Introduction
There is growing concern about marked declines in wetland areal extent, with more than a fifth of global wetlands lost since 1700 CE (Fluet-Chouinard et al., Reference Fluet-Chouinard, Stocker, Zhang, Malhotra, Melton, Poulter and Kaplan2023). In some regions, wetland losses have been much more extensive. For example, in southern Ontario, Canada, more than 87% of wetlands, mostly marshland, have been lost for agricultural conversion (Snell, Reference Snell1987; Ducks Unlimited Canada, 2010; Byun et al., Reference Byun, Finkelstein, Cowling and Badiou2018). Given that wetlands provide several critical ecosystem functions, including groundwater recharge, flood mitigation, improved water quality, and biodiversity protection, as well as carbon sequestration, these losses have major repercussions for the biodiversity and climate crises. As a result, wetland restoration is increasingly practiced to reestablish these wetland functions (Mitsch and Gosselink, Reference Mitsch and Gosselink2015; Environment and Climate Change Canada, 2020).
Paleoecological analyses of wetland sediments, including both the organic and mineral fractions in those sedimentary sequences, have been widely applied to reconstruct past environmental changes and have also been used to inform conservation of habitat and watersheds more broadly (Clarke and Lynch, Reference Clarke and Lynch2016; Finlayson et al., Reference Finlayson, Clarke, Davidson and Gell2016; Boxem et al., Reference Boxem, Davis and Vermaire2018). Past changes in vegetation and sedimentary dynamics provide reference conditions for hydrology, sediment inputs, and ecological succession, which modulate biodiversity and carbon cycling on both short- and long-term timescales in freshwater mineral marsh environments (McNellie et al., Reference McNellie, Oliver, Dorrough, Ferrier, Newell and Gibbons2020; Peteet et al., Reference Peteet, Nichols, Pederson, Kenna, Chang, Newton and Vincent2020; Hu et al., Reference Hu, Mushet and Sweetman2023; Nguyen et al., Reference Nguyen, Hapsari, Saad, Sabiham and Behling2023; Ryu et al., Reference Ryu, Liu, Mccloskey and Yun2023). Thus, paleoecological reconstructions can inform wetland restoration by identifying habitat types, vegetation communities, and watershed conditions before anthropogenic disturbances to help determine restoration benchmarks, and by identifying hydrologic and geomorphic conditions that affect vegetation turnover and sustain critical wetland functions (Manzano et al., Reference Manzano, Julier, Dirk, Razafimanantsoa, Samuels, Petersen, Gell, Hoffman and Gillson2020; McNellie et al., Reference McNellie, Oliver, Dorrough, Ferrier, Newell and Gibbons2020; Loder et al., Reference Loder, Zamaria, Arhonditsis and Finkelstein2023).
While climate changes controlling the length of the growing season and net moisture availability are often the ultimate drivers of change in wetlands, hydrological and sedimentological processes are often influenced strongly by their position in the watershed and are generally more proximate causes of shifts in wetland vegetation communities (Junk et al., Reference Junk, An, Finlayson, Gopal, Květ, Mitchell, Mitsch and Robarts2013; Courtney Mustaphi et al., Reference Courtney Mustaphi, Kinyanjui, Shoemaker, Mumbi, Muiruri, Marchant, Rucina and Marchant2021). Vegetation communities in wetlands are highly sensitive to surface moisture balance, water table position, and its seasonality (hydroperiod), as well as to changes in sediment loading in response to natural processes or anthropogenic disturbances (Delcourt and Delcourt, Reference Delcourt and Delcourt1991; Peteet et al., Reference Peteet, Nichols, Pederson, Kenna, Chang, Newton and Vincent2020). Reconstructing regional paleoclimate alongside analyses of proxies for local hydrological and geomorphic conditions is, therefore, an effective approach to identifying the extent to which larger-scale watershed processes and/or more local processes drive changes in a wetland system (Byun et al., Reference Byun, Cowling and Finkelstein2022; Willard et al., Reference Willard, Jones, Alder, Fastovich, Hoefke, Poirier and Wurster2023). This approach also helps to identify the conditions under which disrupted wetland functions may be effectively recovered to ensure successful restoration outcomes.
The objectives of this study are to devise habitat baselines over the Holocene and reconstruct wetland plant community changes since wetland initiation in a regional biodiversity hotspot where restoration is a major focus. We evaluate the roles of regional paleoclimate, hydrology, and sedimentary dynamics in wetland vegetation change, and place the recent changes over the European settlement era into a long-term context. We include analyses of sediment grain size and geochemistry to reconstruct fluvial processes, sediment sources, and changes in stream base level resulting from shifts in the level of Lake Erie (Camill et al., Reference Camill, Umbanhowar, Geiss, Hobbs, Edlund, Shinneman, Dorale and Lynch2012; Stewart and Desloges, Reference Stewart and Desloges2014; Peteet et al., Reference Peteet, Nichols, Pederson, Kenna, Chang, Newton and Vincent2020). These results are integrated with pollen analysis to reconstruct local vegetation communities and habitat types (Nguyen et al., Reference Nguyen, Hapsari, Saad, Sabiham and Behling2023) and with diatom assemblages to reconstruct aquatic environments. Diatoms are widely used paleolimnological proxies and have also been applied successfully to tracking changes in epiphytic, benthic, and open-water habitats in coastal, wetland, and floodplain contexts (Finkelstein and Davis, Reference Finkelstein and Davis2005; Hargan et al., Reference Hargan, Rühland, Paterson, Holmquist, Macdonald, Bunbury, Finkelstein and Smol2015).
Study site
The study site is a riparian mineral marsh on a floodplain within the central reaches of Big Creek in Norfolk County, southern Ontario, Canada (42.65°N, 80.54°W; Fig. 1A and B). The watershed is the largest in the Long Point region with an area of 750 km2 (Lake Erie Region Source Protection Committee, 2022), and is situated within Ecoregion 7E in Ontario’s provincial Ecological Land Classification system (https://www.ontario.ca/page/ecosystems-ontario-part-1-ecozones-and-ecoregions), which is part of the Carolinian Life Zone. This region contains the highest biodiversity and the greatest number of threatened species in Canada, which has led to the establishment of several protected areas and several internationally significant designations, including the Long Point World Biosphere Reserve, the Long Point Walsingham Forest Priority Place, and the Big Creek National Wildlife Area (Environment and Climate Change Canada, 2020).
