity to other processes?

Reactivity of DOM to different biogeochemical processes depends on its composition. The fact that adsorption to mineral surfaces may change the composition of the DOM that remains in solution, raises the question whether this remaining DOM is more or less photolabile and/or bioavailable. In addition, our understanding of DOM processing would be improved if we knew whether the same DOM fractions are susceptible to different biogeochemical processes (i.e. do different processes ‘compete’ for the same DOM compounds).

The DOM fraction that is selectively adsorbed to the reference clay in Paper I-III resembles the change resulting from photodegradation, which also re- moves coloured (Moran et al. 2000), aromatic (Minor et al. 2007) and low H/C compounds from solution (Stubbins et al. 2010, Ward and Cory 2016).

This can be observed if we compare the ∆EEM that shows which parts of the EEM disappear as a result of adsorption to the reference clay (Fig. 7a) with the ∆EEM that shows which parts of the EEM are photodegraded (Fig. 7c).

While a substantial fraction of the fluorescent DOM thus appears to be susceptible to both adsorption and photodegradation, these two processes have different effects on the carbon cycle. Photodegradation of DOM may result in mineralisation to CO2 or in partially photo-oxidised compounds that are generally more available for microbial degradation (Granéli et al. 1996, Bertilsson and Tranvik 1998, Cory et al. 2014). Adsorption to mineral surfaces, on the other hand, moves DOM from solution to the sediment, where, unless it is mineralised there, it will be removed from the short-term carbon cycle (Kor- telainen et al. 2004, Tranvik et al. 2009).

In Paper III and IV, we addressed the question of how DOM adsorption to mineral surfaces affects the bioavailability of the remaining DOM. In Paper III, adsorption to the reference clay selectively removed DOM in a sample from the river Fyrisån that was highly coloured and of aromatic, ‘humic-like’

character, which is generally not seen as readily bio-available (Kalbitz et al.

2003, Fellman et al. 2008, Hansen et al. 2016). During the 53-day bio-incubation, 29% of the DOC that remained after adsorption was removed, compared to a 16% DOC loss in the control (Fig. 8a). Through the removal of this ‘re- calcitrant’ material, the remaining DOM appears to have become more available, or accessible, to the microbial community. In Paper IV, adsorption of DOM from the palsa mire Storflaket to glacial suspended sediment decreased the amount ‘protein-like’ fluorescent DOM. During the 35-day bio-incubation, 14% of the DOC that remained after adsorption was removed, compared to a 16% DOC loss in the control (Fig. 8b). In this case, adsorption slightly decreased bioavailability of the remaining DOM. Results from the fluorescence analysis show that some labile compounds in the lower left part of an EEM (around Ex270/Em310) which is associated with ‘protein-like’ DOM, appear to be preferentially adsorbed (Fig. 7b). The ∆EEM for adsorption to

(26)

glacial suspended sediment (Fig. 7b) and the ∆EEM for the control after the bio-incubation (Fig. 7d) both show a loss of fluorescence in the ‘protein-like’

DOM. Taken together, the results from Paper III and IV suggest that selectively adsorption can both enhance and decrease the bioavailability of the remaining DOM, depending on which DOM fraction is removed.

Figure 7. ∆EEMs for the Storflaket palsa mire sample exposed to different processes.

∆EEMs show which part of the EEM was removed for each of the processes (i.e. the red parts were subject to the greatest removal), negative values in c) indicate pho- toproduction.

(27)

Figure 8. a) Effect of adsorption to the reference clay, visualised as the ∆EEM, on the bioavailability of the remaining DOM in the river Fyrisån. b) Effect of adsorption to glacial suspended sediment, visualised as the ∆EEM, on the bioavailability of the remaining DOM in the palsa mire surface water.

(28)

Conclusion and Perspectives

This thesis has investigated the effects of adsorption to mineral surfaces on DOM composition and reactivity in aquatic ecosystems. The main conclu- sions are the following:

• There is a widespread potential for DOM adsorption in aquatic ecosystems. The thesis focused on inland waters in a boreal landscape, that are often characterised by substantial inputs of terrestrial DOM.

The studied systems varied widely in DOC concentration, DOM composition and water chemistry parameters. Here, adsorption to a reference clay, as an example of mineral particles with a high affinity for DOM, removed 22-75% of the DOC. Considerable changes in DOM composition were also observed in stream samples from a coastal moorland in the United Kingdom, suggesting that there is widespread potential for adsorption of freshwater DOM in inland waters.

