Analysis of phosphorus retention variations in constructed wetlands receiving variable loads from arable land

(1)

Linköping Studies in Science and Technology Thesis No. 1482

Analysis of phosphorus retention variations in constructed

wetlands receiving variable loads from arable land

Karin Johannesson

LIU-TEK-LIC-2011:21

Department of Physics, Chemistry and Biology Linköpings universitet, SE-581 83 Linköping, Sweden

(2)

Front cover: Genarp wetland, photo by the author. © Karin Johannesson 2011 ISBN: 978-91-7393-168-7 ISSN: 0280-7971 Printed by LiU-tryck Linköping, Sweden, 2011

(3)

Abstract

Seven wetlands, constructed on agricultural land in the south of Sweden, were investigated with respect to phosphorus (P) retention. The overall aim was to increase the understanding of P retention and find possible

explanations for the variations in retention that have been observed in previous studies. This was done by i) investigating P retention in wetlands receiving various water and P loads, ii) investigating the effect of variations in water flow on P transport, iii) comparing how well retention estimates based on water quality data agreed with measurements of the amount of P accumulated in the sediment.

Results showed that P retention was positive in all wetlands, but it was

variable (1–58 kg ha-1_yr-1_{) and months with negative retention were}

observed in nearly all wetlands. Such monthly negative retention coincided with i) high flow periods, when particulate P was either flushed straight through the wetlands or resuspended from the bottoms, and ii) warm low flow periods, in which case dissolved P was probably released from wetland sediments due to anoxic conditions.

The results from the two methods for estimating P retention differed. Based

on water quality data, the total P load during four years was 65 kg ha-1_and

the mean P retention 2.8 kg ha-1_yr-1_{, or 17% of the total P load. In contrast,}

the amount of P accumulated in the inlet zone alone amounted to 78% of the P load, and the P content in the upper sediment of the whole wetland area exceeded the P load with a factor four. This discrepancy showed the need to add studies of sediment accumulation to inflow-outflow estimates for an improved understanding of wetland P retention.

(4)

(5)

List of papers

The following papers are included in the thesis. They are referred to in the text by their Roman numerals.

I: Johannesson, K.M., K.S. Tonderski, P.M. Ehde & S.E.B. Weisner. Phosphorus load variations and retention in non-point source wetlands in southern Sweden. Manuscript.

II: Johannesson, K.M., J.L. Andersson & K.S. Tonderski. Efficiency of a constructed wetland for retention of sediment associated phosphorus. Hydrobiologia, in press.

(6)

Eutrophication is a problem in Swedish lakes and in the Baltic Sea, and can cause algal blooms, anoxia and dead bottoms with severe ecological and economical consequences. In freshwater and brackish systems, phosphorus (P) is often the limiting nutrient (Kalff, 2002). Eutrophication is caused by an excess of nutrients because of e.g. intensified land use and an increasing

population. Since the mid-19th_{century P inputs to the Baltic Sea have}

increased eightfold (Larsson et al., 1985). One reason for this increase is the population growth that has occurred in Sweden during the last 150 years – from approximately four million 1860 to nine million today. With a growing population, increases in the agricultural productivity were crucial, and the use of fertilizer mirrors this need. In 1930’s, mineral fertilizer use was less

than 5 kg P ha-1_{, and in the 70’s it had increased to 20 kg P ha}-1_{(Löfgren et}

al. 1999). However, the use of mineral P fertilizers has decreased since the 1990’s, mainly because of economical reasons (Barbro Ulén, personal comment). According to recent estimates, P loads from agriculture in the

south of Sweden was 156 ton P yr-1_{to the southern Baltic Sea, which}

represents 44 % of the total load of P from Sweden (Stolte et al., 2009). In addition to the increased use of fertilizers, naturally occurring buffer

systems, such as wetlands, have been drained since the middle of 19th

century in order to increase the agricultural areas. In southern Sweden, where the majority of agricultural land is found, the wetland area has decreased by 40-90% due to drainage (Jansson et al., 1994). When these natural ‘filters’ in the landscape disappeared, nutrients were transported more or less straight from the fields to the receiving water body.

Globally, the present use of P fertilizers is not sustainable. Mineral P used for fertilizer production is a finite resource, and a recent study showed that the available P reserves will last for 50-100 years with an expected global peak in P production around 2030 (Cordell et al., 2009). Hence, there is a need to use P more efficiently in agriculture and close the ‘P loop’. This can be achieved in various ways, e.g. plowing crop residues back into the soil; using food waste from households, food processing plants and food retailers for compost or biogas production; and using human and animal excreta as fertilizers (Cordell et al., 2009). However, some leakage of P will always occur from arable land, and constructed wetlands can capture some of the P

(8)

that is lost from the fields. The construction of wetlands has been advocated as one of several measures to reduce the transport of P from agricultural land, which is part of the Swedish Environmental Objective “No eutrophication” and of the Baltic Sea Action Plan (EPA 2009; HELCOM 2009). Through excavation and reuse of wetland sediments on farmland, some of the P can be recycled and the detrimental effects of excess P in sensitive waters, e.g. the Baltic Sea, may decrease.

Many studies have shown that constructed wetlands can function as sinks for P from non-point sources, but the retention efficiency is highly variable, and occasional releases of P have been observed (Jordan et al., 2003; Braskerud et al., 2005; Kovacic et al., 2006). A challenge when constructing wetlands for P retention is that we have incomplete understanding of how factors such as variable water flows, wetland design and location in the landscape affect P retention. This is crucial to obtain a cost efficient retention of P using constructed wetlands. Some of the difficulties lie in the gaps in knowledge regarding P behavior in catchment and wetlands, i.e. in which form P is transported, and also how different seasons affect P retention in wetlands.

Objectives of the study

The overall aim of this research was to increase the understanding of P retention and find possible explanations for the variations in retention that have been observed in previous studies. This was done by i) investigating P retention in wetlands receiving various water and P loads, ii) investigating the effect of variations in water flow on P transport, and iii) comparing how well retention estimates based on water quality data agreed with

measurements of the amount of P accumulated in the sediment of one wetland.

For Paper I, the aims were to i) estimate P retention on annual and monthly basis and ii) investigate the dynamics of inflow P concentrations in relation to water flow and different seasons for seven constructed wetlands. In Paper II, one wetland located in an area with clay soils was investigated in detail. The aims were to i) estimate the P retention and identify the dominating retention processes; and ii) investigate how well estimates of P retention based on inflow-outflow measurements compared with the amount of P accumulated in the sediment.

By collecting and studying data on P concentrations and P forms that enters and exits wetlands situated in different catchments, looking at different

(9)

water flows and different seasons as well as studying P accretion in the sediments of a constructed wetland, the outcome will be an improved understanding of P retention in constructed wetlands.