Figure 1. (A) Location of the study site in the North American Great Lakes region (red circle). (B) Location of the CBC3-01 core site (red circle) in comparison with the lake pollen records locations at Decoy Lake and Hams Lake (blue circles) in the southern Ontario region. (C) Location of the study site (CBC3-01) and the physiographic and surficial geologic units within the Big Creek watershed (Chapman and Putnam, Reference Chapman and Putnam1980; Ontario Geological Survey, 2010). (D) Vegetation communities surrounding the CBC3-01 core site and common in the watershed more broadly (base map is Google Earth Image @ 2024 Terra Metrics, NOAA Airbus; vegetation units provided by Ontario Ministry of Natural Resources and Forestry (MNRF) and the Nature Conservancy of Canada (NCC) and mostly follow Lee et al. (Reference Lee, Bakowsky, Riley, Bowles, Puddister, Uhlig and McMurray1998). The maps are drawn using the GCS NAD 1983 datum and the Great Lakes and St. Lawrence Albers projection. Acronyms for vegetation units: RM = Reed Canary Grass Mineral and Organic Meadow Marsh Bulrush Organic Shallow Marsh; BR = Buttonbush Organic Thicket Swamp–Reed Canary Grass Organic Meadow Marsh; BODS = Black Ash Organic Deciduous Swamp; DW = Deciduous Woodland; DFHWF = Dry and Fresh Black Oak–White Oak Tallgrass Woodland–Dry Fresh Hardwood–Hemlock Mixed Forest–Hardwood Deciduous Forest–Fresh Moist and Dry Fresh Sugar Maple–Dry Fresh White Pine–Oak and Sugar Maple Mixed Forest; FMM = Forb Mineral Meadow Marsh; FMHF = Fresh Moist Hemlock–Hardwood and Fresh Moist Sugar Maple Hemlock Mixed Forest; MPSF = Managed Red Pine Plantation–Fresh Moist Sugar Maple Deciduous Forest; MDMS = Mixed Forb Mineral Meadow Marsh–Dogwood Mineral Thicket and Green Ash Mineral Deciduous Swamp; NSMM = Narrow-leaved Sedge Graminoid Mineral Meadow Marsh; RMDSM = Red Osier Mineral Deciduous and Silky Dogwood Mineral Thicket Swamp, Sedge Mineral and Mixed Forb Mineral Meadow Marsh; SMMT = Swamp Maple Mineral Deciduous Swamp–Silky Dogwood Mineral Thicket Swamp; WMDS = Willow Mineral Deciduous Swamp–Silky Dogwood and Red Osier Dogwood Mineral Thicket Swamp; YBMS = Yellow Birch Mineral Deciduous Swamp.
Big Creek is a sand-bedded meandering river located partially within the Norfolk Sand Plain, a glacial feature formed by deltaic and lacustrine deposits associated with glacial Lakes Whittlesey and Warren. This sand plain was deposited ∼13,000 years ago and reaches >25 m thickness in parts of the Big Creek watershed (Barnes, Reference Barnes1967; Sibul et al., Reference Sibul, Wang and Vallery1977; Chapman and Putnam, Reference Chapman and Putnam1980; Megens, Reference Megens2015; Lake Erie Region Source Protection Committee, 2022; Fig. 1). Tills, glaciolacustrine clays, and organic deposits constitute the remaining surficial deposits within the watershed (Barnett, Reference Barnett1993; Ontario Geological Survey, 2010; Fig. 1). The underlying bedrock is mainly limestone and dolostone from the Dundee Formation of the Middle Devonian Period (Paleozoic Era), although outcrops are rare due to the thick glacial overburden (Barnett, Reference Barnett1993). Humic gleysols and organic soils are common soil types in the watershed, due to the prevalence of poorly drained glaciolacustrine clays and the generally low gradients (Soil Landscapes of Canada Working Group, 2010; Pennock, Reference Pennock, Krzic, Walley, Diochon, Paré and Farrell2021; Warren, Reference Warren, Krzic, Walley, Diochon, Paré and Farrell2021).
Local climate is humid-temperate with daily mean temperatures of 21.1°C in July and −5.4°C in January, and average total annual precipitation of 1036 mm (Delhi, ON, 1981–2010; Environment and Climate Change Canada, 2024). Owing to significant land-use change, this ecoregion has retained only 16% of its pre-settlement vegetation, which consisted of deciduous forest, prairie, savannah, and wetlands (Oldham, Reference Oldham2017; Henson and Brodribb, Reference Henson and Brodribb2005). As a result, several conservation and restoration initiatives have been ongoing in the Big Creek watershed over the last few decades to protect and restore critical habitat functions (Harrison et al., Reference Harrison, Reisinger, Cooper, Brady, Ciborowski, O’Reilly, Ruetz, Wilcox and Uzarski2020). These initiatives include protecting inland and coastal forested wetland and marshland, planting open and tall grass vegetation on former agricultural lands, removing invasive species that are decimating biodiversity in the region (e.g., Phragmites australis), and restoring wetland functions. Because much of the Big Creek watershed has been subject to intensive agricultural activity (Byun et al., Reference Byun, Finkelstein, Cowling and Badiou2018), active restoration measures are being used to recover mineral-based marsh habitat, which was widespread before settlement (Loder et al., Reference Loder, Zamaria, Arhonditsis and Finkelstein2023). Measures include removing drainage infrastructure, excavating soils to create depressions and berms for water retention, and sowing native species around newly restored wetlands on former farmland in both riparian zones and uplands of the watershed (Loder et al., Reference Loder, Zamaria, Arhonditsis and Finkelstein2023). Given the strong focus on marsh restoration in the Big Creek watershed, we chose a reference site that currently supports mineral-based marsh habitat and has not been previously drained to reconstruct ecological succession and changes in hydrologic and geomorphic conditions for identifying a suitable baseline for restoration.
The core site (CBC3-01) is a mineral marsh located on the floodplain within the Big Creek watershed, ∼2 m above mean lake level of Lake Erie (174 m asl), and 8 km upstream from the Lake Erie coast (Fig. 1). The core site is located centrally within the meander belt, ∼130 m from the river channel, and is connected to the river system via overbank flooding. The specific core site was selected to maximize the depth of organic-bearing sediments, as this increases the resolution and detail of paleoecological records. An Oakfield auger was used to probe sediment depths within the meander belt and on the floodplain. At the time of sampling in the month of October, the water table was about 30 cm below surface; however, this varies seasonally, and standing water up to a depth of 50 cm has been observed in the spring. Marsh sediments at the site consist of a mix of mineral sediments with organic content; the organic fraction makes up about 20% by mass, although this varies with depth. A highly clastic layer with low organic matter content was noted between the depths of 20 and 40 cm, interpreted to have resulted from watershed-scale disturbance, including large-scale deforestation associated with the European settlement era (Loder et al., Reference Loder, Gillespie, Haeri Ardakani, Cordero Oviedo and Finkelstein2025). Vegetation at the core site at the time of sampling consisted of an extensive cover of mono-dominant reed canary grass (Phalaris arundinacea) with other wetland emergents, and aquatics including cattails (Typha spp.), waterlilies (Nymphaea), diverse sedges (Cyperaceae), duckweed (Lemna), and bur-reed (Sparganium). Thicket swamp dominated by Cephalanthus occidentalis fringes the marshy sections (Fig. 1).
