Plant Community Structure within a Reclamation Field Trial and Forested Reference Sites in a Post-Mine Environment

Abstract

Early successional plant community assemblage within a reclamation field trial at the Detour Lake Mine in northeastern Ontario is assessed, and compared with reference forested and historically reclaimed sites. The reclamation field trial examines eight amendment treatment combinations that include treatments with a winter kill cover crop of oats, fertilizer, biosolids, peat, and combinations thereof. The objectives of this study are to: (1) Investigate how soil amendments influence plant functional group establishment and growth in mine overburden; and (2) Explore the amendment properties that best support the establishment and growth of a plant community that resembles the baseline reference sites. Currently, the presence of non-native species and a dominant woody plant community explains the largest proportion of variance between the forested upland and lowland reference sites and all reclaimed sites. Similar to non-native species, graminoids were absent from the upland forested reference sites. The difference in the graminoid community explains much of the variance between the forested reference sites and all reclaimed sites. The cumulative additions of fertilizer and peat increased alpha diversity of non-native and graminoid plants within the amendment treatments, which had greater alpha diversity of these plant functional groups than the forested reference sites. Within the amendment treatments, non-native and graminoid alpha diversity was initially greater in the nutrient treatments, but by 2019 there was no significant difference in non-native or graminoid alpha diversity between amendment treatments. The results indicate that applications of nutrients through fertilizer or biosolids may increase graminoid alpha diversity and abundance within reclamation units in year 1. The results also confirm that the vascular plant community composition present within the historically reclaimed sites and amendment treatments does not resemble the forested reference sites. The plant community present within the amendment treatment sites is best described as early successional, with the presence of non-native herbaceous legumes dominating the historically reclaimed sites. Despite this, the results indicate that fertilizer and biosolids-based treatments have developed a vascular plant community, excluding woody species that is more similar to the forested reference sites than the peat-based treatments. Further research and long-term monitoring are needed to determine which amendment treatment will best support a plant community that resembles the forested reference sites. In addition, future studies of this nature might consider including wildfire affected and post-harvested forest stands as additional reference sites, to better capture possible plant community trajectories of a severely disturbed environment.

1. Introduction

A key component of restoration planning is the identification of a reference ecosystem [1,2,3]. This provides detailed information on the community structure, diversity, and attributes of the pre-disturbance environment [3], and can ultimately inform a restoration model and evaluation criteria for restoration success [1,4]. The reference system is often a site of advanced ecosystem development, but at a point along the intended trajectory of the restoration site [1]. The reference site may be a site with a natural disturbance regime (e.g., wildfire, forests pests, blowdown), or one that has undergone anthropogenic disturbance (cultural practices, forest harvest, introduction of invasive species), or some combination of both. Selecting just one reference site is unlikely to reflect the range of variation that could be expressed by the restored system, and further, the practitioner must consider that all available references have had undergone some “human-mediated impacts [1]”.

An ecosystems trajectory is determined by the environment and unique characteristics of the biota involved [5]. In the context of ecological restoration, research indicates that understory plant communities in reclaimed ecosystems are often not representative of native boreal forest biodiversity [3,6,7]. While the lack of a well-defined reference state is considered a weakness of many restoration projects [5,8,9], defining a reference state is made more difficult when there is evidence of multiple alternative states that can result following natural succession [10,11,12]. This shortcoming underscores the importance of incorporating multiple reference sites for planning and evaluating restoration success.

While leaving reclamation to processes of natural or spontaneous succession can be considered [13], concerns over erosion, scale, and distance of reclaimed sites from sources of native propagules typical of many post-mine environments may necessitate the use of technical intervention to initiate revegetation. The prevailing evidence indicates that on sites that undergo severe impairment, such as those expected in post-mine environments, “human intervention may be needed to prevent unacceptable transformation to an alternative and probably less ecologically vigorous state [14]”. In mining disturbed lands in Canada, studies have examined plant community development following disturbance and restoration [15,16,17]. However, few studies have compared these developing plant communities with reference sites from the region [3,6,18]. What these studies do indicate, however, is that within boreal systems, the plant community composition on restored sites is often distinct from, and not representative of the native forest biodiversity of reference sites with a natural disturbance regime [3,18,19]. Reclamation treatments often have more non-native species, more grasses and forbs, and fewer shrubs and trees [3,18,19].

