Controls on organic carbon stocks among restored wetland soils in the Long Point region of southern Ontario, Canada

Freshwater marsh restoration can be a viable natural climate solution; however, the extent to which marsh soils bury and preserve organic carbon within policy‐relevant timescales remains highly uncertain. Here, we compare organic carbon masses and accumulation rates from an undrained reference marsh, a passively restored freshwater marsh (reflooded after 1954) and a chronosequence of actively restored freshwater marshes (<10 years in age) situated in Lake Erie watersheds in the Long Point Biosphere Reserve of Ontario, Canada. The reference site has sustained the highest rates of short‐term organic carbon accumulation (235 g C m−2 yr−1) over the last four decades and has the highest mass of soil organic carbon (122 tC/ha) at 0–30 cm depth. Organic carbon masses are highly variable among all restored wetlands (16–115 tC/ha) at 0–30 cm depth and are not strongly related to time since restoration at least over the last 10 years. Nonetheless, we show that passive wetland restoration generates high rates of organic carbon accumulation (144 g m−2 yr−1) on a multidecadal scale where sites are low‐lying, underlain by alluvial deposits and connected to larger ground and surface water networks. Active restoration measures (e.g. excavation, installation of berms) may promote organic carbon preservation, particularly where fine‐grained soil texture promotes waterlogging. We demonstrate the importance of substrate, topographic gradient, and hydrology in selecting sites for marsh restoration to maximize carbon sequestration, and argue that the presettlement context and reference paleorecords provide necessary baselines for directing successful wetland restoration.