While there are several protected areas and numerous wetland restoration projects in this watershed, land use in the Big Creek watershed is mostly agricultural. This poorly drained flat terrain frequently required drainage to permit productive agricultural activities, resulting in significant losses of wetland extent in the Long Point region since European settlement in the early 1800s (Heffernan, Reference Heffernan1978; Dakin and Skibicki, Reference Dakin and Skibicki1994). The remaining wetlands within the watershed consist of marshes and swamps overlying both mineral and organic soils. Many of these wetlands have been heavily impacted by surrounding land use, including hydrological disturbances related to water extraction for irrigation, dam installations, and widespread tile drainage, as well as biophysical disturbances related to excessive nutrient runoff and erosion from tillage, and extensive spread of invasive species (Wilcox et al., Reference Wilcox, Petrie, Maynard and Meyer2003; Lake Erie Region Source Protection Committee, 2022; Yuckin et al., Reference Yuckin, Howell, Robichaud and Rooney2023; Hrynyk et al., Reference Hrynyk, Behnamian, Banks, Chen, Harmer, Kirby, White, Pasher and Duffe2024). Because of these impacts, and the combined biodiversity and climate crises, there is rapidly increasing interest in wetland restoration in the Long Point region to address threats to biodiversity, mitigation of nutrient runoff, regulation of streamflow, and natural climate solutions (Kraus et al., Reference Kraus, Norman, McFarlane, Lemieux, Jacob and Gray2021).
Methods
Site CBC3-01 was selected for analysis because it has not previously been drained and contains a continuous sediment record extending at least 6000 years, to the Middle Holocene (Loder et al., Reference Loder, Gillespie, Haeri Ardakani, Cordero Oviedo and Finkelstein2025). The CBC3-01 core was collected using a Russian-style peat corer with a barrel length of 50 cm and diameter of 5 cm. Further details on the retrieval of the 430 cm core, calculations of long- and short-term carbon accumulation using laboratory measurements of bulk density and elemental analysis on every second sample of 2 cm thickness through the core, and total organic carbon to total nitrogen ratios (TOC/TN) are described in Loder et al. (Reference Loder, Gillespie, Haeri Ardakani, Cordero Oviedo and Finkelstein2025). Subsamples were removed from the core using a custom angle bracket cleaned between samples; once removed from the core, the faces of each prism were cleared to remove potential contamination, and they were then cut to known volume using a subsampler cleaned between samples. An age–depth model developed using ten 14C dates and the Ambrosia pollen rise is presented in Loder et al. (Reference Loder, Gillespie, Haeri Ardakani, Cordero Oviedo and Finkelstein2025) and is included in the Supplementary Material (Supplementary Fig. S1, Supplementary Tables S1 and S2).
Pollen analysis
Pollen counts were undertaken at 10 cm intervals through the 430 cm core, with an additional 16 samples to improve detail in sections of interest, to track vegetation change. After radiometric dating results were obtained, a resolution interval of every 10 cm was chosen as representative for the ecological states through time. A total of 59 samples were processed using standard methods of acid digestion and sieving (Faegri and Iversen, Reference Faegri and Iversen1989). Palynomorphs were identified to the highest taxonomic level possible using a transmitted-light microscope at 400× and 1000× magnifications. Pollen keys (McAndrews et al., Reference McAndrews, Berti and Norris1973) and reference collections from the laboratory and the Royal Ontario Museum were used to identify the pollen grains. An average of 400 grains were counted per sample, and given the focus on local wetland vegetation dynamics, percentages were calculated using all taxa recorded. Percentages were plotted stratigraphically using C2 (Juggins, Reference Juggins2022); pollen zones (Stages P1, P2, P3, and P4) were defined based on a stratigraphically constrained hierarchical clustering and broken stick model (Gordon and Birks, Reference Gordon and Birks1972; Grimm, Reference Grimm1987; Bennett, Reference Bennett1996) with R (R Core Team, 2021). Alpha diversity of pollen types for each sample was reported based on the number of distinct pollen taxa observed in the sample. This metric can be a useful indicator of broad changes in vegetation diversity (Connor et al., Reference Connor, Van Leeuwen, Van Der Knaap, Akindola, Adeleye and Mariani2021) but must be interpreted with caution due to variation in the possible taxonomic resolution between samples and taxa and differences between plant taxa in terms of pollen production and dispersal (Peros and Gajewski, Reference Peros and Gajewski2008; Bhatta et al., Reference Bhatta, Cao, Felde, Grytnes, Birks and Birks2024).
Regional paleoclimate reconstructions were developed from two available lake pollen records in southern Ontario. Note that the CBC3-01 pollen record in this study was not used for the modern analog technique (MAT) paleoclimate reconstruction, given the frequent high proportions of local aquatic pollen (i.e., after excluding the aquatic pollen counts, the remaining pollen sums were deemed not enough to represent the regional forest pollen percentages). Pollen data were accessed through the Neotoma Paleoecology Database (https://apps.neotomadb.org/explorer/), and the two lake sites, Hams Lake (Neotoma Site ID: 971) (Bennett, Reference Bennett1987) and Decoy Lake (Neotoma Site ID: 656) (Szeicz and MacDonald, Reference Szeicz and MacDonald1991), were selected based on their proximity to the CBC3-01 site, their location in upland environments to capture a representative regional pollen rain, and available radiocarbon age data (Fig. 1B). Pollen-based paleoclimate reconstruction was performed on these two lake pollen records using the MAT (Chevalier et al., Reference Chevalier, Davis, Heiri, Seppä, Chase, Gajewski and Lacourse2020), following a detailed preestablished procedure (Willard et al., Reference Willard, Jones, Alder, Fastovich, Hoefke, Poirier and Wurster2023) that includes the exclusion of Ambrosia and the regional split of continental-scale pollen taxa such as Pinus and Picea to prepare the modern reference relationships between climate variables and pollen abundance (Whitmore et al., Reference Whitmore, Gajewski, Sawada, Williams, Shuman, Bartlein and Minckley2005). Although the calibrated age information for each pollen sample was available with the original dataset, the age–depth relationships of both records were constructed again using a Bayesian chronology model implemented in R (Haslett and Parnell, Reference Haslett and Parnell2008), to update calibration curves for the measured radiocarbon ages and to avoid the previously assumed event ages such as Tsuga decline (Booth et al., Reference Booth, Brewer, Blaauw, Minckley and Jackson2012). Four climate variables—average annual temperature (tave), mean temperature of the coldest month (tmin), mean temperature of the warmest month (tmax), and total annual precipitation (annp)—were extracted from five closest modern analogs by squared-chord distance from the fossil pollen assemblages. The extracted climate values from the analog matches were averaged and standardized to z-scores for comparison with a regionally downscaled time-series extraction of the recent paleoclimate model simulation output from the Community Climate System Model v. 3 (Fordham et al., Reference Fordham, Saltré, Haythorne, Wigley, Otto‐Bliesner, Chan and Brook2017).