Mine closure in Ontario, Canada, requires the restoration of cover materials, vegetation and surface water features to the quality, quantity and appearance of the pre-disturbance conditions and baseline conditions [20]. This standard for success underscores the importance of establishing both reference sites, and long-term monitoring to determine when indicators of success are met [21]. There continues to be uncertainty surrounding the long-term trajectory of restored lands. This includes a lack of long-term plot networks to investigate this question and continue progressing the field of restoration ecology [5,6].

Successful restoration is often linked to overcoming environmental conditions, which are effectively filters that can limit plant species from establishing [22]. Most often, these filters are overcome through the introduction of propagules, matching candidate species with suitable moisture regimes on a site, and the incorporation of both physical and biological amendments to improve soil structure and the bioavailability of nutrients [15,23,24]. This has led to studies examining how soil amendments and salvaged material can improve the establishment of native plant species in post-mine environments in the boreal region [17,25,26]. The process of open pit mining includes the stripping, removal and stockpiling of overburden, including overlying organic layers. Soil disturbance, stripping and stockpiling is understood to degrade the physical, chemical, and biological properties of soil [27,28,29]. To overcome these challenges, seeding, tree planting, soil roughening and the addition of organic and chemical soil amendments are common methods of increasing plant available nutrients in post-mine substrates [19,23,30]. Existing research indicates that facilitating nutrient availability and cycling, as well as incorporating multiple plant functional groups, may lead towards more sustainable and resilient post-mine landscapes [16,24,31].

This present study seeks to evaluate the progression of plant functional groups within a reclamation field trial at the Detour Lake Mine, and compare these results with reference forested and historically reclaimed sites. The objectives of this study are to: (1) Investigate how soil amendments influence plant functional group establishment and growth in mine overburden; and (2) Explore the amendment properties that best support the establishment and growth of a plant community that resembles the baseline reference sites. These results will inform mine restoration strategies in restoring lands to pre-disturbance conditions.

2. Materials and Methods

The Detour Lake mine is operated by Kirkland Lake Gold, and is a conventional open pit gold mine located 185 km northeast of Cochrane, Ontario, Canada. The mine is on the Abitibi Greenstone Belt of the Canadian Shield [32,33]. The overburden on the property, referred to as mine till, is often calcareous, fine grained silt-clay to silt texture [33]. To better understand the influence of soil amendments and treatments on plant community establishment, our team established a reclamation field trial at the Detour Lake Mine in 2015.

2.1. Experimental Design

The experiment is arranged in a randomized block design (Figure 1). Eight treatment combinations are evaluated, each of which is replicated in five blocks. The blocks are oriented in a north south direction to help control for the influence of the forest edge and a low gradient slope across the trial area. Experimental units measure 6 m × 6 m (36 m²).

2.2. Amendments

Eight amendment treatment combinations are evaluated. This includes Control, Control + Fertilizer, Control + Oats, Control + Peat, Control + Biosolids, Control + Oats + Fertilizer, Control + Peat + Fertilizer, Control + Biosolids + Peat.

The Control treatment is represented by the placement of 30 cm of mine till, with seeding and planting treatment applied. The till is a fine sandy loam containing on average 40% silt and 4% clay. Both Organic Matter (OM) and Total Carbon (C) are low (0.1% and 0.026% respectfully) and the soil pH is approximately 8.4. Total nitrogen (N) is 0.026% dry, Phosphorus (P; 3.3 mg/L dry soil) and Potassium (K; 59.2 mg/L dry soil).

