Jukuri, open repository of the Natural Resources Institute Finland (Luke) 
   
 
   
All material supplied via Jukuri is protected by copyright and other intellectual property rights. Duplication 
or sale, in electronic or print form, of any part of the repository collections is prohibited. Making electronic 
or print copies of the material is permitted only for your own personal use or for educational purposes.  For 
other purposes, this article may be used in accordance with the publisher’s terms. There may be 
differences between this version and the publisher’s version. You are advised to cite the publisher’s 
version. 
 
This is an electronic reprint of the original article.  
This reprint may differ from the original in pagination and typographic detail. 
 
Author(s): Markku Yli-Halla, Timo Lötjönen, Jarkko Kekkonen, Seija Virtanen, Hannu Marttila, 
Maarit Liimatainen, Markus Saari, Jarmo Mikkola, Raija Suomela & Erkki Joki-Tokola 
Title: Thickness of peat influences the leaching of substances and greenhouse gas 
emissions from a cultivated organic soil 
Year: 2022 
Version: Published version 
Copyright:   The Author(s) 2022 
Rights: CC BY 4.0 
Rights url: http://creativecommons.org/licenses/by/4.0/ 
 
Please cite the original version: 
Yli-Halla M., Lötjönen T., Kekkonen J., Virtanen S., Marttila H., Liimatainen M., Saari M., Mikkola J., 
Suomela R., Joki-Tokola E. (2022). Thickness of peat influences the leaching of substances and 
greenhouse gas emissions from a cultivated organic soil. Science of The Total Environment 806, 
Part 1, 150499. https://doi.org/10.1016/j.scitotenv.2021.150499. 
Science of the Total Environment 806 (2022) 150499
Contents lists available at ScienceDirect
Science of the Total Environment
j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenvThickness of peat influences the leaching of substances and greenhouse
gas emissions from a cultivated organic soilMarkku Yli-Halla a,⁎, Timo Lötjönen b, Jarkko Kekkonen b, Seija Virtanen c, Hannu Marttila d,
Maarit Liimatainen b,d, Markus Saari d, Jarmo Mikkola e, Raija Suomela b,1, Erkki Joki-Tokola b
a Department of Agricultural Sciences, P.O. Box 56, FI-00014, University of Helsinki, Finland
b Natural Resources Institute, Paavo Havaksen tie 3, FI-90014, University of Oulu, Finland
c Drainage Foundation sr., Simonkatu 12 B 25, FI-00100 Helsinki, Finland
d Water, Energy and Environmental Engineering Research Unit, P.O. Box 8000, FI-90014, University of Oulu, Finland
e Natural Resources Institute, Latokartanonkaari 9, FI-00790 Helsinki, FinlandH I G H L I G H T S G R A P H I C A L A B S T R A C T• Loads of N, P, S, C and GHGs from an or-
ganic agricultural soil were monitored.
• Peat thickness increased N leaching.
• Dissolved P leaching increased upon in-
creasing peat thickness.
• The peat layer seemed to slowdown the
oxidation of sulfidic material in the sub-
soil.
• CO2 emissions were high and not
significantly dependent on peat
thickness.⁎ Corresponding author.
E-mail address:markku.yli-halla@helsinki.fi (M. Yli-H
1 Present address: Oulu University of Applied Sciences,
https://doi.org/10.1016/j.scitotenv.2021.150499
0048-9697/© 2021 The Authors. Published by Elsevier B.Va b s t r a c ta r t i c l e i n f oArticle history:
Received 31 May 2021
Received in revised form 16 September 2021
Accepted 17 September 2021
Available online 23 September 2021
Editor: Ashantha GoonetillekeThe off-site effects of agricultural organic soils include the leaching of N, P, and organic carbon (OC) to wa-
tercourses and CO2, CH4, and N2O emissions into the atmosphere. The aim of this study was to quantify how
the thickness of organic layers affects these loads. A 19.56-ha experimental field drained by subsurface
pipes was established in Ruukki, northwestern Finland. Three plots had a 60–80 cm-thick sedge peat
layer and three others had a thickness of 20 cm or less. The drainage pipes lie in mineral soil that, in this
field, contains sulfidic material. This study documents the experimental settings and reports on the leaching
of substances in the first two years, as well as CO2, CH4 and N2O emissions during eight weeks in one
summer. Total N (TN) and OC loads were higher from the thicker peat plots. The mean TN loads during a hy-
drological year were 15.4 and 9.2 kg ha−1 from the thicker and thinner peat plots, respectively, with organic
N representing 36% of TN load. Total P (TP) load averaged 0.27 kg ha−1 yr−1. Dissolved P load represented
63 and 36% of TP in the thicker peat area and only 23 and 13% in the thinner peat area, and was thus in-
creased upon peat thickness. These N and P loads through the subsurface drainage system represented
roughly 83% of TN and 64% of TP loads from this field. There were no clear differences in greenhouse gas
emissions among the plots during the eight-week monitoring period. Slowly oxidizing sulfide in the subsoilKeywords:
Acid sulfate soils
AnaEE network
Nitrogen
Phosphorus
Organic carbon
Water qualityalla).
Yliopistokatu 9, P.O. Box 222, FI-90101 Oulu, Finland.
. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499resulted in annual leaching of 147 kg S ha−1, almost ten times that of non-sulfidic soils. Our first results em-
phasize the effect of the peat thickness on the leaching of substances andwarn about considering all organic
soils as a single group in environmental assessments.
© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://
creativecommons.org/licenses/by/4.0/).1. Introduction
Globally, the organic soils of cool and humid climates form a major
carbon (C) stock (Batjes, 1996; Scharlemann et al., 2014) and com-
monly serve as farmland (Berglund and Berglund, 2010; Buschmann
et al., 2020). Many studies have highlighted the leaching of major nutri-
ents and high greenhouse gas (GHG) emissions from cultivated organic
soils (Duxbury and Beverly, 1978; Kasimir-Klemedtsson et al., 1997;
Kløve et al., 2010; Tubiello et al., 2016; Regina et al., 2019). According
to soil tests carried out in Finland between 1998 and 2002, 13.6%, or
300,000 ha, of agricultural land in the country consisted of organic
soils (Myllys and Sinkkonen, 2004), and in the northern half of the
country, almost 30% of agricultural land represented organic soils.
They are mostly used for spring cereal and grass cultivation to provide
fodder for dairy production (Kekkonen et al., 2019). The farming of or-
ganic soils is vital for local rural societies and food production, especially
in north-western Finland. Therefore, there is a need for focused studies
to accommodate the environmental challenges related to cultivation
and to promote sustainable agricultural production.
According to the Intergovernmental Panel on Climate Change (IPCC,
2014), “organic soil” has at least a 10-cm thick horizon consisting of ma-
terial that, whenmixed to the depth of 20 cm, has at least 12–18% organic
carbon (i.e., at least 20–30% soil organic matter (SOM), depending on the
clay content). This definition matches well with the traditional Finnish
criteria, where organic soils are principally divided into two categories
(i.e., 1) peat soils that have ≥40% organic matter and 2) other organic
soils that contain 20–39.9% of organic matter (“mull soils”) in the plough
layer, often with a mineral subsoil). These criteria are applied to the
plough layers of agricultural land, with no other depth requirements.
Many of the soils in themull category have been formed either as a result
of the ploughing of shallow peat soil, mixing the topsoil peat with the
mineral subsoil, or by a decrease of the organicmatter content of the orig-
inal peat soil due to long-term cultivation. These definitions differ from
the criteria of the Histosols of Soil Taxonomy (Soil Survey Staff, 2014) or
the World Reference Base for Soil Resources (IUSS Working GroupWRB,
2014), where, in an agricultural context, at least a 40-cm thick horizon
consisting of organicmaterial is required.Many Finnishmull soils thus be-
long to the mineral soil classes of Soil Taxonomy or WRB, with a shallow
histic horizon.
Organic soils are usually located in flat parts of the landscape where
water-logged conditions have favored the growth of mosses or sedges
and the formation of peat. Wetness prevents organic matter oxidation
and stores it as peat (e.g., Maljanen et al., 2007).When reclaimed for ag-
riculture, these soils need to be artificially drained, which leads to the
aerobic decomposition of organic matter coupled with N mineraliza-
tion. Before being productive, these fields usually require liming,
which promotesmicrobial activity and further enhances decomposition
(Ivarson, 1977). Tillage also enhances decomposition via an increased
aeration of the soil. Thus, agricultural use has various effects through
which organic matter is lost from organic soils. Nitrogen is first released
as ammonia upon mineralization, and part of it is consumed by plants
and other organisms, while other parts (especially nitrate) can be
leached to the drainage system and further to surface waters. Transport
of C and N from the soil also takes place in the formof humic substances
that are dissolved or suspended in drainage water (Mattsson et al.,
2005; Manninen et al., 2018). As for emissions into the atmosphere,
fluxes of carbon dioxide (CO2) and nitrous oxide (N2O) increase upon
agricultural use of organic soils whereas methane (CH4) emissions
commonly decrease (Maljanen et al., 2007).2On the flat western coast of Finland, the land is still emerging from
the sea as a consequence of post-glacial rebound following the
Weichselian glaciation that ended about 10,000 years ago; the annual
rebound is about 2 mm in the south and 8 mm in the northern part of
the coast (Tikkanen andOksanen, 2002). Thus, the time for the accumu-
lation of peat has been rather short on these coastal areas. Soil surveys
conducted about 50 years ago (Soini and Virri, 1968; Urvas, 1976) doc-
umented that sedge peat was themost common soil type of agricultural
land inmany areas of Ostrobothnia on thewestern coast of Finland, and
the horizons that consist of organic material were thin. Even though the
SOM content of the plough layer may be as high as 60% (organic carbon
35%) (Soini and Virri, 1968), the total SOM stock can be rather small,
owing to the shallow depth of the organic material. In these soils, the
subsurface drainage pipes, usually installed at depths of 1.2m, lie almost
exclusively in the mineral soil layers. However, the only experimental
fieldwhere the leaching of nutrients from anorganic soil has beenmon-
itored in Finland represents a soil that has a deep (3–4 m) peat layer
(Huhta and Jaakkola, 1993), and it is questionable whether or not
those results can be extrapolated to thin organic soils. In addition, the
thickness of organic horizons has not been explicitly considered in stud-
ies where GHG emissions have been measured.
Organic matter as such possesses a low capacity to retain dissolved
phosphorus (P); therefore, cultivated organic soils can be sources of P
loading to waters (Longabucco and Rafferty, 1989; Huhta and Jaakkola,
1993). However, organic soil material commonly contains plenty ofmin-
eral material and associated iron (Fe) and aluminum (Al) oxohydroxides
which can effectively retain P. The P retention capacity of peat samples is
also strongly dependent on the concentration of Fe and Al (Kaila, 1959;
Nieminen and Jarva, 1996; Ronkanen et al., 2016). Phosphorus leaching
from organic soils likely depends on whether subsurface pipes have
been installed in the mineral or organic subsoil. Therefore, the thickness
of the organic horizons probably has an impact on P leaching as it influ-
ences water flow paths.
Besides peat soils, acid sulfate (AS) soils are also prevalent on the
western coast of Finland (Purokoski, 1959; Yli-Halla et al., 1999; GTK,
2021). Organic soils in that area commonly have fine- or medium-
textured AS subsoils that contain sulfidic material. The subsoils have
been partly oxidized upon artificial drainage, resulting in acidification
of the soil and drainage water (Edén et al., 1999; Beucher et al., 2015).
Studies carried out in peat mining areas have shown that a peat layer
thicker than 0.5 m may inhibit the oxidation of sulfidic materials of
the deeper horizons (Nystrand et al., 2019); however, the effect of
peat thickness on acid loadings has not been studied in cultivated sub-
surface drained peat fields. In these kinds of fields, in addition to the de-
composition of peat, a substantial load of N can be produced by the
oxidation of the large mineral N stock in the subsoil (Yli-Halla et al.,
2020). Therefore, the N load from organic soils that contain sulfidic ma-
terial in the subsoil can originate from themineralization of SOM in the
upper soil horizons and the AS horizons deeper in the soil.