Diatoms
Diatom analysis was used to track changes in water table position or standing water level and in wetland nutrient status. A few samples from every 50 cm drive were initially processed to assess diatom preservation; subsequently, every second 2 cm increment of the 50 cm drives where diatoms were well preserved was processed using standard acid digestion (Battarbee et al., Reference Battarbee, Jones, Flower, Cameron, Bennion, Carvalho, Juggins, Smol, Birks, Last, Bradley and Alverson2002). Samples were soaked in HCl (10%) for 24 hours, then brought to neutral pH with repeated washes in reverse osmosis water. Afterward, samples were digested in a 1:1 molar mixture of concentrated sulfuric acid (98%) and concentrated nitric acid (63%) for 24 hours, during which they were placed in a hot bath at 75°C for 2 hours. Once brought to a neutral pH again with repeated washes, diatom slurries were pipetted onto cover slips, left to evaporate over 24 hours, and mounted onto glass slides using Naphrax. Counts were conducted on a Zeiss light microscope at 1000× magnification under oil immersion and differential interference contrast optics.
At least 300 diatom valves were counted per sample and identified to the genus and/or species. Several sources were used to identify diatoms and assign diatoms to habitat types (Patrick and Reimer, Reference Patrick and Reimer1966; Krammer and Lange-Bertalot, Reference Krammer and Lange-Bertalot1986; Bunting et al., Reference Bunting, Duthie, Campbell, Warner and Turner1997; Reavie et al., Reference Reavie, Axler, Sgro, Danz, Kingston, Kireta, Brown, Hollenhorst and Ferguson2006; Lavoie et al., Reference Lavoie, Hamilton, Campeau, Grenier and Dillon2008). Raw diatom counts were converted to relative abundances.
Granulometry
As the sediments at the core site consist of mineral material with some organics, and the site is closely connected to the stream channel by overbank flooding, sediment grain-size analysis was applied to evaluate hydrological and sedimentological changes in the record, as sediment size distributions reflect the energy at which sediments were deposited (Sahu, Reference Sahu1964; Hazermoshar et al., Reference Hazermoshar, Lak, Espahbood, Ghadimvand and Farajzadeh2016). A total of 75 samples ranging in wet volumes from 3 to 9 ml were analyzed, including all samples at 10 cm intervals coeval with pollen samples, with additional samples at transition points. Grain-size distributions were quantified using a Malvern Mastersizer 3000 laser diffraction particle-size analyzer using the volume-based principle following the Mie theory (Wriedt, Reference Wriedt2012; Malvern, 2013). Pretreatment included removal of organic matter by digestion in 30% hydrogen peroxide (Allen and Thornley, Reference Allen and Thornley2004) and overnight shaking in 5% sodium hexametaphosphate solution. Each sample was measured in triplicate, and the quality for each sample was evaluated with the weighted residuals (wr) to assess how well the calculated data fit the measurement data; a good fit is considered if the residual is <1% (Wriedt, Reference Wriedt2012). Only the data with wr < 1% were considered in this study (Supplementary Material). The analyzer output consists of percentage composition from each of the grain sizes using the Wentworth distribution scale (Wentworth, Reference Wentworth1922).
Sediment composition
We estimated organic matter content in core CBC3-01 by loss-on-ignition (LOI) on every second 2 cm increment of the core by combustion of 0.5 g samples at 550°C for 4 hours (Chambers et al., Reference Chambers, Beilman and Yu2011). Concentrations of the major elements Al, Si, K, and Ca and minor element S were measured using a portable X-ray fluorescence (pXRF) analyzer (Bruker TRACER 5i/S1 TITAN 600-800), detection limit of 0.1–3% (Al, Si), 100–1000 ppm (Ca, K). The Geo Exploration calibration supplied with the instrument was used along with two reference sediment samples from the Geological Survey of Japan: lake sediments (JLk-1) and stream sediments (JSd1) from which major (above 1%), minor (between 1% and 0.1%), and trace elements (<0.1%) were evaluated (Imai et al., Reference Imai, Terashima, Itoh and Ando1996). The standards were measured in triplicate and were highly correlated to the standard values of the reference material.
A total of 43 samples were selected every ∼10 cm for analysis, as well as samples at transition points for additional detail. Ground, un-combusted samples were placed into 0.7 g pellets of 1 cm diameter. Each sample was measured at three time ranges: 30, 60, and 90 minutes. We focus on the elements present above the detection limit (Al, Si, K, Ca, and S). These can be used to understand the broad composition of the sediments and infer their possible origin, as well as erosion, transportation, and deposition processes (Boës et al., Reference Boës, Rydberg, Martinez-Cortizas, Bindler and Renberg2011; Davies et al., Reference Davies, Lamb, Roberts, Croudace and Rothwell2015). A z-score normalization was applied to the dataset using the transform function in R. These normalized data were used in a principal component analysis (PCA) to summarize the changes through the record (Van der Weijden, Reference Van Der Weijden2002; Rollinson, Reference Rollinson2013).
Results
Pollen analysis
Basal sediments in CBC3-01 date to 5710 cal yr BP (details provided in Loder et al. [Reference Loder, Gillespie, Haeri Ardakani, Cordero Oviedo and Finkelstein2025] and in Supplementary Fig. S1), corresponding to the Nipissing I Lake Erie high stand when widespread wetland initiation took place in coastal zones of the lower Great Lakes (Finkelstein and Davis, Reference Finkelstein and Davis2006). A total of 86 distinct taxa were identified in the CBC3-01 pollen record. Mean alpha diversity was 37 taxa (alpha-diversity metrics are provided at https://doi.org/10.5683/SP3/AKNMBR, Borealis, V1 in Cordero Oviedo et al. [Reference Cordero Oviedo, Loder, Byun and Finkelstein2025]), with maximum values recorded during Stage 3, and lows in taxonomic diversity associated with the transitions between Stages P2 and P3 and between Stages P3 and P4. Zonation by stratigraphically constrained cluster analysis resulted in four statistically significant pollen zones (Supplementary Figs. S2 and S3). Pollen zone 1 (P1) showed high abundances of trees, with some key swamp or mesic indicators including Fagus, Fraxinus, Juglans, Nyssa, Tsuga, and Ulmus (Fig. 2). Shrubs, herbs, and aquatic vegetation showed relatively low abundances, except for Rumex at ∼5%. Pollen zone 2 (P2) is marked by establishment of emergent wetland taxa, including Cyperaceae, Nuphar, Nymphaea, Sagittaria, and Typha/Sparganium. The dominant trees from P1 mostly decline, with only selected tree taxa remaining stable or increasing, including Juniperus/Thuja, Pinus, Picea, Quercus, and Tilia. Shrubs and herbaceous taxa increase, especially Cephalanthus, Ilex, Salix, Asteraceae: Tubuliflorae, and Poaceae. Most of the ferns (notably Polypodiaceae), horsetails, and clubmosses showed their highest abundances through this zone (Fig. 2).