Introduction
Natural climate solutions have potential to help mitigate climate change by providing carbon sequestration and storage and by offsetting greenhouse gas emissions through protection, restoration, and improvement of land management (Griscom et al. 2017;Strassburg et al. 2020;Drever et al. 2021). Wetlands in particular have garnered attention due to high rates of soil carbon sequestration (Euliss et al. 2006;Mitsch et al. 2013;Drever et al. 2021). Although wetlands are the largest natural source of methane, their capacity to support low rates of decomposition and accumulate and preserve organic carbon in their soils enables them to act as long-term net carbon sinks (Yu 2011;Kolka et al. 2018). Thus, identifying wetland restoration measures that promote net carbon uptake to help achieve emission reduction targets over policy-relevant timescales is an ongoing area of active research (Moreno-Mateos et al. 2012;Strassburg et al. 2020).
Freshwater (inland) wetlands contain significant carbon stocks due to their extensive area globally and their potential to accumulate organic carbon at high rates in surface and deep soils (Nahlik & Fennessy 2016;Loder & Finkelstein 2020;Uhran et al. 2021). However, they have been highly susceptible to disturbance and conversion to farmland (Mitsch & Gosselink 2015;Harrison et al. 2020). As a result, they are at risk of losing their carbon stocks and capacity to sequester carbon (Byun et al. 2018;Spivak et al. 2019). Avoided conversion of at-risk freshwater mineral wetlands has been identified as the second largest opportunity among wetland-related natural climate solutions for providing net carbon uptake and mitigation against greenhouse gas emissions in Canada by 2030 (Drever et al. 2021). Yet, the net cumulative potential of freshwater mineral wetlands, such as freshwater marshes, for climate change mitigation may be underestimated given the lack of data on organic carbon stocks in deeper soils (Loder & Finkelstein 2020). Restoration of freshwater mineral wetlands has also been identified as an opportunity for mitigation in agriculturally intensive regions of Canada (Drever et al. 2021), but available data suggest that net carbon uptake and organic carbon stocks in surface soils of restored wetlands are markedly lower than in undrained sites (Ballantine & Schneider 2009;Brown & Norris 2017;Yu et al. 2017). Consequently, the time period over which a restored freshwater mineral wetland transitions from being a net carbon source to a net carbon sink (also known as the "switchover time" ;Mitsch et al. 2013;Bridgham et al. 2014;Neubauer 2014) may span decades to centuries (Ballantine & Schneider 2009;Badiou et al. 2011;Yu et al. 2017). More field data are required over annual to century timescales to accurately quantify their mitigation potential.
Alternatively, the recovery of soil carbon stocks among restored wetlands has been shown to be highly variable and driven by a multitude of factors related to landscape and land use (e.g. Drexler et al. 2019;Tangen & Bansal 2020). Poorly drained substrates are often targeted as underlying parent material for wetland restoration sites because of their capacity to sustain hydric soils and reduce rates of decomposition, which promotes organic matter accumulation (Zoltai 1988;Richardson & Vepraskas 2000). Some substrates also promote retention of organic carbon due to adsorption on clay and silt particles in wetland soils, thereby protecting organic matter from microbial decomposition (e.g. Van Veen & Kuikman 1990;de Telles et al. 2003;Yu et al. 2017). While research projects continue to examine the relationship between the speed of recovery of organic carbon stocks versus soil composition or amendments (e.g. Scott et al. 2020), there has also been emphasis on how landscape position (Zoltai 1988;Bedard-Haughn et al. 2006;Tangen & Bansal 2020) and topographic gradient (Richardson & Vepraskas 2000) affect organic carbon burial in restored wetlands. Areas that are low lying and flat often support wetland functions because of their capacity to receive and retain surface and groundwater flow and remain waterlogged (e.g. National Wetlands Working Group 1997; Richardson & Vepraskas 2000;Daniel et al. 2022). However, land management practices for wetland conversion and restoration, including soil compaction via heavy equipment (Hossler & Bouchard 2010;Brown & Norris 2017), removal of top soils containing organic carbon (e.g. Yu et al. 2017), and/or soil drainage (Hossler & Bouchard 2010;, can alter substrates and may ultimately reduce organic carbon masses upon restoration. Wetlands are commonly drained when creating agricultural land, which requires the reestablishment of waterlogged soils to restore wetland functions (Bortolotti et al. 2016;Winikoff et al. 2020). It is often most feasible to promote water and organic matter accumulation by removing drainage mechanisms and restoring or creating small depressions on former farmland. In this process, small depressions are either shaped in low-lying land with dikes or berms, or connected to natural flow regimes by enabling outflow or installing a spillway to sustain shallow water levels (Bortolotti et al. 2016;Loder et al. 2018;Winikoff et al. 2020). These types of wetlands can support high levels of biological production and plant cover within 10 years of restoration (Bortolotti et al. 2016;Brown & Norris 2017;Winikoff et al. 2020), which can lead to high concentrations of soil organic carbon (e.g. Badiou et al. 2011;Yu et al. 2017). In contrast, other studies show that organic carbon retention may be particularly low in soils of depressional restored wetlands (e.g. Ballantine & Schneider 2009;Brown & Norris 2017). More research is required to better constrain the recovery of organic carbon masses in restored depressional freshwater marshes situated in agriculturally intensive watersheds.
In this study, we use spatial and temporal analyses to identify drivers of organic carbon burial in a series of restored freshwater marsh sites and an undrained reference site in adjacent watersheds in the Lake Erie Basin in southern Ontario, Canada. Because rates of organic carbon accumulation are challenging to measure in restored wetlands due to their young age, we use paleorecords from a reference site and a passively restored wetland to calculate and compare apparent rates of organic carbon burial. We compare these baseline values with soil organic carbon stocks in 14 actively restored wetlands, which are <10 years in age and characterized by different landscape attributes. Our research objectives are to determine whether (1) organic carbon densities and masses increase in surface soils of restored wetlands with time since active restoration took place; (2) landscape attributes that drive the formation of natural wetland soils (i.e. underlying parent material, vegetation cover, elevation in watershed, slope) significantly affect organic carbon masses in restored wetland soils; and (3) recent apparent rates of organic carbon accumulation differ between soils in a passively restored wetland versus a reference site paleorecord in the same watershed and other sites in the literature. Using available predictions of wetland extent prior to the extensive losses occurring after European settlement (Byun et al. 2018), we also examine whether organic carbon masses in surficial soils of restored wetlands are higher at sites where presettlement wetlands existed.

Study Region
The Long Point Biosphere Reserve (42.6 N, 80.5 W; Fig. 1) is located in Norfolk County, Southern Ontario, Canada on the Restoration Ecology July 2023 northeast shore of Lake Erie, and was designated a UNESCO Biosphere Reserve because of its unique habitat that supports species at risk (Long Point Biosphere Reserve Foundation 2022). This study region is located on the traditional lands of the Attawandaron, Haudenosaunee, and the Anishinaabe First Nations and encompassed within the Treaty lands of the Mississaugas of the Credit. It is situated near the northern limit of the Carolinian Life Zone, which supports provincially rare tree species of eastern deciduous forest including Asimina triloba, Magnolia acuminata, and Sassafras albidum (Waldron 2003;Riley 2013). Furthermore, it is influenced by a temperate climate with daily mean temperatures of 21.1 C in July and À5.4 C in January, and annual rain and snowfalls of 906 mm and 130 cm, respectively (Environment and Climate Change Canada 2021).