Accurate information concerning nutrient loads and GHG emissions
from shallow agricultural peat soils is needed when developing policies
for the sustainable use of organic soils. The primary aim of this study
was to quantify the loads of N, P, acidity and total organic carbon
(TOC) into the watercourses and GHG emissions into the atmosphere
from a shallow agricultural sedge peat soil with an AS subsoil, a set of
characteristics typical of the western coast of Finland. In this study, we
document the characteristics of a newly established experimental field
at the Ruukki field station and report on hydrology and the leaching of
substances during the first two hydrological years after establishing
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499the field. GHG emissions were measured in one summer. We hypothe-
sized that the thickness of the peat layer affects leaching and atmospheric
emissions from the peat fields 1) in terms of P leaching— shallow peat
soils differ from thick peat soils and behave more like mineral soils,
retaining P strongly onto the abundant Fe and Al oxide surfaces; 2) the
thickness of the peat layer affects N and TOC loads to discharge water;
3) GHG emissions into the atmosphere are increased upon increasing
peat thickness; and 4) the acidity released from the subsoil is related to
the thickness of the peat layer.
2. Materials and methods
2.1. Experimental field and weather observations
The experimental field is located at the Ruukki research station
(25.00°E, 64.42°N) in the Siikajoki municipality, about 60 km southwest
of Oulu, Finland (Fig. 1A). Its elevation is circa 45m above themean sea
level, and the distance from the current seashore is 26 km. The station
belongs to the Natural Resources Institute Finland (Luke). The current
experimental field was established in 2016 to allow monitoring of
GHG emissions and the leaching of substances. The surface soil is
sedge peat, mixed with different amounts of coarse silt, and the subsoil
consists of actual and potential AS soil material. The present experimen-
tal area was already entirely in agricultural use about 100 years ago, as
proven by a soil map drawn in 1933. The experimental field has an
area of 19.56 ha, divided into six plots, 2.97–3.77 ha each (Fig. 1B),
and it has a slope of less than 0.5%. The field belongs to the European
network of platforms for Analysis and Experimentation on Ecosystems
(AnaEE, 2021).
The average annual temperature at Ruukki between 1981 and 2010
was 2.6 °C, with January being the coldest (−9.3 °C) and July the
warmest (15.9 °C) month. The average precipitation during the same
30-year period was 541 mm. Snow cover in the area usually lasts from
late November until mid-April, about 130–140 days. The growing sea-
son (daily mean temperature > 5 °C) commonly starts in the first days
of May and ends at the beginning of October (Kersalo and Pirinen,
2009; FinnishMeteorological Institute, 2021). Air temperature and pre-
cipitation data for the present monitoring period were obtained from
the Finnish Meteorological Institute which has a weather station
800m from themidpoint of the experimentalfield. The depth of the fro-
zen ground was measured monthly using methylene blue tubes where
the color change indicates the depth of frost (Soveri and Varjo, 1977).
2.2. Soil chemical properties and spatial variation of peat thickness
Before setting up the experimental field, the area was surveyed for
the thickness of the peat layer, sampled at depths of 0–40 cm,
40–80 cm and 80–100 cm and analyzed for pH at a soil-to-water ratio
of 1:2.5 using a Radiometer GK 2401B electrode. The electrical conduc-
tivity (EC) of the supernatantwasmeasuredwith a ProMinentDulcotest
TyP LFT 1 FE electrode. Easily soluble calcium (Ca), magnesium (Mg),
potassium (K), P, and sulfur (S) were extracted with a mixture of
0.5 M ammonium acetate and 0.5 M acetic acid solution buffered at
pH 4.65 (Vuorinen and Mäkitie, 1955), which is the soil testing method
used in Finland. Phosphorus concentration was measured colorimetri-
cally with a Foss – Tecator FIAStar 5000 Flow Injection Analyser, while
Ca, Mg, K, and S were measured with an ICP-OES Thermo Fisher Scien-
tific iCAP6500 Duo Inductively Coupled Plasma Optical Emission Spec-
trometer.
In Plot 3 and a fewmeters from the edge of Plot 6, drillings extending
to the depth of 3m (drilling sites P1 and P2 in Fig. 1B)weremade by the
Geological Survey of Finland during a survey of acid sulfate soils (GTK,
2021). Mineral subsoil samples were analyzed at intervals of 20 or
40 cm for total S after aqua regia digestion. Subsoil samples were tested
for sulfidic material measuring the fresh pH(H2O) right after sampling
and after eight weeks of aerobic incubation using a Hamilton Flatrode3electrode and a minimal amount of water. To meet the definition of
sulfidic material, soil pH has to drop at least 0.5 pH units to a final pH
value of 4.0 or less during the incubation (Soil Survey Staff, 2014).
Eleven additional drillings (locations not shown, distributed evenly)
were made within and in the vicinity of the plots to the depth of 2 m,
to visually assess the textural class and the depth of sulfidic material.
A ground penetrating radar (GPR) survey to determine the depth of
the organic horizons was conducted in July 2020. The GPR lines with a
total length of 17 km form an even grid with a transect gap of 30 m.
The setup for measurement consisted of a Malå Geoscience ProEx unit
with a 250 MHz shielded antenna. The data were validated with 32
manual depth measurements. Data were processed and interpreted
using ReflexW software (Sandmeier, 2016).
The soil layers consisting of organic material (thickness 20–80 cm at
different plots) were sampled in the middle of every plot between the
laterals (subsurface drainage pipes that conduct water to the collector
pipes) at intervals of 10 cm. Below the organic material, the uppermost
10 cm of themineral subsoil and the deeper mineral soil layers at inter-
vals of 50 cmwere also sampled. These samples were analyzed for total
C and N using a Leco CHN628 or Leco TruMac CN dry combustion appa-
ratus. To calculate the C andN stocks, bulk densitywas determinedwith
a core method using 150 cm3-cylinders with individually determined
volumes. The soil of each plot was classified according to the WRB sys-
tem (IUSS Working Group WRB, 2014).
Additional soil samples were taken at 0–20, 20–50, 50–100,
100–120 (or 150) and 150–200 cm in the middle of every plot in
May 2020 before fertilization. The soil samples were stored frozen
until mineral N (Nmin) was extracted with 2 M KCl (Esala, 1995).
The concentrations of ammonium-N (NH4+-N) and nitrate-N (NO3−-
N) were determined colorimetrically with an autoanalyzer (Lachat
QuickChem AE, Hach, Loveland, CO, USA or Skalar San++ System,
Skalar Analytical B.V., Breda, The Netherlands).
2.3. Cultivation of the experimental field
The crops and agricultural practices are presented in Supplementary
Table S1. Grasses harvested for silage and cereals were grown in the
field. In 2016, when the experiment was established, the whole field
was under a mixture of timothy and meadow fescue grass. In spring
2017 basic slag was applied at a rate of 8 tons per ha to decrease soil
acidity, and the field was seeded with cereals. Plots 5 and 6 were
undersown with grass seeds simultaneously with the sowing of cereal.
The aboveground vegetation was harvested as whole-crop silage in
September. In summer 2018 spring barley for grain was grown in
Plots 1–4 and the grass was harvested as silage in Plots 5 and 6. The
fertilization was carried out with compound NPKS and NPS fertilizers.
The ley received surface application while the fertilizers for cereals
were drilled into the soil. All farming activities were carried out using
conventional farm machinery.
2.4. Land drainage of the experimental field
The experimental plots were originally drained by open ditches.
Subsurface drainage was subsequently installed, using wooden box
drains in 1900 (Plot 6), clay pipes in 1960 (Plot 5) and plastic pipes in
the 1980s (Plots 1 and 2). The subsurface drainage was renewed with
plastic pipes in the 1980s (Plots 1 and 2), in 2005 (Plots 5 and 6), and
in 2007 (Plots 3 and 4). For this experiment, the existing subsurface
drainage was largely used in Plots 3, 4, 5, and 6, to avoid unnecessary
disturbance of the soil. It is known that during the first few years after
the installation of subsurface drainage, elements can be mobilized be-
cause of soil disturbances during trenching (Äijö et al., 2016), which
may also promote the oxidation of sulfidic material and consequent
acidification in AS soils.
In Plots 1 and 2, new subsurface drainage had to be installed because
the old drainage did not function well. Owing to the gradual oxidation
Fig. 1. A) The location of the Ruukki experimental station. The shaded area showswhere acid sulfate (AS) soils can potentially exist in Scandinavia. B) The drainage and soil surveymap of
the experimental field. C) The thickness of the organic horizons. D) Schematic cross sectional drawing of the experimental field.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499
4
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499and compaction of the surface peat, the drainpipes were only at a depth
of 80 cm. The laterals (corrugated PVC pipes, Ø 65mm)were installed at
a depth of 1.1–1.3 m, with a drain spacing of 12 m. Gravel (0.2–4 mm)
was used about 6.5 m3 per 100m above the drainage pipes. The laterals
were connected to collectors (diameter Ø 160mm) that convey thewa-
ters to a control well (Ø 800mm, height 1800mm) in the corner of each
plot. The groundwater level of the plots can be controlled by adjusting
the height of the outlet pipe in the control well, but during themonitor-
ing period in this study, the control system was not in use. From each
control well, a separate outlet pipe conveys water to the sampling sta-
tion at the lower endof thefield. In this area, rust precipitates constantly
clog the drainage pipes; therefore, every lateral has a joint tubing that
extends to the soil surface and can be used for flushing rust out of the
pipes.
To prevent the ponding of water on the soil's surface, the depres-
sions within the plots were levelled with an RTK-GPS steered grader
during autumn 2016 and spring 2017. The target was to carry out grad-
ing so that every plot inclines to the edges of the field. Plot 6 was
levelled so that its inclination was marked toward the two surface run-
off gutters in the corner (Fig. 1B) from where the runoff is channelled
into the sampling station. This arrangement appeared not to be func-
tional during the snowmelt periods, and the measured surface runoff
remained low. Therefore, the surface runoff gutters were replaced in
2019, and preliminary results of the surface runoff volumes of 2020
were obtained for this paper.
2.5. Monitoring of groundwater level and drain discharge
Perforated groundwater (GW) pipes (diameter Ø 50 mm) were
installed in each field reaching a depth of 2 m. In all, four pipes were
installed to each plot between the laterals. The GW levels were moni-
tored manually from all pipes between July 5 and August 28, 2017 and
continuously (at 15-min intervals) using Solinst Levelogger sensors
installed in two pipes at each plot between March 26 and December
31, 2018. The drain discharge was collected at the sampling station
from each plot individually, and the water volume was measured
using a V-dam and water pressure sensor (type: STS PTM/N) (Supple-
mentary Fig. S1A–D). These data were collected on-line and stored in
a cloud service, so it could be viewed via the internet. The runoff ratio
(RR) was calculated individually for the plots by dividing the discharge
amount of the respective plot by the precipitation measured for the
Ruukki station.
Discharge water needs to be pumped away from the sampling sta-
tion, because the water level outside the station is often higher than
the outlets for sampling. The theoretical design flow for the whole
area (19.56 ha) is 20 l s−1. The electrical discharge pump (2.5 kW)
was chosen so that its capacity would be enough to pump away the
whole design flow. However, during times of extremely high flow,
there has beenflooding in the sampling station, and a tractor-driven hy-
draulic pump was used for flood control. Sampling pumps, sampling
cans, and command electronics were all placed above the flood line.
The discharge waters are rich in dissolved Fe, and rust tends to pre-
cipitate onto the lower corner of the V-dam opening, resulting in erro-
neous results of the discharge volume. Therefore, a system was
developed to remove the precipitates at intervals of 6 h automatically
with pressurized (3–4 bar) air.
2.6. Water quality measurement
Water samples were automatically taken by hose pumps propor-
tional to the discharge volume. About 0.1-liter subsamples were
drawn into 10-liter cans from every 2000 l of drainage water. From
every plot, 1 l of the collectedwaterwas delivered to a laboratory for pe-
riodic analysis in 2016–2017. In 2018, the samples were stored in a
freezer at −20C° until analyzed in spring 2019. According to Dore
et al. (1996) and Fellman et al. (2008), the freezing of water samples5does not have a marked effect on the concentrations of elements that
we measured. Altogether, 23 water samples from each plot were
taken between October 2016 and October 2018. The samples repre-
sented the periods of rainy autumn plus early winter and snow melt
in spring. During the discharge periods, the intervals between sam-
plings were 5–29 days, with a mean of 12 days. Hydrological years,
starting in October, are used as the basis for presenting the flow-
weighted characteristics of water samples and substance loads.