Figure 2. Pollen diagram for core CBC3-01. Taxa recorded at >2.5% abundance in any one sample are plotted (47 out of 86 taxa). The y-axis shows both depth (in cm) and mean age from the bacon model output (in cal yr BP). Taxa are classified into vegetation groups as follows: dark green: trees; orange: shrubs; yellow: herbs; light green: ferns; blue: aquatic vegetation. The diagram is divided into four significant pollen zones: P1, P2, P3, and P4 (details provided in “Methods” and in Supplementary Material).
Pollen zone P3 is marked by an increase in number of taxa recorded to maximum values for the CBC3 record, and a shift among wetland taxa towards a greater dominance of pollen of thicket swamp shrubs, mainly Cephalanthus, with Cornus stolonifera appearing as well (Fig. 2). Herbaceous taxa Cornus canadensis, Rumex, and Poaceae increase. Ferns were somewhat less abundant than in P2; however, aquatic vegetation types remain as indicated by presence of taxa such as Lemna, Sagittaria, and Cyperaceae. Pollen concentrations reach maximum values in P3. Finally, pollen zone P4 is delineated by an initial rapid decline in the number of pollen taxa recorded. Diversity then rebounds, associated with important increases in the disturbance indicators Ambrosia, Poaceae, Polygonum, and Typha angustifolia. The thicket swamp shrub Cephalanthus and emergent wetland indicator Sagittaria, which dominated in P3, both decline, while Tsuga, Juglans and Juniperus/Thuja show small increases toward the upper part of P4.
Owing to the strong dominance in the CBC3-01 pollen record by local wetland plants, regional paleoclimate reconstructions were produced from nearby lake pollen records to assess the role of regional Holocene paleoclimatic changes in wetland dynamics in the study area. These reconstructions confirm that lake pollen records in this region successfully reconstruct broad trends in Holocene paleoclimate, including postglacial warming into the mid-Holocene warm period and Late Holocene cooling, which is also associated with declines in total annual precipitation in parts of the Great Lakes region (Shuman and Burrell, Reference Shuman and Burrell2017). Further, our analysis of regional lake pollen records and downscaled paleoclimate simulations (Fig. 3) indicates that patterns of seasonal variability differ through the Holocene and that individual lake records show some heterogeneity in reconstructed temperature and precipitation, despite the proximity of the two lakes (Calcote, Reference Calcote2003). For example, the onset timing of the mid-Holocene warm period by average annual temperature (tave) was earlier in Hams Lake than in Decoy Lake (by ∼3 ka BP), while the climate-model extraction places this warming in between the timing reconstructed from the two lake sites. The temperature of the warmest month (tmax) was reconstructed with a larger variability among the three cases, and only the climate model output follows the known physical factor of decreasing summer insolation over time since the mid-Holocene. The coldest month temperature (tmin) was generally similar to the average annual temperature, suggesting the sensitivity of vegetation to winter climate conditions in this region. The reconstructed precipitation changes from the climate model and the Hams Lake pollen record are similar in terms of the general increase in the Early to Middle Holocene transition followed by a plateau and a declining trend to recent time. The Decoy Lake pollen is less sensitive to the model-simulated recent decline of annual total precipitation by comparison (Fig. 3). This insensitivity is perhaps because the Decoy Lake record has lower rates of sediment accretion and thus lower temporal resolution from the Middle Holocene to the present day, and Quercus is consistently dominant throughout this part of the record because of local edaphic factors (Szeicz and MacDonald, Reference Szeicz and MacDonald1991).
Figure 3. Regional paleoclimate. The top two panels show the pollen-based paleoclimate reconstructions for the study site using adjacent pollen records (Decoy Lake [Bennett, Reference Bennett1987] and Hams Lake [Szeicz and MacDonald, Reference Szeicz and MacDonald1991] sites in Ontario) downloaded from the Neotoma paleoecology database (https://apps.neotomadb.org/). The bottom panel shows the statistically downscaled outputs of the TraCE-21 k model runs for the user-defined grid using the PaleoView application (Fordham et al., Reference Fordham, Saltré, Haythorne, Wigley, Otto‐Bliesner, Chan and Brook2017). The results of tave (temperature annual average), tmax (average temperature of the warmest month), tmin (average temperature of the warmest month), and annp (average annual total precipitation) are shown in order from the left at each site or by the climate model runs. The y-axis is shown by z-scores centered on the mean of the plot time period (from 8300 years ago to present).
Diatoms
Diatom preservation in the CBC3-01 paleorecord is limited to the upper 1 m of the core, through pollen zones P4 and P3, with declining recovery between 100 and 200 cm depth. Diatom frustules are absent in the lower 2 m of the core. In the disturbed section between 20 and 40 cm, diatom frustules are scarce and insufficient for counting. During the 600 year period represented in the upper 1 m of the core, 58 diatom taxa were identified, many of which are commonly found around the Great Lakes region (Reavie et al., Reference Reavie, Axler, Sgro, Danz, Kingston, Kireta, Brown, Hollenhorst and Ferguson2006). Between 600 and 135 cal yr BP, >80% of the diatom taxa enumerated were tychoplanktonic and included small Fragilarioid taxa in the genus Staurosira (S. construens, S. venter, and S. elliptica) with more minor proportions of Staurosirella pinnata and Pseudostaurosira brevistriata (Fig. 4). Epiphytic and benthic species, including Achnanthidium spp., are scarcer through this section, with relative abundances of <5%.
Figure 4. Relative abundances of diatom taxa in the CBC3-01 paleorecord. Diatom taxa present in any one sample at >5% are plotted. The y-axes shows depth (in cm), mean age from the bacon model output (in cal yr BP) (Supplementary Figure S1) and calendar year (CE). Diatoms were absent below 95 cm depth. Colors indicate the habitat of each species: yellow, epiphytic; green, benthic; red, tychoplanktonic.
A marked change in the diatom assemblages occurred in the post-European settlement era (post–1860 CE; Dakin and Skibicki, Reference Dakin and Skibicki1994). At this time, tychoplanktonic taxa declined markedly as the assemblages were replaced with predominantly epiphytic and benthic taxa. The epiphytic taxon Achnanthidium minutissimum comprises between 27% and 36% of all diatoms since 1935 CE. Some diatom taxa in CBC3-01 were only abundant in the post-European settlement period, including Fragilaria capucina, F. ulna, Gomphonema micropus, Navicula seminulum, Nitzschia amphibia, and N. palea. Some diatom taxa that were abundant before European settlement are largely absent post-settlement, including P. brevistriata and Punctastriata lancettula (Fig. 4).