Big
Creek and Dedrick-Young watersheds are located within the Long Point Biosphere Reserve and drain into Long Point Bay adjacent to the Long Point Sand Spit-the longest freshwater sand spit in the world and one of the most ecologically diverse areas in Canada ( Fig. 1; Chapman & Putnam 1984;Riley 2013). The main courses of Big Creek and Dedrick-Young are situated on low-lying topography and alluvial deposits from the postglacial and are surrounded by more elevated terrain underlain predominantly by glaciolacustrine sand deposits or clay and silt till ( Fig. 1A; Barnett 1993;Bunting et al. 1997;Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry 2012). Marshland was widespread across the Big Creek and Dedrick-Young watersheds prior to European settlement (1790 AD), particularly in the lower reaches where (1) the topographic gradient was low; (2) underlying parent materials (B) Presettlement wetlands, referring to wetland that has been drained and altered since European settlement in the region (Source: Byun et al. 2018). All present-day wetlands in Byun et al. (2018) are also considered to have been wetlands during the presettlement period. (C) Postsettlement wetlands, referring to wetland that remains intact at present day (Source: Byun et al. 2018). Soil types are shown for the Big Creek and Dedrick-Young watersheds as characterized by Soil Landscapes of Canada Working Group (2010). ONPFD: Ontario Plainfield soil (97% sand, 2% silt, and 1% clay). ONGOB: Ontario Gobles soil (29% clay, 52% silt, and 19% sand). of postglacial origin had low permeability; and/or (3) riverine and/or lacustrine influence (via flooding or heightened water table) were most prominent ( Fig. 1B; Byun et al. 2018).
The first significant period of European settlement took place between 1790 and 1820 AD (Chapman & Putnam 1984). Shortly after, large-scale deforestation and intensive landscape alterations occurred primarily for agriculture (Bunting et al. 1997;Riley 2013). Marshland was specifically targeted for ease of conversion to farmland because of its flat relief, herbaceous vegetation, and fertile soils (Fig. S1). Consequently, the proportion of wetland cover in the Big Creek Watershed declined to 9% as of 2015 (Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry 2019). Moreover, the land spanning the Big Creek and Dedrick-Young watersheds is largely agricultural (up to 71%) at present day and has now been heavily farmed, tilled, and fertilized (Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry 2019). The Big Creek and Dedrick-Young watersheds are also subject to intensive surface and groundwater extractions for agricultural irrigation and have dozens of designated points for permitted water extractions of >50,000 L per day (Ontario Ministry of Environment, Conservation and Parks 2021). There have been increasing efforts over the past two decades to restore freshwater marshes and biodiversity in Lake Erie watersheds due to issues related to eutrophication and decimation of native species (Harrison et al. 2020).

Site Descriptions
We analyzed 14 actively restored sites in the Big Creek and Dedrick-Young watersheds that range between 0.1 and 1.3 ha in areal extent and were restored by the Nature Conservancy of Canada (NCC) 1-8 years before the time of sampling (Table 1; Figs. 1, 2C, & 2D). These sites range between circumneutral to alkaline and comprise fresh waters with conductivity measurements predominantly <400 μS/cm (Table S1). The NCC restored these sites on former agricultural fields in the Long Point Biosphere Reserve where poorly drained soils were present and where waterlogged areas existed or had existed. To recover water retention, heavy equipment was used to move formerly farmed soils and remove drainage tiles (perforated pipes) or prevent tiles from draining water. All actively restored wetlands were excavated, except for site LBC5-W, which was enclosed by the creation of berms and has a spillway to maintain wetland hydrology. A mixture of native grass, herbaceous, and woody species (e.g. Andropogon gerardii, Asclepias spp., Aster spp., Carya spp., Cornus spp., Elymus spp., Juglans nigra, Juncus spp., Lupinus perennis, Platanus occidentalis, Quercus spp., Sorghastrum nutans) was sowed in the upland area around the newly created wetlands to establish early successional meadows and forest (NCC 2022). To date, over 790 ha of agricultural land has been restored to terrestrial and wetland habitat by the NCC in Norfolk County (L. Monck-Whipp, NCC, personal communication, 2022). Table 1. Characteristics of the reference, passively restored, and actively restored sites sampled in this study, which are located in southern Ontario, Canada. All sites are excavated depressions except for LBC5-W, which is enclosed by berms, and CBC3-01 and UBC4-01, which are the reference and passively restored sites, respectively. All sites are situated in the Big Creek watershed except for sites BB4-N, BB4-S, BB5-E, BB6-C, and BB6-E, which are situated in the Dedrick-Young watershed. *Calculated in ArcGIS using shapefiles produced by the Nature Conservancy of Canada. Areal coverage includes other wetland types that are hydrologically connected to marsh site at the surface. **According to presettlement (prior to 1850 CE) wetland map in Byun et al. (2018). We identified a passively restored floodplain marsh that has not been subject to active restoration in Upper Big Creek Marsh of the Big Creek Watershed (Table 1; Figs. 1 & 2B). The site was farmed before 1954 AD, but was reverted to wetland by 1964 AD when the adjacent river channel was straightened, former farmland was abandoned, and drainage infrastructure was no longer actively maintained ( Fig. S2; Norfolk County Geographic Information Systems, 2018). At present day, the passively restored marsh supports large stands of Typha spp., and an abundance of Lemna spp. and Nuphar lutea where standing water is present. This marsh is surrounded by thicket and treed swamp.