• Adsorption can substantially alter DOM composition by selectively removing a certain fraction, but the effect varies with the minerology and surface properties of the adsorbent. Adsorption to the reference clay removed DOM with a predominantly terrestrial signature, which manifested itself as a decrease in coloured, aromatic, oxygen-rich compounds. Compounds with oxygen containing functional groups appeared to be susceptible to salinity-induced flocculation. Adsorp- tion to suspended sediment derived from a glacial stream removed

‘protein-like’ fluorescent DOM from a palsa mire surface water.

• On both a spatial and a temporal scale, DOM composition and water chemistry regulate DOM adsorption to the reference clay. Since DOM of a more terrestrial origin is selectively adsorbed, a larger relative abundance of terrestrially-derived material leads to a larger change in the DOM composition. Water chemistry parameters such as pH and the concentration of cations also affect adsorption.

• Since effect of adsorption on DOM composition varies with the adsorbent, DOM reactivity after adsorption depends on which DOM fraction is removed. Selective adsorption of ‘humic-like’ DOM resulted in an increase in the bio-availability of the remaining DOM, while removal of ‘protein-like’ DOM decreased the bio-availability of the remaining DOM.

(29)

In addition to the main results above, this thesis illustrates the potential but also some of the limitations of commonly used analytical methods for study- ing DOM composition. Since each method targets a specific DOM sub-pool, it is important to consider what can be seen through these different analytical windows when designing and interpreting the results of studies on DOM composition.

The quantitative role of adsorption and flocculation of DOC is a substantial gap in our understanding of the inland water carbon cycle. This thesis demon- strates that there is a widespread potential for adsorption of DOM to mineral surfaces. Adsorption to a reference clay allowed controlled comparison between sites of widely different DOM composition. However, in order to quan- tify the actual contribution of adsorption, subsequent studies should be designed to investigate adsorption of DOM from specific sites to the particles it naturally gets exposed to at the same location. The aim should be to develop a range of studies from highly controlled laboratory experiments to field studies that more closely resemble the natural conditions.

Future research should focus on identifying potential adsorption hotspots along the aquatic continuum which could strongly impact DOM cycling. Can- didate systems that may experience high inorganic suspended sediment loads include aquatic ecosystems affected by agriculture, forestry and mining practices, waterbodies that receive glacial meltwater, areas affected by permafrost thaw where e.g. river bank or coastal erosion and thermokarst slumps may supply particulate material, and deltas and estuaries were high turbidity in combination with a reduction in flow may allow for adsorption and settling of particles on the sediment.

Qualitative changes as a result of selective DOM adsorption have the potential to alter DOM reactivity to other processes, such as microbial degradation and photodegradation. The effect of adsorption on DOM bioavailability and photolability should be studied in more detail and in a wider context, fo- cusing on sites where adsorption is known or suspected to affect the local carbon budget, as suggested above. Changes in reactivity as a result of selective adsorption are also important from a carbon budget perspective. If photodegradation and adsorption target the same coloured, terrestrial molecules, know- ing which process takes precedence is vital for quantifying carbon fluxes to the sediment and atmosphere. Similarly, whether DOM becomes more or less bioavailable after selective adsorption of a certain fraction may affect miner- alization rates. In addition, more information is needed on whether the adsorbed organic matter is protected from degradation by enzymatic hydrolysis in aquatic ecosystems.

During the last decades, research on organic matter in the water column of inland waters has developed tremendously, not in the least through the ad- vancement of analytical techniques. The emphasis has been on the dissolved fraction of the organic matter, and on how this fraction is affected by microbial and photochemical degradation. The role of formation of and adsorption to particles, as well as the fate of both particle-bound organic matter and the organic matter that escapes association with particles, has received surprisingly limited attention. This thesis is a step to fill this knowledge gap. Especially

(30)

since human disturbances, either through land use or climate change, have the potential to increase particle loads into aquatic ecosystem, it becomes increas- ingly important to direct more research effort to this topic.