Sediment properties
The CBC3-01 core is composed of mixed mineral and organic sediments with organics increasing up-core from <10% in the basal and near-basal sediments to maximum values of ∼20% around 50 cm depth (Fig. 5). Lower organics and increased rates of sediment accretion take place above 40 cm, associated with significant disturbances across the watershed due to land clearance at the time of European settlement. Silt-sized sediments dominate the mineral fraction of the CBC3-01 core, followed by the sand fraction; clay was less abundant (<10%), and mainly present through zone P2 (Fig. 5). Silty sediments dominate the two initial pollen zones (P1 and P2), while sandier sediments predominate through zones P3 and P4. Accretion rates are lowest in the silt-dominated P2 and are generally higher in the sand-dominated portions of the record. Rates of sediment accretion increase by an order of magnitude in the post-settlement zone (P4). Loss-on-ignition indicates sediment organic matter contents of 5–20% with a general trend of gradual up-core increases. An inverse trend of gradual decrease up-core is shown by the total organic carbon/total nitrogen ratio.
Figure 5. Sediment composition for core CBC3-01 including grain-size distributions, accretion rates using the mean from the age–depth model (Supplementary Figure S1), and sediment geochemistry. The y-axis shows both depth (in cm) and mean age from the bacon model output (in cal yr BP). Results are divided by the four significant pollen zones. LOI, loss-on-ignition; TN, total nitrogen; TOC, total organic carbon.
The major elements measured by XRF included Al, Si, K, and Ca and minor element S (Fig. 5). Al and Si percentages are low in P1, increase through P2, followed by a gradual decrease from P3 to P4. The trends for K and Ca are inverse to Al and Si, with high percentages in P1, decreases through P2 followed by a gradual increase from P3 to P4 (Fig. 5). The opposite tendencies between Al, Si and Ca, K are further supported by the PCA analysis on axis 1 (Supplementary Figs. S4 and S5, Supplementary Table S3). The tendency for S is a general slow up-core increase, similar to that of the organic matter by LOI (Fig. 5).
Discussion
We identify four pollen zones with distinct vegetation communities and show how changes in vegetation communities and wetland habitats since initiation ∼6000 years ago were primarily driven by shifts in water level of Lake Erie and fluvial processes associated with channel migration. We demonstrate how intensive anthropogenic land uses have had a significant impact on the present-day marsh habitat at the CBC3-01 site, especially as they relate to vegetation, hydrology, and sedimentation. Later, we justify a recommendation to use Stage P3 as a baseline for restoration while highlighting the role of whole-watershed rehabilitation in improving the success of targeted wetland restoration.
Stage P1 (6000–5300 cal yr BP): swamp initiation during Lake Erie high stands
Wetland establishment had taken place at the forested CBC3-01 site by ∼5960 cal yr BP, during a significant flooding episode associated with a 6.5 m rise in water levels known as Nipissing I (Coakley and Lewis, Reference Coakley and Lewis1985; Pengelly et al., Reference Pengelly, Tinkler, Parkins and McCarthy1997; Finkelstein and Davis, Reference Finkelstein and Davis2006; Lewis et al., Reference Lewis, Cameron, Anderson, Heil and Gareau2012). The initiation and preservation of continuously deposited organic- and pollen-bearing sediments from this time forward are indicative of wetland establishment. The pollen assemblages through Zone P1 suggest the CBC3-01 wetland site began as mixed-wood treed swamp habitat, with broadleaved taxa such as Nyssa, Juglans, Fraxinus, Ulmus, and the conifer Tsuga present as well. The broadleaved tree communities were likely well established in surrounding upland forests with the arrival of the mid-Holocene warm period, and those flood-tolerant taxa were readily encroaching on the lowland swamp soils when Lake Erie water levels rose in the Nipissing I. Higher concentration of calcium (Ca) through this zone suggests influence of carbonate bedrock and/or possibly more alkaline lake water, reflecting complex interactions between weathering reactions and hydroclimate (Ewing and Nater, Reference Ewing and Nater2002). Sulfur (S) content gradually increases, likely linked to the organic matter within the sediments (Mitsch and Gosselink, Reference Mitsch and Gosselink2015; Kemp et al., Reference Kemp, Sadler and Vanacker2020), while higher TOC/TN and potassium (K) suggest terrestrial inputs associated with overbank flooding at the site from the Big Creek River channel, located today ∼130 m from the core site. The mineral fraction is mostly silt-dominated but with an important sand fraction, particularly through the late stages of P1, corresponding with the peak of the Nipissing I Lake Erie high stand (Holcombe et al., Reference Holcombe, Taylor, Reid, Warren, Vincent and Herdendorf2003) and likely indicative of a back-swamp environment in the context of a meandering stream (Golet et al., Reference Golet, Calhoun, Deragon, Lowry and Gold1993; Megens, Reference Megens2015). Wetland establishment at this site reflects the dynamic interplay between changing drainage and water levels in Lake Erie, as well as soil substrates and fluvial processes (Fig. 1). The paleoclimate, as reconstructed from the nearby lake pollen records, shows minimal fluctuations through this phase after the initial rise in temperature and precipitation associated with the mid-Holocene warm period (Fig. 3) and was probably not a major driver of local wetland vegetation change for the well-established swamp. Wetland communities and hydric soils at this time were sustained more proximally by the higher water tables originating from the higher Lake Erie base level.
Stage P2 (5300–2000 cal yr BP): shifting hydrological influences and marsh establishment
This stage is characterized by a shift toward a riverine marsh environment that took place in the context of ongoing high-water levels in Lake Erie associated with the Nipissing I and II rises and the formation of the Long Point sand spit at ∼5000 cal yr BP (Coakley and Lewis, Reference Coakley and Lewis1985; Coakley, Reference Coakley, Fletcher and Wehmiller1992; Pengelly et al., Reference Pengelly, Tinkler, Parkins and McCarthy1997; Holcombe et al., Reference Holcombe, Taylor, Reid, Warren, Vincent and Herdendorf2003; Lewis et al., Reference Lewis, Cameron, Anderson, Heil and Gareau2012). The paleoclimate remained stable after mid-Holocene warmth and again was not likely a significant driver of the wetland vegetation shifts recorded in the CBC3-01 pollen record at the transition to Stage P2. A shift from sandy to silt-dominated sediments with high Si and Al content occurs in this stage, with an increase in clay content, suggesting a more proximal relationship and the development of a hydrological connection between the site and the river channel. The pollen record indicates a shift from a mixed-wood swamp to a marsh wetland, likely in a floodplain pond or wetter back-swamp environment. This transition is marked in the pollen record by the increase in aquatic and emergent taxa, including Cyperaceae, Nymphaceae, Sagittaria, Sparganium, and ferns (Fig. 2), accompanied by increases in sediment organic matter and S content (Fig. 5). Swamp taxa remain present through Stage P2, although at lower abundances. A notable decline in Tsuga pollen is noted around 5000 cal yr BP, coeval with regional declines linked to pathogen outbreaks and potentially to drought (Bennett and Fuller, Reference Bennett and Fuller2002; Booth et al., Reference Booth, Brewer, Blaauw, Minckley and Jackson2012).