A 19-19-19 NPK fertilizer (urea-based nitrogen) was applied at a rate of 250 kg/ha on 24 May 2016, a 16-16-16 NPK fertilizer (urea-based nitrogen) was applied at a rate of 300 kg/ha on 7 June 2017, and a 16-4-8 NPK fertilizer (urea and ammonium-based nitrogen) was applied at a rate of 300 kg/ha on 9 June 2018. Each year, fertilizer was surface broadcast by hand only in experimental units requiring the fertilizer treatment, at rates based on those reported in the literature [19].

The winter kill cover crop treatment of oats (Avena spp.) was surface broadcast at a rate of 38 lbs/acre on 5 August 2015, to experimental units requiring this treatment. The oats had 56 frost free days and reached a height of 10–15 cm before frost. Sporadic plants (3–5 individual plants) were observed within two experimental units in August 2016, each of which was found in a Control + Oats + Fertilizer treatment. No living oat stems were observed within the experimental units or trial area by 25 August 2017. Peat was applied to experimental units requiring this treatment at an average depth of 10 cm. The peat used in this trial had been stockpiled for approximately 4 years before application. The peat was partially decomposed (hemic) and smearable in the hand, with no identifiable moss fragments in the material. Treated biosolids were applied on 3 October 2015. Experimental units requiring this treatment received 120 gallons of biosolids over two passes. When applied to bare till, infiltration was low, and pooling occurred next to the silt fence that was installed down slope (west edge) of each experimental unit.

2.3. Seeding and Planting Experimental Units

Experimental units were seeded by hand with ticklegrass (Agrostis scabra), fringed bromegrass (Bromus ciliatus), bluejoint reedgrass (Calamagrostis canadensis) and American vetch (Vicia americana) on 6 October 2015. Each 36 m² experimental unit, as well as 30 cm beyond each experimental unit edge, received 175 g of seed. This equates to a rate of 35 lb/acre. Germination of these seeds relied on the natural freeze-thawing cycles in the winter of 2015/2016 to break dormancy, resulting in germination in spring 2016.

Tree seedlings were planted by hand in each experimental unit on 26 May 2016. Two seedlings of each of the following were planted in each experimental unit, for a total of 10 seedlings per unit: green alder (Alnus crispa; 20–30 cm with 1 × 4 inch plug), white spruce (Picea glauca; 10–15 cm with 1 × 4 inch plug), jack pine (Pinus banksiana; 10–20 cm with 1 × 4 inch plug), prairie willow (Salix humilis; 20–40 cm with 2 × 5 inch plug), and lowbush blueberry (Vaccinium angustifolium; 10–15 cm with 1 × 4 inch plug). All seedlings were grown from seed in greenhouses, and fertilizer was applied to the seedlings during the 2015 growing season while under the care of the respective nurseries.

2.4. Data Collection

Data collection occurred each year during the following time periods: Year 1 from 23–24 August 2016; Year 2 from 25–26 August 2017; Year 3 from 20–22 August 2018; and Year 4 from 6–9 August 2019. A photograph was taken in each year, a subset of which from 2019 is provided in Figure 2. Technicians estimated and recorded physical variables of erosion, deposition, and surface roughness on a scale from 1–5, with 1 being nill, and 5 being highest incident of the variable encountered within an experimental unit. Percentages of coarse wood and rock within each experimental unit were estimated as <1, in 1% increments from 1–10% and in 5% increments from 10–100%.

Plant surveys followed the methods outlined by Rapai et al. [3]. Technicians measured the height of each woody plant that established within the experimental units at the top of the living branch or needle. If a specimen was dead, it was given a height of 0 cm. Within the center of each experimental unit, technicians established a circular plot (2.83 m radius, 25 m²), and estimated the percentage cover of all woody shrubs and trees, herbaceous plants, and grasses within the plot. Technicians estimated plant percent cover as <1, in 1% increments from 1–10%, and in 5% increments from 10–100%. Thirteen reference sites (60 plots) were selected to be surveyed near the mine site. A coarse classification of these sites includes: four forested lowland sites (DLM 1, 2, 8, and 28); four mining disturbed and historically reclaimed sites (DLM 9, 12, 26, 27); and four forested upland sites (DLM 13, 14, 15, and 29). These sites represent the reference systems to which the reclamation trial results are compared.