(31)

Summary in English

Organic matter, which is essentially all living organisms and their remains, needs to be recycled so that it can be used by other organisms. In addition to being vital for the production of new biomass, organic matter degradation also contributes to greenhouse gas emissions such as carbon dioxide (CO2) and methane (CH4), since carbon is the main constituent of organic matter. If we want to understand element cycles such as the carbon cycle, we need to understand the processes that move the elements between the different compart- ments of the cycle. Inland water, such as streams, rivers and lakes, play an important role in the global carbon cycle, in that they receive large amounts of terrestrial organic carbon from the soils and also have an internal production from algae and bacteria in the water. Inland waters transport organic matter towards the ocean, but a large fraction gets processed on the way. Dissolved organic carbon may be degraded to CO2 and CH4 under the influence of sunlight and as a result of bacterial respiration. CO2 in the water can then be out- gassed to the atmosphere. Organic carbon can also be converted to particles that sink to the sediment. Here, one part of the carbon will still be consumed by bacteria and transformed into CO2 and CH4. The remaining part, however, will be stored in the sediment and is thus removed from the short-term carbon cycle.

The change from dissolved organic matter to particulate organic matter can occur if the dissolved molecules stick together, forming particles that become too heavy to stay suspended in the water column. The dissolved organic matter molecules generally have a negative charge, and thus repel each other. How- ever, if there are positively charged ions present in the water, these ions can neutralise the negatively charged organic molecules and have them stick together to form aggregates. This is called flocculation. Sedimentation of organic matter can also occur when molecules stick to mineral particles that are suspended in the water column. These mineral particles, particularly fine- grained clay particles that are often very suitable for binding organic molecules, may have ended up in the water as a result of erosion of the surrounding land. The process of organic molecules sticking to inorganic surfaces is called adsorption. Once the organic molecules are adsorbed onto the mineral particles, they may be buried in the sediment, where they might even be less available to the microbes that degrade them, due to the strong attachment of the organic matter to the mineral particles.

Adsorption is a well-known mechanism for the stabilisation of organic matter in soils, where mineral particles are very abundant, and to some extent it has also been studied in marine and lake sediments. In inland waters, mineral

(32)

particles are often in short supply, and the process of adsorption has therefore been little researched. However, there are situations in which high suspended particle loads can occur, for example as a result of erosion. This can be caused by human activities, such as agriculture, forestry and mining, but it also occurs naturally. Glacial streams tend to have high particle concentrations because the glacier erodes the underlying bedrock, and rainstorms may cause erosion by washing away soil in catchments with steep slopes. Finally, erosion can also be a consequence of climate change, for example when warming causes permafrost thaw, which releases particles during riverbank and coastal erosion.

Consequently, adsorption of organic matter is an important ecosystem process, but there is limited knowledge of this process from inland waters. The aim of this thesis was therefore to investigate the potential for dissolved organic matter adsorption to mineral surfaces in inland waters. Dissolved organic matter is a complex mixture of different molecules, and processes that transform it generally don’t affect all the different compounds evenly, but are selective towards certain chemical characteristics of the individual molecules.

If certain compounds are thus removed from solution, we can see a change in the composition, or quality, of the remaining dissolved organic matter. This is important because other processes that may alter or remove organic matter from the water column, such as photochemical degradation and microbial degradation, also selectively target specific compounds. Therefore, the next aim of the thesis was to investigate what the effect of organic matter adsorption is on the composition of the remaining composition and how this change in composition changes the reactivity to other processes.

In Paper I, I tested how widespread the potential for adsorption is, using a set of water samples collected from 30 widely different lakes, streams, rivers and peat surface waters distributed throughout Sweden. A commercially available reference clay, that had shown high affinity for organic matter in a pilot experiment, was added to all of these samples at a fixed concentration. I meas- ured that between 22 and 75% of the dissolved organic carbon could be removed by adsorption to the particles. Analyses of the organic matter composition showed that compounds of terrestrial origin, rather than internally produced material, were selectively removed. These compounds were highly coloured, aromatic in nature and had oxygen containing functional groups through which they adsorbed to the clay particles. Analyses of dissolved organic matter quality parameters as well water chemistry parameters for all the sites revealed that both the quality and water chemistry parameters such as pH and cation concentrations played a role in regulating adsorption.

In Paper II, we looked at the potential for adsorption to the reference clay and salt-induced flocculation in surface waters of a coastal moorland in Ex- moor, the United Kingdom. Water samples were collected from one pond on the moor and eight streams running from the moor to the coast. Even though these waters differed from the Swedish waters in Paper I, similar compounds were adsorbed to the references clay. Salt-induced flocculation removed compounds with oxygen containing functional groups, regardless of aromaticity.

The combined clay and salt treatment greatly resembled the clay treatment.