Phosphorus retention in constructed wetlands

P enters a wetland in inorganic and organic form, and in particulate and dissolved form (PP and DP, respectively). Dissolved inorganic P is

considered bioavailable, whereas organic and particulate P forms generally must undergo transformations to inorganic forms to be considered

bioavailable (Reddy et al., 1999). Wetland P retention can be defined as the result of a number of physical and biogeochemical processes leading to removal of P from the water column, and storage in a non-bioavailable form in the sediment. PP is retained by sedimentation as the water velocity drops when running water enters a wetland and particles can settle on the bottom. DP is retained by both chemical and biological processes. It can be sorbed to particles or form chemical precipitates with metal cations. Uptake of DP by biota is also an important retention process (Fig.1a) (Reddy et al. 1999). However, all the processes described above are reversible. For instance, particles that have settled on the bottom could be resuspended due to high flows or bioturbation by fish, birds and invertebrates. DP can be released from the chemical bonds should the chemical status (e.g. redox potential or pH) of the wetland change (Fig. 1b). Furthermore, even though biological uptake can be fast and effective, there is a temporal heterogeneity because of the different life cycles of the organisms. Most of the assimilated P is

released back into the water column after the death of the organism; according to Richardson (1985), 35-75% of the plant P is rapidly released again. The balance between the internal processes sedimentation and resuspension, adsorption and desorption, and biological uptake and

decomposition will determine whether the wetland is a sink or a source of P. The desired effect of constructed wetlands is long-term retention of P, i.e. the risk for leakage back to the water column should be minimized. In conclusion, several retention and release processes are involved in determining how efficient a wetland is in retaining P. Net P retention in wetlands is usually expressed in two different ways; specific (kg P retained per ha and year) and relative (% of total P load retained). Previous studies of 22 constructed wetlands from five countries have shown that the P retention was positive, i.e. more P entered the wetlands than what exited (Tab.1). The wetlands differed with respect to e.g. hydraulic load, size and inflow P

(10)

species and concentrations. The P retention efficiency varied between the wetlands, and retention of DP was sometimes negative.

Fig.1. A simplified picture of the different mechanisms for a) retention and b) release of particulate phosphorus (PP) and dissolved phosphorus (DP) in wetlands.

Methods of estimating P retention

A transport study can be used to assess the retention of P. Data on water flow and concentrations of inflow and outflow P is collected, and P retention estimated by subtracting the amount of P leaving the wetland from the amount entering. This is of course a simplified picture of P retention, since it only takes into account in–out and assumes that the difference represents retention by the processes described in Fig.1a. In reality, P retention is more complex, and involves internal processes that may contribute to the P export (Fig.1b). Performing a transport study is difficult not only theoretically, but also practically, since it is difficult to obtain accurate data on mass inflow and outflow of P. Automatic water sampling at the inlet and outlet often results in samples that are not accurate with respect to PP concentrations. The concentrations can be both over- and underestimated, as it is difficult to

(11)

capture the event-based movement of particles (Jarvie et al., 2002). For instance, Kronvang and Bruhn (1996) showed that in two Danish streams, TP transport was nearly always underestimated, especially for PP. Also, simplifications with respect to the water balance of constructed wetlands are usually made. For example, water flow is often measured only at one point

(inlet or outlet)– hence, not all inputs and outputs of P and water are

identified and measured (Reddy et al., 1999). This can have importance when calculating P transport, especially in low-loaded constructed wetlands. For instance, Kovacic et al. (2006) showed that in two low-loaded wetlands

(hydraulic loads of 11 and 16 m yr-1_{respectively) the outlet only represented}

64 and 68 % of the total water volume leaving the wetlands. The rest of the water was lost either by evaporation (7 and 6 %, respectively) or seepage (29 and 28 %, respectively). Hence, if mass transport data in such low-loaded wetlands are based on water flow measurements at the outflow, nutrient budgets will be inaccurate.

In addition, transport studies do not give information about internal release processes in wetlands such as resuspension, as mentioned above. Studies on net and gross sedimentation can provide information on resuspension processes, as was done by Braskerud (2001). Sedimentation plates

represented net retention and traps gross retention, and the difference hence represented the amount of sediment that was resuspended (Braskerud, 2006). In conclusion, it is difficult to obtain accurate transport data in wetlands, because of the problems with water budgets and water sampling. Studies of in- and out flowing masses of P in wetlands receiving agricultural runoff often show a positive retention (Tab.1). Some of the variations in P retention can probably be explained by differences in sampling technique and

estimates of water and P budgets. To assess the longevity of wetlands as P sinks more information about the different forms and mobility of the retained P is needed. A combination of water quality analyses and e.g. sedimentation studies could provide information closer to the truth.

(12)

Ta b. 1. P hos pho rus c onc ent ra tions , r et ent ion and w et land char ac ter is tic s f or 22 w et lands in the N or di c c ount ries , Sw itz er lan d and U .S , r ank ed ac cor di ng t o m ean i nc om ing tot al p hos pho rus (T P) c onc ent ra tions . ( n. d. = no da ta) Site Co un try Na m e Su rf ac e ar ea Aw /Ac ratio Cul tiv at ed H ydr aul ic loa d M ea n T P in TP re te nt io n D P/ TP D P re te nt io n Re fe ren ce (m 2) (%) (%) (m yr -1) (m g l -1) (kg ha -1 yr -1) (%) (%) (%) 1 Fi nl and Fl yt trä sk 600000 3 35 14 0. 067 2 15 14 10 (K os ki ah o et a l. 2 003 ) 2 Sw ed en Vä st er by 400 0. 01 64 4190 0. 07 39 1 19 3 (Ul én , 2 004) 3 U. S We tla nd 1 1600 4 n. d. 11 0. 1 10 68 35 78. 5 (K ov aci c et al ., 2006 ) 4 Fi nl and Al ast ro 4800 0. 53 90 47 0. 12 6 7 19 -22 (K os ki ah o et a l., 200 3) 5 U. S We tla nd B 3000 6 90 9 0. 12 2 23 100 36 (B ra ske rud et al . 2 005 ) 6 U. S We tla nd 2 4000 3 n. d. 16 0. 125 6 44 32 42. 5 (K ov aci c et al ., 2006 ) 7 Sw itz er lan d Son nh of 2350 1. 15 80 34 0. 14 11 23 90 51 (Re in ha rdt e t al ., 2 005 ) 8 U. S O RW RP 2 10000 n. d. n. d. 54 0. 164 52 58 11 n. d. (N ai rn & M its ch, 2000 ) 9 U. S O RW RP 1 10000 n. d. n. d. 54 0. 169 56 62 11 n. d. (N ai rn & Mi tsc h, 2000) 10 N or wa y Be rg 904 0. 06 17 620 0. 18 513 43 14 -3 (B ra ske rud et al ., 2005 ) 11 N or wa y Skut er ud 2300 0. 05 61 917 0. 19 272 16 29 n. d. (B ra ske rud et al ., 2005 ) 12 U. S We tla nd A 6000 4 90 15 0. 2 4 23 100 32 (B ra ske rud et al ., 2005 ) 13 N or wa y Fl at ab ek ken 885 0. 09 14 648 0. 22 373 27 38 0 (B ra ske rud et al ., 2005 ) 14 N or wa y K in n 347 0. 07 27 683 0. 35 578 29 15 0 (B ra ske rud et al ., 2005 ) 15 N or wa y G ra ut ho le n 843 0. 38 99 285 0. 39 462 42 20 8 (B ra ske rud et al ., 2005 ) 16 U. S We tla nd 2 73255 0. 32 100 45 0. 481 595 56 32 61 (M ayn ar d et a l., 2009) 17 Fi nl and H ovi 6000 5 100 7 0. 51 24 62 8 27 (K os ki ah o et a l., 200 3) 18 U. S We tla nd 1 23211 5. 2 100 70 0. 564 398 60 36 63 (M ay na rd et a l., 2 009 ) 19 N or wa y Li er 1200 0. 15 88 241 0. 6 269 20 34 n. d. (B ra ske rud et al ., 2005 ) 20 N or wa y Le irv ollb ek k 2000 1 90 124 1. 53 1562 83 92 89 (B ra ske rud et al ., 2005 ) 21 U. S Ta rd iff 6100 8.7 99 2 2. 15 18 88 4 n. d. (B ra ske rud et al ., 2005 ) 22 U. S Ba rn sta bl e 1 13000 9 82 6 n. d. 8 27 n. d. 18 (J or da n et al ., 2003 )