Soil samples were collected from the trial area on 28 August 2016, 29 August 2017, 22 August 2018, and 5 August 2019. Soil samples were collected from the reference sites on 6–8 August 2019, and analyzed for soil pH, OM, Calcium (Ca), C, Copper (Cu), Iron (Fe), K, Magnesium (Mg), Manganese (Mn), Sodium (Na), N, P, and Zinc (Zn). For analysis of soil chemistry, an aggregate sample consisting of five subsamples were pooled for each experimental unit, and collected along a profile from 0–15 cm in depth. Prior to sampling, the vegetation layer was removed. The shovel was wiped clean between each sample until all visible soil material was removed. For reference locations, the aggregate soil sample was collected from the center plot only.

2.5. Statistical Analysis

All statistical analysis was conducted using “R” software [34]. Species richness (alpha diversity) for the amendment treatments was analyzed using a series of Poisson GLM models for each year and functional group combination, with species counts for each experimental unit modeled against the amendment treatments. The Poisson model was used instead of a traditional ANOVA model as the Poisson model was determined to be a more appropriate model for the species richness count data in this study. The overall significance of treatment for each model was tested using a Chi-Squared Analysis of Deviance test. Pairwise comparisons (using Tukey’s Honestly Significant Difference method; HSD) between treatments within each analysis were conducted using the R package ‘emmeans’. Letter codes indicating groupings for pairwise tests were automatically assigned using the ‘cld’ function from the R package ‘multcomp’.

All multivariate analyses were conducted using the ‘rda’ function of the R package ‘vegan’. The percent cover data were converted to presence/absence to generate the sites × species matrices used in the RDA analyses.

A series of RDA models were fit in order to explore how the amendment treatments, soil chemistry (OM, C, P, Mg, K, Na, Ca, Mn, Zn, Cu, Fe, pH, N) and physical variables (roughness, coarse wood, coarse rock, erosion, and deposition) were related to community composition. For each year, an RDA model was fit where the sites × species matrix for the amendment trial was modeled against the soil chemistry and physical variables. For each RDA model, species loadings (% variance related to each species on each axis) were calculated and summed across functional groups in order to determine how much variance in community composition was related to each functional group.

To explore the amendment site communities in relation to the reference communities, an additional RDA analysis was conducted. A combined site × species matrix for the species data from the reference sites (obtained in 2019) and the 2019 species data from the amendment trial sites was modeled against the soil chemistry data for each of the amendment trial sites and reference sites. Species loadings (% variance related to each species on each axis) were calculated and summed across functional groups in order to determine how much variance in community composition was related to each functional group.

Lastly, the vegetation data set was analyzed by four plant functional groups. These were graminoids, forbs, woody plants, and non-natives. Graminoids were considered to be all herbaceous plants with a grass-like morphology (e.g., grass, sedge, or rush). Forbs were considered those herbaceous plants that are not graminoid. Woody plants were all native plants that have hard stems, with stems and buds that survive the winter above ground (e.g., woody shrubs and trees). While non-native species were considered those to be introduced by human activity, and outside of their natural range (e.g., Trifolium spp.), and inclusive of non-native forbs, graminoids and woody shrubs.

3. Results

The result of the overall Analysis of Deviance test comparison test for the effect of treatment on functional group species richness (alpha diversity) is provided in

Table 1. Results of Analysis of Deviance tests for treatment effect on functional group species richness. The asterisk (*) indicates a statistically significant effect (p < 0.05).