Site
Finally, we selected an undrained reference site located in Central Big Creek Marsh on the floodplain of Big Creek (

Sample Collection
A total of 75 soil samples were collected from 14 actively restored wetlands in July 2019 following protocols used in the Prairie Pothole Region (e.g. Euliss et al. 2006;Tangen & Bansal 2020). In doing this, we collected soil samples from three sampling locations within each site (Euliss et al. 2006;Tangen & Bansal 2020). If zonation was apparent, we collected soil from each of the following three zones: deep marsh (0.5-1 m water depth; open water, and submerged, attached and floating vegetation, e.g. algae, Brasenia); shallow marsh (<0.5 m in water depth; emergent and herbaceous vegetation, e.g. Typha, Cyperaceae); and wet meadow (waterlogged soils, and herbaceous and woody vegetation, e.g. Aster, Cornus, Equisetum, Populus, Salix). If zonation was not apparent, samples were taken from three locations at equidistance across the inundated area. We used an open-chambered Oakfield soil probe with a chamber diameter of 2.5 cm and length of 42 cm to collect soil samples at 0-15 cm and 15-30 cm depth. Where sediments were too loose to entirely fill the probe chamber or too stiff to penetrate to 30 cm depth, we collected the soil from the partially filled chamber (n = 5) or one surface soil sample representative of 0-30 cm depth (n = 9).
We recovered a 36-cm sediment core (UBC4-07) from the open-water area of the passively restored site in June 2019 using a Glew gravity corer (Glew et al. 2001). Standing water was about 50 cm in depth at the time of collection. The sediment core was extruded on site at 1-cm intervals for a total of 36 samples.
For our baseline measurements, we recovered a long sediment core (CBC3-01; 429 cm length) from the center of the reference marsh in October 2019 using a Russian peat corer (Eijkelkamp Soil and Water, Giesbeek, the Netherlands). Subsectioning of the CBC3-01 reference core is discussed in Loder et al. (2023). All samples from the actively restored, passively restored, and reference sites were stored in dark refrigeration at 4 C until analyses began.

Organic Carbon Densities and Masses
Subsamples of known volume were dried and weighed to constant mass for bulk density from the following increments: 0-15 cm and 15-30 cm depths for the actively restored sites, at contiguous 1-cm increments for the passively restored site, and for every second 2-cm increment for the reference site. Extractions from these subsamples were subsequently combusted at 550 C for 4 hours to estimate organic matter content via loss-on-ignition (LOI). We calculated organic carbon content as 50% of organic matter content. This conversion factor has been estimated for freshwater coastal wetlands in the Great Lakes region (Braun et al. 2020) and freshwater mineral marshes in the St. Lawrence lowlands in southern Québec (Magnan et al. 2023). It has also been used to estimate soil organic contents from LOI data in other studies in temperate North America (Loder & Finkelstein 2020).
We calculated organic carbon densities (g/cm 3 ) for each sample by taking the product of bulk density and organic carbon content for each increment. Then, we calculated and summed organic carbon masses of these increments to 30 cm depth at all study sites. For the reference site, we calculated organic carbon masses for 4-cm contiguous samples using the corresponding organic carbon density from every second 2-cm increment and summed these organic carbon masses (tC/ha) down to 30 cm depth. For the actively restored sites, we calculated and reported the average organic carbon mass of the three sample locations.

Chronology and Rates of Organic Carbon Accumulation
Dried sediment (approximately 0.4 g) from each 1-cm increment of UBC4-07 was ground to fine powder using a Retsch MM220 ball mill for trace metal analyses using inductively coupled plasma mass spectrometry at the Department of Physical Environmental Sciences at the University of Toronto Scarborough. Blanks, spikes, replicate samples, and reference materials were analyzed for quality assurance and control. We examined trace metal concentrations (As, Cd, Cu, Pb, and Zn) to track industrial particulate contamination for the purpose of establishing chronology in core UBC4-07 and estimating rates of accretion in the passively restored site. Lead (Pb) has been used as a chronostratigraphic marker in North American lake sediments as Pb concentrations and deposition peaked in the mid-1970s (e.g. Nriagu 1990; Health Canada 2013), then subsequently declined due to the phase out of leaded gasoline (Cheyne et al. 2018;Dunnington et al. 2020). Furthermore, we examined concentrations of aluminum (Al; an inert metal in mineral sediments), which we plotted against Pb concentrations (ratio of Pb/Al) to normalize for flood events that could increase the deposition of trace metals through minerogenic influxes (Dunnington et al. 2017).
Dating methods and other analyses of the CBC3-01 reference core are discussed in Loder et al. (2023). We calculated downcore rates of organic carbon accumulation at the passively restored and reference sites by taking the product of organic carbon density and the rate of accretion derived from the chronological data of UBC4-07 (total Pb peak) and CBC3-01 (age-depth models).