(13)

Factors affecting P retention in constructed wetlands

Catchment characteristics

Wetlands constructed on arable land are exposed to varying weather events which result in highly variable water flow and nutrient concentrations. In Scandinavia, the effect of different seasons probably influences P retention in wetlands. The highest runoff takes place during snow melt in spring and during heavy rainfall periods in the autumn (Fig.2). In contrast, summer runoff is usually very low and sometimes drops to zero, which can also

occur during the winter period if temperatures are below 0o_{C during a}

prolonged period. During low-flow periods the water in the wetlands become stagnant which alters the biogeochemical properties of the sediment and can affect P retention. The seasonal changes in water flow result in very variable P loadings to the wetlands, and to include seasonal variation into the evaluations of P transport would improve the understanding of different retention processes.

(14)

Wetland size in relation to its catchment (Awetland/Acatcment) has been shown to affect P retention efficiency, since this determines the amount of water that enters the wetland, i.e. the hydraulic load (Kadlec & Knight, 1996). There is

no ‘rule of thumb’ regarding the optimum Aw/Ac ratio. In Tab.1 data from

several wetland studies have been compiled, and the results show that when studying specific retention (expressed as kg P removed per hectare wetland and year) wetlands that are small in relation to their catchment area had

higher P retention (Fig.3a, R2_{=0.28, p=0.02). However, when studying the}

relative retention (expressed as a percentage of the total P load to the

wetland) the reverse relationship was found, i.e. larger wetlands had a higher

relative retention (Fig.3b, R2_{=0.23, p=0.04). There was no significant trend}

for relative DP retention (Fig.3c), although larger wetlands should theoretically have a higher DP retention efficiency due to the higher residence time which give more time for the biogeochemical processes involved in removing DP. For instance, Reinhardt et al. (2005) suggested that the residence time should not be lower than seven days to retain at least 50% of DP.

Fig.3. Relationship between wetland area in relation to catchment area (Aw/Ac) and a) absolute P retention, b) relative P retention and c) DP retention in 21 wetlands. Data from Tab.1.

The high specific retention in small wetlands shown by e.g. Braskerud (2002) was attributed to an efficient retention of suspended material and PP. Hence, the form in which P enters a wetland also affects retention, since PP is

(15)

retained primarily through sedimentation and DP through sorption or assimilation in biota (Fig.1). Several studies have shown that in runoff from agricultural fields in clay and silt dominated areas in Scandinavia, P is

transported predominately as PP (Uusitalo et al., 2000; Koskiaho et al., 2003; Uusitalo et al., 2003; Ulén, 2004). In Sweden, PP is sometimes transported as colloids which, because of their small size, do not sink to the bottom in ditches, drainage pipes and wetlands since the settling velocity is too low (Ulén, 2004).

In conclusion, various factors are known to affect P retention efficiency in constructed wetlands, such as flow variations, hydraulic load (related to

Aw/Ac), residence time and the form and amount of P transported from the

catchment. These factors all determine how P enters a wetland, but which factor that influences the P retention the most has not been clearly identified and is an important area for further investigations.

Wetland characteristics

In order to retain DP efficiently, a wetland needs to be large in relation to its catchment area (Fig.3c). Wetland construction in agricultural areas is often limited by space and wetlands in these areas are often small in relation to the catchment. Since P is often transported from arable land as PP, one of the most important retention mechanisms in such wetlands is sedimentation. Sedimentation occurs when P bound to particles and aggregates enters a wetland and the water velocity is reduced. Sedimentation is determined by the incoming water flow velocity and the size distribution of the particles. Sedimentation velocity can be predicted using Stoke’s law (1), which states that the settling velocity depends on the density of the particle, i.e. more dense (or larger) particles sink faster than less dense (or smaller) particles.

n PL PS g D v 18 ) ( 2 ₋ = (1)

Where v is the settling velocity of a spherical particle with diameter D and density PS, PL is the density of the fluid and n is the viscosity of the fluid. g is the gravitational force (Kroetsch & Wang, 2007). For example, according to Stoke’s law, it will take the coarsest clay particles (diameter 2 μm) approximately 88 h to sink 1 m (15°C). Hence, in order for sedimentation of clay particles to occur, the residence time in the wetland needs to be quite long. However, in a study of constructed wetlands in Norway, Braskerud (2003) showed that fine clay particles from arable land had sedimentation velocities similar to coarse clay or silt, indicating a high degree of

(16)

(2008). Since conditions differ between regions regarding runoff, erosion and proportion clay, Braskerud’s results may not be applicable for wetlands in other catchments. For example, Ulén (2004) has shown that a majority of particles in drain flow from a clay soil were colloids, and their settling

velocity was only 0.08 cm day−1_.

In addition to the difficulty in predicting clay particle sedimentation, particles on the bottom could be resuspended during high flow or through bioturbation (e.g. Croel & Kneitel, 2011). This could lead to a recycling of the P, and it could be re-used biologically (Reynolds & Davies, 2001). Hence, factors affecting sedimentation and resuspension of clay particles in constructed wetlands need to be further investigated.