This section is characterized by silty sediments, which in the pXRF analysis are primarily composed of Al and Si with increased low-permeability clay content. Coupled with high water tables associated with persistently high Lake Erie water level, these conditions likely contributed to sustained high water at the core site, and marsh establishment. Following the Nipissing high stands, Lake Erie water level gradually declined until it reached its modern configuration at ∼3000 cal yr BP (Pengelly et al., Reference Pengelly, Tinkler, Parkins and McCarthy1997; Finkelstein and Davis, Reference Finkelstein and Davis2006; Lewis et al., Reference Lewis, Cameron, Anderson, Heil and Gareau2012). Thus, the gradual lowering of lake level toward the end of this stage may have resulted in enhanced upstream erosion and transport, leading to renewed mineral sediment input. This is evidenced by the highest percentages of silt in the record and is also associated with a decrease in the number of pollen taxa recorded (Fig. 2).
Stage P3 (2000 cal yr BP–80 cal yr BP): development of thicket swamp–marsh mosaic
The hydrological changes in this stage reflect the modern influence of the Norfolk sand plain as a major geomorphic agent within the Big Creek catchment (Megens, Reference Megens2015). Fluvial processes associated with floodplain development and overbank flooding result in accumulation of sandier sediments at increasing accretion rates at the CBC3-01 site, likely reflecting a more proximal position and more prominent hydrological connection to the river channel or increased activation of upstream sediments (Pierce and King, Reference Pierce and King2008; Stewart and Desloges, Reference Stewart and Desloges2014). The shift to coarser textures from P2 to P3 may reflect decreasing distance between the CBC3-01 core site and the Big Creek channel as a result of lateral migration of the meander channel across the floodplain. The composition of the mineral fraction is steady through Stage P3, with gradual decreases in Al and Si, as the proportions of K and Ca increase modestly and coincide with an increase in the accretion rate (Fig. 4). Sediment organic matter and S content increase as a thicket swamp wetland is established, dominated by Cephalanthus (buttonbush), which is well adapted to sand–silty soils and tolerates seasonal flooding (Finkelstein and Davis, Reference Finkelstein and Davis2006). Aquatic pollen taxa including Sagittaria and Cyperaceae persist through this zone (Fig. 2), and diatoms appear in the record. The diatom assemblages through Stage P3 are mainly tychoplanktonic taxa recorded at other Lake Erie coastal marshes associated with standing water of >1–2 m in depth (Yang and Duthie, Reference Yang and Duthie1995).
Taken together, the combined diatom and pollen records suggest the persistence of a floodplain pond with standing water and marsh vegetation through Stage P3 as thicket swamps expanded, approaching the modern configuration of a thicket marsh–swamp mosaic on the floodplain. Stage P3 also corresponds to the highest number of pollen taxa recorded. A notable change in the paleoclimate during this period was the inferred decline in the annual total precipitation as shown by the Hams Lake pollen reconstruction corresponding to the climate model output. Assuming it was influential at the regional scale, the thicket swamp expansion might be attributed to the lowering water table, particularly at coarser-grained microsites in response to decreasing precipitation. This time period also includes the establishment of early palisaded villages in this part of southwestern Ontario (Fox et al., Reference Fox, Conolly, Stewart and Timmins2023), although further research is required to determine any impacts on local wetland ecosystems.
Stage P4 (80 cal yr BP to present): settlement era and major human influence on the wetland
This stage is marked by highly significant changes in ecological communities, sediment inputs, and water levels associated with human impact. Although we were not able to extract a clear paleoclimate signal using the MAT on pollen assemblages from the CBC3 pollen record due to dominance of wetland pollen taxa, these changes are nevertheless taking place in the context of anthropogenically warming climate, land clearance, and other widespread human impacts, well documented in southern Ontario (McCarthy et al., Reference McCarthy, Patterson, Head, Riddick, Cumming, Hamilton and Pisaric2023). Simultaneously, and consistent with markedly reduced water levels, the diatom record shows a major shift in assemblages with steep declines in tychoplanktonic taxa, increases in benthic and epiphytic diatoms, and appearance of novel diatom taxa, all indicative of siltation of the marsh, with post-settlement alluvium widely recorded in North American floodplain records (Stewart and Desloges, Reference Stewart and Desloges2014; Kemp et al., Reference Kemp, Sadler and Vanacker2020). This change is corroborated by the pollen data, which show a significant decline in overall number of taxa recorded at the onset of the zone, followed by a recovery in alpha diversity associated with large increases in upland herbaceous taxa associated with land clearance, especially Ambrosia and Polygonaceae. Sediment accretion rates increase several-fold, consistent with widely reported pulses in erosion associated with land clearance and the emergence of humans as dominant geomorphic agents (Kemp et al., Reference Kemp, Sadler and Vanacker2020). The high influx of terrigenous material is also signaled by increasing K concentrations; a rapid pulse of undated, silt-dominated sediments (20–40 cm); and the appearance of benthic diatom taxa, including A. minutissimum, Cocconeis placentula, F. capucina, and Nitzschia palea, some of which are well adapted to turbidity (Köster et al., Reference Köster, Lichter, Lea and Nurse2007; Sivarajah et al., Reference Sivarajah, Paterson, Rühland, Köster, Karst-Riddoch and Smol2018). Taken together, these results are consistent with significant human modification of the watershed through land clearance, road construction, and hydrological alteration associated with widespread tile drainage.
Cephalanthus, the thicket swamp shrub dominant through P3, declines in P4 as Sparganium/Typha pollen increases, likely consisting mainly of the wetland invasive Typha angustifolia. High abundances of epiphytic diatoms in the recent record correlate with increased vegetation cover and much lower water levels following European settlement and may also reflect the more recent invasion of Phalaris arundinacea, which dominates the plant cover at site CBC3-01 today. The disturbed conditions associated with hydrological alteration and sedimentation in this stage are known to be readily tolerated by invasive species (Galatowitsch et al., Reference Galatowitsch, Anderson and Ascher1999; Kercher and Zedler, Reference Kercher and Zedler2004; Rothman and Bouchard, Reference Rothman and Bouchard2007) and may contribute to the proliferation of invasive species recorded both in the watershed and in the recent paleorecord (e.g., Typha angustifolia, Phalaris arundinacea, Phragmites australis). The proliferation of these invasives may also further contribute to high rates of sediment accretion because of high biomass production, restriction of surface water flow, sediment trapping, and enhanced evapotranspiration (Kercher and Zedler, Reference Kercher and Zedler2004; Mitsch and Gosselink, Reference Mitsch and Gosselink2015; Weilhoefer et al., Reference Weilhoefer, Jakstis and Fischer2017).