(33)

In Paper III, I followed the river Fyrisån, which was one of the study sites in Paper I, over the course of a year to capture seasonal variability in adsorption. Water samples were collected once per month and exposed to the reference clay in the same manner as in Paper I and II. Many dissolved organic matter quality parameters followed a wave-like pattern throughout the sea- sons, with organic matter of terrestrial origin increasing in late autumn and winter and decreasing in late spring and summer. The adsorption potential followed this pattern, as more dissolved organic carbon was removed in autumn, winter and spring. Just like in Paper I, concentrations of cations such as calcium and magnesium were also important in regulating adsorption. In a simple biodegradability test on one of the samples, the remaining dissolved organic matter appeared to be more available to microbes after some of the terrestrial material had first been adsorbed.

In Paper IV, I designed an experiment with naturally occurring mineral particles. In the north of Sweden, permafrost thaw releases terrestrial organic matter that was previously locked away in frozen soils to surface waters where it may be processed. There is a concern that organic carbon released from permafrost soils worldwide contributes to CO2 and CH4 emissions to the atmosphere once it is degraded by sunlight and/or microbes. At the same time, streams draining melting glaciers contain high concentrations of suspended particles. I explored a scenario where the stream transporting the glacial suspended sediment mixes with the organic-rich waters in the lowlands, to investigate whether the glacial suspended sediment could adsorb the newly released organic matter in thaw areas. Additional questions were what the effect of adsorption is compared to that of photodegradation by sunlight, and how these two processes affect the bio-availability of the remaining dissolved organic matter. Unlike the reference clay, the particles derived from a glacial stream did not adsorb aromatic, oxygen-rich compounds, but ‘protein-like’ dissolved organic matter. After the adsorption to particles of the ‘protein like’ material, the bio-availability of the remaining dissolved organic matter decreased relative to the control with unmodified dissolved organic matter. This was likely because the ‘protein-like’ material is a good food source for microbes. UV- light, on the other hand, removed and altered terrestrial organic matter and thereby improved the remaining dissolved organic matter as a food source for microbes.

Taken together, my results demonstrate that adsorption has a high potential to selectively remove dissolved organic matter from inland waters, with con- sequences for reactivity towards microbial degradation. Moreover, I found that the extent and type of organic matter removal is a complex function of the composition of the adsorbent (e.g. mineral properties of suspended particles), the composition of the dissolved organic matter, and water chemistry parameters (pH and cation concentrations). Future studies should investigate envi- ronments where high particle concentrations occur, either naturally or because of human disturbances, and aim to determine the role of dissolved organic matter adsorption to mineral particles in the global carbon cycle.

(34)

Sammanfattning på Svenska

Organiskt material, vilket omfattar i princip alla levande organismer samt res- ter av dem, behöver brytas ned i mindre beståndsdelar innan det kan användas igen som energi och byggstenar för nya organismer. Utöver sin avgörande roll för produktion av ny biomassa bidrar nedbrytning av organiskt material även till växthusgasutsläpp, som koldioxid (CO2) och metan (CH4), då kol är den viktigaste beståndsdelen i organiskt material. För att förstå ett ämneskretslopp som kolcykeln måste vi därför först förstå de processer som transporterar grundämnen mellan de olika segmenten i cykeln. Inlandsvatten, så som bäckar, floder och sjöar, spelar här en viktig roll i den globala kolcykeln då de tar emot stora mängder organiskt kol från skog och mark (alloktont kol) och även har egen ny produktion av organiskt kol via alger och bakterier (autoktont kol). Inlandsvatten transporterar det organiska materialet mot ha- vet, men en stor andel omarbetas på vägen. Organiskt kol som är löst i vattnet kan brytas ned till CO2 och CH4 av solstrålning och som ett resultat av bakte- riell respiration. Inlandsvatten mättat med dessa växthusgaser avger dem se- dan till atmosfären. Det organiska kolet kan också omvandlas till partiklar som sjunker till bottensedimentet. Här kan en del av kolet fortfarande konsumeras av bakterier och omvandlas till CO2 och CH4, men återstående kol förblir lag- rat i sedimentet och avlägsnas därför från den kortsiktiga kolcykeln till en mer långsiktig kolsänka.