Landscape Attributes
We identified underlying parent material (above bedrock) for each site using soil characterizations from the Soil Landscapes of Canada database (Soil Landscapes of Canada Working Group 2010) and Quaternary Geology of Ontario dataset from the Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry (2012). Due to differences in spatial resolution of the two datasets, any discrepancies between these datasets were resolved based on observations in the field and laboratory. Second, we categorized sites into the following groups based on in situ vegetation cover at the time of sampling: "no growth" due to a lack of vegetation in the newly restored marshes; "partial vegetation cover"; and "dense vegetation cover" (Fig. S3). Third, we obtained a ground slope ( ) value representing the gradient of each site by extracting a slope raster value at 30-m resolution corresponding to the center location of each wetland from the Provincial Digital Elevation Model for Ontario (Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry 2013). Last, we obtained an elevation above the mean annual level of Lake Erie (approximately 174 m above sea level; NOAA 2022) for each site using a Garmin GPSMAP field device and Google Earth.

Statistical Analyses
For the actively restored sites, we first examined whether organic carbon densities were greater in more recent deposits between 0-15 cm and 15-30 cm depth increments using nonparametric Mann-Whitney U tests (at p < 0.05). Then, we evaluated the relationship between each landscape attribute and soil organic carbon in the actively restored wetlands. Because the ages are not normally distributed, we categorized the age of restored wetlands using the following groups: newly created (1 year in age; n = 5), 5-6 years in age (n = 5), and 7-8 years in age (n = 4). Given the small sample size and similarity among variances, we conducted separate nonparametric Mann-Whitney U or Kruskal-Wallis tests ( p < 0.05) to identify significant differences in organic carbon masses at 30 cm depth among age, soil type, and vegetation cover categories. If a significant difference was detected in the Kruskal-Wallis test, we conducted subsequent Dunn's multiple comparisons tests to identify which categories are significantly different (at 95% confidence level; p < 0.05). Because slope data are continuous, we conducted a Spearman correlation test to determine if a significant relationship (at p < 0.05) exists between slope values and organic carbon masses.

Organic Carbon Densities and Masses
We found no significant difference between organic carbon densities at 0-15 cm and 15-30 cm depths in the actively restored marsh soils (W = 483; p = 0.77; Fig. S4). Instead, organic carbon densities in the restored wetland soils are highly variable among sites and sometimes highly variable within sites (Table 1). Some newly restored sites have similar or even markedly higher organic carbon densities in comparison to sites 5-6 and 7-8 years in age (Fig. 3A). Derived from these densities, average organic carbon masses at 0-30 cm depth for the actively restored site soils ranged between 16.0 and 115.0 tC/ha. Site LBC5-W has the highest organic carbon mass of all actively restored sites and is the only wetland that was not excavated upon restoration.
Organic carbon densities in core UBC4-07 from the passively restored marsh are highest in the bottom 10 cm, and gradually decrease up-core where they reach a minimum in the upper-most sample (Fig. S5a). While the passively restored marsh has a similar organic carbon mass in its surface soils at 0-30 cm depth to those of the actively restored sites (Table 1), its average (AE standard deviation) bulk density (0.19 AE 0.12 g/cm 3 ) is several-fold lower than those of the actively restored marshes (1.3-1.6 g/cm 3 ).
The reference marsh has the highest organic carbon mass at 0-30 cm depth (Table 1). Throughout the 4.3 m soil profile of the reference marsh, organic carbon densities range between 0.01 and 0.12 g C/cm 3 (Fig. S5b). Densities are generally highest in the most recently deposited sediments and decrease downcore due to long-term decay (Loder et al. 2023).