Conclusions
The sedimentary record from the CBC3-01 marsh site on the floodplain of Big Creek in southern Ontario records a dynamic interplay between shifting Lake Erie water levels in response to postglacial isostatic adjustment and paleoclimatic change at a regional scale. We demonstrate how changes vegetation and wetland habitat types were driven by fluvial processes in relation to hydrological connectivity, channel migration and floodplain accretion, and most recently, intensification of anthropogenic land uses. Initially, a forested wetland established upon the marked rise in Lake Erie water levels in the Middle Holocene during the Nipissing high stands. This stage was followed by the establishment of a marsh dominated by aquatic and emergent plants as high water levels persisted and supported a hydrological connection with Big Creek river channel. Later, as Lake Erie water levels converged on the near-modern configuration, a mosaic of marshes and thicket swamps established across the floodplain while mineral sediment inputs were sustained. The European settlement era is marked by a rapid onset of highly significant changes to vegetation communities, sediment accretion, and water levels in the marsh.
The paleoecological record provides baselines in vegetation communities and demonstrates how these communities have been affected by both natural variability and anthropogenic land use as they relate to river channel connectivity, water levels, and mineral sediment influxes at the CBC3-01 site (Kraus et al., Reference Kraus, Norman, McFarlane, Lemieux, Jacob and Gray2021; Burge et al., Reference Burge, Richardson, Wood and Wilmshurst2023). We infer that the role of regional climate was mainly in the way it shaped the regional available pool of vegetation while showing that hydrological and fluvial changes were the more proximate causes of subsequent wetland vegetation changes. This differs from upland vegetation dynamics, which are more strongly driven directly by paleoclimate change (Shuman et al., Reference Shuman, Newby and Donnelly2009). Our work confirms that Lake Erie water level is fundamental to wetland processes in this watershed, as it sets the base level, and thus fluvial dynamics, including stream channel migration, sediment delivery, and hydrological connectivity. Collectively, these dynamics support local wetland water tables and plant communities in the riparian zone over thousands of years. We recommend the use of Stage P3 as a baseline for restoration, as it includes the time period when Lake Erie water levels are most similar and provides an analog that is hydrologically consistent with modern conditions. Further, Stage P3 represents a maturing phase of floodplain development with the establishment of a highly biodiverse mosaic of wetland types and plant taxa. While extensive drainage, land-use change, and the proliferation of invasive species present challenges for returning to Stage P3 baselines, this paleorecord provides critical benchmarks for restoration, including biodiversity targets that encompass both specific plant taxa and habitat types..
The record from Central Big Creek Marsh confirms the need for riparian and coastal zone marshes to be sustained by water inputs and mineral sediment delivery. Our Holocene baselines show how fluctuations in the water levels of Lake Erie impact marsh development and vegetation communities up to several kilometers upstream from the coast. Future projections of Lake Erie water levels under climate change scenarios are important to consider when implementing wetland conservation measures and planning restoration projects.
Additionally, the pollen record spanning the last several thousand years provides a list of more than 40 plant taxa that can be used to support planting and seed-sowing efforts under suitable habitat conditions. Further, the pollen record shows significant increases in invasive species over the course of the settlement era, with negative consequences for biodiversity. The marked post-settlement changes in habitat conditions, under which native species cannot thrive, have likely favored the proliferation of invasive species, which require consideration in restoration plans.
Finally, the paleorecord demonstrates the importance in hydrological connectivity of the CBC3-01 site to the Big Creek channel in supporting marsh plant communities. Although the site has been resilient to natural fluctuations related to Lake Erie water levels and sediment delivery, the marked decline in water levels and decreased influxes of excessive mineral sediments due to land clearance resulted in a complete shift in diatom communities, declines in plant diversity, and novel vegetation assemblages. Loder et al. (Reference Loder, Zamaria, Arhonditsis and Finkelstein2023) found that depressional wetlands that were actively restored at higher elevations in the Big Creek watershed lack inundation and hydro-fluvial events and the delivery of inorganic and organic materials, which are known to increase the speed of recovery of wetland functions at altered sites (Moreno-Mateos et al., Reference Moreno-Mateos, Power, Comín and Yockteng2012; Poppe and Rybczyk Reference Poppe and Rybczyk2021). We conclude that attaining pre-settlement hydro-fluvial events and rates of sedimentation through restoration of targeted wetland sites may be challenging, especially on lands at higher elevations in the watershed. Focusing on whole-scale watershed approaches that sustain water flows and reduce excessive sediment delivery to riparian sites may lead to more successful wetland restoration outcomes in the long term.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2025.10038.
Acknowledgments
Norfolk County is on the traditional lands of the Attawandaron, Haudenosaunee and the Anishinaabe First Nations, and encompassed within the Treaty lands (“Between the Lakes Purchase”) of the Mississaugas of the Credit. We respectfully acknowledge the original peoples of this land and their long stewardship of it. We thank the Nature Conservancy of Canada (NCC) for permission to access the sampling sites; all sampling was conducted under NCC research license G-ON-2018-153246. We thank J. Desloges, D. Gregory, and M. Gorton for assistance with laboratory equipment, D. Hiler for field assistance, and J. Desloges for helpful comments on an earlier version of the article. We also thank D. Metsger for providing access to critical pollen reference material at TRT (The Royal Ontario Museum Green Plant Herbarium). This research was funded by a Natural Sciences and Engineering Research Council of Canada Discovery Grant (grant no. RGPIN-2017-06759) and support from the Ontario Ministry of Natural Resources and Forestry.
Data availability statement
All data associated with this study have been deposited in the Borealis Repository hosted by the University of Toronto (Cordero Oviedo et al., Reference Cordero Oviedo, Loder, Byun and Finkelstein2025). https://doi.org/10.5683/SP3/AKNMBR. Pollen and diatom biostratigraphies will also be available on the Neotoma Paleoecology Database (https://www.neotomadb.org/, with the following dataset IDs: Chronology, 65850; Diatoms, 65823; Pollen, 65849).
Competing interests
The authors declare no competing interests, and that there are no conflicts of interest including any financial, personal or other relationships, that have inappropriately influenced, or be perceived to have influenced, this work.
Author contributions
CCO: Conceptualization, Formal Analysis, Method- ology, Writing – original draft; ALL: Conceptualization, Formal Analysis, Writing – Review & Editing; EB: Formal analysis, Writing – Review & Editing; SAF: Conceptualization, Data curation, Funding, Supervision, Writing – original draft.
Open access