Omvandling från löst organiskt material till partikelformigt organiskt material kan ske om de lösta molekylerna klumpas ihop och bildar partiklar som är för tunga för att förbli suspenderade i vattnet. De lösta organiska molekylerna har i allmänhet en negativ laddning och stöter därför bort varandra, men om positivt laddade joner är närvarande i vattnet kan de neutralisera de negativt laddade organiska molekylerna och hålla ihop dem så att de bildar aggregat. Detta kallas flockulering. Sedimentering av organiskt material kan också uppstå när molekyler håller fast vid mineralpartiklar som är suspenderade i vattenpelaren. Dessa mineralpartiklar kan hamna i vattnet som följd av erosion av den omgivande marken. Speciellt gäller detta finkorniga lerpartik- lar som ofta är väl lämpade för att binda organiska molekyler. En sådan process, där organiska molekyler fastnar på icke-organiska ytor, kallas för adsorption. När de organiska molekylerna har adsorberats på mineralpartiklarna kan de begravas i sedimentet, och där bli än mindre tillgängliga för nedbry- tande mikrober på grund av det organiska materialets hårda bindning till partiklarna.

Adsorption är en välkänd mekanism för stabilisering av organiskt material i jord där det finns rikligt med mineralpartiklar, och har i viss utsträckning

(35)

även studerats i havs- och sjösediment. Inlandsvatten har dock generellt låga halter av dessa partiklar, och som en konsekvens är adsorptionsprocessen inte lika väl studerad här. Det finns ändå förhållanden då höga halter av suspenderade partiklar kan påträffas i inlandsvatten, som till exempel vid erosion.

Denna kan orsakas av mänskliga aktiviteter som jordbruk, skogsbruk och gruvdrift men sker också på naturlig väg; Bäckar som avvattnar glaciärer har ofta höga partikelkoncentrationer då glaciären eroderar underliggande berg- grund. Även kraftiga regn kan orsaka erosion genom att skölja bort mark i avrinningsområden med branta sluttningar. Erosion kan också ske till följd av klimatförändringar, som till exempel då global uppvärmning tinar permafrost, vilket frigör partiklar genom erosion av flodbankar och kuster.

Följaktligen är adsorption av organiskt material en viktig process i ekosy- stemen, men det finns begränsad kunskap om den i inlandsvatten. Målet med denna avhandling var därför att undersöka potentialen för adsorption av upp- löst organiskt material på mineralytor i inlandsvatten. Löst organiskt material som är en komplex blandning av olika molekyler, och processer som omvand- lar materialet påverkar vanligtvis inte de olika ingående föreningarna på samma sätt utan är i sin tur beroende av kemiska egenskaper hos de enskilda molekylerna. Om vissa föreningar avlägsnas från lösningen kan vi därför se en förändring i sammansättningen eller kvaliteten på det återstående lösta organiska materialet. Detta blir viktigt eftersom andra processer som kan för- ändra eller avlägsna organiskt material från den fria vattenmassan, som foto- kemisk nedbrytning och mikrobiell nedbrytning, också påverkar specifika för- eningar selektivt. Nästa syfte med avhandlingen var därför att undersöka vil- ken effekt adsorptionen av organiskt material har på sammansättningen av det lösta organiska material som inte adsorberas, och hur denna förändring i sam- mansättning förändrar hur reaktivt det organiska materialet är med avseende på andra processer.

I den första studien (studie I) testade jag hur spridd potentialen för adsorption är med hjälp av en uppsättning vattenprover insamlade från 30 högst olika sjöar, bäckar, floder samt ytvatten från torvmark, fördelade över olika delar av Sverige. En kommersiellt tillgänglig referenslera visade hög affinitet för organiskt material i ett pilotförsök, och tillsattes till alla prover i samma kon- centration. Jag fann att mellan 22 och 75% av det lösta organiska kolet kunde avlägsnas genom adsorption till partiklarna. Analys av det organiska materialets sammansättning visade vidare att organiskt material importerat från omgivande mark avlägsnades selektivt, medan material som producerats på plats av alger och växter behölls. Dessa föreningar var starkt färgade, aromatiska till sin natur och hade syreinnehållande funktionella grupper genom vilka de adsorberades till lerpartiklarna. Analys av det lösta organiska materialets kva- litet samt vattenkemi för alla provtagningsplatser visade att både kvalitets- och vattenkemiparametrarna, såsom pH och katjonkoncentrationer, påverkade adsorptionen.

I studie II undersökte vi potentialen för adsorption till referensleran samt saltinducerad flockulering i ytvatten. Undersökningen utfördes vid ett hedland på kusten av Exmoor, i Storbritannien, där vattenprover samlades in från en damm på heden samt åtta bäckar som ledde från heden till kusten. Även om

Adsorption of dissolved organic matter in aquatic ecosystems: Effects on composition and reactivity