Relationships With Landscape Attributes
The actively restored sites are located at elevations between 21 and 37 m above Lake Erie, as compared to the passively restored and reference sites which are nearly at lake level ( Table 1). The reference and passively restored marshes are located on presettlement wetland, while 4 of the 14 actively restored marshes are located on presettlement wetland according to Byun et al. (2018; Table 1). We found no significant relationship between organic carbon masses and age in actively restored Figure 3. Organic carbon densities amalgamated over collection sites and depth increments in the actively restored marshes shown by (A) age, (B) soil type, and (C) vegetation cover. Box plots show the median values, first and third quartiles, maximum and minimum values, and outliers. Panel (D) shows organic carbon masses and topographic slopes (degrees). A significant difference or trend that was detected for organic carbon masses is marked with an asterisk (*) in the given panel. No significance is denoted as "NS." wetland soils. Instead, we found that underlying parent material is the strongest factor driving masses of soil organic carbon. Sites underlain by clay and silt have significantly greater organic carbon masses (107 AE 9 tC/ha) than sites underlain by sand (39 AE 19 tC/ha; W = 49; p < 0.001; Table 2; Fig. 3B). Most of the older restored wetlands had partial vegetation cover (n = 7), while only two had dense vegetation cover (n = 2). Nonetheless, we found high variability in organic carbon masses among sites with similar vegetation cover (Fig. 3C).

Chronology and Rates of Organic Carbon Accumulation
We recorded a peak in Pb concentrations (approximately 37 ppm) and Pb/Al values at 23 cm in the passively restored marsh core (UBC4-07), and subsequent decline in concentrations up-core (Figs. 4 & S6). Given Pb emissions and consumption of leaded gasoline peaked in the mid-1970s, we assign the 23 cm depth mark an age of 1975 AD, and estimate that rates of accretion were approximately 5 mm/year. These rates are comparable to the short-term rates of accretion measured using radioisotope dating in the reference site (Loder et al. 2023).
Using 5 mm/year as the rate of accretion, we estimate that the mean short-term rate of organic carbon accumulation is 144 g C m À2 yr À1 and a range of rates between 44 and 290 g C m À2 yr À1 at the passively restored site (Fig. 5A). These rates were highest shortly after restoration and gradually decreased after 1990 AD reaching a minimum rate at present day. Basal sediments in the reference site (core CBC3-01) date to 5,710 calendar years before present in relation to 1950 AD (hereafter cal. yr BP; Fig. 5B) and are associated with the development of marshland in coastal watersheds of Lake Erie (Finkelstein & Davis 2006;Lewis et al. 2012). Since formation, rates of Table 2. Comparison of organic carbon masses among our study sites and other freshwater marshes in North America at 0-30 cm depth. Time since restoration is provided in parentheses where available. *Organic carbon mass calculated for 0-20 cm, assuming organic carbon is 50% of organic matter content. **Organic carbon mass calculated for 0-15 cm, assuming organic carbon is 50% of organic matter content. + Equivalent soil mass procedure (Ellert & Bettany 1995) applied to organic carbon masses in restored site soils.

Region
Wetland accretion have ranged between 0.2 and 6.7 mm/year at Central Big Creek Marsh. Rates of accretion notably increased between 630 and 80 cal. yr BP, and reached maximum values between 1983 and 2019 AD. The mean long-term rate of organic carbon accumulation through the paleorecord is 41 g C m À2 yr À1 , while the mean short-term rate of organic carbon accumulation since 1983 is 235 g C m À2 yr À1 .

Discussion
Our study underscores the notion that organic carbon retention in wetland soils takes place over several decades. We demonstrate that organic carbon masses of the restored soils do not increase with time since active restoration took place in sites <10 years in age and at many sites do not yet approach those of natural wetlands. We use paleorecords, presettlement wetland maps, and spatial analyses to demonstrate that substrate, topography, and hydrological influence from larger surface and groundwater networks have a strong influence on organic carbon stocks and burial in the freshwater marsh sites, and drive their potential to be net carbon sinks.

Substrate Strongly Affects Organic Carbon Masses Among Actively Restored Marshes
Regardless of age and vegetation cover, parent material has the strongest relation to organic carbon masses in soils 0-30 cm in depth among the actively restored wetlands sampled in the Long Point Biosphere Reserve. Actively restored wetlands underlain by clay and silt till have imperfect drainage and low infiltration which likely drives their higher organic carbon masses compared to restored wetlands underlain by well-drained sands with high infiltration and situated at high elevations in the Big Creek and Dedrick-Young Watersheds. While the sample size for comparison here was limited, similar results have been recorded elsewhere. Clays and silts are also characteristic of postglacial parent material beneath productive freshwater marshes in the Prairie Pothole Region (e.g. Euliss et al. 2014) and were found  Figure 5. Rates of organic carbon accumulation measured for the (A) passively restored site (UBC4-07) assuming an average rate of accretion of 5 mm/year based on peak abundance of total Pb, and for the (B) reference site (CBC3-01) using rates of accretion derived from age-depth models. Organic carbon contents were estimated as 50% of organic matter contents measured via LOI.
to increase the recovery of soil organic carbon in restored and created wetlands across the United States (Yu et al. 2017). Clay and silt particles may further enhance organic carbon burial in restored wetland soils because they are more nutrient rich and have the capacity to adsorb on and stabilize organic carbon (e.g. Yu et al. 2017;Huang et al. 2023). Our results are consistent with several lines of evidence describing parent material and its permeability as key drivers in sustaining hydric soils and soil carbon masses (Richardson & Vepraskas 2000;Mitsch & Gosselink 2015;Byun et al. 2018). Nonetheless, we cannot discount the role of groundwater discharge zones in supporting saturation in the upper soil profiles of wetlands underlain by sandy soils (Richardson & Vepraskas 2000). Given our limited sample size, further exploration is required to assess the relationship between groundwater discharge zones underlain by sandy soils, and wetland formation and organic carbon burial.