In conclusion, wetlands are used in the agricultural landscape as a measure to reduce the transport of both nitrogen (N) and P to the sea, but the efficiency of wetlands as P traps in Swedish conditions have not been fully investigated (Bergström et al., 2007). Studies of 22 constructed wetlands showed that the P retention is often positive but also very variable (Tab.1). The reasons for these large variations are probably i) variations in water flow and P concentrations, ii) variations in wetland size in relation to the

catchment, and iii) variations in P speciation (PP or DP) due to variations in soil type and land use.

(17)

Summary of methods and results

Study sites

The wetlands included in this study are situated in the south of Sweden, and are all constructed on arable land, although the percentages of the

catchments that are used for agriculture differ between the sites (Tab.2). They were all excavated on mineral soils, and vary in size but are all relatively small compared to the catchment area (only Stene was >1% of catchment area). The wetlands were selected based on availability of time series of water flow and P concentration measurements.

Tab.2. Summary of the seven wetlands investigated in this study. *Mean clay content in municipality, data from SLU.

Construction

year Wetland area Awetland/Acatchment

Agricultural land in catchment Mean clay content* Wetland (ha) (%) Lilla Böslid 1991 0.41 0.06 93 16 Slogstorp 1997 0.65 0.07 85 16 Råbytorp 1992 0.75 0.2 100 15 Bölarp 2002 0.22 0.1 85 8 Genarp 1997 1 0.33 65 15 Edenberga 2001 0.22 0.23 100 8 Stene 2003 2.1 2.19 35 34

Water quality and transport estimates

Monitoring data on water flow and nutrient concentrations were obtained from Ekologgruppen (for Råbytorp, Genarp and Slogstorp), the Wetland Research Centre at Halmstad University (for Lilla Böslid, Bölarp and Edenberga) and WRS Uppsala AB (for Stene). In all wetlands, the water flow had been continuously measured for the sampling periods (for details, see Tab.2 in Paper I). Water flow was measured in the outlet in all wetlands except for Stene, where flow measurements were performed both at the inlet and outlet. Hydraulic loads were calculated for each wetland by dividing the

cumulative annual inflow (m3_{) by the area of the wetland (m}2_{), and varied}

between 7 and 725 m yr-1_{(Tab.3). Water samples for determining P}

concentrations – both total P (TP) and DP – were taken time or flow proportionally (Tab.2 in Paper I). Additional grab samples were taken regularly as supplements to the automatic sampling for Lilla Böslid, Bölarp and Edenberga. In Stene, Genarp, Råbytorp and Slogstorp grab samples were taken only when the automatic sampling failed for technical reasons.

(18)

Results from the grab samples revealed a great variation in inflow P

concentrations, both between wetlands and between different seasons in the individual wetlands. Flow-weighted mean concentrations varied between 38

μg l-1_{(Bölarp) and 244 μg l}-1_{(Stene). The higher concentration in Stene was}

probably explained by the higher clay content in the soils in that part of Sweden. Flow-weighted mean concentrations and mean clay content in the

area were correlated (R2_{=0.9, p=0.005). Generally, the inflow P}

concentrations were higher than the outflow concentrations for all wetlands. Exceptions occurred during low flow periods in the warm half of the year (Apr-Sep), when outflow TP concentration exceeded inflow in Slogstorp, Råbytorp, Edenberga and Stene (Fig.4 in Paper I). There were correlations between inflow TP concentrations and water flow (Tab.5 in Paper I), but the strength of the relationship varied between wetlands, and the trend was positive for some (Bölarp, Edenberga and Stene) and negative for some (Lilla Böslid, Slogstorp and Råbytorp). In Genarp there was no significant relationship between water flow and inflow TP concentration.

Fig.4. Proportion of particulate phosphorus for in- and outflow water in Stene wetland during three intensive sampling periods. (Fig.2 in Paper II).

In Stene, the results from analyses of grab samples during two high-flow and one low-flow period showed that PP was the dominant form both in the

(19)

inflow and outflow water (Fig.4), with PP constituting 69±19% and 78±13% of TP, respectively. At the inlet the proportion of PP was low during low-flow and high in high-low-flow periods, whereas at the outlet there was a constant and high proportion of PP in the water.

Daily TP transport was calculated for wetlands with time proportional sampling (Genarp, Råbytorp and Slogstorp), by multiplying daily water flow

(m3_d-1_{) with the TP concentrations in and out from composite samples (for}

approximately one week periods). For wetlands with flow proportional sampling (Lilla Böslid, Edenberga, Bölarp and Stene), TP transport was calculated for each sampling period, by multiplying TP concentration in the collected sample with the total amount of water measured by the logger in the wetland during the period. The P load, i.e. the total amount of P that

entered the wetlands, varied between 16 and 725 kg ha-1_yr-1 _{(Tab.3). The}

area specific P retention was calculated as the difference between inflow and outflow P, and summarized for individual years and months for each

wetland. Specific retention varied between wetlands (1-58 kg ha-1_yr-1_{), but}

also between individual years for each wetland. When investigating relationships between P retention and different wetland and catchment characteristics (from Tab.2 and 3), only the P load was significantly

correlated with the P retention (R2_{=0.9, p=0.005). Mean inflow TP}

concentrations were not correlated at all with the P retention, and neither

was percentage clay in the catchment, age of wetland or Aw/Ac ratio. There

was a tendency towards a positive relationship between P retention and

hydraulic load (R2_{=0.73, p=0.06). Release of P occurred during specific}

months in nearly all wetlands except Genarp and Edenberga (Fig.5).

Tab.3. Sampling period, hydraulic load, P load and retention (average ± standard deviation) for the seven wetlands included in this study. The wetlands are listed according to the P load.

Sampling period Flow weighted mean inflow TP concentration Hydraulic

load P load P retention Wetland (years) (ug l-1₎ _{(m yr}-1₎ _{(kg ha}-1_yr-1₎ _(%)

Lilla Böslid 3 162 437 625 ± 272 58 ± 54 9 Slogstorp 6 67 725 405 ± 146 49 ± 55 12 Råbytorp 9 112 128 144 ± 58 15 ± 28 11 Bölarp 2 38 285 112 ± 32 1 ± 0.7 1 Genarp 5 134 85 91 ± 21 27 ± 10 29 Edenberga 1.5 56 56 35 ± 23 13 ± 2 38 Stene 4 244 7 16 ± 10 3 ± 2 17

(20)

Fig.5. Monthly P retention and hydraulic load for the seven wetlands included in this study. Wetlands are grouped according to P load; high: Lilla Böslid and Slogstorp, medium: Råbytorp, Genarp and Bölarp, and low: Edenberga and Stene. (Fig.2 in Paper I).