Effect	p-Value
	2016	2017	2018	2019
Forbs	0.0000 *	0.0000 *	0.0000 *	0.0000 *
Graminoids	0.0003 *	0.3916	0.2098	0.1867
Non-native	0.0000 *	0.0000 *	0.0000 *	0.0364 *
Woody	1.0000	0.9994	0.9890	0.7751
Total	0.0000 *	0.0000 *	0.0000 *	0.0000 *

. The effect of the amendment treatments on functional group species richness was significant in all years (p < 0.05), from 2016 (year 1 following establishment) to 2019 (year 4 following establishment).

The effect of treatment on the species richness of forbs, was significant in all years. For graminoids, the effect of treatment was significant in 2016, but not from 2017–2019. For non-native species richness within the field trial, treatment was significant from 2016–2019 (p < 0.05). Lastly, the effect of treatment on woody plant species richness was not significant in any year, and neither was the block effect (p > 0.1).

Figure 3 and Figure 4 provide the results of pairwise comparisons (using Tukey’s HSD method) of functional group species richness between treatments and reference sites. When examining total species richness, inclusive of all functional groups (Figure 3), the Control and the Control + Oats treatments are found to have significantly lower species richness in all years when compared to other Amendment treatments (p < 0.05). However, the species richness within the Control + Oats treatments is similar to richness in the upland forested reference sites. The treatments that include the peat amendment (Control + Peat, Control + Peat + Fertilizer, and Control + Biosolids + Peat) have significantly greater species richness than the Control and Control + Oats in all years (p < 0.05), but not always more significant than Control +Fertilizer, Control +Biosolids and the Control + Oats + Fertilizer treatments. Whereas the non-peat treatments with fertilizer (Control + Fertilizer and Control + Oats + Fertilizer) had significantly greater species richness than the Control and Control + Oats in 2016–2018 (p < 0.05) but not 2019 (p > 0.05).

When examining the effect of treatment on functional group species richness (Figure 4), the Control and Control + Oats treatments have significantly lower forb species richness in all years (p < 0.05), whereas treatments that contain peat have the greatest forb species richness in all years. Treatments without peat but containing fertilizer (Control + Fertilizer, Control + Oats + Fertilizer) maintain a moderate level of forb species richness, whereas the Control + Biosolids treatment has variable species richness, ranging from low to moderate levels of forb species richness. When examining graminoid species richness, there are significant differences in 2016 (Year 1 following establishment; (p < 0.05)), and no significant differences in richness from 2017–2019 (Years 2 through 4 of the trial; (p > 0.05)). Similarly, there is no significant difference in woody plant species richness between treatments inclusive of all years (p < 0.05).

Non-native species richness is variable between treatments and years. The Control and Control + Oats treatments were found to have the lowest non-native species richness in 2016 and 2017, whereas the Control + Peat + Fertilizer and Control + Biosolids + Peat treatments had the greatest non-native species richness in 2016 and 2017. However, non-native species richness between these treatments was not significant in 2019 (p > 0.5). Community composition differs between treatment and reference sites, as non-native species are abundant within the treatments, whereas the upland and lowland reference sites have a dominant woody shrub community. In contrast to the historically reclaimed and lowland reference sites, graminoids are nearly absent from upland forest reference sites.

The results of the multivariate RDA analyses is presented in

Table 2. Summary of RDA eigenvalues for the first two RDA axes for the amendment trial from 2016–2019.

	2016		2017
	RDA1	RDA2	RDA1	RDA2
Eigenvalue	1.313	0.487	1.716	0.718
Proportion Explained	0.256	0.095	0.298	0.125
Cumulative Proportion	0.256	0.352	0.298	0.422
	2018		2019
	RDA1	RDA2	RDA1	RDA2
Eigenvalue	1.169	0.449	1.334	0.346
Proportion Explained	0.193	0.074	0.228	0.059
Cumulative Proportion	0.193	0.267	0.228	0.287

. The results indicate that there is less variance associated with the first two axes (x and y) in 2018–2019, compared to 2016–2017.

