Economic valuation of suspended sediment and phosphorus filtration services by four different wetland types: A preliminary assessment for southern Ontario, Canada

Wetlands are known for their water filtration (or purification) function. Although different wetland types differ in their filtration capacity, they are usually aggregated together in economic valuation studies. Here, we explicitly separate the valuation of the suspended sediment and phosphorus (P) filtration services of the four major wetland types—bogs, fens, marshes and swamps—found in southern Ontario, Canada. The areal extents of the four wetland types are derived from the Canadian Wetland Inventory (CWI) progress map, while the sediment accretion rate is used as the key variable regulating the suspended sediment and P filtration functions. Based on available literature data, we assess the relationship of the sediment accretion rate to wetland size. Because only weak positive correlations are found, we assign a mean (average) sediment accretion rate to each wetland type. The sediment accretion rates are combined with mean soil P concentrations to estimate Pretention rates by the wetlands. The replacement cost method is then applied to valuate the sediment and P filtration services. The unit values for both sediment and P retention decrease in the order: marshes > bogs ≈ swamps > fens. The total value of sediment plus phosphorus removal by all wetlands in southern Ontario amounts to $4.2 ± 2.9 billion per year, of which about 80% is accounted for by swamps. We further assess the costs of different options to offset the additional P loading generated in a hypothetical scenario whereby all wetlands are converted to agriculture. The results demonstrate that replacing the P filtration function of existing wetlands with conventional land management and water treatment solutions is not cost‐effective, hence reinforcing the importance of protecting existing wetlands.