(21)

Comparison of methods for estimating P retention

In one of the wetlands, Stene, sediment studies were performed to compare with water quality estimates (Paper II). For the sediment studies, the wetland was divided into five areas, representing the inlet, open area near the outlet, and the shallow areas dominated by Typha latifolia L. (Ts at a sewage pipe discharge, T1 near the inlet and T2 near the outlet; Fig.1 in Paper II). Six sampling sites were randomly selected in each area. In each site, two sub-samples were collected using a core sampler. For each sample, the thickness of the accumulated sediment was measured, and additional measurements of sediment thickness were done in the shallow areas. The volume of the sediment in T1, T2, Ts and the outlet area was calculated by multiplying the respective surface area with the mean sediment thickness. For the inlet, P content of each sediment sample was first calculated, and the mean was multiplied with the inlet surface area. Bulk density and TP content was analyzed in all sediment samples.

Sediment thickness was more than four times higher at the inlet than in the other parts of the wetland, and sedimentation rate in this area was estimated

to 22 kg m-2_yr-1_{. Water quality analyses and flow measurements showed that}

the average P retention was 2.8 kg ha-1_yr-1_{or 17% of the load. When}

comparing that result with the P stored in the accumulated sediment, the amount of P in the inlet area alone represented almost 80% of the P load (Fig.6). In addition, the uppermost newly formed organic sediment in the T. latifolia dominated areas contained an additional 390 kg P.

0 20 40 60 80 100 120 140 160

Total P load Retention Inlet sediment T1 sediment T2 sediment Ts sediment

P

(k

g)

Fig.6. Phosphorus load and calculated P retention based on inflow-outflow calculations for four years (white bars), compared to the amount of P in the sediment accumulated in the inlet area (0.11 ha) and the top 4 cm of sediment in

(22)

Sediment characteristics

In order to determine the longevity of a wetland as P sink, the sediment P composition is important. For instance, P bound to calcium carbonates is generally considered to be stable, and will not be released to the water column. On the other hand, P bound to iron oxides and hydroxides is sensitive to changes in redox conditions, and should anoxia occur, P can be released to the water and leave the wetland (e.g. Richardson &

Vaithiyanathan, 2009).

To separate the different fractions of inorganic P in the Stene wetland sediments, sequential fractionation was used, where different forms of P are separated based on their different solubility in various chemical extractants

(Reddy et al., 1999). Inorganic P was divided into 1) NH4Cl-P =

bioavailable P, 2) NH4F-P = aluminum-bound P, 3) NaOH-P = P bound to

iron-oxides, and 4) H2SO4-P = calcium-bound P. The P left in the soil after

the fractionation procedure is considered biologically unavailable, hence representing a long-term storage pool in a wetland (Diaz et al., 2006). This insoluble fraction was given by the difference between TP and the sum of the fractions (Rydin, 2000).

On average, 22% of the P in the accumulated sediment in Stene was found in stable forms i.e. bound to calcium or the residual fraction not extracted in the

sequential fractionation. The NH4Cl-P (bioavailable P) fraction was small

(0.2%). The Fe/Al-P fraction was large (almost 40%), though the proportion bound as Fe-P could not be successfully separated with the chosen

fractionation scheme. At the inlet, where almost all the inflow P settled, 24% of the P was found in stable forms, and 43% was bound to organic material. Here, the Fe/Al-P fraction was 33%.

Another important factor when discussing sediment stability in wetlands is grain size. Clay particles can contain 12 times the amount of P found associated with sand particles (Pacini & Gächter, 1999). A recent study of sediments in constructed wetlands showed that clay content correlated well with the potentially bioavailable P (Maynard et al., 2009). Also, settling velocities for clay particles are slower, and they are more easily resuspended. Hence, wetland retention of clay particles is of prime importance for

downstream water bodies prone to eutrophication.

In an ongoing study (not included as a manuscript in this thesis), sediment P content and particle size distribution have been analyzed for eight

(23)

constructed wetlands in the south of Sweden. The main objective was to investigate whether or not clay particles settle in constructed wetlands situated in areas with a high proportion of clay soils. Grain size distribution was analyzed in a MasterSizer 2000 which uses a laser diffraction technique to measure particle size. It can detect particles from 0.01 to 10000 μm. Preliminary results from sediments sampled in Genarp wetland showed that no particles smaller than 0.5 μm were found (Fig.7). Coarse clay is usually defined as being < 2 μm and fine clay < 0.2 μm (Kroetsch & Wang, 2007). The results from the size fraction analyses showed that there was no

sedimentation of fine clay particles in any of the eight wetlands investigated in that ongoing study.

Fig.7. Grain size distribution in a sediment sample from Genarp wetland measured with a laser diffraction techniques. The arrows denote the size limits for coarse and fine clay, respectively.

(24)

General discussion

Annual P retention estimates

The positive but variable (1-58 kg ha-1_yr-1_{) P retention observed for all the}

seven wetlands agree with the results from other studies investigating P retention in constructed wetlands receiving diffuse pollution from

agriculture. For instance, Koskiaho et al. (2003) investigated three wetlands

in Finland, and the retention varied from 2 to 24 kg ha-1_yr-1_{. In one of the}

wetlands TP retention was negative the second study year. Braskerud et al. (2005) compiled data from 16 wetlands (three of them included in this study; Genarp, Råbytorp and Slogstorp), and the P retention varied from 2 to 1562

kg ha-1_yr-1_{. The variations in P retention in Braskerud’s study were only}

partly explained by differences in P loads. In the seven wetlands included in this study, P load was strongly correlated with the P retention. However, when investigating the relationship between the P retention and the two components making up the P load, i.e. hydraulic load and inflow P concentrations, water flow dynamics seemed to determine P retention efficiency, and not the P concentrations.

Estimating P retention in wetlands is difficult since the sampling strategies and methods are dependent on accurate information about water inflows to the wetlands. Assumptions on water balance are usually made, since water flow is usually measured only at one point (inlet or outlet) – hence, not all inputs and outputs of P and water are identified and measured (Reddy et al., 1999). Water input from other sources than the measured inlet can affect the water balance, either by dilution effect by groundwater, or by additional inputs of P to the wetland from the local catchment. As an example, the local catchment of the Bölarp wetland was approximately 2 ha or 1% of the catchment area. This local catchment provided additional water to the wetland. Water flow was measured at the outlet (Tab.2 in Paper I); hence, the inflow from the local catchment was included when making the transport calculations for P. If there was 1% more water in the outlet than in the inlet, the P loads were probably overestimated, since a lower water volume at the inlet would lead to a lower transport of P to the wetland. A comparison was made between monthly P retention (for January 2005) using the original data, and monthly P retention using an adjusted value of the inflow water volume (1% less than outflow). This resulted in a decrease in P retention from 4.3 to

1.8 kg ha-1_month-1_{. This theoretical calculation illustrates the possible}

(25)

field. In order to obtain accurate estimates of water and P budgets, water flow should be monitored both at the inlet and the outlet or the importance of the local catchment estimated from ground water studies. The former approach was used in Stene, and the results showed that the sampled inlet contributed 79% of the P load, and the rest was attributed to the local catchments (Karlsson, 2005). Stene wetland was large in comparison to the other wetlands included in this study, so the local catchment was more important there. Presumably, the importance of the local catchment decreases as the ratio wetland area to catchment area decreases; hence the assumptions that inflow more or less equals outflow may be justified at least for the high loaded wetlands included in paper I.

