The Sandhill Wetland (SW) and Nikanotee Fen (NF) are two wetland research projects designed to test the viability of peatland reclamation in the Alberta oil sands post-mining landscape. To identify effective approaches for establishing peat-forming vegetation in reclaimed wetlands, we evaluated how plant introduction approaches and water level gradients influence species distribution, plant community development, and establishment of bryophyte and peatland species richness and cover. Plant introduction approaches included seeding with a Carex aquatilis-dominated seed mix, planting C. aquatilis and J. balticus seedlings, and spreading a harvested moss layer transfer. Establishment was assessed six years after introduction at SW and five years after introduction at NF. A total of 51 species were introduced to the reclaimed wetlands, and 122 species were observed after five and six years. The most abundant species in both reclaimed wetlands was C. aquatilis, which produced dense canopies and occupied the largest water level range of observed plants. Introducing C. aquatilis also helped to exclude marsh plants such as Typha latifolia that has little to no peat accumulation potential. Juncus balticus persisted where the water table was lower and encouraged formation of a diverse peatland community and facilitated bryophyte establishment. Various bryophytes colonized suitable areas, but the moss layer transfer increased cover of desirable peat-forming mosses. Communities with the highest bryophyte and peatland species richness and cover (averaging 9 and 14 species, and 50% to 160% cover respectively) occurred where summer water level was between -10 and -40 cm. Outside this water level range, a marsh community of Typha latifolia dominated in standing water and a wet-meadow upland community of Calamagrostis canadensis and woody species established where the water table was deeper. Overall, the two wetland reclamation projects demonstrated that establishing peat-forming vascular plants and bryophytes is possible, and community formation is dependent upon water level and plant introduction approaches. Future projects should aim to create microtopography with water tables within 40 cm of the surface and introduce vascular plants such as J. balticus that facilitate bryophyte establishment and support development of a diverse peatland plant community.

Study Region

The Sandhill Wetland (SW) and Nikanotee Fen (NF) are located north of Fort McMurray, Alberta, Canada. Mean total annual precipitation in Fort McMurray is 419 mm, with mean annual rainfall of 316 mm and snowfall of 134 cm. Daily mean temperatures dip to -17.4^o C in January and peak at 17.1^o C in July (1981-2010 climate normal from weather station at 56° 39’ N, 111° 13’ W; Environment Canada 2018a). The data reported here were collected in the summer of 2017, when total annual precipitation was 312 mm and daily mean July temperature averaged 22.9^o C. (daily data report for 2017 from weather station at 56° 39’ N, 111° 13’ W; Environment Canada 2018b).

Vegetation Sampling

Vegetation composition was sampled and depth to water measured in 79 plots at SW and 42 plots at NF in July 2017. Canopy cover by species was assessed by visual estimation to the nearest 5 % in an 8 m² circular plot at SW and to the nearest 1 % in a 4 m² square plot at NF. The sampled area represents 0.37% of the total wetland area at SW (632 m² of 17 ha) and 0.76% of the total fen area of NF (168 m² of 2.2 ha).  Plot sizes differed due to original sampling protocols at each fen, but are both within recommended plot sizes for low-growing (bryophyte) and herbaceous vegetation (4 – 16 m²; Chytrý and Otýpková 2003). Species-area relationships assume that species richness increases with increased sampling area (Connor and McCoy 1979); however, species richness was higher in 4 m² plots at NF (mean = 16 species, se = 1.2) than at SW in 8 m² plots (mean = 9 species, se = 0.7). In addition, Otýpková and Chytrý (2006) showed that ordinations containing differently sized plots (of less than a factor of four) do not produce patterns associated with plot size.

The data represent the sixth-year post-introduction at SW (2012 to 2017) and fifth-year post-introduction at NF (2013 to 2017). Despite the difference in number of growing seasons, establishment outcomes by 2017 are relatively equivalent because of the different propagule growth stages.

Latin and common names for vascular plants follow the Alberta Conservation Information Management System (ACIMS 2017, 2018), mosses follow Flora of North America Editorial Committee (FNA 2007, 2014), and Stotler and Crandall-Stotler (1977) for liverworts.

Depth to water was measured in an open pit dug adjacent to each vegetation plot during the 2017 sampling. Total precipitation in Fort McMurray during the days between the survey dates was negligible at 9 mm (local weather station at 56o 39’ N, 111^o 13’ W; Environment Canada 2018b).