| INTRODUCTION
Wetlands, which are among the most productive terrestrial ecosystems, provide huge economic benefits through a variety of functions (Gallant et al., 2020). Among these, their hydrological (e.g., flood control), biogeochemical (e.g., nutrient retention) and ecological (e.g., nursery plants) functions deliver socio-economic benefits known as ecosystem services (Aziz & Van Cappellen, 2020). Differences in hydrological and geomorphological characteristics distinguish the various wetland types (Warner & Rubec, 1997; Table 1) that, in turn, results in variable provisioning of ecosystem services (Turner et al., 2000).
Wetlands have long been recognized for their key function of filtering pollutants out from water (Gopal & Ghosh, 2008). Increased sediment loads and nutrient enrichment are major threats to the quality of aquatic ecosystems (Dordio et al., 2008;Fennessy et al., 2004).
Therefore, the role of wetlands in improving water quality is a primary argument for their preservation and restoration across the world (Bring et al., 2020;Dordio et al., 2008). Freshwater wetlands trap sediment and sequester nutrients (Craft & Casey, 2000) and filter water through physical (sedimentation), chemical (adsorption, precipitation, chelation) and biological (plant uptake) processes (Fennessy et al., 2004;Kadlec & Wallace, 2009;Kidd et al., 2015;Reddy et al., 1999;Settlemyre & Gardner, 1977). Sediment deposition depends on wetland type with some types more efficiently retaining sediment than others (Bruland, 2008;Loaiza & Findlay, 2008). The sediment filtering effectiveness also depends on watershed size, land use and the wetland's connectivity to the stream and groundwater network (Craft & Casey, 2000). Sediment accumulation in wetlands is heavily affected by human activity in the watershed.
For example, in the Murray-Darling Basin in Australia sedimentation rates doubled after European settlement and are now 80 times higher than the mean rate in the Late Holocene (Gell et al., 2009). Sediment accretion is the net balance between sediment deposition and resuspension (Neubauer et al., 2002), and an important indicator of the functioning of restored wetlands (Takekawa et al., 2010).
It is influenced by, among other things, the amount of suspended material delivered to the wetland, the composition and distribution of vegetation, flooding and waterlogging patterns, depth and bottom morphology and biomass production (Cahoon & Turner, 1989;Goodman et al., 2007;Jarvis, 2010). Sediment accretion rates for wetlands, however, are often difficult to estimate and data are relatively sparse (Loaiza & Findlay, 2008).
Through sediment retention, wetlands can be helpful in mitigating excess nutrients and pollutants (Mitsch & Gossilink, 2000). Hence, a wetland's sediment accretion rate is a critical parameter regulating water quality improvement (Bhomia et al., 2015;Gustavson & Kennedy, 2010). Wetlands remove phosphorus (P) from the water through physical and biological processes (Reddy et al., 1999). Phosphorus may accumulate in sediments by settling of allochthonous particulate P and autochthonous biomass P, the precipitation of aqueous T A B L E 1 Major wetland types and their characteristics (from National Wetlands Working Group, 1997;Smith et al., 2007;Zoltai & Vitt, 1995) (Mitsch & Gossilink, 2000). Wetlands generally trap phosphorus although sometimes they may release aqueous P under anoxic conditions (Johnes et al., 2020). While uptake by vegetation can temporally remove P from water, P in accreted sediments represents the longterm sink in wetlands (Mitsch & Gossilink, 2000). Therefore, the sediment accretion rate is the key parameter used to estimate P retention in wetlands (Griffiths & Mitsch, 2020).
Wetlands are complex and diverse ecosystems, and therefore valuation of their ecosystem services is challenging. Economic valuation of some services relies on perceived benefits and people's preferences, which can vary significantly. Thus, there is no standard valuation framework as yet to value ecosystem services generated from different wetland types (Lambert, 2003). Nonetheless, studies imply that, in many areas of the world, ecosystem services have been declining due to draining of wetlands (Zedler, 2003 (Gustavson & Kennedy, 2010).
The economic valuation of wetland ecosystem services can help inform a balanced assessment of the importance of ecosystems for human wellbeing and the economy (Gleason et al., 2008). Wetlands are described as the kidneys of the landscape because of the chemical and hydrological processes they perform (Barbier et al., 1997). Most wetland services are public goods and their consumption is nonexcludable. Despite being the only ecosystems with an international treaty calling for their protection (the Ramsar Convention), the degradation of wetlands continues to be exacerbated by ignorance about the values of their ecosystem services and, in particular, that of their non-market environmental services (Ajibola, 2012).
The values of ecosystem services generated by different wetland types are expected to vary. This is certainly true for the services that are closely linked to sediment retention dynamics (Loaiza & Findlay, 2008). However, in most watershed-scale economic valuations of ecosystem services, the same unit value for the water filtration service is assigned to all wetland types (Anielski & Wilson, 2010;Hotte et al., 2009). The purpose of this paper is to present a valuation framework for the filtration services for suspended sediment and phosphorus that explicitly distinguishes between the broad categories of wetlands.
The framework is applied to southern Ontario, a region characterized by intensive agriculture that is home to roughly one third of Canada's population. To our knowledge, this is the first study that separately valuates the water filtration functions of different wetland types.

| Southern Ontario
The study area comprises the most southerly Mixedwood Plains Ecozone in Ontario, Canada ( Figure 1). It is the country's region most affected by human activity (Taylor et al., 2014) and covers 5.33 million hectares, that is, 4.9% of Ontario's total surface. The region experiences high population growth, urban development and intensive farming. Agriculture is presently the dominant land use with natural vegetation reduced to 3% of its historic area. Aquatic ecosystems have deteriorated due to sediment loading and pollution from intensive agriculture, including excess nitrogen (N) and P (Taylor et al., 2014). Wetland area has declined by more than 70% since European settlement (c.1800). Southern Ontario is completely mapped in the Canadian Wetland Inventory (CWI). Based on the Southern Ontario Land Resource Information System (SOLRIS) land use data (MNR, 2008), the areas of the four major wetland type are: bogs (0.85%), fens (0.58%), marshes (11.72%) and swamps (86.85%; Table 2). The total area of all wetlands is 896 149 hectares.

| Valuation methodology
The water filtration services (i.e., sediment and P retention) are valu-  (1) and (2): where V si is the total value (in $ per year) of sediment retention by the i-th wetland type, R i is the mean sediment accretion rate (cm/year) of the i-th wetland type, A i is the total surface area of the i-th wetland type in southern Ontario (ha), and SR C is the sediment removal cost (in $ per m 3 ); and: where V pi is the total value (in $ per year) of phosphorus retention by the i-th wetland type, D i is the mean soil density (g/cm 3 ) in the i-th wetland type, Pr i is the mean phosphorus soil concentration (mg/kg) in the i-th wetland type, and PR C is the phosphorus removal cost ($/kg).

| Sediment accretion rates
We relied on literature data to estimate representative sediment accretion rates of the different wetland types. The two methods commonly applied to measure sediment accretion data are mass balancing and geochemical tracers. The mass balancing method involves monitoring suspended matter inflow to and outflow from a wetland. Geochemical tracer analysis involves the isotopic dating of sediment cores F I G U R E 1 Wetland types in southern Ontario, Canada. The area in grey is the selected/study region (MNR, 2008) T A B L E 2 Economic valuation of sediment and phosphorus (P) filtration services by the four major wetland types in southern Ontario  (Demissie & Fitzpatrick, 1992). In radiometric dating, radionuclides are used as chronological markers. The two natural radionuclides that are most frequently employed are 210 Pb and 14 C (Church et al., 1987;Walker et al., 2007). Additional artificial radionuclides ( 137 Cs and 14 C) pads for a few years tend to give higher rates of sediment accretion compared to long term dating methods (e.g., using 137 Cs and 210 Pb) because short-term measurements do not account for shallow subsidence within the top layer of sediment (Ensign et al., 2014). In our analysis, most of the values listed in Tables A1-A4 were taken from studies that applied long-term measurement techniques (Church et al., 1987;Craft, 2007;Neubauer et al., 2002).
We investigated whether the sediment accretion rates in the different types of wetlands are significantly correlated with the wetland surface area ( Figure A1). While we found positive correlations, these ranged statistically from insignificant to weak, in part because of the limited number of rates that could be obtained from the literature. Hence, in the valuation calculations, we assigned a constant sediment accretion rate to each wetland type, which was calculated as the arithmetic mean of the values in Tables A1-A4. Note that the majority of the values used to compute the arithmetic means were taken from studies on wetlands in the United States (see Tables A1-A4).
Below 5-10 cm, the concentration tends to stabilize, indicating that P turnover processes have ceased (Wang et al., 2008).
Therefore, we used mean values of the total P concentrations measured on soil samples taken at depth of at least 10 cm as representative for long-term P retention in a wetland (Pinder et al., 2014). The mean total P soil concentrations are then multiplied by the estimated sediment accretion rates to calculate P retention rates (Equation 2).  Table 2). The mean rates are calculated by averaging the individual accretion rates, which are extracted from the literature and listed in Table A1-A4. Admittedly, the use of a constant mean sediment accretion rate per wetland type is a strong simplification and represents a source of uncertainty in the economic valuation of the filtration functions. Future work should explore in more detail the variability of sediment accretion rates in wetland systems in order to refine the assessment of their role in sediment retention.

| Phosphorus retention
The average total P concentrations given in Table 2  (USA) were also included in the calculation of the mean P retention in marshes in Table 2 (Bruland & Richardson, 2006). Reported average dry bulk densities of wetlands in Ontario and Alberta are as follows: 1.49 (bogs), 1.54 (fens), 2.0 (marshes) and 1.57 g/cm 3 (swamps; Redding & Devito, 2005). However, for consistency, we systematically F I G U R E 2 A framework for valuation of ecosystem services from different wetland types (modified from Turner et al., 2000) imposed a dry bulk density of 1.75 g/cm 3 , because this is the value used by Fennessy et al. (2004) to estimate the total P concentrations shown in Table 2.
Existing studies generally point to the efficient retention of total P by wetlands. For example, a mass balance study of the Hidden Valley wetland, Ontario, found that 50% of total phosphorus is trapped by the wetland (Shane et al., 2001). However, for the same wetland the export of bioavailable P (i.e., dissolved orthophosphate) was 22% higher than the corresponding input. Thus, in-wetland transformation processes can significantly alter the chemical speciation and, hence, the bioavailability of P, not unlike those caused by in-reservoir processes (Van Cappellen & Maavara, 2016).
The average P retention rates used here vary by a factor of three between the lowest (fens) and highest (marshes) values ( Creek marsh in the western basin of Lake Erie (Mitsch et al., 1989;Shane et al., 2001). The latter research also concluded that the restoration of one-fourth of the original Old Woman Creek marsh area alone could reduce P loading to the western basin of Lake Erie by 25%-30%.
Overall, P retention rates in wetlands are highly variable across landscapes. Here, relatively high mean values are used because the wetlands of southern Ontario are all located in agricultural watersheds and thus receive high P loads, which in turn results in higher retention rates than for non-agricultural watersheds (Johnston, 1991;Riemersma et al., 2006). The high standard deviations in Table 2 imply that the mean P retention rates yield preliminary, order of magnitude, estimates of the values of the corresponding service.