Another difficulty in estimating P retention in constructed wetlands is the problem of getting representative samples in automatic sampling, especially in water rich in particles (Jarvie et al., 2002). However, the estimated P loss

from the catchments of Stene wetland (0.26 kg ha-1_yr-1_{when sewage pipes}

and background leakage from forest had been accounted for) corresponded fairly well with the P loss from a clay soil situated in the same geographical

area (0.29 ha-1_yr-1_{, Ulén & Persson, 1999). Hence, at least for Stene,}

automatic sampling seemed to give representative samples even for PP.

Monthly P retention

When investigating P retention with higher temporal resolution, i.e. monthly retention, it was clear that in almost all wetlands there were periods with negative retention (Fig.5). Monthly negative retention was probably caused by two main mechanisms. First, high flow that led to an increased P

transport from the catchments and this in combination with a reduced sedimentation and possible resuspension probably caused an export of P from the wetlands. Release of TP during high flow periods has often been observed in inflow/outflow studies of constructed wetlands receiving non-point source runoff (e.g. Reinelt & Horner, 1995; Braskerud, 2002; Hoffmann et al., 2009). Second, low oxygen levels in the sediments could explain negative retention during months with low flow. Generally, as P load to the wetland decreases, the role of internal loading of P becomes

increasingly important. Internal loading of P can be mediated either by biological or chemical processes. DP is assimilated by biota immediately upon entering a wetland. Most of the assimilated P is released back into the water column after the death of the organism; according to Richardson (1985), 35-75% of the plant P is rapidly released. Chemical changes of e.g.

(26)

instance, Søndergaard et al. (2003) showed that for shallow lakes, internal loading of P continued to reduce water quality although external P loading had decreased. In small streams in southern Sweden, periods with no flow commonly occur during the warm months. In such periods, the water in constructed wetlands becomes stagnant. As decomposition processes deplete the oxygen concentrations near the bottom, anoxic conditions may develop. During anoxia iron oxides are reduced and P sorbed to these molecules can be released into the water column and exported from the wetlands. Similarly, anoxic conditions can also occur during low flow periods in winter if the wetland is covered by an ice layer that prevents oxygen diffusion from the atmosphere (e.g. Devito & Dillon, 1993).

The importance of correct sampling and analyzing techniques increases as the temporal resolution decreases, e.g. when calculating monthly P retention as opposed to annual. When grab samples were compared to composite samples in Stene wetland the mean inflow and outflow concentrations were higher in the composite samples (Tab.6 in Paper I). This could imply that the P transport to and from the wetland is overestimated. During periods with high mean flow, TP concentrations were high, and variations were greater in the grab samples compared to the composite samples. Since grab samples provide a ‘snap-shot’ of the P dynamics, the fact that the variation was large during high flow periods shows that P transport varies more than what can be seen from calculations based on composite samples. P retention during the three periods also differed depending on sampling technique. For instance, during one sample period with high water flows, P retention based on composite samples was negative (-3 kg) and when retention was

calculated based on grab samples it was positive (1 kg). In conclusion, during high flow periods P retention estimated from composite and grab samples, respectively, are the most different, which can affect the estimations of P retention on a shorter time scale.

When investigating grab samples taken at the inlet in relation to the water flow, varying patterns emerged. There was a relationship between water flow and inflow TP concentration in almost all wetlands (Tab.5 in Paper I). However, the strength of the relationship varied, and was both positive and negative depending on wetland. In wetlands receiving high hydraulic and P load the relationship between flow and TP concentration was negative, i.e. high concentrations of TP occurred during low flow periods, e.g. during summer months. Hence, in these wetlands, other factors than water flow determines inflow P concentrations. One explanation could be anoxic

(27)

standing water upstream during low flow periods, and another explanation could be that inflow water is affected by rural wastewater. Inflow TP concentrations were sometimes very high in these wetlands, but since they coincided with low water flow, the total transport of P (kg) was small. In wetlands receiving low hydraulic and P load, the relationship between water flow and TP concentration at the inlet was positive.

Sediment accretion and characteristics

Results from grab samples taken in Stene showed that the dominating form of P in both inflow and outflow water was PP (69% and 78%, respectively). Similar inflow proportion of PP (78%) has been shown by e.g. Braskerud (2002), and reflects properties of the catchment area in respect to soil type, land use and topography.

There was a clear accumulation of material closest to the inlet (Fig.8a); hence, the main retention mechanism in Stene was sedimentation. The

estimated sedimentation rate in Stene was 22 kg m-2_yr-1_{, which is lower than}

the 40–90 kg m-2_yr-1_{observed in small wetlands receiving high loads of}

particle rich water in Norway (Braskerud et al., 2000). The P amount in the newly accumulated sediment corresponded to almost 80% of the total P load to the wetland during four years. However, P retention efficiency based on water quality studies showed P retention of only 17%. Hence, despite the high sedimentation rate of inflow PP, there were a lot of particles leaving the wetland (Fig.4), and a substantial part of these must have originated from within the wetland. The outflow zone in Stene consisted of an open pond dug in clay; hence, internal erosion of fine particles caused by e.g. high water flow could be an explanation for the PP export. Also, there was a lot of biological activity in the wetland with numerous invertebrates and different species of birds feeding and nesting. Bioturbation caused by their activity could possible explain some of the PP export.

In several other wetlands included in this study, there was a visible layer of accumulated sediments, especially close to the inlet (e.g. Bölarp and Genarp, Fig.8b and c). For instance, in Bölarp a berm of material had accumulated 2-3 meters from the inflow to the wetland (area of the berm was approximately

10 m2_{, based on visual estimation). This berm was 83 cm thick and consisted}

of various grain sizes (<0.2 mm to >2 mm). A rough estimate of sediment

growth at the inlet was approximately 10 cm yr-1_{(83 cm in 8 years). In four}

(28)

sedimentation unit. The P content of the sediment in the berm has not yet been analyzed; hence, no comparison on P retention between water quality studies and sediment studies can be performed (as was done for Stene). When P retention was estimated using water quality and water flow data, the

results were low, 1 kg ha-1_yr-1_{, or 1% of the total P load. Since there was an}

obvious removal of suspended material, and probably PP as well, it is possible that the low P retention efficiency could be explained by a release of P due to internal processes such as resuspension, bioturbation or

desorption from sediments. However, since P analyses of the berm material have not yet been performed it is too early to draw any conclusions.