The electrical conductivity (EC) of water samples was measured
with a conductivity meter (SFS-EN 2788:1994). The pH was measured
according to SFS 3021:1979. Acidity was analyzed with a potentiomet-
ric titration with 0.02 M sodium hydroxide (NaOH) up to pH 8.3 (SFS
3005:1981). Total organic carbon (TOC) was determined according to
SFS-EN 1484:1997 using a Shimazdu TOC-LCSH analyzer where the
CO2 generated by oxidation is detected using an IR gas analyzer. Sulfur
was analyzed with the method SFS-EN ISO 11885:2009 from unfiltered
samples acidified with nitric acid (HNO3) using a Thermo Fisher
Scientific iCAP6500 Duo Inductively Coupled Plasma Optical Emission
Spectrometer. Phosphorus was determined colorimetrically using a
Foss – Tecator FIAStar 5000 Flow Injection Analyser (FIA) (SFS-EN ISO
15681-1:2005). For total P, unfiltered samples were digested with
peroxodisulfate before determination, and for dissolved P determina-
tion, the original samples were filtered using 0.45 μm Polyethersulfone
(PES) filters. Particulate P was calculated as the difference of total P and
dissolved P. Total Nwasmeasured colorimetrically with FIA after an ox-
idative digestion with peroxodisulfate (ISO 11905-1:1997). Concentra-
tions of NH4+-N and NO3−-N were determined colorimetrically with FIA
according to SFS-EN ISO 11732:005 and SFS-EN ISO 13395:1997(FIA),
respectively. Organic N was calculated as the difference of total N and
inorganic N (i.e., the sum of NH4+-N and NO3−-N). Water analyses
were carried out at Suomen Ympäristöpalvelu, which is currently part
of Eurofins Ahma Oy. The laboratory holds accreditations for all of
these methods except the titration of acidity.
The loads of substances for each plot were calculated using the fol-
lowing equation:
Annual load kg ha−1

 
¼ Cw  Q
where, Cw is the runoff weighted average concentration (mg l−1) in
water samples and Q is the cumulative mean discharge (l ha−1)
during the period represented by the water sample.
2.7. Measurements of GHG emissions
Tomeasure GHG emissions, onemetal collar (60 cm× 60 cm)was
installed permanently into the soil next to each GW pipe, which
made four replicates within every six plots. In summer 2017, GHG
emissions were measured six times between July 7 and August 30
using the closed chamber technique (Nykänen et al., 1995). The
dark chamber (60 cm × 60 cm × 40 cm) was placed on top of the col-
lar, and the groove of the collar was filled with water to make the
chamber airtight. During the 30 min closure, four gas samples of
20 ml were taken with a 60 ml plastic syringe and injected immedi-
ately into pre-evacuated 12 ml gas vials. The gas samples were ana-
lyzed with a gas chromatograph (HP 7890 Series, GC System,
Agilent, USA) equippedwith flame ionization (FID) and electron cap-
ture detectors (ECD) and a nickel catalyst for converting CO2 to CH4.
The precolumn and analytical columns were 1.8- and 3-m long steel
columns packed with Hayesep Q (80/100 mesh). The GC had a 10-
way valve with a 2 ml sample loop and a backflush system for flush-
ing the precolumn between the runs. A 6-way valve passed the flow
to either the FID or the ECD. The temperatures of the GC oven, FID,
and ECD were 70 °C, 300 °C, and 350 °C, respectively. Nitrogen was
used as the carrier gas and a mixture of argon and methane (5%) as
a make-up gas (1.4 ml min−1) to increase the sensitivity of the
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499ECD. An autosampler (222 XL Liquid handler, Gilson Medical
Electronics, France) fed the samples to the loop of the GC.
2.8. Statistical methods
Quantities of precipitation and discharge were aggregated into
monthly values and allocated to the corresponding water samples. Dif-
ferences in the chemical characteristics of soil layers were tested with
one-way ANOVA. Linear correlation coefficients were calculated for
the characteristics of the discharge water samples.
Differences of N, P, and TOC concentrations of water samples were
tested after grouping the plots into those with a thicker (Plots 1, 2,
4) or thinner (Plots 3, 5, 6) layer of organic material. Another grouping
was carried out based on theAS soil characteristics (Plots 1, 3, 5 and 6 vs.
Plots 2 and 4). The soil properties of each plot that are the basis for these
groupings will be presented in detail in the Results section. Testing of
the different concentrations was done by estimating analysis of covari-
ance (ANCOVA) models for the repeated measurements. These models
were analyzed in the SAS program's procedure for generalized linear
mixed models (GLIMMIX). The version of base SAS was 9.4 and the
GLIMMIX procedure that was used in the modelling was from the ana-
lytical package SAS/STAT 15.1. Before the actual model building was
done, transformations of themodel variableswere considered by study-
ing the relationships among dependent variables and possible covari-
ates. Either levels or logarithms of the variables were used in the final
models. Logarithmic transformation was chosen if it seemed to guaran-
tee a better model due to a statistically more normal relationship be-
tween a dependent and an independent variable. Time effects were
taken into account by fitting a time trend or period dummies into a
model. The applied within-subject covariance structure was selected
from several spatial structures by comparing values of information
criteria. The spatial covariance structures allow for autocorrelation and
can be applied to data that have unequally spaced repeated measure-
ments. Degrees of freedom and standard deviation approximations for
the F tests were calculated according to Kenward and Roger (2009).
3. Results
3.1. Carbon and nitrogen contents of the experimental field
The thickness of the horizons rich in soil organic carbon (SOC) varied
from 20 to 80 cm (Fig. 1C, D). Plots 1, 2, and 4 had thicker, and plots 3, 5,
and 6 thinner, topsoil horizons rich in SOC. This is the basis for the divi-
sion of the plots into two groups. Below these horizons, there was a
sharp decline of SOC to values ≤1% SOC (Fig. 2A, Supplementary
Table S2). The subsurface drainage pipes of all plots were thus clearly
installed in themineral soil at least 60 cmbelow the horizons consisting
of organic material. The mineral subsoil underneath the organic mate-
rial consisted of coarse silt and fine sand, and the mineral material inFig. 2. A) The concentration of soil organic carbon (SOC) in the experimental plots. Plots 3, 5, a
error bars show the standard deviation. B) The pH of soil profiles in Plot 3 (drilling site P1) and
incubation (inc). The exact locations of P1 and P2 appear in Fig. 1B. The pH data were retrieve
6the topsoil rich in SOC likely also consisted of these particle size frac-
tions, amounting to 42–59% in Plots 1–5 and about 70% in Plot 6. Clay
was detected deeper, 0.9 m in Plots 1 and 2, with a gradual and consis-
tent deepening down to 1.9 m in Plot 6. In the study area, the clay frac-
tion is dominated bymica, chlorite, and amorphous material (Fe and Al
hydroxides), while the coarser fractions mostly consist of quartz and
feldspars (Sippola, 1974).
SOC stocks of each plot and horizon are shown in Supplementary
Table S2. Within the uppermost 100 cm of Plots 1, 2, and 4, the SOC
stock was 627–666 tons ha−1, while in Plots 3, 5, and 6, the stock was
237–263 tons ha−1. These are rough estimates because the thickness
of the organic horizons varied within each plot, but they well illustrate
the difference between the two parts of the field.
The total nitrogen (TN) concentration of soil was associated with
SOC, and the TN stock (Supplementary Table S2) in Plots 1, 2, and 4
within the top 100 cm of soil ranged from 35 to 39 tons ha−1, while in
Plots 3, 5, and 6, it ranged from 14 to 16 tons ha−1. Between 100 and
200 cm, the TN stock in Plots 1–5 was 6–10 tons ha−1, and in Plot 6, it
was only 2 tons ha−1.
In spring 2020, before fertilization, there were large differences in
the amounts of Nmin between the different soil layers (Supplementary
Fig. S2).When studied at 50 cm-intervals, the soil at 50–100 cmwas the
lowest in Nmin in all plots, while the water-logged soil at 150–200 cm
contained at least as much Nmin as the topsoil at 0–50 cm. In all plots,
the NO3−-N stock, concentrating in the surface layers, amounted to
only around 14 kg ha−1, because it was depleted by leaching in autumn
and spring. The NH4+-N stockwasmuch larger at all depths. In Plots 1, 2,
and 4,which had a larger TN stock, the amount of NH4+-Nwas as high as
208 kg ha−1, but it was only 62 kg ha−1 in Plots 3, 5, and 6.
3.2. Soil pH and readily soluble mineral elements
The measurement of pH throughout the entire soil profile in two lo-
cations (drilling P1 at the edge andmore centrally located P2 in Fig. 1B)
suggests that the subsoil of the whole area had acid sulfate (AS) soil
characteristics. The measurements down to 3 m (Fig. 2B) showed that
Plot 3 (drilling P1) had a fresh pH of 3.5–3.9 at 0.4–1.0 m, indicative of
the thionic horizon. In drilling P2, close to Plot 6, the thionic horizon
with a fresh pH 3.7 occurred at 1.0–1.2 m. Based on the pH drop in the
incubation test, the soils contained sulfidic material below 1.0–1.2 m.
According to the results from GTK (2021), total S concentration be-
tween 0.8 and 2.0 m ranged from 0.32% to 1.02% in P1 and from 0.45
to 0.79% in P2,with themaximumoccurring at 1.2–1.4mat both drilling
sites.
For the presentation of soil test results (Supplementary Table S3),
the plots were divided into two groups based on the AS soil properties
(S, pH, EC) because, in terms of other characteristics, the field was
more uniform. In Plots 1, 3, 5, and 6, stacked in the north-western part
of the field (Fig. 1B), AS characteristics were more clearly pronouncednd 6 represent plots of thinner, and Plots 1, 2, and 4 thicker layers of organic material. The
close to the edge of Plot 6 (drilling site P2) upon sampling (fresh) and after eight week's
d from GTK (2021).
Table 1
The amount of precipitation (mm) in eachhydrological year, amount of drainagewater (D,
mm) and the runoff ratio (RR, %) at each plot.
Hydrological year
and precipitation
Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Plot 6
D RR D RR D RR D RR D RR D RR
1: 544 348 64 171 31 219 40 312 57 226 41 221 41
2: 458 226 49 123 27 214 47 198 43 210 46 143 31
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499than in Plots 2 and 4 in the south-eastern part of the field. Concentra-
tions of chemical elements are presented in mg l−1 of soil, because the
material varied in bulk density from 0.15mg l−1 in the horizons richest
in SOC to 1.6 kg l−1 in themineral subsoil. As determined in soil testing,
regular fertilization and liming had elevated the exchangeable Ca, Mg,
and easily soluble P concentrations and the pH of the topsoil. At the
depth of 40–80 cm, fresh pH values lower than 5.0 were measured in
all plots, with a minimum at 3.7 in Plot 6. At the depth of 80–100 cm,
pH values as low as 3.5 were measured in Plots 1 and 6, indicating ac-
tively oxidizing sulfidicmaterials. The lowest fresh pH valueswere asso-
ciated with elevated EC and concentrations of easily soluble S. The
conversion of measured EC to EC in saturated paste extract, according
to Richards (1954), resulted in estimates of 3.2–12.5 dS m−1 at the
depth of 80–100 cm in the north-western end of Plots 1, 3, 5, and 6, in-
dicating salinity caused by the oxidation of sulfidic material. Soluble S
concentration correlated closely (r= 0.98, p < 0.001) with EC. Where
the EC was the highest, soluble S concentration exceeded the threshold
value of 500 mg kg−1 for a thionic horizon (IUSS Working Group WRB,
2014), with the maximum concentration of 1200 mg kg−1 (converted
from the soil test result of 1620 mg l−1 in Supplementary Table S3
using the bulk density of 1.35 kg dm3) being found in Plot 1.
3.3. Soil classification
Based on SOC content and the thickness of the organic horizons,
Plots 1, 2 and 4, and part of Plot 3, had a histic horizon thick enough
(≥40 cm) to qualify as Histosols, while Plot 5 and part of Plot 3 had a
thin (10–20 cm) histic horizon. Furthermore, according to the WRB
criteria, the topsoil of Plot 6 did not consist of organic material, but
had an umbric horizon. The AS soil properties were indicated with
Thionic and Sulfidic qualifiers. The organic material was sapric, and ar-
tificial drainage justified the drainic qualifier. Classification of the plots
according to the WRB system (IUSS Working Group WRB, 2014) was
as follows:
1 and 3 (partly): Drainic Sapric Thionic Histosols (Sulfidic)
2 and 4: Drainic Sapric Histosols (Sulfidic)
5 and 3 (partly): Gleyic Histic Umbrisols (Drainic, Thionic, Sulfidic)
6: Gleyic Umbrisols (Drainic, Thionic, Sulfidic).
3.4. Hydrology and soil temperature
In the first hydrological year the precipitation was 544 mm, very
close to the average annual for 1981–2010 (541mm), while the second
hydrological year was less rainy (458 mm). The discharge peaked in
April–May (max 99 m3 ha−1 day−1) and in autumns (Supplementary
Fig. S3), even though in autumn 2018 the discharge remained low
(max 4 m3 ha−1 day−1). The monitoring started in October 2016 and
weather statistics indicate that themeasureddischargepeak inNovember
2016 (25 m3 ha−1 day−1) represented the major outflow period in that
autumn; this was hydrologically comparable with the autumn peak in
October 2017 (19 m3 ha−1 day−1). In summer periods, there was hardly
any discharge from the plots. In the three summermonths, the precipita-
tionwas 240mmand only 156mm in the two hydrological years, respec-
tively, and the potential evaporation in May–August of the second year
(386mm)wasmuchhigher than in thefirst summer (266mm), resulting
in plenty of empty pore space in the soil at the end of the second summer.
As a consequence, no water samples could be obtained for almost six
months, between early May and late October of the second hydrological
year. In the dry (second) year, the mean RR was 41% (range 27–49%),
and in the average (first) year it was 46% (31–64%) (Table 1). However,
the RR varied among the plots from 27 to 64%, suggesting that Plot 1,
in the corner of the field (see Fig. 1B), may have received surface runoff
from adjacent Plot 2 in the first year. In the dry year, the RR in Plot 2
was very low, suggesting seepage alongside the main drain. Because
the surface runoff among plots could not be measured and the RR
calculated from the discharge varied inconsistently among them, the7average runoff through the drainage pipes during a given hydrological
year was used in the calculation of loads.
In the present monitoring period of two years, the duration of frost,
from late autumn until the beginning of melting, was six months for
2016–2017 and four months and 10 days for 2017–2018 (Supplemen-
tary Fig. S3). During the first winter, the frozen soil extended deeper
(36 cm) than in the second winter (19 cm). It is noteworthy that half
(12) of the discharge water samples (altogether 23) were obtained
from soil that was frozen at shallow depths. In 2017, 37–51% of dis-
charge cameduring the periodwhen the surface soil layers were frozen,
and in 2018 the proportion was 26–45%, except for Plot 6, where only
14% of discharge was collected when there was frost in the surface soil
layers. According to the long-term (1989–2002)monitoring of soil tem-
perature by the Finnish Meteorological Institute (data not shown), soil
temperature at the depth of subsurface drainage pipes varied between
1 and 11 °C within a year and was above 5 °C for five months.
Monitoring of GW level in July–August 2017 (Fig. 3A) showed that,
in most plots, the GW level was only slightly below the subsurface
drainage pipes, while in Plot 6 the GW level was deeper. The increased
GW level at the end of August 2017 in all plots was due to abundant rain
(56 mm) during the three last days of the month. During the dry sum-
mer of 2018 (Fig. 3B), GW was lower, except in Plot 1, where GW did
not decrease below the subsurface drainage system. In Plots 2, 3, 4,
and 5, the GW level went below the drainage pipes in early July for
about three months and reached a maximum depth of 156 cm in early
August. In Plot 6, GW decreased below the GW pipes (200 cm) for
more than three months in 2018 and was elevated to only 150 cm by
the end of the year, explaining the very small amount of discharge
from this plot in autumn 2018.
According to the GW results, we estimated two possible GW flow
path patterns, a high and a low GW pattern (see Fig. 1D). During high
GW, subsurface drains function efficiently and they dominate the
groundwater flow. When GW drops below the subsurface drain pipes,
the flow pattern changes and percolation to the large open main drains
around the field is supposed to be the primary GW flow pattern.
Regarding conditions conducive to the oxidation of sulfidic material
and the production of acidity and sulfate in the drainage water, at least
the upper parts of the sulfidic horizons were above the GW level in
2017, with the possible exception of Plot 1. During a severe drought in
2018, GW dropped so low that sulfidic materials were no longer
water-saturated in Plots 2–6. Deep GW likely allowed the sulfidic sub-
soil to be oxidized in Plots 2–6. According to GW results we estimated
two possible groundwater flow path patterns, a high ground water pat-
tern and a low groundwater pattern (see Fig. 1D). During the high GW
subsurface drains function efficiently and they dominate groundwater
flow pattern. However, when GW drops below subsurface drain pipes
the flow pattern changes and percolation to deeper main drains is sup-
posed to be the primary GW flow pattern. It should be noted that our
preliminary estimation must be verified for example by stable water
isotopes in further studies.
3.5. Characteristics of water samples
The concentrations of TOC and N in discharge water and the
amounts that were leached differed from plots with large SOC stock
(1, 2, 4) and small SOC stock (3, 5, 6) (Supplementary Table S4). The dif-
ferences were statistically significant for TOC, total N (TN), and organic
Fig. 3. The groundwater level in the experimental plots in A) July and August 2017 (manualmeasurements) and B)March throughDecember 2018 (logger data). The error bars in A) stand
for ±standard deviation on each measuring day in 2017 and 10-day intervals in 2018. Where the error bars are not presented, the results are based on data from one logger only.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499N (Norg), as confirmed by ANCOVA tests. The TOC concentration (Sup-
plementary Fig. S4) peaked at the beginning of the high flow period in
the autumn, and the minimum values with little difference between
the two groups of plots were reached during the most intensive dis-
charge, in the spring. Altogether, 190 kg of TOC ha−1 leached from
Plots 1, 2, and 4, while only 94 kg ha−1 leached from Plots 3, 5, and 6.Fig. 4. The concentrations of different N (A–B) and P (C–D) fractions in the discharge and th
phosphorus.
8The discharge from the plots with a larger SOC stock (1, 2, 4) had
substantially higher TN concentrations compared to Plots 3, 5, and 6
(Fig. 4A–B). The highest TN concentrations were measured during the
high flow period, in spring 2017, and the lowest concentrations during
low discharge. The fluctuation in TN concentration was mostly caused
by NO3−-N, and the peak values of TN predominantly consisted ofe average monthly amounts of discharge. PP = Particulate phosphorus, DP = Dissolved
Fig. 5. A) The pH, B) sulfur concentration, and C) Fe concentration of the discharge water.
The average amount of the monthly discharge is expressed in all figures. The error bars in
A) and B) express the standard deviation.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499NO3−-N. For example, in April–May 2017, during the abundant dis-
charge, about 70% of TN consisted of NO3−-N while in the winter
months of 2017, only about 20% of TN in the discharge consisted of
NO3−-N. The flow-weighted average NO3−-N concentration of individ-
ual Plots was at 1.1–4.8 mg l−1, and no values above 10 mg l−1 were
measured. Among the plots with thicker organic layers, Plots 1 and 2
produced water with higher NO3−-N concentrations (3.0 and
4.7 mg l−1) than Plot 4 (2.0 mg l−1). The concentrations of Norg or
NH4+-N were much more stable within the two groups of plots than
those of NO3−-N. Norg concentration correlated with TOC (r = 0.67,
p < 0.001).
The proportions of NO3−-N,NH4+-N and organic N for the entiremon-
itoring period were 41, 23, and 36%, respectively. It is noteworthy that
the proportion of NH4+-N of TN was highest in the water samples of
Plot 6 (40%) and those peaks coincided with the lowest pH in water
(r=−0.66, p= 0.0011), while in the whole material this correlation
was non-significant.
The average TN load in the rainier hydrological year was
19.6 kg ha−1, from Plots 1, 2, and 4, with a high SOC stock, and
10.9 kg ha−1, from Plots 3, 5, and 6 with a smaller SOC stock. In the
drier hydrological year, the TN loads were smaller than in the rainier
year but again higher from plots with a high SOC stock (Supplemen-
tary Table S4).
We are aware that during this monitoring period surface runoff
was not collected comprehensively; therefore, the loads calculated
based on the discharge alone are underestimates. We have prelimi-
nary results of runoff volumes for the year 2020, when the surface
runoff was properly collected. In that year, surface runoff repre-
sented 30% of the total drainage of the plot next to the surface runoff
collector. During the monitoring period, we could obtain two surface
runoff samples. Assuming those concentrations are representative,
we can calculate that during the monitoring period loads of different
substances through the subsurface pipes represented 83% of total N
and 88% of TOC loads.
The flow-weighted mean total P (TP) concentration of discharge
from different plots ranged between 98 and 170 μg l−1 (Fig. 4C–D),
with no consistent connection to soil properties. Instead, dissolved P
(DP) was higher (51–74 μg l−1) in Plots 1, 2, and 4, which had a larger
SOC stock compared to 12–32 μg l−1 in Plots 3, 5, and 6, while particu-
late P concentration behaved in the opposite manner. When calculated
for the entire monitoring period, the proportion of DP was 49–54% in
plots rich in SOC, while only 12–22%was in dissolved form in thewaters
of Plots 3, 5, and 6 (F= 101.9, p< 0.001). There was a tendency of high
TP and DP concentrations during the peak discharge periods, with de-
creasing concentrations toward the end of the high flow. The higher
proportion of DP in Plots 1, 2, and 4, compared to the rest of the plots,
seems to be associated with the different thicknesses of organic mate-
rials in the two groups of plots and not to tillage operations. In particu-
lar, no tillage was applied to Plots 1 and 2 in autumn 2016, while Plot 4
was ploughed, but the three plots had similar proportions of DP in all
discharge samples until spring 2017. According to the ANCOVA tests,
the difference between the plot groups was statistically significant for
DP, but insignificant for TP.
The annual TP load in the first hydrological year ranged from 0.17
to 0.50 kg ha−1 (Supplementary Table S5) Even though the highest
load came from Plot 1, which had the thickest layer of organic mate-
rial, the load was dependent on the amount of discharge and there
was no consistent relationship between the SOC stock of a given
plot and the TP load (r =−0.227, p = 0.665). In the second hydro-
logical year, the range of TP load was 0.13–0.37 kg ha−1. The dis-
solved P load in the entire monitoring period was highest (0.43
and 0.26 kg ha−1) from Plots 1 and 4, which had the thickest hori-
zons of organic material, while 0.05–0.17 kg ha−1 were leached
from the other plots. In the same manner, as for N, we can roughly
estimate that TP in drain discharge represented 64% of TP and 68%
of DP load.93.6. Characteristics related to acid sulfate soil
Despite the firm evidence of the AS subsoil in all plots, the pH of
the discharge water from Plots 1, 2, 4, and 5 was above 6.0 through-
out the experimental period and above 5.0 in water samples from
Plot 5, except for one sample (Fig. 5A), suggesting that the sulfidic
horizons were oxidizing only to a moderate extent due to the rather
high GW levels in most of the plots. In Plot 6, the discharge water ex-
hibited pH 3.8 in autumn 2017, and the pH values were just above 4.0
in the springs of 2017 and 2018. It is noteworthy that this plot also
often produced discharge that had a pH of more than 6, particularly
at the beginning of the high flow periods. Plot 3 also occasionally
produced discharge of pH <5.5. More clearly than by pH, the plots
differed by a load of titratable acidity. While Plots 2 and 5 yielded
only 1400 mol ha−1 acidity during the entire monitoring period,
the other four plots (1, 3, 4, 6) yielded 2300–2800 mol ha−1. Electri-
cal conductivity, reflecting the concentration of charged solutes in
the discharge, ranged from 30 to 110 mS m−1, which, according to
Fig. 6.A) Carbon dioxide, B) nitrous oxide, and C)methane fluxes from the different study
plots of the cultivated peatland at the Ruukki research station. Carbon dioxide results are
represented as kg CO2-C ha−1 d−1 and nitrous oxide and methane results as g N2O-
N ha−1 d−1 and g CH4-C ha−1 d−1. Error bars represent standard deviation (n = 4). In
Plots 1, 2, and 4, the peat depth is approximately 70 cm, and in Plots 3, 5, and 6, it is
between 15 and 30 cm.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499the equation presented by Griffin and Jurinak (1973), correspond to
the ionic strength of 0.004–0.015 mol l−1. The highest values in the
water samples of all plots occurred in autumn 2017, and the lowest
values were in April–May, when snowmelt water diluted the dis-
charge.
The overall mean flow-weighted S concentration of the water sam-
ples was 71 mg l−1, while the corresponding results of individual
plots ranged from 46 (Plot 4) to 106 mg l−1 (Plot 6). In all the other
plots (1, 2, 3, 5), S concentrations were within a narrow range of
62–76 mg l−1. The concentration of S was smallest at the beginning of
the snowmelt or rainy periods (Fig. 5B), but the concentration increased
over time when the high flow continued. There was a strong positive
correlation (r= 0.898, p < 0.001) between the S concentration of the
discharge water and EC, but a moderate negative correlation (r =
−0.61, p < 0.001) with pH. Plots 1 and 6 produced the highest S
loads, altogether 449 and 394 kg ha−1 during the monitoring period,
owing to the high discharge (Plot 1) and high S concentration in the
water (Plot 6). Plots 2 and 4 produced the lowest S load (232 and
246 kg ha−1), in accordance to the easily soluble S in the soil (Supple-
mentary Table S3).
The concentrations of Al correlated with the discharge water pH
(r=−0.645, p < 0.001). The highest Al concentrations were measured
in water from Plot 6, which was associated with the discharge of the
lowest pH values. In the other plots, Al concentrations were rather sta-
ble and low, with the median values of these four plots being only
0.6–1.4 mg l−1 (Supplementary Table S6). As for Fe, there was a strong
fluctuation of the concentration in every plot, and it was common dur-
ing high flowperiods for there to be sudden increases or decreases of Fe
concentrationwithin oneweek (Fig. 5C); the peaks were observed spo-
radically in different plots. The peak concentrations commonly occurred
at the beginning of the high flow and decreased rapidly. Contrary to Al,
there was no negative correlation between the discharge Fe concentra-
tion and pH (r= 0.008, p= 0.926), and the highest Fe concentrations
commonly occurred in water samples that had a pH around 6, not in
the most acidic water samples. The Fe load was about 8-fold compared
to Al. Even though the highest Fe loads were produced from Plots 1 and
4 (which had thicker layers of organic material) and the smaller loads
came from Plots 3, 5, and 6 (which had thinner layers of organic mate-
rial), Plot 2,with a thick cover of organicmaterial, produced the smallest
load in both hydrological years.
3.7. Greenhouse gas emissions
During the eight-week monitoring in July and August 2017, CO2
emissions were highest in the beginning of August for all study
plots (Fig. 6A). Toward the end of August and the growing season,
the CO2 emissions decreased similarly in all plots. There was a
trend that the highest CO2 emissions occurred from Plots 1 and 3,
and the lowest emissions were from Plots 5 and 6, but the
difference was not statistically significant. The average amount of
CO2 emitted for the whole eight-week measurement campaign for
Plots 1, 2, and 4 was 120 ± 53 kg CO2-C ha−1 d−1, while for Plots 3,
5, and 6, it was 117 ± 40 kg CO2-C ha−1 d−1, respectively. Nitrous
oxide fluxes were highest in early July, when the measurement
period for GHGs started (Fig. 6B). On the third measurement day,
August 2, 2017, the N2O fluxes were slightly increased in some
study plots. Thereafter, N2O fluxes decreased similarly in all plots
toward the end of August. Like the CO2 emissions, N2O fluxes
seemed to be the smallest in Plots 5 and 6 and highest in Plots 3
and 4, but there was no statistical significance. The average N2O
emission for the whole measurement campaign for Plots 1, 2, and 4
was 49 ± 68 g N2O-N ha−1 d−1 and for Plots 3, 5 and 6 it was
29 ± 46 g N2O-N ha−1 d−1, respectively. CH4 uptake was usually
measured in all study plots. In some cases, a small flux was
measured, but on average the CH4 fluxes were close to zero
(Fig. 6C). For Plots 1, 2, and 4, the average CH4 emission for the10whole measurement period was −2.4 ± 4.4 g CH4-C ha−1 d−1,
while for Plots 3, 5, and 6, it was 0.8 ± 4.0 g CH4-C ha−1 d−1.
During the period when GHG fluxes were measured, the GW
level ranged between 70 and 175 cm below the soil surface
(Fig. 3A). The GW levels were, on average, deepest when the GHG
flux measurements started at the beginning of July 2017 and when
the growing season was most active. At the end of August, the GW
level was increased closer to the soil surface, and CO2 and N2O
fluxes decreased simultaneously. The GW level did not affect CH4
fluxes. The CO2 fluxes were smallest when the GW level was less
than 60 cm from the soil surface and highest when GWwas between
70 and 130 cm. Further, the N2O fluxes were highest when GW was
between 70 and 130 cm. Outside this range, the N2O fluxes were
rather small.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 1504994. Discussion
4.1. Substantial differences in soil properties among the study plots
Our 19.6-ha experimental field contains substantial heterogeneity,
in spite of its flat and uniform appearance from the soil surface. The
major factors of heterogeneity among the six plots include 1) thickness
of the horizons consisting of organic material, 2) uneven manifestation
of acid sulfate (AS) soil characteristics, and 3) hydrological differences
exhibited by different amounts of drainage flow. This spatial variability
has impacts on environmental loading from different parts of this field.
The thickness of organic material varies notably and is reflected in soil
classification. Plots 1, 2, and 4 are Histosols, and Plots 5 and 6 are min-
eral soils with histic or umbric topsoil, while Plot 3 is partly Histosol
and partly Umbrisol with a thin (<40 cm) histic horizon. The plots
with a thicker peat layer (1, 2, 4) may have been natively wetter, partly
attributable to the clayey subsoil closer (0.9 m) to the soil surface than
in the rest of the plots (1.9m); this has been conducive to the decreased
decomposition of organic matter. The thickness of the organic material
may also reflect the agricultural drainage history, because the plots
that have an older subsurface drainage are characterized by a thinner
peat layer. Several studies (e.g., Maljanen et al., 2007) reported large
CO2 emissions from cultivated organic soils, which gradually results in
a marked loss of SOM and a thinner, or even disappearing, peat layer.
In Plot 6 which was subsurface drained in the early 1900s, the peat
layer of unknown original thickness has practically disappeared, but
Plots 1 and 2, which were subsurface drained for the first time in the
1980s still have 60–80 cm of peat. The division of plots based on the
thickness of organic material comprises an experimental factor associ-
ated with the topsoil. As for the subsoil, all plots contained sulfidic ma-
terial, but its effect on drainage water quality varied temporarily and
from plot to plot. Since uniform water management was applied to all
plots, variable drainage flow and GW levels indicated native hydrologi-
cal differences among the plots. The abundant stock of SOM above the
drainage pipes constitutes a release potential of dissolved organic C and
mineral N that is enhanced by wetting and drying cycles. Furthermore,
whenGW is low, sulfidicmaterial at and below the drainage pipes can ox-
idize to sulfate (SO42), oxyhydroxysulfates, and ferrihydrite. Environmental
loading from this field is impacted both by histic and AS characteristics,
and the resultsmust not be interpreted as an effect of either of these char-
acteristics alone.
Soil temperature, moisture, acidity, the availability of decomposable
organic matter, and redox conditions have an impact on biological,
chemical and transport processes taking place in the soil. In our experi-
ment, half (n=12) of the drainagewater samples (n=23)were taken
during periods when the topsoil was frozen. However, soil tempera-
tures below 50 cm are above 0 °C throughout the year, enabling wintry
drain discharge, which reflects the porewater characteristics of the sub-
soil. A summermean soil temperature of 12.8 °Cwasmeasured at 20 cm
depth between 1976 and 1990 at the Ruukki site (Heikinheimo and
Fougstedt, 1992), indicating favorable conditions for aerobic microbial
activity in the summer, in spite of occasional droughts that may impede
biological activity. The mean summer soil temperature decreases with
depth, settling at 9.2 °C at the depth of the drainage pipes and 5.7 °C
at 200 cm (Heikinheimo and Fougstedt, 1992). This slows down, but
does not stop, chemical andmicrobiological reactions. Water saturation
hinders aerobic reactions, but promotes reducing reactions, in large
parts of the subsoilmost of the year. The lowpH (<4) of the thionic sub-
soil horizons caused by the oxidation of sulfidic material also decreases
the biological activity of subsoil horizons in AS soils (Šimek et al., 2011).
Furthermore, the low concentration of organicmatter in the entiremin-
eral subsoil likely limits the activity of heterotrophicmicrobiota. In sum-
mer, reaction products are accumulated and not transported out of the
soil because of the lack of discharge. In autumn and spring, rain ormelt-
ingwater fills the pores and carries dissolved substances to the drainage
pipes. Sulfate is probably transported first, followed by Fe as soon as11ferrihydrite has dissolved during anaerobic periods. The transport of
Fe can also be enhanced by complexation with organic matter in pore
water. Thus, the loads are dependent on the production rate of soluble
substances and the effectiveness of the transport processes, both of
which are spatially and temporally variable.
4.2. Differences in nitrogen fractions of discharge water among soil types
The proportions of N fractions in the discharge water in Ruukki dif-
fered from those of mineral soils. In the Ruukki field, Norg comprised
a substantial part (two-year average of different plots 33–44%) of TN,
and there was no systematic difference in the species distribution be-
tween the plots of small or large soil SOM stock. In turn, the proportion
of NO3−-N was only 30–52%, while NO3−-N commonly dominates in the
discharge ofmineral soils, comprising 94% of TN fromamineral acid sul-
fate soil in 2011–2017 (Yli-Halla et al., 2020), 85% from a clay soil in
Finland in 1991–1993 (Turtola and Paajanen, 1995), and 78% from a
449-ha catchment dominated by cereal cropping in south-eastern
Norway (Chen and Bechmann, 2019). The proportion of NH4+-N
(12–24% at Plots 1–5) was substantially higher than in mineral soils.
In that respect, our shallow peat soil is similar to the 3–4 m-thick peat
soil of Huhta and Jaakkola (1993), where NH4+-N comprised 5–16% of
TN in the discharge. In our study, the NH4+-N proportion in Plot 6
(37%) represented an even larger share (37%) than NO3−-N, although
the NH4+-N stock was not the most abundant at this plot. Simultaneous
peaks of low pH and high NH4+-N concentrations at this plot suggest
that the peaks were associated with the transport of solutes from
deeper horizons where sulfidic material had been oxidized and that ni-
trification had been prevented by acidity or anoxia. The high proportion
of NH4+-N in the discharge water from Plot 6 may also be explained by
the reduction of NO3−-N to NH4+-N coupled with the oxidation of sul-
fides in the actively oxidizing AS soil layer (Brunet and Garcia-Gil,
1996). An abundance of NH4+-N seems to be typical of peat soils, as re-
ported in the discharge waters of several peat mining areas in Finland
(Heikkinen et al., 2018). These results show that, in peat soils, the deter-
mination of NO3−-N alone is not sufficient when estimating N leaching,
but TN needs to be explicitly measured.
4.3. Soil properties influence nitrogen and carbon concentrations in
discharge waters
A major observation from our field was that the thicker the layer of
organic material and larger the SOC stock, the higher the TOC and TN
concentrations were in the subsurface drainage waters. However, the
leaching of TOC and TN seem to have rather different temporal patterns.
TOC concentration peaked after dry periods at the beginning of the high
flow in autumns 2016 and 2017. It is likely that the previous dry periods
promoted SOM transport, since air drying substantially increases the
water solubility of SOM(Kaiser et al., 2015, with references). In autumn,
the leaching of TN occurred slightly later during the runoff peaks, sug-
gesting that amajor part of theN originates from the interiors of soil ag-
gregates, from where N slowly diffuses into macropores in a rewetted
soil. Furthermore, TOC concentrations were low during the peak flow
in spring, while N concentrations, particularly in the form of NO3−-N,
were high in these water samples. More abundant leaching from Plots
1, 2, and 4 can be partly explained by the mineralization of the large
TN stock in soil. These same plots contained a higher Nmin stock, both
in the surface layers and in the sulfidic subsoil, likely contributing to
the discharge of N as well. The higher NO3−-N concentrations in Plots 1
and 2 may also be partly attributed to the soil disturbance caused by
the extensive ditching activities recently carried out in these two plots.
According to the results compiled in Table 2, the TOC concentra-
tions from all plots were higher compared to agricultural mineral
soils (Manninen et al., 2018), where the maximum concentrations
(25–30 mg l−1) were similar to the median in our field. Instead, the
TOC range in our water samples was similar to that of runoff waters
Table 2
Results of TOC, N, and P concentrations and annual loads in subsurface discharge water compiled from various studies. TOC= total organic carbon, TN= total nitrogen, TP= total phos-
phorus, DP = dissolved phosphorus.
Reference TOC
mg l−1
TN
mg l−1
NO3−-N
mg l−1
NH4-N
mg l−1
TN
kg ha−1
TP
mg l−1
DP
mg l−1
TP
kg ha−1
Soil, crop and average annual discharge (mm)
This study 21.7 4.2 1.5 1.3 9.3 0.13 0.022 0.29 0.2–0.3 m peat, 218 mm
43.5 7.1 3.4 1.1 15.4 0.12 0.059 0.26 0.6 m peat, 218 mm
Huhta and Jaakkola, 1993 n.d. 8.5 n.d. 0.6 38.2 0.31 0.19 1.4 3–4 m peat, barley hay; 429 mm, 1983–87
n.d. 4.8 n.d. 0.7 19.1 0.43 0.24 1.7
Kløve et al., 2010b 20.8 2.8 1.2 0.24 21.7 0.30 0.229 2.32 0.6 m peat, 763 mm, perennial grass, 2003–04
Yli-Halla et al., 2020 n.d. 21.1 19.8 0.15 63 n.d. n.d. n.d. Silty clay loam AS soil, 298 mm, cereals, 2011–17
Jaakkola, 1984 n.d. n.d. 9.0 n.d. 13d n.d. n.d. n.d. Clay, 145 mm, cereals, 1976–82
Turtola and Paajanen, 1995a 10.3e 7.9 6.8 n.d. 16.4 0.20 0.028 0.59 Clay, 328 mm, cereals, 1991–93
n.d. 6.2 5.2 n.d. n.d. 0.23 0.027 n.d. Clay, 159 mm, ley, 1991–93
Manninen et al., 2018c 15.2 n.d. n.d. n.d. n.d. n.d. n.d. n.d. Fine sand, cereals, mineral fertilization, 2012–14
a Period after improved subsurface drainage.
b Plot drained with subsurface pipes. Loads during a period of 13 mo 12 d.
c Site Toholampi.
d NO3−-N load.
e Data of the same field by Manninen et al. (2018) for two years (March 2012–March 2014).
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499from five peat extraction areas in central Finland (18–81 mg l−1,
Heiderscheidt, 2016). Accordingly, the TOC load from our field was
higher than (plots of thicker organic horizons) or similar to (plots of
thin organic horizons) the loads from mineral soils (25–52 kg ha−1,
Manninen et al., 2018), but lower than from a peat soil in Norway
(Kløve et al., 2010)where a high TOC load of 159 kg ha−1, despite a sim-
ilar or lower TOC concentration, can be explained by a 3.5-times higher
discharge than in our field.
The discharge from a 3–4 m thick peat soil grown for barley (Huhta
and Jaakkola, 1993) had a somewhat similar TN concentration as the
waters from our experimental field's 0.6 m peat layer, but the higher
discharge from their field was conducive to the higher load. Despite a
much lower N concentration in the discharge from a Norwegian peat
soil (Kløve et al., 2010), a higher TN load was also measured there be-
cause of the abundant amount of discharge. There was no systematic
difference in TN concentration in the discharge water between our
peat soil and a heavy clay soil from southern Finland (Turtola and
Paajanen, 1995). Indeed, our findings suggest that TN leaching from
the present shallow peat field was at the same level as or slightly
lower than that from mineral soils (15.5 kg ha−1, Tattari et al., 2017;
10–17 kg ha−1, Turtola and Paajanen, 1995). Nitrogen leaching from
our experimental field was also much less than that from a mineral
acid sulfate soil in Finland, where the TN concentration in discharge
water was three- to five-times higher and 50–63 kg N ha−1 was annu-
ally leached in 2011–2017 (Yli-Halla et al., 2020), even though both
fields had a large stock of Nmin in soil horizons below the subsurface
drainage pipes. As for the crop, Huhta and Jaakkola (1993) reported in
their thick peat soil that TN leaching from perennial hay was only
about half of that measured from spring barley plots, suggesting a
strong positive influence of a permanent soil cover crop that catches
the Nmin that is mineralized. Even though we also grew spring cereals
and grass in our field, the experimental set-up and the short duration
of themonitoring do not allow us to make conclusions about the effects
of different crops onN leaching. A compilation of results thus shows that
even though organic soils have a large N stock, it is not always reflected
as an abundant N loading, but the concentration in discharge water and
the load from peat soil can be at the same level as, or even lower than,
that found in mineral soils.
4.4. Phosphorus leaching was controlled by mineral soil contact
The flow-weighted mean TP concentration (0.10–0.16 mg l−1) was
quite similar in all plots, but there was a tendency for Plots 1, 2, and 4,
of high SOC stock, to produce drainage water with a higher percentage
of DP (49–57%) than Plots 3, 5, and 6 (12–23%). Even a higher propor-
tion of DP (50–80% of TP) has been found in a 3–4 m thick peat soil
where the drainage pipes were in the peat layer (Huhta and Jaakkola,121993). While in our field 0.17–0.34 kg TP ha−1 were leached annually
through the subsurface drainage pipes, 1.3–2.0 kg TP were leached in
the experiment by Huhta and Jaakkola (1993). Our results are compara-
ble to those measured in mineral soils, e.g., 0.59 kg ha−1 by Turtola and
Paajanen (1995). Compared to the average TP load of 1.1 kg ha−1, in-
cluding surface runoff and discharge, from agricultural land in Finland
(Tattari et al., 2017), the present field seems not to be a source area of
large P loading. Even taking into account that our results do not include
surface runoff andmay represent only about 64% of the total P load from
the field, an estimate of 0.27–0.53 kg ha−1 for the total annual TP load is
quite low.
In the Ruukki field, subsurface drainage pipes have been installed in
themineral soil, which likely has a high P sorption capacity, while in the
study by Huhta and Jaakkola (1993) the drainage pipes were in the peat
layer. Saarela (2008) demonstrated that the risk of P leaching is much
higher in soil that has a peaty subsoil compared to mineral soil, even
with a peaty plough layer.When evaluating the P loading risk of agricul-
tural peatlands, there is a need to separate shallow peat soils with
mineral subsoils from thickpeat soils. Our results confirm that the abun-
dance of organicmaterial contributes to a higher DP load, increasing the
eutrophication effect of a unit of TP that is transported to watercourses.
4.5. Acid sulfate subsoil causes a risk of acid discharge water
Our experimental field contains a thionic horizon and sulfidic mate-
rial starting just below the subsurface drain pipes at a depth of
1.0–1.2 m. The oxidation of sulfide, occurring in summer when the
GW is low, is indicated by a high SO42−-S concentration. However, low
SO42−-S concentrations were observed at the very beginning of dis-
charge peaks due to rain or snowmelt water, but after the oxidation
products had dissolved anddiffused from the interiors of soil aggregates
into themacropores and further into the drainage system, high concen-
trations of SO42−-S weremeasured. The strongest peaks of pH and SO42−-
S, measured in Plot 6, can be attributed to the remarkable lowering of
GW in this plot compared to the others, allowing effective oxidation of
sulfidicmaterial. The lack of a peak in autumn2018 in Plot 6 is explained
by the hydrological conditions. In October 2018, GW levels were still
low after the exceptional drought, and themeager amount of water en-
tering the drain pipes likely represented rainwater from above and not
the horizons where sulfide had actively oxidized in summer. Therefore,
it was likely that plenty of SO42−-S and aciditywould leach in the follow-
ing spring, or whenever the GW level reached the drainage pipes.
In all plots, SO42−-S concentrations (median 65 mg l−1) were much
higher than commonly measured in non-AS soils (5 mg l−1, Turtola
and Jaakkola, 1986). However, the SO42−-S concentrations of water sam-
ples were much lower than measured in the discharge water of two
mineral AS fields in Ilmajoki and Mustasaari about 300 km south,
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499where the average concentrations were 441 and 2312 mg l−1 in five-
and three-yearmonitoring (Bärlund et al., 2004). Accordingly, the aver-
age S load (166 and 127 kg ha−1 in the two hydrological years) was
much lower than in an AS soil in Rintala, about 300 km south of Ruukki
(630 kg ha−1, Österholm and Åström, 2004); however, these were al-
most 10 times higher than is commonly measured in clayey non-
sulfidic soils (15–20 kg ha−1, Turtola and Jaakkola, 1986; Eriksen and
Askegaard, 2000). Despite a similar total S concentration as in the AS
soilsmentioned above, Plots 1, 2, 4, and 5 in this AS field did not produce
severe loading of acidity, and the dischargewater from these plots had a
pH of 6.0 or above during thismonitoring period. Titratable acidity (me-
dian 0.4mmol l−1) and dissolved Al concentration (median 0.9 mg l−1)
were low, when compared to Ilmajoki andMustasaari (mean for acidity
3.9 and 11.0 mmol l−1 and Al 20 and 88 mg l−1, respectively), and the
mean pH of discharge waters at those sites were as low as 4.3 and 3.7.
The moderate loading of acidity and sulfur, despite a thionic horizon
and sulfidic material, may be associated with the protective peat cover-
age. The peat layer above the mineral soil horizons may consume most
of the oxygen via biological decomposition and so inhibit the oxidation
of sulfidic material, as reported in AS soils in Australia (Cook et al.,
2004). In addition, higher pH values can be explained by the high TOC
concentration of drain discharge water, which favors formation of
organo-Fe complexes (Herzog et al., 2020) and the co-precipitation of
organo-Fe aggregates (Ingri et al., 2018) instead of the precipitation of
ferrihydrite with a concomitant release of acidity. This explanation is
in agreement with the high dissolved Fe concentrations (median
7.7 mg l−1) compared to mineral AS soils in Ilmajoki and Mustasaari
(mean 0.9 and 3.7 mg l−1, respectively). The sporadically peaking Fe
concentrations occurring at the beginning of high flow periods may
refer to the flushing of rust precipitates from the drainage pipes by
high velocity runoff after a stagnant period. Furthermore, the highest
Fe concentrations (40–100 mg l−1) in our field were associated with
pH values around 6, suggesting that those waters may have originated
from the reduced soil layers and not from the acidified aerobic layers,
because the most acidic water samples usually had low dissolved Fe
concentrations.
4.6. GHG emissions from the Ruukki experimental plots
There were no significant differences in GHG emissions among the
study plots. The measurement period for GHG emissions was only
eight weeks, extending from the beginning of July to the end of August
2017, and it did not contain fertilization events. The aim of the GHG
measurements was to get rough estimates for emissions after establish-
ing the site and to see if therewere differences among the plots. In 2017,
spring barley was grown in Plots 1–4, and triticale in Plots 5–6. Both
crops had received the same fertilization in May. In the study by Yli-
Halla et al. (2017), conducted in boreal AS soils, N2O emissions were
higher in sites with 30 cm peat depth, compared to sites with 60 cm
peat depth. At the Ruukki site, the GHG emissions were slightly smaller
in Plots 5 and 6,with the thinnest peat compared to other sites butwith-
out statistical significance. When comparing the average fluxes for Plots
1, 2, and4 (thicker peat)with those fromPlots 3, 5, and 6 (thinner peat),
the differences were within the standard deviation.