Actively Restored Marshes Differ From Presettlement Marshland in the Long Point Biosphere Reserve
We found that most actively restored sites in the Long Point Biosphere Reserve are not situated where wetlands (larger than 0.5 ha) were present prior to European settlement according to Byun et al. (2018). Instead, we found they are small in areal extent and similar to actively restored freshwater wetlands in other regions (e.g. Moreno-Mateos et al. 2012;Bortolotti et al. 2016;Loder et al. 2018). Although geographically isolated wetlands provide a multitude of important functions (e.g. Cohen et al. 2016), our series of very small and relatively isolated restored wetlands is functionally dissimilar from the extensive, contiguous marshland that was present in this area of southern Ontario prior to European settlement (Byun et al. 2018). Collectively, this suggests that most of the actively restored sites in the Long Point Biosphere Reserve may be better characterized as created wetlands because they may not have supported wetland functions prior to European settlement. The presettlement map of Byun et al. (2018) has a minimum mapping unit of 0.5 ha and thus may not capture small depressional wetlands that were present or adjacent to larger areas of wetland. Additional sampling is required to determine how area and connectivity to larger presettlement wetland systems affects organic carbon stocks in actively restored marshes. Organic carbon masses in the upper 30 cm of the actively restored marsh soils are markedly lower than that of our reference site and those of reference and restored sites in other agriculturally intensive areas in North America. Furthermore, we did not detect higher organic carbon densities in surficial soils (0-15 cm) where most deposition would have occurred since active restoration took place than that in deeper soils (15-30 cm) of the actively restored marshes. These results indicate that soil development and carbon stock recovery are slow in the actively restored marshes as shown in other restored depressional wetlands (e.g. Ballantine & Schneider 2009;Hossler & Bouchard 2010;Brown & Norris 2017).
We also found that the bulk density of actively restored soils is markedly higher than that in the reference and passively restored soils. We further suspect that active restoration has caused soil compaction and resulted in higher soil masses and possibly lower organic carbon masses (e.g. Hossler & Bouchard 2010;Brown & Norris 2017) in comparison to the reference site which has not been subject to human-induced drainage or soil movement with heavy equipment. These kinds of discrepancies should be recognized when comparing organic carbon masses in actively restored wetlands to those of the reference and passively restored sites (e.g. Ellert & Bettany 1995;Tangen & Bansal 2020).

Fluvial Influences on Carbon Burial at Reference and Passively Restored Marshes
We found that the reference and passively restored sites in the Long Point Biosphere Reserve have high potential to receive and hold larger volumes of water because of their location on the floodplains of the Big Creek watershed (Moreno-Mateos et al. 2012;Peteet et al. 2020;Daniel et al. 2022). Using the reference paleorecord, we demonstrate that the rates of organic carbon accumulation have been influenced strongly by inundation and hydro-fluvial events during which the water table and sediment supply were high at the reference marsh (Loder et al. 2023). These factors have been shown to increase the speed of recovery of wetland functions at restored sites (e.g. Moreno-Mateos et al. 2012;Yu et al. 2017;Poppe & Rybczyk 2021) and may have promoted the recovery of wetland functions at the passively restored site. Deposits from the Big Creek channel likely elevate burial of allochthonous organic carbon and minerogenic sediments and help preserve autochthonous organic matter in deeper sediment layers (Kemp et al. 2019;Spivak et al. 2019), in part explaining the high rates of carbon accumulation at the reference and passively restored marshes. We recognize that the reference site is the only site in this study that supports dense monotypic stands of Phalaris arundinacea. This requires further consideration because P. arundinacea can produce high yields of biomass (Galatowitsch et al. 1999;Lavergne & Molofsky 2004) and may consequently increase rates of carbon uptake and burial at the reference site like other invasive wetland plants (e.g. Phragmites australis, Typha angustifolia, Typha Â glauca; e.g. Rothman & Bouchard 2007;Bills et al. 2010).
Restoration of freshwater marshland in optimal locations can lead to rates of organic carbon accumulation as high as those in other reference or intact freshwater marshes (e.g. Graham et al. 2005;Bernal & Mitsch 2012;Tangen & Bansal 2020). Rates in the passively restored marsh were particularly high shortly after soils ceased to be actively drained and upon reestablishment of waterlogged soils and riverine influence from the Big Creek channel. While we recognize the limitation in measuring rates of organic carbon accumulation based on one assigned date (i.e. total Pb peak), we found that rates have declined over the past 30 years and are now several-fold lower than short-term rates at the reference marsh as found elsewhere (e.g. Drexler et al. 2013). In marine-influenced watersheds, high rates of accretion and organic carbon accumulation may occur shortly after restoration of tidal marshland that has been previously diked and drained, and has consequently subsided, due to the large amount of available accommodation space between the marsh elevation and tidal frame (Drexler et al. 2019). These rates may eventually decline as inorganic and organic materials vertically accrete and the accommodation space decreases (Drexler et al. 2019). Similar processes likely occur in riverine-influenced freshwater marshes that have been farmed and restored on lowlying floodplains. This type of wetland restoration site in coastal watersheds of Lake Erie likely has better restoration potential for organic carbon burial than that of created sites.
In comparison to the passively restored marsh and reference site, the actively restored marshes are likely influenced predominantly by autogenic processes and rely on the production and preservation of in situ organic matter for sustaining organic carbon burial (Ballantine & Schneider 2009). Despite the small sample size, we suspect that some of the actively restored wetlands are not predisposed to the same conditions that drive high water tables and rates of organic carbon accumulation in the reference and passively restored sites in this particular watershed. They may have lower potential to accumulate organic carbon at high rates.