Fig.8. Sediment cores collected in a) Stene, b) Genarp and c) Bölarp wetlands. There was a clearly visible layer with accumulated material that was

distinguishable from the underlying clay bottom (indicated by arrows in a) and b)). The core from Bölarp was taken close to the inlet and there are several layers of varying grain size. (Photos: Karin Johannesson).

Results from sediment analyses showed that 22% of the P in the

accumulated sediment in Stene wetland was found in stable forms (Ca-P or

residual P). The bioavailable P fraction (NH4Cl-P) was small (0.2%) and the

Fe/Al-P fraction was large (almost 40%). At the inlet, where almost all the inflow P settled, 24% of the P was found in stable forms, and 43% was bound to organic material. Here, the Fe/Al-P fraction was 33%. Although the directly available P fraction was small, there is a risk of release of P from the sediments, should the chemical status of the sediment change. For instance, it is possible that during December 2005 and January 2006 the

(29)

observed net loss of P was caused by a release of dissolved P bound to Fe (III) oxides (Fig.5); however this was not investigated in detail. Release of soluble P from sediment storage pools has been observed but commonly when flooding agricultural soils rich in organic matter (reviewed by Reddy et al., 1999; Kadlec, 2005; Hoffmann et al., 2009). The Stene wetland was constructed on a mineral soil, and as much of the topsoil was removed there was little risk for release of mobile P from the soil. A recent study by Tanner and Sukias (2011) showed a negative P retention overall in three constructed wetlands in New Zealand. These wetlands were excavated on fertilized soil, and the topsoil from the excavation was returned into the bottom of the wetlands as a carbon source for denitrifying bacteria. Hence, the release of P observed in their study was probably because of release from the sediments saturated with P. In Stene, as more litter and organic sediment accumulate in the vegetated areas, the risk for development of anoxic conditions, with subsequent release of P bound to Fe (III) oxides, may increase.

(30)

Conclusions and further studies

The results showed that wetlands constructed on arable land in Sweden functioned as P traps, and could help reducing transport of P from agricultural areas. P release occurred at various occasions and could be explained either by high flow or anoxic conditions.

Sediment studies in one of the wetlands showed that the stores of P in the upper sediment layer were substantially larger than the annual load of P (based on water flow and quality measurements). P fractionation analyses indicated that there is a risk for future release of P from those pools. The discrepancy between the results showed that there is a need to add studies of sediment accumulation to inflow-outflow estimates for an improved

understanding of wetland P retention.

Water flow and quality measurements are expensive and are depending on electricity. Also, the assumptions on water balance discussed above (e.g. no groundwater inflow and no inflow from local catchments) are not always realistic. In an ongoing study we are investigating P retention in the sediments of eight constructed wetlands situated in clayey areas. Here, P is assumed to be transported predominately as PP, and sedimentation is assumed to be the main retention mechanism. Hence, P retention was estimated by measurements of P amount in sediments accumulated in the wetlands during one year. Preliminary results showed a net sedimentation in

all wetlands, and the P retention varied between 5 and 149 kg ha-1_yr-1_{. One}

of the eight wetlands is monitored both through sediment studies and water quality, and preliminary results show a similar P retention efficiency with

both methods (49 kg ha-1_yr-1_{according to sediment studies, and 56 kg ha}-1

yr-1_{according to water quality studies). Hence, estimating P retention by}

studying sediment accumulation could be a cost effective alternative to water quality studies in wetlands constructed on clayey soils.

(31)

Acknowledgements

I would like to thank my supervisor Karin Tonderski, for your guidance, support and encouragement, and for always being in a good mood and knowing exactly what to say to raise my spirits when they are low! Also, thank you Stefan Weisner, my second supervisor at Halmstad University, for good advice and healthy skepticism. Thanks for the lodging while in Halmstad! Thank you Barbro Ulén, aka the Phosphorus Lady at SLU, Uppsala, for providing ideas and advice during our meetings. Thank you Pia

Kynkäänniemi at SLU, for being a partner in nerd! Thanks for all the good

times spent in Berlin, Sevilla, Venice and Lancaster and for all the laughs in the lab and the muddy field! Thanks to my colleagues at the Biology

department and especially my fellow PhD students for interesting

discussions in the fika room regarding everything from lace to naked bird-watching!

My warmest thank you to my family and friends, who have kept me sane and helped me keep a healthy distance to my work. My mum Marie-Louise and my three brothers Per-Martin (with his lovely family; Emma, Stella and baby Julia), Patrik and Kristoffer – you are awesome! Thanks for at least trying to look interested when I talk about my research . To my childhood friends; Bisse and Malin P – thanks for always being there with your endless support and encouragement, through thick and thin! To Mia and Malin A – thanks for countless dinners (and desserts!), movie and game nights. And finally, to Mikael – thanks for providing a whole different view on natural sciences, and for bringing some art and beauty into my life!

In loving memory of my dad, Roland (1948-2000), who died before his time. Wishing you were somehow here again

(32)

References

Bergström, L., F. Djodjic, H. Kirchmann, I. Nilsson & B. Ulén. 2007. Phosphorus from farmland to water. Uppsala, Sweden: Swedish University of Agricultural Sciences, Report Food 21, no. 4.

Braskerud, B. C. 2006. Mass flux from catchments and retention in constructed wetlands: Sampling techniques. In 24th Nordic hydrological conference 2006: Experiences and challenges in implementation of the EU water framework directive, eds. J. C. Refsgaard, A. L. Højberg. 49: 147-154.

Braskerud, B. C. 2003. Clay particle retention in small constructed wetlands. Water Research 37: 3793-3802.

Braskerud, B. C. 2002. Factors affecting phosphorus retention in small constructed wetlands treating agricultural non-point source pollution. Ecological Engineering 19: 41-61. Braskerud, B. C. 2001. The influence of vegetation on sedimentation and resuspension of soil

particles in small constructed wetlands. Journal of Environmental Quality 30: 1447-1457. Braskerud, B. C., H. Lundekvam & T. Krogstad. 2000. The impact of hydraulic load and

aggregation on sedimentation of soil particles in small constructed wetlands. Journal of Environmental Quality 29: 2013-2020.

Braskerud, B. C., K. S. Tonderski, B. Wedding, R. Bakke, A. G. Blankenberg, B. Ulén & J. Koskiaho. 2005. Can constructed wetlands reduce the diffuse phosphorus loads to eutrophic water in cold temperate regions? Journal of Environmental Quality 34: 2145-2155. Cordell, D., J. O. Drangert & S. White. 2009. The story of phosphorus: Global food security and

food for thought. Global Environmental Change 19: 292-305.