| Wetland value functions (V si ,V pi )
To determine the unit values of the filtration services, we used the average cost for sediment removal and disposal (SR C = $170 ± 78 per cubic meter) compiled from data from 10 stormwater management facilities in Ontario (Aziz, 2018).  (Schuyt & Brander, 2004).
The large discrepancies in unit values between our and other studies illustrates the lack of a unified approach in the valuation of the water filtration service of wetlands, which in turn may cause ambiguities and misunderstandings. These discrepancies emphasize the need to clearly outline the basis of the cost estimates. Our estimates are the highest, because they require that the full capacity to trap sediment and P by the existing wetlands be conserved and accommodated entirely by improved conservation practices and built infrastructure. That is, we valuate the water filtration service of wetlands by matching the original benefits (see also Breaux et al., 1995;Lambert, 2003). Other approaches, including those in the studies mentioned above, estimate the downstream increase in treatment costs that would result from the loss of the existing natural retention capacity. These costs, however, are attenuated by in-stream dilution, retention and transformation processes and therefore only represent a fraction of the value of the lost ecosystem service.
From a sustainable management perspective, the high values of the sediment and phosphorus filtration functions must be assessed in conjunction with the many other ecosystem services provided by wetlands and the interlinkages between these services. In a worst case scenario, a high sediment trapping caused by excessive sediment loading to a wetland may result in ecological degradation, for example by causing habitat instability and loss (Sileshi et al., 2020). Similarly, a high phosphorus filtration efficiency can lead to the undesirable eutrophication of a wetland. In the long run, these negative impacts may even cause a reduction in the sediment and phosphorus filtration functions themselves. Thus, when using the estimated values of the filtration functions to inform environmental decision making, the finite filtration capacities and resilience of the affected ecosystems need to be considered.

| Total wetland filtration service value
The unit values for the water filtration service provided by each wetland type are applied to the respective wetland areas in southern Ontario to obtain the total values of phosphorus and sediment retention by all wetlands (Table 2). These total values are strongly dependent on the relative surface areas covered by the different types of wetlands in southern Ontario. For instance, even though the unit values of swamps are approximately half those of marshes, swamps dominate the total value because they make up most (87%) of the total wetland area in the region (Figure 4). The total value of water filtration service (sediment plus P removal) performed by all wetland types in southern Ontario amounts to $4.2 ± 2.9 billion per year (CAD 2016). Furthermore, the value of sediment retention by wetlands is about six times higher than that of phosphorus retention.