Since the measurement period did not cover even one whole grow-
ing season, and because the most active growing phase earlier in sum-
mer (Lohila et al., 2003) was not captured by the measurements,
annual emissions cannot be calculated, but the results can be compared
to studies that havemeasured summer timefluxes. Furthermore, recent
soil disturbances, from establishing the study field, most likely affected
GHGdynamics by producing higher emissions, becausemore airwas in-
troduced to the peat profile than in normal agricultural practices. In our
study, the CO2 fluxes were between approximately 50 and 250 kg CO2-
C ha−1 d−1. These are of the same magnitude or slightly higher than in
other studies on organic soils. In the study by Lohila et al. (2003), the
total ecosystem respiration on barley and grass plots was between 4513and 210 kg CO2-C ha−1 d−1 and 45–260 kg CO2-C ha−1 d−1,
respectively. In the same study, the total ecosystem respiration for
sandy and clay soil growing barley or grass ranged between
approximately 10 and 165 kg CO2-C ha−1 d−1. In the study by
Nykänen et al. (1995), CO2 emissions were between 10 and
110 kg CO2-C ha−1 d−1 in a peat plot with grass cultivation. The N2O
fluxes were also in the same range as reported in other studies. The
N2O fluxes on grass plots in cultivated peat soil have been between
zero and approximately 100 g N2O-N ha−1 d−1 and from barley plots
between close to 0 and 200 g N2O-N ha−1 d−1 with few higher
occasions (Regina et al., 2004), whereas in our study the N2O fluxes
from Plots 1–4 growing barley were between close to 0 and
150 g N2O-N ha−1 d−1 with few higher fluxes in the first
measurement day. In the study by Syväsalo et al. (2006), N2O fluxes of
grass and cereal on sandy soil ranged between approximately 0 and
20 g N2O-N ha−1 d−1 under grass cultivation and 0 and 50 g N2O-
N ha−1 d−1 under cereal cultivation. This study field has been
effectively drained and, due to low GW levels, the CH4 fluxes are
usually negative or close to zero (Regina et al., 2007), as was the case
in our study.
Methane fluxes were small in all plots, with a range of approxi-
mately −10 and +10 g CH4-C ha−1 d−1 with quite large variation.
Because the research field is well-drained and peat is aerated from the
surface, the CH4 emissions are naturally small compared to pristine
peatlands. There were no systematic differences among plots when it
came to CH4 fluxes, but there were indications that, on some days,
CH4 fluxes were highest in Plots 5 and 6 (thin peat). However, CH4
fluxes in all plots were so small that they had a negligible effect on the
overall GHG budget of the research field.
5. Conclusions
In this paper, we introduce a 19.6-ha experimental field that was
established in 2016 at the Ruukki field station to monitor the leaching
of substances and GHGs from a cultivated peatland. We report results
from thefirst two years after establishing thefield. TheN load in subsur-
face discharge waters increased upon increasing peat depth. Owing to
the average and low amounts of discharge in the two years, the TN
load was similar or lower than is commonly reported from mineral
soils, but the proportion of NO3−-N was lower and NH4+-N and Norg
was higher than in mineral soils. The total P load was low, compared
to thick peat and even mineral soils, because the coarse silty and clayey
mineral subsoil likely effectively adsorbed P in the seepage water, the
retention being more effective the thicker the mineral soil horizons
are before the pore water reaches the drainage pipe. However, the pro-
portion of dissolved PO4-P increased upon increased peat depth. This
study tentatively suggests that N and P loads transported through sub-
surface drainage pipes from a shallow peat field are lower than from a
thick peat soil where the drainage pipes have been installed in the
peat. Our results indicate that peat cover canmitigate the negative envi-
ronmental impacts of AS soils by hindering the oxidation of sulfidicma-
terial, thus preventing the formation of acidity, but the required
thickness of the peat cover to achieve this effect should be studied in
greater detail. In cultivated peatlands, like our experimental field, the
GHG emissions, and especially those of CO2 and N2O, are higher than
in comparable mineral soils. Due to the short measurement campaign
and recent establishment of the study field, the effects of peat depth
or crops cannot be clearly seen from this data set. The CO2-C emissions
into the atmosphere were abundant during the monitored summer
weeks, and the daily losses were equal to or higher than the amounts
of TOC transported through the subsurface drainage pipes during the
whole year, but N loss in the form of N2O, at least in the summer, was
small compared to N loss contained in discharge water. This study
points out the diversity of cultivated peatlands and highlights the im-
portance of their monitoring to obtain trustworthy estimates of the en-
vironmental loading in northern conditions.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499CRediT authorship contribution statement
Yli-Halla: investigation, methodology, visualization, writing the
original draft. Lötjönen: methodology, investigation, management.
Kekkonen: investigation, management. Virtanen: conceptualization,
visualization, writing the original draft. Marttila: investigation, man-
agement (soil measurements), visualization, funding acquisition.
Liimatainen: investigation, visualization, writing the original draft
(GHG), funding acquisition. Saari: investigation (GPR).Mikkola: statisti-
cal methodology and analysis. Suomela: conceptualization, investigation
(establishment of the field). Joki-Tokola: conceptualization, investiga-
tion, project supervision and administration, funding acquisition. All au-
thors contributed to the reviewing and editing of the manuscript.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgments
The authors thank the Council of Oulu Region for allocating funds orig-
inating from theEuropeanRegional andDevelopment Fund (ERDF) for the
design and implementation of the experimental field. The grants received
in 2012–2017were A32186, A71390 and A71403.We also thank the ERA-
NET GAS program which funded the PEATWISE project in 2017–2020
(grant number 4400T-102), Kone Foundation for a financial support in
2018–2022 (grant number 201802192) and Suoviljelysyhdistys ry –
Finnish Peat Cultivation Association for financial support in 2019–2021.
The authors thank professor Kristiina Regina, Natural Resources Institute
Luke, Finland, for organizing the GHGmeasurement campaign and the re-
spective analyses.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.scitotenv.2021.150499.
References
Äijö, H., Paasonen-Kivekäs, M., Myllys, M., Nurminen, J., Turunen, M., Salo, H., Warsta, L.,
Koivusalo, H., Sikkilä, M., Alakukku, L., Puustinen, M., 2016. The effect of additional
subsurface drainage on water discharge and nutrient load on clay soil. In: Strock,
J.F. (Ed.), Proceedings of the 10th International Drainage Symposium, 7-9 Sept.,
2016 Minneapolis, Minnesota. American Society of Agricultural and Biological Engi-
neers ASABE, St Joseph, Minnesota, USA.
AnaEE, 2021. European network of platforms for analysis and experimentation on ecosys-
tems. https://platforms.anaee.eu/research-platform-index/10052-5/.
Bärlund, I., Tattari, S., Yli-Halla, M., Åström, M., 2004. Effects of sophisticated drainage
techniques on groundwater level and drainage water quality on acid sulfate soils:
final report of the HAPSU project. Finn. Environ. 732, 1–68. https://helda.helsinki.fi/
handle/10138/40557.
Batjes, N.H., 1996. Total carbon and nitrogen in the soils of the world. Eur. J. Soil Sci. 47,
151–163.
Berglund, Ö., Berglund, K., 2010. Distribution and cultivation intensity of agricultural peat
and gyttja soils in Sweden and estimation of greenhouse gas emissions from culti-
vated peat soils. Geoderma 154, 173–180.
Beucher, A., Fröjdö, S., Österholm, P., Auri, J., Martinkauppi, A., Edén, P., 2015. Assessment
of acid sulfate soil mapping utilizing chemical indicators in recipient waters. Bull.
Geol. Soc. Finl. 87, 5–23.
Brunet, R.C., Garcia-Gil, L.J., 1996. Sulfide-induced dissimilatory nitrate reduction to am-
monia in anaerobic freshwater sediments. FEMS Microbiol. Ecol. 21, 131–138.
Buschmann, C., Röder, N., Berglund, K., Berglund, O., Lærke, P.E., Maddison, M., Mander, U.,
Myllys, M., van den Osterburg, B., Akker, J.J.H., 2020. Perspectives on agriculturally
used drained peat soils: comparison of the socioeconomic and ecological business en-
vironments of six European regions. Land Use Policy 90 13 pp.
Chen, X., Bechmann, M., 2019. Nitrogen losses from two contrasting agricultural catch-
ments in Norway. R. Soc. Open Sci. 6 (190490). https://doi.org/10.1098/rsos.190490.
Cook, F., Dobos, S., Carlin, G., Millar, G., 2004. Oxidation rate of pyrite in acid sulfate soils:
in situ measurements and modelling. Aust. J. Soil Res. 42, 499–507.
Dore, J.E., Houlihan, T., Hebel, D.V., Tien, G., Tupas, L., Karl, D.M., 1996. Freezing as a
method of sample preservation for the analysis of dissolved inorganic nutrients in
seawater. Mar. Chem. 53, 173–185.14Duxbury, J.M., Beverly, J.H., 1978. Nitrogen and phosphorus losses from organic soils.
J. Environ. Qual. 7, 566–570.
Edén, P., Weppling, K., Jokela, S., 1999. Natural and land-use induced load of acidity,
metals, humus and suspended matter in lestijoki, a river in western Finland. Boreal
Environ. Res. 4, 31–43.
Eriksen, J., Askegaard, M., 2000. Sulfate leaching in an organic crop rotation on sandy soil
in Denmark. Agric. Ecosyst. Environ. 78, 107–114.
Esala, M.J., 1995. Changes in the extractable ammonium-nitrogen and nitrate-nitrogen
contents of soil samples during freezing and thawing. Commun. Soil Sci. Plan. 26,
61–68.
Fellman, J.B., D'Amore, D.V., Hood, E., 2008. An evaluation of freezing as a preservation
technique for analyzing dissolved organic C, N and P in surface water samples. Sci.
Total Environ. 392, 305–312.
Finnish Meteorological Institute, 2021. Ilmasto. https://www.ilmatieteenlaitos.fi/ilmasto.
Griffin, R.A., Jurinak, J.J., 1973. Estimation of activity coefficients from the electrical con-
ductivity of natural aquatic systems and soil extracts. Soil Sci. 116, 26–30.
GTK, 2021. Geological Survey of Finland. Data sets and on-line services. Acid sulfate soils.
https://gtkdata.gtk.fi/hasu/index.html.
Heiderscheidt, E., 2016. Evaluation and Optimisation of Chemical Treatment for Non-
point Source Pollution Control Purification of Peat Extraction Runoff Water. 580. Uni-
versity of Oulu Graduate School; University of Oulu, Faculty of Technology Acta Univ.
Oul. C.
Heikinheimo, M., Fougstedt, B., 1992. Statistic of Soil Temperature in Finland 1971–1990.
22. Finnish Meteorological Institute. Meteorol. Publ. Finland.
Heikkinen, K., Karppinen, A., Karjalainen, S.M., Postila, H., Hadzic, M., Tolkkinen, M.,
Marttila, H., Ihme, R., Kløve, B., 2018. Long-term purification efficiency and factors af-
fecting performance in peatland based treatment wetlands: an analysis of 28 peat ex-
traction sites in Finland. Ecol. Eng. 117, 153–164.
Herzog, S.D., Persson, P., Kvashnina, K., Kritzberg, E.S., 2020. Organic iron complexes en-
hance iron transport capacity along estuarine salinity gradients of Baltic estuaries.
Biogeosciences 17, 331–344. https://doi.org/10.5194/bg-17-331-2020.
Huhta, H., Jaakkola, A., 1993. Viljelykasvin ja lannoituksen vaikutus ravinteiden
huuhtoutumiseen turvemaasta Tohmajärven huuhtoutumiskentällä v. 1983-87.
Maatalouden tutkimuskeskus. Tiedote 20/93.
Ingri, J., Conrad, S., Lidman, F., Nordblad, F., Engström, E., Rodushkin, I., Porcelli, D.J., 2018.
Iron isotope pathways in the boreal landscape: role of the riparian zone. Geochim.
Cosmochim. Acta 239, 49–60. https://doi.org/10.1016/j.gca.2018.07.030.
IPCC, 2014. In: Hiraishi, T., Krug, T., Tanabe, K., Srivastava, N., Baasansuren, J., Fukuda, M.,
Troxler, T.G. (Eds.), 2013 Supplement to the 2006 IPCC Guidelines for National Green-
house Gas Inventories: Wetlands. IPCC, Switzerland.
ISO 11905-1:1997, 1997. Water Quality — Determination of Nitrogen — Part 1: Method
Using Oxidative Digestion With Peroxodisulfate. International Organization for Stan-
dardization.
IUSS Working Group WRB, 2014. World Reference Base for Soil Resources 2014. World
Soil Resources Reports 106. FAO, Rome.
Ivarson, K.C., 1977. Changes in decomposition rate, microbial population and carbohy-
drate content of an acid peat bog after liming and reclamation. Can. J. Soil Sci. 57,
129–137.
Jaakkola, A., 1984. Leaching losses of nitrogen from a clay soil under grass and cereal crops
in Finland. Plant Soil 76, 59–66.
Kaila, A., 1959. Retention of phosphate by peat samples. J. Sci. Agric. Soc. Finl. 31, 215–225.
Kaiser, M., Kleber, M., Berhe, A.A., 2015. How air-drying and rewetting modify soil organic
matter characteristics: an assessment to improve data interpretation and inference.
Soil Biol. Biochem. 80, 324–340.
Kasimir-Klemedtsson, Å., Klemedtsson, L., Berglund, K., Martikainen, P., Silvola, J.,
Oenema, O., 1997. Greenhouse gas emissions from farmed organic soils: a review.
Soil Use Manag. 13, 245–250.
Kekkonen, H., Ojanen, H., Haakana, M., Latukka, A., Regina, K., 2019. Mapping of cultivated
organic soils for targeting greenhouse gas mitigation. Carbon Manag. 10 (2),
115–126. https://doi.org/10.1080/17583004.2018.1557990.
Kenward, M.G., Roger, J.H., 2009. An improved approximation to the precision of fixed ef-
fects from restricted maximum likelihood. Comput. Stat. Data Anal. 53, 2583–2595.
Kersalo, J., Pirinen, P., 2009. Suomenmaakuntien ilmasto. Abstract: The Climate of Finnish
Regions. 8. Finnish Meteorol. Inst. Rep.
Kløve, B., Sveistrup, T.E., Hauge, A., 2010. Leaching of nutrients and emission of green-
house gases from peatland cultivation at Bodin, northern Norway. Geoderma 154,
219–232.
Lohila, A., Aurela, M., Regina, K., Laurila, T., 2003. Soil and total ecosystem respiration in
agricultural fields: effect of soil and crop type. Plant Soil 251, 303–317.
Longabucco, P., Rafferty, M.R., 1989. Delivery of nonpoint-source phosphorus from culti-
vated mucklands to Lake Ontario. J. Environ. Qual. 18, 157–163.
Maljanen, M., Hytönen, J., Mäkiranta, P., Alm, J., Minkkinen, K., Laine, J., Martikainen, P.J.,
2007. Greenhouse gas emissions from cultivated and abandoned organic croplands
in Finland. Boreal Environ. Res. 12, 133–140.
Manninen, N., Soinne, H., Lemola, R., Hoikkala, L., Turtola, E., 2018. Effects of agricultural
land use on dissolved organic carbon and nitrogen in surface runoff and subsurface
drainage. Sci. Total Environ. 618, 1519–1528.
Mattsson, T., Kortelainen, P., Räike, A., 2005. Export of DOM from boreal catchments: im-
pacts of land use cover and climate. Biogeochemistry 76, 373–394.
Myllys, M., Sinkkonen, M., 2004. Viljeltyjen turve- jamultamaiden pinta-ala ja alueellinen
jakauma suomessa. Summary: the area and distribution of cultivated organic soils in
Finland. Suo – Peat Mires 55, 53–60.
Nieminen, M., Jarva, M., 1996. Phosphorus adsorption by peat from drained mires in
southern Finland. Scand. J. For. Res. 11, 321–326.
M. Yli-Halla, T. Lötjönen, J. Kekkonen et al. Science of the Total Environment 806 (2022) 150499Nykänen, H., Alm, J., Lång, K., Silvola, J., Martikainen, P.J., 1995. Emissions of CH4, N2O and
CO2 from a virgin fen and a fen drained for grassland in Finland. J. Biogeogr. 22,
351–357.
Nystrand, M., Auri, J., Bollström, F., Hadzic, M., Ihme, R., Österholm, P., 2019. Minimum
peat thickness to prevent oxidation of underlying sulfidic mineral soil in peat extrac-
tion sites. Geophys. Res. Abstr. 21, EGU2019-8066.
Österholm, P., Åström, M., 2004. Quantification of current and future leaching of sulfur
and metals from boreal acid sulfate soils, western Finland. Aust. J. Soil Res. 42,
547–551.
Purokoski, P., 1959. Rannikkoseudun rikkipitoisista maista. Referat: Ûber die
schwefelhaltigen Böden an der Küste Finlands. 74. Agrogeological Publ, Helsinki,
Finland.
Regina, K., Syväsalo, E., Hannukkala, A., Esala, M., 2004. Fluxes of N2O from farmed peat
soils in Finland. Eur. J. Soil Sci. 55, 591–599.
Regina, K., Pihlatie, M., Esala, E., Alakukku, L., 2007. Methane fluxes on boreal arable soils.
Agric. Ecosyst. Environ. 119, 346–352.
Regina, K., Heikkinen, J., Maljanen, M., 2019. Greenhouse gas fluxes in agricultural soils in
Finland. In: Shurpali, N., Agarwal, A.K., Srivastava, V.K. (Eds.), Greenhouse Gas Emis-
sions: Challenges, Technologies and Solutions. Book Series Energy, Environment and
Sustainability. Springer, pp. 7–22.
Richards, L.A., 1954. Diagnosis and improvement of saline alkali soils. Agriculture Hand-
book. 60. US Department of Agriculture, Washington DC.
Ronkanen, A.-K., Marttila, H., Celebi, A., Kløve, B., 2016. The role of aluminium and iron in
phosphorus removal by treatment. Ecol. Eng. 86, 190–201.
Saarela, I., 2008. Estimating bioavailable reserves and potential leaching of soil P by sim-
ple chemical tests. NJF Seminar PhosphorusManagement in Nordic-Baltic Agriculture
– Reconciling Productivity and Environmental Protection. NJF Reports. 4, 4 ,
pp. 166–170. https://orgprints.org/15044/1/njffosfor.pdf.
Sandmeier, K.J., 2016. ReflexW (Computer Software). Sandmeier Software . https://www.
sandmeier-geo.de/.
Scharlemann, J.P.W., Tanner, E.V.J., Hiederer, R., Kapos, V., 2014. Global soil carbon: under-
standing and managing the largest terrestrial carbon pool. Carbon Manag. 5, 81–91.
SFS 3005: 1981, 1981. Veden alkaliteetin ja asiditeetin määritys. Potentiometrinen titraus.
(Determination of the Alkalinity and Acidity of Water. Potentiometric Titration).
Finnish Standards Association.
SFS 3021:1979, 1979. Veden pH-arvon määritys. (Determination of the pH of Water).
Finnish Standards Association.
SFS-EN 1484: 1997, 1997. Water Analysis. Guidelines for the Determination of Total Or-
ganic Carbon (TOC) and Dissolved Organic Carbon (DOC). Finnish Standards Associ-
ation.
SFS-EN 27888:en, 1994. Water Quality. Determination of Electrical Conductivity (ISO
7888:1985). Finnish Standards Association.
SFS-EN ISO 11732:en, 2005. Water Quality. Determination of Ammonium Nitrogen.
Method by Flow Analysis (CFA and FIA) and Spectrometric Detection (ISO 11732:
2005). Finnish Standards Association.15SFS-EN ISO 11885:2009, 2009. Water Quality. Determination of Selected Elements by In-
ductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) (ISO 11885:
2007).
SFS-EN ISO 13395:en, 1997. Water Quality. Determination of Nitrite Nitrogen and Nitrate
Nitrogen and the Sum of Both by Flow Analysis (CFA and FIA) and Spectrometric De-
tection (ISO 13395:1996). Finnish Standards Association.
SFS-EN ISO 15681-1:2005, 2005. Water Quality. Determination of Orthophosphate and
Total Phosphorus Contents by Flow Analysis (FIA and CFA). Part 1: Method by Flow
Injection Analysis (FIA) (ISO 15681-1:2003).
Šimek, M., Virtanen, S., Krištůfek, V., Simojoki, A., Yli-Halla, M., 2011. Evidence of rich mi-
crobial communities in subsoil of boreal acid sulfate soil conducive to greenhouse gas
emissions. Agric. Ecosyst. Environ. 140, 113–122.
Sippola, J., 1974. Mineral composition and its relation to texture and to some chemical
properties in Finnish subsoils. Ann. Agric. Fenn. 13, 169–234.
Soil Survey Staff, 2014. Keys to Soil Taxonomy. 12th ed. USDA-Natural Resources Conser-
vation Service, Washington, DC.
Soini, S., Virri, K., 1968. Soil map of Oulu-Liminka. Ann. Agric. Fenn. 7 (Suppl. 2).
Soveri, J., Varjo, M., 1977. Roudan muodostumisesta ja esiintymisestä Suomessa vuosina
1955-1975. English summary: On the formation and occurrence of frost in Finland
1955-1975. Publ. Water Environ. Res. Inst. 20.
Syväsalo, E., Regina, K., Turtola, E., Lemola, R., Esala, M., 2006. Fluxes of nitrous oxide and
methane, and nitrogen leaching from organically and conventionally cultivated sandy
soil in western Finland. Agric. Ecosyst. Environ. 113, 342–348.
Tattari, S., Koskiaho, J., Kosunen, M., Lepistö, A., Linjama, J., Puustinen, M., 2017. Nutrient
loads from agricultural and forested areas in Finland from 1981 up to 2010 - can the
efficiency of undertaken water protection measures seen? Environ. Monit. Assess.
189 (95). https://doi.org/10.1007/s10661-017-5791-z.
Tikkanen, M., Oksanen, J., 2002. Late weichselian and holocene shore displacement his-
tory of the Baltic Sea in Finland. Fennia 180, 9–20.
Tubiello, F.N., Biancalani, R., Salvatore, M., Rossi, S., Conchedda, G., 2016. A worldwide as-
sessment of greenhouse gas emissions from drained organic soils. Sustainability -
Basel 371 (8). https://doi.org/10.3390/su8040371 13 pp.
Turtola, E., Jaakkola, A., 1986. Viljelykasvin, lannoituksen ja sadetuksen vaikutus kaliumin,
kalsiumin, magnesiumin, natriumin, sulfaattirikin sekä kloridin huuhtoutumiseen
savimaasta. Maatalouden tutkimuskeskus. Tiedote 17/1986.
Turtola, E., Paajanen, A., 1995. Influence of improved subsurface drainage on phosphorus
losses and nitrogen leaching from a heavy clay soil. Agric.Water Manag. 28, 295–310.
Urvas, L., 1976. Soil map of Ruukki-Lumijoki. Ann. Agric. Fenn. 15 (Suppl. 1).
Vuorinen, J., Mäkitie, O., 1955. The Method of Soil Testing in Use in Finland. 63.
Agrogeological Publ, Helsinki, Finland, pp. 1–14.
Yli-Halla, M., Puustinen, M., Koskiaho, J., 1999. Area of cultivated acid sulfate soils in
Finland. Soil Use Manag. 15, 62–67.
Yli-Halla, M., Virtanen, S., Mäkelä, M., Simojoki, A., Hirvi, M., Innanen, S., Mäkelä, J.,
Sullivan, L., 2017. Abundant stocks andmobilization of elements in boreal acid sulfate
soils. Geoderma 308, 333–340.
Yli-Halla, M., Virtanen, S., Regina, K., Österholm, P., Ehnvall, B., Uusi-Kämppä, J., 2020. Ni-
trogen stocks and flows in an acid sulfate soil. Environ. Monit. Assess. 192 (751).