Analysis

Species were classified as ‘peatland’, ‘marsh-swamp’, ‘upland’, or ‘widespread’ based on habitat type designations obtained primarily from the Alberta Wetland Classification System (AWCS; ESRD 2013) for vascular plants, and Vitt and House (2015) and Vitt et al. (2023) for bryophytes. The Reclamation Criteria for Well Sites and Associated Facilities for Peatlands (Environment and Parks 2017), the National Wetland Plant List (USACE 2018), and the Flora of North America (FNA 2007, 2014) were used in support. Habitat classifications are not static as many plants occur across multiple habitat types. Priority was given to the ‘peatland’ classification even if the species was not an obligate peatland species and may occur in other habitat types. Species designated as ‘marsh-swamp’ typically are found in mineral soil wetlands and are not known to occur in peatlands. Wet meadow species occur on mineral soil in the transition area between an upland and marsh were also classified as ‘marsh-swamp’. Species designated as ‘upland’ are not known to occur in wetlands. If the species is not strongly associated with one particular habitat type and can occur from uplands to wetlands, it was classified as ‘widespread’.

Rank abundance curves were calculated to assess the relative proportion of each species within each plant introduction approach (Kindt and Coe 2005). Best fit regressions for species distribution along the water level gradient at each wetland were determined by selecting the most parsimonious model with the lowest Akaike information criterion (AIC) value (Akaike 1987, Burnham et al. 2011). Analyses were performed in R using the Stats and BiodiversityR  packages (R Core Team 2018).

For multivariate analyses, species abundance data were relativized by plot total and square root transformed to downweight the contribution of dominant species. Plots were then compared using Bray-Curtis similarity (Bray and Curtis 1957) and analyzed using a non-metric multidimensional scaling ordination (NMDS). Two NMDS ordinations were produced, one NMDS to evaluate change over time from the initial plant composition in each plant introduction approach and the plant composition data from 2017, and a second NMDS with only the data from 2017 to compare plots and communities across the reclaimed wetlands. A solution for the NMDS ordination with all data years was reached in 2 dimensions after 50 iterations with a stress value of 0.12. The trajectory distance between plots across years was calculated on a 2D coordinate plane. A solution for the 2017 NMDS ordination was reached in 2 dimensions after 50 iterations with a stress value of 0.14. A permutational multivariate analysis of variance (PERMANOVA; Anderson et al. 2008) using Type II sums of squares was conducted to assess multivariate differences between plant introduction approaches and in response to water level. Communities were identified using a group average cluster analysis with a cut-off level of 47% similarity, determined by similarity profile analysis (SIMPROF) (Appendix 1: Figure S1). The SIMPROF test provides statistically significant evidence of groups in samples. The 47% cut-off level was ultimately selected to obtain the most parsimonious solution that minimized the number of outlier samples and maximized the similarity between clustered groups. A similarity percentages analysis (SIMPER) identified the species contributing most to the similarities and differences between groups. Diversity within communities was calculated as species richness, the total number of species in each plot, and the average multivariate dispersion of plots from the group (community) centroid (PERMDISP), where higher average dispersion (Bray-Curtis distance) indicates more variability in the assemblage composition (Anderson et al. 2008). Multivariate dispersions from the group centroid are directly interpretable as beta diversity among groups, which may be defined as the variability in species composition among sampling units for a given area at a given spatial scale (Anderson 2006, Anderson et al. 2006). Differences between variables were not determined statistically due to unequal sample sizes within communities. All multivariate analyses were conducted in Primer 7, Version 7.0.21, and with PERMANOVA +1 (Primer-E Ltd.; Anderson et al. 2008, Clarke and Gorley 2015).

We used published reclamation criteria for peatlands in Alberta (Environment and Parks 2017) to assess whether the vegetation composition six- and five-years post-introduction would achieve reclamation certification from AER. For reclaimed fens from bare soil, the target criteria for each relevant vegetation assessment component include:

• bryophyte canopy cover ≥ 50 %; 
• peatland (desirable fen) species canopy cover ≥ 50 %; 
• non-peatland (marsh-swamp, upland, or widespread) species canopy cover ≤ 20 %; and 
• peatland plant species richness of ≥ 4 for saline fens.

Species introduction approaches include: 

• Spreading a native seed mix dominated by Carex aquatilis (CaSdMx-SW)
• Carex aquatilis seedlings (Ca-NF),
• Juncus balticus seedlings (Ju-NF),
• C. aquatilis seedlings and MLT (moss layer transfer) (Ca-MLT-NF),
• J. balticus seedlings and MLT (Ju-MLT-NF), and
• MLT (MLT-NF). 

Dataset and metadata that describe the abbreviated terms used in the dataset are available.