Recommendations for Conservation and Restoration
Our study demonstrates the importance in prioritizing conservation of intact freshwater marshes, such as Central Big Creek Marsh, where high rates of organic carbon burial have occurred over thousands of years and carbon stocks several meters in depth are present. While wetland restoration can increase organic carbon burial and preservation in soils, masses and rates in restored freshwater marshes are nonetheless lower and do not approach levels of natural marsh soils in the Long Point Biosphere Reserve within 10 years since active restoration took place. Thus, conservation of undrained freshwater marshes has higher potential as a natural climate solution.
Although restoration of freshwater mineral marshes has been identified as a pathway for reducing greenhouse gas emissions over the next decade (e.g. Drever et al. 2021), we show that active restoration of depressional marshes does not recover organic carbon stores in soils within this time frame. The lengthened time over which restored and created freshwater marshes may need to become net carbon sinks requires careful consideration when implementing wetland restoration and creation as natural climate solutions and incorporating their greenhouse gas inventories into regional and national carbon budgets. Despite the slow functional recovery of carbon accumulation, restoration sites may be selected where underlying soils have low infiltration, topographic gradient is low, and connectivity to a larger waterbody (e.g. river channel) or zone of groundwater discharge is apparent to maximize organic carbon preservation. Restoration sites may also be more favorable where wetland functions can be restored and sustained passively with minimal human intervention. In these cases, excavation and soil movement are not required, which would minimize disturbance to preexisting organic carbon stocks.
Finally, we demonstrate how paleorecords are valuable in understanding functions of organic carbon burial and creating pre-and postsettlement baselines for wetland restoration as a natural climate solution in a given landscape. We show that freshwater marshes in the Long Point Biosphere Reserve formed in low-lying, flat floodplains underlain by impermeable sediments and evolved from inundation events during which organic carbon and inorganic materials were deposited. We also show that major landscape alterations after European settlement (e.g. agricultural activity, habitat fragmentation, proliferation of invasive species, soil alteration) may affect and in some cases limit organic carbon burial in the contemporary record of both reference and restored sites. Consideration of pre-and postsettlement conditions can help prioritize restoration sites for optimizing the recovery and sustenance of organic carbon burial and inform management strategies that support wetland conservation over short and long periods of time.

Supporting Information
The following information may be found in the online version of this article: Figure S1: Land-use classification in the Big Creek watershed of the Long Point region in southern Ontario, Canada. Figure S2. Aerial photos showing and highlighting the passively restored site by the red boxes. Figure S3. On-the-ground photos of four actively restored wetland sites. Figure S4. Organic carbon densities measured at 0-15 cm versus 15-30 cm depth increments, and amalgamated from all collection sites in each of the actively restored marshes. Figure S5. Organic carbon densities down core in the (a) passively restored marsh (UBC4-07) and (b) reference marsh (CBC3-01). Figure S6. Concentrations (ppm) of lead (Pb), arsenic (As), cadmium (Cd), copper (Cu), iron (Fe), and zinc (Zn) down core at 1-cm increments in UBC4-07. Table S1. Average water quality measurements (AE standard error).