Figure 5 provides a summary of the 2019 RDA analysis by treatment and functional group, with the soil and site variables included. Across all years, variance associated with the first two axes is lower in 2018 and 2019 when compared to 2016 and 2017. In all years, a greater proportion of the variance explained on the first axis was attributed to forbs (2019; RDA 1 variance proportion explained = 0.518) and non-native species (2019; RDA 1 variance proportion explained = 0.434), whereas the greatest amount of the explained variance on the second axis is attributed to graminoids (2019; RDA 2 variance proportion explained = 0.418) and forbs (2019; RDA 2 Variance proportion explained = 0.397).

In all years, the non-nutrient amended treatments (e.g., Control, Control + Oats) tended to cluster, as did the treatments containing fertilizer (Control + Fertilizer, Control + Oats + Fertilizer). Similarly, treatments containing peat (Control + Peat, Control + Peat + Fertilizer, and Control + Biosolids + Peat) created a cluster, while the Control + Biosolids treatment tended to fall between the non-nutrient amended treatments and treatments containing fertilizer. The Control and Control + Oats treatments cluster is associated with greater erosion and deposition, and greater Na and pH. The Control + Fertilizer, Control + Oats + Fertilizer, and Control + Biosolids treatments cluster is associated with higher pH and Mn, whereas the Control + Peat, Control + Peat + Fertilizer, and Control + Biosolids + Peat treatments cluster is associated with the physical variables of roughness, coarse rock, coarse wood, and soil chemical variables that included greater OM, TC, Ca, Cu, Fe, Mg, N, Zn, and lower pH.

The resulting influence of soil chemistry and the physical variables on plant functional groups is less distinct. The presence of certain graminoid species is associated with the Control + Fertilizer, Control + Oats + Fertilizer, and Control + Biosolids treatments and therefore higher pH, P and Mn. While certain woody species appear to be associated with the treatments containing peat, only a very small amount of variance between plots/treatments is explained by the presence of woody species (RDA 1 and RDA 2 variance explained = <0.04). Woody species did not indicate any significant differences between reclamation treatment, as they are associated with all treatments. By 2019, the presence of forbs and non-native species was not associated with a particular soil amendment treatment.

When comparing the 2019 treatments to the 2019 reference site conditions (upland, lowland, and historically reclaimed), the soil physical and chemical variables explained approximately 50% of the variance (R² = 0.54; Adjusted R² = 0.47). Most of this variance is explained on the first axis (31%), while 7% of the variance is explained on the second axis. The first axis is associated with non-native and woody species (RDA 1 variance proportion explained = 0.36 and 0.37 respectfully). The second axis is associated with forbs (RDA 2 variance proportion explained = 0.44) and woody species (RDA 2 variance proportion explained = 0.32). The presence of both woody species and non-native species explains a significant portion of the variance that separates the forested upland and lowland reference sites from the amendment trial treatments and historically reclaimed reference sites.

The results of the RDA ordination for the 2019 treatment and reference site conditions with soil variables is provided in Figure 6. In Figure 6A, Axis 1 shows significant separation between the lowland and upland reference sites (on the left) and the treatment sites (on the right). When examining the influence of the soil physical and chemical variables on the plant communities present, the first axis aligns with pH. When examining Axis 2 of the RDA (Figure 6A), treatments containing peat appear more similar to the lowland sites.

The results show that the Control and the Control + Oats treatments are most similar to the historically reclaimed sites. When examined by functional group, many woody species are associated with the forested upland and lowland reference sites, whereas select non-native species are associated with the historically reclaimed and amendment treatments. This result is consistent with the results of the variance partitioning that woody species and non-native species explain the greatest amount of variance along Axis 1.

4. Discussion

In this study, the presence of non-native species and a dominant woody plant community explains the largest proportion of variance between the forested reference sites, and the amendment treatments and historically reclaimed sites. