INITIAL NUTRIENT RETENTION CAPACITY
IN A CONSTRUCTED WETLAND
Evaluating the effectiveness of a newly constructed wetland to reduce eutrophication symptoms in a Baltic Sea bay in northern Sweden
Elin Eriksson Supervisor: Jenny Ask
Abstract
Since the turn of the last century, a substantial increase in nutrient load to the Baltic Sea is apparent. Adding the ongoing environmental change with raising temperatures and increased precipitation, this will continue to have a prominent environmental impact on our coastal ecosystems, especially in northern latitudes. Constructed wetlands are becoming more important as a mitigation measure to retain nutrients, however, they are until this day not well studied in northern latitudes. In this paper, nutrient retention in a newly constructed wetland is studied during its first month after activation, as well as potential downstream effects in associated sea bay. An additional literature study compiles information about the current knowledge, use and functionality of wetlands surrounding the Baltic Sea. This is done to widen knowledge regarding effectiveness of wetlands as nutrient traps in general, as well as to compare with the studied wetland. A net retention of 30 % for dissolved organic carbon (DOC) and total phosphorus (TP) was found, as well as 27 % for total nitrogen (TN), 25 % for phosphate (PO43-) and 21 % for nitrate (NO3-). TP was found to be within range of expected retention capacity, when comparing with wetlands included in the synthesis. TN retention, however, seemed to be somewhat greater than in other wetland studies. Furthermore, the retention varied and seemed to be highest during an increased discharge, in the beginning and end of March. This was partly reflected by greater inlet concentrations and transports in most of the parameters during the initial time period. Decreasing temporal trends was seen in concentrations of DOC, total nutrients and NO3- concentrations in the sea bay, indicating an immediate downstream effect of the wetland installation. Findings from the synthesis indicate that there are very few studies in, and thus little knowledge about, wetlands in northern climate. Overall, the results from the pioneer northern wetland in Sörleviken suggest that net retention is possible during its first month post-activation.
Key words: retention, constructed wetland, total phosphorus, total nitrogen, nutrient dynamics, northern latitudes, nutrient mitigation measure
Table of contents
1 Introduction ... 1
1.1 Purpose and hypotheses ... 3
2 Method ... 4
2.1 Site description ... 4
2.2 Experimental setup... 4
2.3 Water sampling and chemical analysis ... 6
2.3.1 Nutrient and DOC sampling ... 6
2.3.2 Water velocity measurements ... 7
2.3.3 Physiochemical measurements ... 7
2.4 Hydrochemical calculations and statistical analysis ... 7
2.5 Literature study ... 8
3 Results ... 8
3.1 Physicochemical properties of the study area ... 8
3.2 Concentrations... 10
3.3 Transports... 13
3.4 Retention ...14
3.5 Literature study ... 15
4 Discussion ... 20
5 Conclusions ... 24
Acknowledgements ... 25
References ... 26
Appendix A ... 29
Appendix B ... 30
Appendix C ... 34
Appendix D ... 35
Appendix E ... 36
Specific terms
Relative retention: the amount of P or N that is retained in the wetland in relation to the total amount flowing in, expressed in percent (%).
Specific retention: the amount of P or N that is retained per area unit in the wetland, expressed in kg per ha wetland area and year (kg * ha-1 * yr-1).
Hydraulic load: waterflow per area (Q/A), expressed as average runoff in the area (Q, m3 * day-1) divided by wetland area (A, m2).
1 Introduction
The Baltic Sea is a semi-enclosed brackish water located in the north-east of Europe, surrounded by highly industrialized countries. The catchment area is four times larger than the sea area, within which 85 million people are living and thus influencing the water discharging into the sea. As the only inlet is located in between Denmark and southern Sweden, its salinity levels, oxygen levels and species richness naturally differ (Mårtensson, 2021b), giving it a north-south biogeochemical gradient. Since the turn of the last century, the amount of phosphorus (P) and nitrogen (N) in the Baltic Sea has increased by 8 and 4 times, respectively (Mårtensson, 2021a). This has caused eutrophication symptoms in many areas and has had a significant impact on the marine ecosystems.
Agriculture is identified as the greatest contributor to the eutrophication of the Baltic Sea (HELCOM, 2018). P and N originating from fertilizers, leaches from the farmlands into nearby waterways, to eventually be discharged in the coastal area. This induces algal blooms, oxygen depletion and a decreased biodiversity among macrophytes (Bokn et al., 2002). Different actions to reduce eutrophication symptoms has been tested in the southern parts of the Baltic Sea (Johannesson, 2011; Kumblad & Rydin, 2019; Naturvårdsverket, 2006) while they are rare in the northern parts. However, according to HELCOM (2018) latest assessment of eutrophication in the sea, the Quark, the Bothnian Bay and the Bothnian Sea all hold nutrient levels (total P and total N) exceeding threshold values for ‘not good status’. This calls for a considerable increase in measures to reduce eutrophication symptoms in the northern parts of the Baltic. Even more so, since research indicates a stronger environmental impact from climate change and rising temperatures on higher latitudes (Collins et al., 2013). This could potentially increase microbial activity and precipitation, with successive elevated nutrient loads in our coastal waters.
To decrease nutrients leaching from agricultural land to the sea, different measures can be implemented depending on local prerequisites. Common measures include;
• structure liming of the soil, which improves plant uptake of P instead of leakage (Bergström et al., 2015);
• covered drainage ditches with added lime to bind the P (Wesström et al., 2017);
• two-stage-ditches, enabling different water flow paths depending on hydraulic load, which lowers the water velocity, thus increasing particle and nutrient retention (Davis et al., 2015);
• sedimentation ponds, with or without a constructed well with a lime-filter, to lower the amount of dissolved P in the water (Uusi-Kämppä et al., 2000);
• constructed wetlands. The wetlands are almost exclusively bigger than the sedimentation ponds and without wells but uses similar nutrient removal processes (Rocksén et al., 2019).
Between 2000-2019, 1236 wetlands were created or restored in Sweden, where the main or one of the main purposes were nutrient removal from agricultural runoff water (SMHI, 2021a).
P and N are transported from agricultural lands in dissolved form (most common forms are phosphate (PO43-) and nitrate (NO3-) or bound to soil particles (Jordbruksverket, 2010). From the fields, nutrients are either transported through surface runoff or leached through the ground to a ditch or drainage pipe. This water can then be directed through a nutrient- retaining wetland, such as the one in the current study.
The study site, Sörleviken, is located in northern Sweden, in the municipality of Kramfors and in the Bothnian Sea. It is included in a planned marine reserve in Gaviksfjärden and is currently showing symptoms of high nutrient load and eutrophication (Wibjörn & Hallén, 2008). The
most recent environmental quality classification of the water in Sörleviken stated bad status (class 5) with respect to phytoplankton, and not satisfying (class 4) with respect to nutrients (P and N) (VISS, 2019). To reduce the leakage of nutrients to Sörleviken, two wetlands were created in the early spring of 2020. They were allowed to vegetate and settle during the following summer and autumn and were connected to the ditch draining the surrounding agricultural land in the early spring of 2021. The larger of the two wetlands is investigated in this study and constructed with a small island benefiting birdlife. Except retaining nutrients, it also is also expected to promote fish migration and thus serve as a multifunctional wetland (Rocksén et al., 2019).
There is research and recommendations from responsible authorities indicating optimal wetland design to achieve a high nutrient removal efficiency, depending on the prioritized purpose being P or N removal. Sediment burial has been reported to be one of the most important P removal processes (van Helmond et al., 2019). Jordbruksverket (2010) state the construction of a deep area in the beginning of the wetland, where the agricultural runoff water enters, followed by a shallow, vegetated area as recommendations for P-filtering ponds. This will lower the water velocity, mitigating sedimentation of bigger particles in the deep part and letting smaller particles and dissolved nutrients to settle in the shallow area. The most important process of N removal in wetlands is nitrification-denitrification (when microbes transform NO3- to N2 or N2O), a process that is mitigated by a warmer climate, where microbial processes are possible during a longer period of time (Kyllmar & Aronsson, 2019). According to Jordbruksverket (2010) wetlands optimal for N removal are large, shallow and have a high sun radiation, contradicting optimal conditions for P removal. However, sedimentation and burial are reported to be an important removal process especially for organic N (Taylor et al., 2005).
Koskiaho and Puustinen (2005) show that a higher filtering capacity is expected for both N and P, the larger the wetland area (W) in relation to catchment area (C). A higher W:C ratio lowers the inflowing water volume per area wetland, thus increasing water residence time, giving soil particles more time to settle. Jordbruksverket (2010) state that the size of the wetland should be minimum 0,1-0,5% of the catchment area. However, for P, high retention has also been found in small wetlands (W:C < 1%) where a high amount of the incoming P was in particulate form (Braskerud, 2002). The angle of the edges is another important design factor influencing the retention. An inclination of 1:3 is preferred to avoid erosion and increase the area of the wetland at high flows, which in turn lowers the water velocity and promote retention (Bioforsk, 2008).
Catchment characteristics, such as soil type, has been proven to play an important role in wetland P removal. Fine textured soils containing high amounts of clay and especially silt, have the highest surface runoff of P since they easily erode (Jordbruksverket, 2010). However, in what way it affects P retention in wetlands is unclear and there are studies showing contradictory results. Johannesson (2011) report that more fine soil types gave a lower P accumulation in wetland sediments, since smaller particles have a longer sedimentation rate and thus a higher probability of being transported downstream. Braskerud, Lundekvam and Krogstad (2000) on the other hand, report the opposite relationship. This can probably be explained by aggregation of clay particles, which enhances their settling velocity. Other catchment characteristics such as livestock density (more livestock gives higher retention), P content in catchment soils and average slope (higher slope gives higher risk of erosion and more accumulation) have been linked to accumulation of both P and particles (Johannesson, 2011). Hence, catchment characteristics and wetland placement within a catchment is decisive in wetland P retention.