Croel, R. C. & J. M. Kneitel. 2011. Ecosystem-level effects of bioturbation by the tadpole shrimp Lepidurus packardi in temporary pond mesocosms. Hydrobiologia 665: 169-181. Devito, K. J. & P. J. Dillon. 1993. Importance of runoff and winter anoxia to the P and N

dynamics of a beaver pond. Canadian Journal of Fisheries and Aquatic Sciences 50: 2222-2234.

Diaz, O. A., S. H. Daroub, J. D. Stuck, M. W. Clark, T. A. Lang & K. R. Reddy. 2006. Sediment inventory and phosphorus fractions for water conservation area canals in the everglades. Soil Science Society of America Journal 70: 863-871.

EPA. Sweden's environmental objectives. in Sweden's environmental objectives secretariat [database online]. http://www.miljomal.se/Environmental-Objectives-Portal/, 2009 [cited November 26 2009].

HELCOM. The Baltic Sea action plan. Helsinki, Finland, 2009. Available from http://www.helcom.fi/BSAP/en_GB/intro/ (accessed October 6, 2009).

Hoffmann, C. C., C. Kjaergaard, J. Uusi-Kämppä, H. C. Bruun Hansen & B. Kronvang. 2009. Phosphorus retention in riparian buffers: Review of their efficiency. Journal of Environmental Quality 38: 1942-1955.

Jansson, M., R. Andersson, H. Berggren & L. Leonardson. 1994. Wetlands and lakes as nitrogen traps. Ambio 23: 320-325.

Jarvie, H. P., P. J. A. Withers & C. Neal. 2002. Review of robust measurement of phosphorus in river water: Sampling, storage, fractionation and sensitivity. Hydrology and Earth System Sciences 6: 113-132.

Jordan, T. E., D. F. Whigham, K. H. Hofmockel & M. A. Pittek. 2003. Nutrient and sediment removal by a restored wetland receiving agricultural runoff. Journal of Environmental Quality 32: 1534-1547.

Kadlec, R. H., 2005. Phosphorus removal in emergent free surface wetlands. Journal of Environmental Science and Health 40:1293–1306.

(33)

Kalff, J. 2002. Limnology - inland water ecosystems. Upper Saddle River, New Jersey: Prentice Hall.

Karlsson, M. 2005. Näringsbelastning på en anlagd våtmark från åker, skog och enskilda avlopp. Uppsala: Swedish University of Agricultural Sciences, Division of Water Quality, Seminarier och examensarbeten nr. 50.

Koskiaho, J., P. Ekholm, M. Räty, J. Riihimäki & M. Puustinen. 2003. Retaining agricultural nutrients in constructed wetlands - experiences under boreal conditions. Ecological Engineering 20: 89-103.

Kovacic, D. A., R. M. Twait, M. P. Wallace & J. M. Bowling. 2006. Use of created wetlands to improve water quality in the Midwest - Lake Bloomington case study. Ecological Engineering 28: 258-270.

Kroetsch, D. & C. Wang. 2007. Particle size distribution. In Soil sampling and methods of analysis., eds. M. R. Carter, E. G. Gregorich. Second ed.CRC Press.

Kronvang, B. & J. Bruhn. 1996. Choice of sampling strategy and estimation method for calculating nitrogen and phosphorus transport in small lowland streams. Abstract. Hydrological Processes 10: 1483-1501.

Larsson, U., R. Elmgren & F. Wulff. 1985. Eutrophication and the Baltic Sea: Causes and consequences. Ambio 14: 9-14.

Löfgren, S., A. Gustafson, S. Steineck & P. Ståhlnacke. 1999. Agricultural development and nutrient flows in the Baltic States and Sweden after 1988. Ambio 28: 320-327.

Maynard, J. J., A. T. O'Geen & R. A. Dahlgren. 2009. Bioavailability and fate of phosphorus in constructed wetlands receiving agricultural runoff in the San Joaquin valley, California. Journal of Environmental Quality 38: 360-372.

Nairn, R. W. & W. J. Mitsch. 2000. Phosphorus removal in created wetland ponds receiving river overflow. Ecological Engineering 14: 107-126.

Pacini, N. & R. Gächter. 1999. Speciation of riverine particulate phosphorus during rain events. Biogeochemistry 47: 87-109.

Reddy, K. R., R. H. Kadlec, E. Flaig & P. M. Gale. 1999. Phosphorus retention in streams and wetlands: A review. Critical Reviews in Environmental Science and Technology 29: 83-146.

Reinelt, R. E. & R. R. Horner, 1995. Pollutant removal from storm water runoff by palustrine wetlands based on comprehensive budgets. Ecological Engineering 4: 77−97.

Reinhardt, M., R. Gächter, B. Wehrli & B. Muller. 2005. Phosphorus retention in small

constructed wetlands treating agricultural drainage water. Journal of Environmental Quality 34: 1251-1259.

Reynolds, C. S. & P. S. Davies. 2001. Sources and bioavailability of phosphorus fractions in freshwaters: A British perspective. Biological Reviews 76: 27-64.

Richardson, C. J. 1985. Mechanisms controlling phosphorus retention capacity in freshwater wetlands. Science 228: 1424-1427.

Richardson, C. J. & P. Vaithiyanathan. 2009. Biogeochemical dynamics II: Cycling and storage of phosphorus in wetlands. In eds. E. Maltby, T. Barker: The wetlands handbook, 1st ed. Blackwell publishing.

Rydin, E. 2000. Potentially mobile phosphorus in Lake Erken sediment. Water Research 34: 2037-2042.

Stolte, W., E. Östensson & N. Holmgren. 2009. Finn de områden som göder havet mest. Swedish River Basin District Authorities.

Sveistrup T.E., Marcelino V. & Braskerud B.C. 2008. Aggregates explain the high clay retention of small constructed wetlands: A micromorphological study. Boreal Environment Research 13:275-284.

(34)

Tanner, C. C. & J. P. S. Sukias. 2011. Multiyear nutrient removal performance of three constructed wetlands intercepting tile drain flow from grazed pastures. Journal of Environmental Quality 40: 620-633.

Ulén, B. 2004. Size and settling velocities of phosphorus-containing particles in water from agricultural drains. Water, Air and Soil Pollution 157: 331-343.

Ulén, B. & K. Persson. 1999. Field-scale phosphorus losses from a drained clay soil in Sweden. Hydrological Processes 13: 2801-2812.

Uusitalo, R., E. Turtola, M. Puustinen, M. Paasonen-Kivekäs & J. Uusi-Kämppä. 2003. Contribution of particulate phosphorus to runoff phosphorus bioavailability. Journal of Environmental Quality 32: 2007-2016.

Uusitalo, R., M. Yli-Halla & E. Turtola. 2000. Suspended soil as a source of potentially

bioavailable phosphorus in surface runoff waters from clay soils. Water Research 34: 2477-2482.

Analysis of phosphorus retention variations in constructed wetlands receiving variable loads from arable land