| Offsetting P retention by existing wetlands
Phosphorus is the ultimate limiting nutrient in streams and lakes in and around southern Ontario (Schindler, 2012). The only method that has so far proven successful in controlling eutrophication of the F I G U R E 4 Total value of sediment and phosphorus removal from water for the four wetland types in southern Ontario (error bars show standard deviations of the mean values) region's lakes, in particular Lake Erie and Lake Ontario, is to reduce P inputs (Schindler, 2012). To address the resurgence of algal blooms in Lake Erie, the United States and Canada have committed to reduce P inputs to the lake by 40% from the year 2008 baseline, which means an annual reduction of 200 metric tonnes of P from the Canadian side (Hanief & Laursen, 2019).
For the cost-effectiveness analyses, we assessed the cost of replacing wetlands' P retention capacity under a scenario where all existing wetlands are converted to agriculture. Using the P retention rates of the four wetland types and their areas, the total annual P retention by wetlands in southern Ontario is close to 30 000 tonnes ( Table 2). The additional P load from converting wetlands to agricultural land was calculated using the average P export rates for row crops, small grains, forage and pasture from local and regional studies (Donahue, 2013;Jeje, 2006;Shaver et al., 1994;Winter, 1998). Using an estimated composite delivery rate of 0.52 ± 0.28 kg P ha À1 year À1 , the additional P load is then 466 ± 251 t P/year. Therefore, the total P loading from wetland loss and additional agricultural P is 30420 ± 11 990 t P/year. We now consider three alternatives to offset this excess P load: (1) best management practices (BMPs), (2) CWs and (3) wastewater treatment plant (WWTP) upgrades. The cost of converting 2 ha of each wetland type plus that of converting all wetlands to agriculture is estimated using these three alternatives (Table 3).

| Best management practices
A generally accepted cost for removing 1 kg P by completed BMPs projects in southern Ontario is $400/year (CAD 2009). This includes the cost of the BMP implementation and project management (Marcano, 2015). When accounting for inflation, the value in 2016 is $447 per kg of P removal per year. The annual total cost of offsetting the lost P retention via BMPs then equals about $13 billion (Table 3).
3.5.2 | Constructed wetlands Kynkäänniemi et al. (2013)  This cost is based on interest on capital investment, operation and maintenance cost, annual depreciation and loss of crop yield on the land (Tousignant et al., 1999). Scaling the cost to the entire area of constructed wetlands required then yields a total cost of $2.9 billion (Table 3).

| WWTPs upgrades
A cost-benefit analysis of phosphorus in the Grand River watershed, Ontario, suggests that, if all the WWTPs are upgraded in the watershed, it will cost $5475 to remove 1 kg of P (CAD 2016; Hanna, 2015). This cost does not include the optimization of operation of current processes in the upgrading option. Using this cost, WWTPs become the most expensive option to offset the lost P from conversion of wetlands to agriculture: $164 billion per year (Table 3).
The results in Table 3 indicate that the options for phosphorus removal considered are not cost effective when compared to the P retention service values of the existing wetlands in Table 2. The least expensive option is constructed wetlands, however it requires that land is made available to install the new wetlands. The areas of constructed wetlands required to counteract the extra P loads generated by the loss of 1 ha of bog, fen, marsh and swamp are 0.62, 0.28, 0.41 and 0.91 ha, respectively. The required area of constructed wetlands is almost equal in the case of marshes because of the very similar P retention rates.

| CONCLUSIONS
This study presents a first valuation of the sediment and P water filtration services of wetlands in southern Ontario. The estimates are based on mean sediment accretion rates for different wetland types as the master variable regulating the water filtration efficiency for suspended sediment and P. The unit values of the water filtration services of the four wetland types in southern Ontario increase in the order: marsh > bog ≈ swamp > fen. Hence, marshes are the most valuable wetland type for water filtration. Our cost-effectiveness T A B L E 3 Costs of three interventions-Best management practices (BMPs), constructed wetlands (CWs), waste water treatment plants upgrades (WWTPUs)-To offset P released from the loss of 1 ha of wetland from the four types, as well as from the loss of all existing wetland area analysis further shows that it would be very costly to replace the existing wetlands' water filtration services by improved land and nutrient management and manmade infrastructure. Further work should refine the valuation estimates presented here by more precisely delineating the relationships between wetland size and sediment accretion rates, and by accounting for the hydrological connectivity of wetlands across the landscape as well as the variability of concentration, speciation and mobility of sedimentary phosphorus. In addition, the filtration functions assessed here are part of a much larger set of ecosystem services provided by wetlands.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Tariq Aziz
https://orcid.org/0000-0003-0749-5870 Philippe Van Cappellen https://orcid.org/0000-0001-5476-0820 F I G U R E A 1 Sediment accretion rates (cm/year) versus wetland size (as surface area in ha) for the four wetland types F I G U R E A 2 Frequency distributions of the sediment accretion rate data for the four wetland types. The data are given in Tables A1-A4