In addition to external processes, there are several internal processes affecting the retention in a wetland. One of them is a the density vegetation coverage, mainly because it increases the sedimentation. In a four-year Norwegian study, the vegetation coverage increased from 5 to 80% during the experiment and they reported a prominent increase in sedimentation during this time (Braskerud, 2001). It is not due to plant uptake of nutrients, which has been shown to only correspond to 3 % of the total retention in the wetland when assuming an average uptake of 1 g P m-2 (Jordbruksverket, 2010), but because the vegetation prohibits sediment resuspension. The plant roots bind the sediment, as well as create a top layer of plant remains.
However, as the vegetation cover exceeds approximately 50 %, factors such as sediment and hydraulic load becomes more important for wetland retention (Braskerud, 2001; Braskerud et al., 2000). In a study by Johannesson (2015), it is reported that also hydraulic load seemed to determine retention of P up to a certain breaking-point, until the water flow was too high for particles (and P) to settle.
Laakso, Uusitalo and Yli-Halla (2016) reported redox conditions in the bottom sediments to be an important factor determining P retention in created wetlands. It is crucial to maintain an oxygenated bottom sediment to avoid unnecessary P-leaching. Ferric iron (Fe3+) is one of the most common P-binding agents, along with aluminium (Al) compounds. When oxygen levels sink too low, Fe3+ will be reduced to ferrous iron (Fe2+) which lacks the P-binding capacity. This leads to a mobilisation of P into the water column which might be transported downstream. The process is not common, but might happen especially during low flow periods, risking the wetlands to become sources of P instead of sinks (Jordbruksverket, 2010). If the wetland sediment is high in Al-complexes however, it indicates a high P-sorption even during anoxic conditions which can thus mitigate the effects of Fe-reduction. Furthermore, sulphur has also been shown to affect the sediment sorption-capacity, by forming complexes with Fe and lowering the P sorption capacity of the wetland sediments (Laakso et al., 2016).
Concerning N removal, sufficient oxygen levels and high pH (7,5-8,0) will favour nitrification, while anoxic conditions increase denitrification (Bastviken, 2006).
Even though studies have been made concerning retention of created wetlands, there is still insufficient knowledge about the mechanisms of retention. Detailed data on temporal and seasonal variations and their underlying processes is lacking, especially for the northern regions surrounding the Baltic Sea. There are also difficulties in finding elaborate data on nutrient dynamics in and downstream wetlands immediately following activation. Several longer studies located in the south report net retention during their study periods, although periods with negative retention was discovered (Braskerud et al., 2005; Johannesson, 2011).
A closer look at wetland function throughout its lifetime will yield important insight on how improvements can be made to enhance wetland efficiency in general, and studies in northern climate will let us acquire knowledge about the need to differentiate and optimize wetlands in different regions along the Baltic Sea coast. The current study seeks to add understanding to both fields of interest.
This master thesis is done in cooperation with the environmental consultant firm Sweco and the department of Ecology and Environmental Science (EMG) at Umeå University and financed by both.
1.1 Purpose and hypotheses
The purpose of this study is to investigate initial nutrient dynamics (of P, N and DOC) in a newly constructed wetland in Sörleviken. The hypotheses are that the nutrient leaching from the wetland will increase immediately after connecting the wetland to the stream and coastal area of Sörleviken, followed by a decrease. Wetland nutrient retention will be low or absent throughout the study period. The nutrient levels in Kåstaviken (the natural wetland located between the wetland and the sea bay) and Sörleviken will initially increase, to gradually
decrease to concentrations lower than pre-wetland construction. However, as the time span in this study is limited to a few weeks, a noticeable decrease in the nutrient levels might not be detected.
Additionally, a synthesis of similar projects around the Baltic Sea and their removal efficiency is done to 1) give an overview of the current use and knowledge concerning nutrient-filtering created wetlands, 2) compare obtained results from Sörleviken with other similar projects, and 3) widen the knowledge regarding effectiveness of wetlands as nutrient traps in general.
Together, the field experiment and synthesis will enable further development of created wetlands as a measure to decrease eutrophication, in both southern and northern parts of the Baltic Sea.
2 Method
2.1 Site description
Sörleviken is an elongated bay located within a semi-mountainous area bordering Gaviksfjärden outside Nordingrå, Kramfors municipality. The bay lies within the High Coast, an UNESCO World Heritage Site, and is also an appointed Baltic Sea Protected Area (BSPA).
The bay area covers 0,51 km2 and its catchment area 21 km2, within which the villages Sörle, Kåsta, Bäckland and Orsta are located. Agricultural (pasture and arable) land and forests are dominating the catchment, with a low proportion of wetlands, heaths and inland waters. The bay has for long been exposed to high nutrient loading originating from the surrounding cultivated land with consequently scattered oxygen depleted areas (Wibjörn & Hallén, 2008).
This is enhanced by the narrow connection to surrounding sea, which only grows smaller by the on-going land lift (Rocksén et al., 2019). Fresh water circulation is thus small, adding to the eutrophication of Sörleviken.
The wetland is located 750 m upstream the main inlet to Sörleviken (Fig. 1). There are two inlets into the wetland, of which the smaller one is connected to another constructed wetland 250m further upstream. Downstream the wetland outlet, water flows through a natural wetland, Kåstaviken, before it drains in the sea bay. The design of the wetland is approximately 65x85 m, with a small bird islet of 30x20 m in it. Across the northern part, where the inlets are, there is a deeper section of 0,6-1 m. Remaining area is about 0,3 m deep.
The area is located within the cold temperate climate zone which covers most of northern Sweden, with boreal forest as dominating forest type. Dominating soil types are moraine and surged soils of varying composition (postglacial sand, gravel, boulders etc). These are interspersed with fine textured clay-silt belts in the valleys (SGU, 2020). Yearly precipitation averages around 800-900 mm, and the average temperature is 3 0C (SMHI, 2020). Yearly runoff within the catchment area is around 450 mm and evapotranspiration 380 mm year-1 (SMHI, 2021b).
2.2 Experimental setup
Sampling was done at seven locations spread out in the constructed wetland and Kåstaviken (hereafter abbreviated “the wetlands”), the adjoining streams and in Sörleviken to monitor water flow, DOC (dissolved organic carbon) and nutrient dynamics (Fig. 1a-b). Nutrients and DOC were sampled in the inlets (V1A, V1B) and outlet (V2) of the wetland, in the outlet of Kåstaviken (K2) as well as in Sörleviken (S0). Pressure loggers were deployed in the water at the main inlet to the wetland (V1A), the wetland outlet and the outlet of Kåstaviken to continuously record water pressure every 10 min hour-1 day-1 during the entire sampling period. An additional pressure logger to measure air pressure were installed on land, close to V1A.
During each sampling occasion, water velocity measurements were taken at the same locations as the loggers. Additional physiochemical water properties were collected at deep points in the wetlands (V0, K0), as well as approximately 50 m outside the inlet to Sörleviken (S0). The first sampling round was done right before connecting the wetland to Kåstabäcken, and the second round the day following connection. After that, a total of 4 sampling rounds were made, with approximately one-week intervals. All in all, the sampling period covered 30 days; 1-30th of March.
Figure 1. Map of a) entire study site with the seven sampling locations (yellow triangles mark SeaGuard sampling points, filled red dots mark sampling points were both nutrients and water velocity measurement were made, and empty red dots where only nutrients were sampled), b) the created wetland with associated four sampling locations, and c) the area for the created wetland before construction. Note the differences in scale. © Lantmäteriet.
Figure 1. Continued from previous page.
2.3 Water sampling and chemical analysis 2.3.1 Nutrient and DOC sampling
In the five inlet and outlets, water samples were collected with a 5L bottle or a Ruttner sampler and analysed for total and dissolved nutrients (total nitrogen, total phosphorus, phosphate, nitrate, nitrite (NO2-), ammonium (NH4+)) and DOC. At Sörleviken, nutrients and DOC were only collected above the pycnocline, in the epilimnion. An ice cap was present during the entire study period at the deep points in the wetland (V0), Kåstaviken (K0) and Sörleviken (S0) and froze over between every sampling occasion. In the outlet of the wetland ice was present, although it kept open from the first sampling occasion onwards due to the flowing water beneath.
TC-flasks top-filled with MilliQ water were prepared beforehand for DOC-sampling. When taking nutrient and DOC samples, AcroDisc filters were attached to a 60 mL syringe, each syringe rinsed with half of the Milli-Q water (approximately 35 mL) found in one TC-flask. The syringe was then top filled with sample water and any air was avoided. All but 30 mL of the sample water was filtered into a falcon tube (for dissolved nutrients) or TC-flask (for DOC) and used for rinsing. Remaining 30 mL of sample was then filtered into respective container. 450 µL of 1.2M HCl was added to acidify the DOC samples and samples were kept dark and cold until analyses. A washed Scint vial was filled with approximately 20 mL of sample water without filtering for analyses of total nutrients. All nutrient samples were frozen within 10 hours of sampling.
All DOC and nutrient samples were analysed at the Swedac accredited lab of Umeå Marine Sciences Centre (Hörnefors, Sweden) by Annie Cox, in accordance with procedures described by (Karlsson et al., 2002).
2.3.2 Water velocity measurements
Water and air pressure were recorded at four locations by TD-Diver DI8xx dataloggers (divers) during the entire sampling period. Additional water velocity measurements were consciously taken at the same locations during each sampling occasion. Measurements were made at ten points in a transect across the stream, 20 and 80% below the water surface. The measurements were taken with a handheld 1-D electro-magnetic current meter for rivers, AEM1-DA, and calibrated before each sampling round.
2.3.3 Physiochemical measurements
A CTD probe (Seabird 19 Plus V2 SeaCat profiler, Sea-Bird Electronics INC.) recorded profiles on dissolved oxygen, temperature, turbidity, salinity and CDOM (coloured dissolved organic matter) at the deep points in the wetlands, as well as in Sörleviken. It took measurements in the entire water column, where depths were approximately 0,8 m in the wetland, and 2 m in both Kåstaviken and Sörleviken.
Ice and snow thickness were noted during all sampling occasions and locations.
2.4 Hydrochemical calculations and statistical analysis
Water velocity was translated to flow by calculating the area of each point measured in the transect, which was then multiplied with corresponding water velocity. A summarized flow was calculated for each location and sampling occasion. To estimate water flow for each day and sampling location during the study period, flow-pressure functions were established using linear regression (Appendix A) for each location (wetland in- and outlet and Kåstaviken outlet).
Average pressure for the day of each water velocity sampling was used in the functions. Next, extrapolation of water pressure to flow was made for the whole study period. Total pressure values collected by the divers were corrected for air pressure.
DOC and nutrient concentrations were interpolated to cover whole study period, assuming linear change between measurements. Nutrient transport was then calculated by multiplying average flow per day and concentration. Based on this, nutrient retention could be estimated.
Transportation and concentrations values for DOC, TP, TN, PO43- and NO3- were tested for statistically significant differences with a two-tailed Mann-Whitey U test, with an alpha-level of 0,05. The data were divided into three time periods, S1 (hereafter referred to as period 1), S2 (period 2) and S3 (period 3), excluding the day before activation. N-values were 10, 10 and 9, respectively. Each time period was tested against corresponding period in all sampling locations. Furthermore, linear regression analysis was performed on the same parameters to evaluate temporal dynamics in and downstream the wetland, and additionally, any gradient patterns.
Pressure data collected from the CTD probe was used to calculate the depth gradient for each sampling occasion. The initial fifteen pressure values at every sampling location were used to calculate an average air pressure (these were taken while the SeaGuard were above-ice). This was then subtracted from each measured pressure value and divided by the gravity constant.
Furthermore, profiles for temperature, O2 concentration, turbidity and CDOM were used to calculate averages for all sampling occasions and locations (Appendix B1 and B2). Some profiles in Sörleviken (S0) exhibited a clear stratification pattern (a pycnocline), whereupon averages were calculated above and below this (Appendix C). Data in the upper part of the water column, affected by the ice cap, were removed (which were 46, 50 and 40 cm for the wetland, Kåstaviken and Sörleviken, respectively). Outliers were removed, e.g., when the CTD
probe were suspected to have stirred the bottom sediment which contaminated the results. No further statistical analysis was performed on the CTD measurements since replicates were lacking.
2.5 Literature study
The literature used in the review was selected in two steps and included both peer-reviewed articles and grey literature. The scientific database Web of Science was used in the initial step, where a search string was set after a swift overview of current terms used within the area of topic. The search string can be found in Appendix D and generated 95 articles for screening.
Articles published 2000 and onward was used in the review and screened in two stages; first by title and abstract, followed by full text for those who were determined relevant in stage one.
In the first stage, articles were determined relevant if they indicated or reported calculated retention within the current study. Both primary and secondary sources were included, hence articles with a publishing date older than 2000 could be included if they were a part of a review.
Number of articles chosen in the first and second stage was 16 and 14 respectively.
Since wetland projects concerning eutrophication in the northern parts of the Baltic Sea was not represented within the article search in Web of Science, an extended search was carried out. Responsible persons at the County Administration Boards of Västernorrland, Västerbotten and Norrbotten were contacted and asked to report about implemented wetland projects (with nutrient retention as part of their purpose) that had been carried out in the past 20 years in each county. Additionally, the LONA (Naturvårdsverket, 2021), LOVA (Havs- och Vattenmyndigheten, 2021) and Wetland (SMHI, 2021a) databases were searched for reported projects within the above-mentioned counties.
3 Results
3.1 Physicochemical properties of the study area
The temperature was quite low during the sampling period and was on average 1,1, 1,1 and 1,2
°C in the wetland, Kåstaviken and Sörleviken (above the pycnocline), respectively (Table 1).
Oxygen concentrations indicate oxygenated conditions in all sampling locations and points in time (Havs- och Vattenmyndigheten, 2020). Although concentrations seem to be lower in Sörleviken than in the wetlands, particularly beneath the pycnocline, they are never below the limit for hypoxia (approximately ≤ 2 mgL-1) or anoxia (SMHI, 2017).
Turbidity patterns throughout the study period are similar in both wetlands and Sörleviken, where higher turbidity is found during the initial phase post-activation and towards the end of the study period (Table 1). In Kåstaviken, turbidity drops right after activation of the wetland, whereas it appears to increase in the other sampling locations. Lowest turbidity values are found in Sörleviken, beneath the pycnocline (0,9-2,3 FTU) and the highest are generally found in Kåstaviken (8,4-22,9 FTU). CDOM in Sörleviken surface water and both wetlands vary within similar concentrations, where the overall average (i.e., for the entire sampling period) is 56,2 µgL-1, 56 µgL-1 and 50 µgL-1 in the wetland, Kåstaviken and Sörleviken above the pycnocline, respectively (Table 1). The low CDOM concentrations (19,7-25 µgL-1) in the bottom water of Sörleviken appear to be relatively lower than the surface water. There is a distinct difference between salinity in the surface and bottom waters of Sörleviken, where the bottom salinity values seem to be higher than surface salinity and similar to levels measured in the surrounding sea water (SMHI, 2021c).
Table 1. Averages and standard deviations for each sampling occasion, as well as overall for the entire sampling period (March 2021), of physiochemical in the constructed wetland (V0), Kåstaviken (K0) and Sörleviken (S0).
Note that averages for the wetland and Kåstavikenwere made including data from the entire water column, while averages above and below the halocline are displayed for Sörleviken.
V0 March
01 March 02 March
08 March
15 March
22 March
30 Overall average
Ice thickness (cm) 55 45 54 40 50 32
Temperature (°C) 1,4±0,8 1,3±1,1 0,2±0,03 1±0,2 1±0,3 1,9±0,8 1,10,7 O2 (mgL-1) 130,3 12,20,3 12,40,2 12,20,1 12,40,1 12,20,2 12,30,3 Turbidity (FTU) 7,7±5,5 12±1,4 7,8±0,4 - 5,1±0,4 14,7±1,7 9,34,2 CDOM (µgL-1) 27,9±6,1 63,8±1,6 63,9±1,3 58,4±1,4 54,9±8,3 55,5±2,5 56,29,2 K0
Ice thickness (cm) 53 49 50 46 54 49
Temperature (°C) 1,3±0,6 2±1,5 0,7±0,3 1,1±1 0,9±0,7 1,2±1 1,11 O2 (mgL-1) 11,72,0 120,4 12,31,1 11,11,8 11,81,0 12,10,3 11,81,3 Turbidity (FTU) 22,9±1,7 14,3±3,7 9,4±5,2 8,4±4,5 9,6±6,4 16,9±4,2 11,86,6 CDOM (µgL-1) 50±3,8 56,1±6,8 53,8±11 55,5±8,5 57,1±2,9 56,6±1,6 565,9 S0
Ice thickness (cm) - - 40 40 42 29
Above pycnocline
Temperature (°C) 0,590,2 0,750,2 0,20,6 2,50,9 0,90,3 2,21,3 1,21,1 O2 (mgL-1) 11,60,1 11,40,0 121,4 10,81,8 - 11,30,5 11,50,8 Turbidity (FTU) 13,81,5 191,4 4,72,7 2,91,2 4,61,2 19,38 12,48,4 CDOM (µgL-1) 57,81,3 56,11,2 46,719 35,710,4 49,612,9 47,712,9 5012,9
Salinity (PSU) 0,00,0 0,00,0 - 1,61,4 0,30,8 0,91,6 0,51,2
Below pycnocline
Temperature(°C) 2,30,5 3,10,4 2,10,3 1,70,1 2,30,6
O2 (mgL-1) 7,31,1 81,1 - 10,90,4 9,11,8
Turbidity (FTU) 1,50,1 10,2 0,90,3 2,30,5 1,30,7
CDOM (µgL-1) 251,3 21,21,4 19,71,4 211,4 20,92
Salinity (PSU) - 4,30,7 4,10,8 4,70,2 4,30,7
Discharge in the in- and outlet of the wetland, as well as in the outlet of Kåstaviken show a similar pattern but different magnitudes throughout the study period (Fig. 2a). However, there is a distinct increase in the outlet of the wetland on the 8th of March, which is not reflected in any of the other locations. The sudden increase results in discharge of the same level as in the inlet, whereas previous discharge was clearly lower. The outlet of Kåstaviken exhibits a greater discharge than the upstream points throughout the study period. The average discharge in the in- (184 L s-1) and outlet (150 L s-1) of the wetland are slightly lower than the average discharge during March, based on measurements done by SMHI (2021b) between 2004-2020. In the outlet of Kåstaviken (264 L s-1), the average discharge during the study period display levels in the expected range compared to the same reference. Average air temperature in March is 0,2 C°, with daily average temperatures above 0C° (Fig. 2b) in the beginning and end of the study period. Total precipitation during March is 15 mm, of which most (67 %) falls between 11-12th of March (Fig. 2c).
3.2 Concentrations
In the constructed wetland, concentration of TP in the inlet is higher (Table 2; Fig. 3b) than in the outlet during time period 1 and 2. Greater concentration of PO43- and TN is also exhibited during time period 1 in the inlet compared to the outlet of the wetland (Fig. 3c-d).
In Kåstaviken, greater TP and PO43- concentrations are displayed in the outlet compared to the inlet during time period 2 (Table 2; Fig. 3b, d). Yet, in period 3, PO43- concentrations switches to displaying greater inlet concentrations. During period 1 and 2, TN concentration is greater in the outlet than the inlet (Fig. 3c). However, the inlet NO3- and DOC concentrations are higher than in the outlet during period 2 (Fig. 3a, e).
Between the wetland inlet and the Kåstaviken outlet, greater TP and NO3- concentrations are displayed in the wetland inlet during time period 2 (Table 2; Fig. 3b, e). PO43- display greater concentrations in the wetland during period 1 (Fig. 3d).
When comparing concentrations in the wetland inlet with Sörleviken, TN and NO3-
concentrations are consistently greater in the wetland inlet (Table 2; Fig. 3c, e). DOC concentration is higher in the wetland inlet compared to Sörleviken during time period 1 and 2. Yet, both TP and PO43- exhibits higher concentration in Sörleviken compared to the wetland inlet (Fig. 3b, d), during period 2 and the two last time periods, respectively.
Figure 2. a) Discharge in in- and outlet of the wetland as well as the outlet of Kåstaviken, b) air temperature at the study site, where the red line denotes daily average, and c) precipitation during March collected from nearby weather station (SMHI, 2021d).
Between the outlet of Kåstaviken and Sörleviken, same pattern as for the wetland inlet and Sörleviken is found for all parameters.
An upstream-downstream regression analysis on each time period, where the main inlet to the wetland (V1A) is the uppermost sampling location and Sörleviken the lowermost, shows decreasing downstream concentrations for DOC, TN and NO3- throughout the study period (Appendix E1). TP concentration is initially decreasing in a downstream gradient. However, during period 2 there is an increase in TP concentration downstream, which during time period 3 disappears and no trend is detected. PO43- concentration increases throughout the study period.
Table 2. Results (U-, p- and r-values) from statistical tests (two-tailed Mann-Whitney U tests) for DOC, TP, TN, PO43- and NO3-concentration, significant values are marked in bold (p≤0,05, U≤23 (n=10) U≤17 (n=9)). S1-3 denotes the groupings of data, where S1 is the initial time period of the field experiment, S2 the middle and S3 the concluding period. S1 and S2: n1=n2=10, S3: n1=n2=9. For more information, see Method.
V1A vs. V2 DOC TP TN
U p r U p r U p r
S1 40 0,45 0,17 20 0,02 0,51 12 0,00 0,64
S2 31 0,15 0,32 6 0,00 0,74 35 0,26 0,25
S3 27 0,23 0,28 28 0,27 0,26 38 0,83 0,05
V2 vs. K2
S1 26 0,07 0,41 46 0,76 0,07 14 0,01 0,61
S2 17 0,01 0,56 17 0,01 0,56 1 0,00 0,83
S3 27 0,23 0,28 36 0,69 0,09 36 0,69 0,09
VA1 vs. K2
S1 34 0,23 0,27 29 0,11 0,35 38 0,36 0,20
S2 25 0,06 0,42 10 0,00 0,68 28 0,10 0,37
S3 36 0,69 0,09 36 0,69 0,09 39 0,89 0,03
VA1 vs. S0
S1 10 0,00 0,68 31 0,15 0,32 5 0,00 0,76
S2 6 0,00 0,74 0 0,00 0,85 0 0,00 0,85
S3 23 0,12 0,36 28 0,27 0,26 0 0,00 0,84
K2 vs. S0
S1 16 0,01 0,57 47 0,82 0,05 6 0,00 0,74
S2 8 0,00 0,71 0 0,00 0,85 0 0,00 0,85
S3 27 0,23 0,28 36 0,69 0,09 0 0,00 0,84
V1A vs. V2 PO3 NO3
U p r U p r
S1 1 0,00 0,83 46 0,76 0,07
S2 24 0,05 0,44 37 0,33 0,22
S3 54 0,05 0,47 36 0,69 0,09
V2 vs. K2
S1 31 0,15 0,32 38 0,36 0,20
S2 22 0,03 0,47 17 0,01 0,56
S3 15 0,02 0,53 40 0,96 0,01
VA1 vs. K2
S1 9 0,00 0,69 34 0,23 0,27
S2 26 0,07 0,41 23 0,04 0,46
S3 36 0,69 0,09 40 0,96 0,01
VA1 vs. S0
S1 39 0,36 0,20 8 0,00 0,71
S2 0 0,00 0,85 0 0,00 0,85
S3 9 0,00 1,40 0 0,00 0,84
K2 vs. S0
S1 24 0,05 0,44 13 0,01 0,63
S2 0 0,00 0,85 0 0,00 0,85
S3 2 0,00 0,80 0 0,00 0,84
Regression analyses on DOC, TN, TP, PO43- and NO3- concentrations for each sampling location over the entire study period shows decreasing temporal trends for all parameters in Sörleviken, except for PO43- (Appendix E2). DOC is also decreasing in all other locations. Except for Sörleviken, there are no temporal patterns in TN concentration. However, NO3- concentrations are increasing in the outlet of Kåstaviken. Lastly, PO43- concentrations are decreasing in the wetland inlet and outlet, as well as in the Kåstaviken outlet.
Figure 3. Temporal variation in concentrations of a) DOC, b) TP c) TN d) PO43- and e) NO3- in all sampling points during the entire study period. Dotted vertical line highlights time of wetland activation.
3.3 Transports
Transport of all parameters in the inlet is higher (Table 3; Fig. 4) than in the outlet of the wetland during period 1. However, during the following time period (2), DOC transport changes to a larger output than input from the wetland (Fig. 4a).
In Kåstaviken, the transport of DOC, TP and TN is greater in the outlet than in the inlet (Table 3; Fig. 4a-c) throughout the study period. PO43- and NO3- display greater transport in the outlet during study periods 1 and 2 (Fig. 4d-e).
When comparing the wetland inlet and outlet of Kåstaviken, DOC transport is higher in the outlet throughout the study period (Fig. 4a). Furthermore, TN, PO43- and NO3- transport is higher in the outlet during time period 1 and 2, and TP during period 2 (Fig. 4b-e).
Table 3. Results (U-, p- and r-values) from statistical tests (two-tailed Mann-Whitney U tests) for DOC, TP, TN, PO43- and NO3- transport, significant values are marked in bold (p≤0,05, U≤23 (n=10) U≤17 (n=9)). S1-3 denotes the groupings of data, where S1 is the initial time period of the field experiment, S2 the middle and S3 the concluding period. S1 and S2: n1=n2=10, S3: n1=n2=9. For more information, see Method.
V1A- vs.
V2
DOC TP TN
U p r U p r U p r
S1 16 0,01 0,57 12 0,00 0,64 8 0,00 0,71
S2 18 0,02 0,54 29 0,11 0,35 42 0,55 0,14
S3 33 0,51 0,16 30 0,35 0,22 34 0,57 0,14
V2 vs.
K2
S1 0 0,00 0,85 0 0,00 0,85 0 0,00 0,85
S2 0 0,00 0,85 0 0,00 0,85 0 0,00 0,85
S3 6 0,00 0,72 17 0,04 0,49 15 0,02 0,53
VA1 vs.
K2
S1 23 0,04 0,46 27 0,08 0,39 19 0,02 0,52
S2 0 0,00 0,85 0 0,00 0,85 0 0,00 0,85
S3 18 0,05 0,47 29 0,31 0,24 26 0,20 0,30
V1A vs.
V2 PO3 NO3
U p r U p r
S1 5 0,00 0,76 19 0,02 0,52
S2 39 0,41 0,19 29 0,11 0,35
S3 29 0,31 0,24 34 0,57 0,14
V2 vs.
K2
S1 0 0,00 0,85 0 0,00 0,85
S2 0 0,00 0,85 0 0,00 0,85
S3 18 0,05 0,47 22 0,10 0,39
VA1 vs.
K2
S1 21 0,03 0,49 21 0,03 0,49
S2 0 0,00 0,85 0 0,00 0,85
S3 32 0,45 0,18 30 0,35 0,22
Through regression analysis, an increasing temporal trend in DOC, TN and NO3- transport is seen in the outlet of the wetland during the study period (Appendix E3). Additionally, decreasing PO43- transport is seen in the outlet of Kåstaviken.
3.4 Retention
Up until around the 9th of March there is a net retention for all parameters, meaning there is a greater inflow than outflow of DOC and nutrients from the wetland (Fig. 5). NH4+ and NO2- are the sole compounds where retention increases shortly after activation. However, they initially display relatively low retention, why the increase results in similar retention capacity as for the other parameters during the first six days. One exception being TN, which exhibit a distinctly greater retention than the other parameters from shortly after activation up until the 13th of March.
Between March 11th and 22nd most of the parameters displays net emanation, indicating that there is a greater outflow than inflow of nutrients and DOC. However, some exceptions are seen during this period. Between the 18th and 20th all parameters increase, indicating net
Figure 4. Daily transport of a) DOC, b) TP c) TN d) PO43- and e) NO3-in sampling points V1A (inlet to wetland), V2 (outlet of wetland) and K2 (outlet of Kåstaviken) during the entire sampling period. Dotted vertical line highlights time of wetland activation.
From the 14th and 15th, TP and NO2-, respectively, displays net retention until the end of sampling. Residuary parameters change from emanation to retention between the 24th and 25th. Monthly relative retention is 30 % for both DOC and TP, 27 % for TN, 25 % for PO43- and 21 % for NO3-.
Figure 5. Temporal variation in relative retention of DOC and nutrients in the wetland during the study period. The grey horizontal line denotes where retention switches to emanation (i.e., inflow and outflow is equal). Dotted vertical line marks the time of wetland activation.
3.5 Literature study
The database Web of Science generated a total of 34 projects where retention has been investigated (Table 4), of which 9 are in Southern Finland, 15 in central and Southern Sweden and 10 in Southern and central Norway. However, not all studies are conducted in different wetlands, some presented projects are conducted in the same wetlands but during different stages of post-activation. Some review articles are found, which presents several projects within the same study. No studies found in the database is in the northern parts of Scandinavia.
Through the extended search focusing on the northern parts of Sweden, 10 additional objects are found. In the LONA (Lokala Naturvårdssatsningen) database and LOVA (Lokala Vattenvårdsprojekt) project catalogue, two objects are reported in Västerbotten with nutrient removal as their main purpose (Table 4). However, both are still under construction, thus no numbers on removal rates or retention are available. The same applies for the LONA project in Västernorrland. In Norrbotten, one project with a sedimentation pond is found through LOVA, which was finished in 2018 where retention capacity has been monitored (Table 4).
Additionally, through communication with personnel at the county administration in Västerbotten, two more projects are discovered, although no data on retention is available.
Lastly, four projects in the counties of Västernorrland, Västerbotten and Norrbotten are reported in the Wetland Database, of which three are in Västerbotten, and one in Västernorrland. As the total amount of projects reported in the database is 33 154, the wetlands reported in the coastal counties of northern Sweden amount to 0,0003 %.
-25 -5 15 35 55 75
Relative retention (%)
DOC TP TN PO4 NH4 NO2 NO3
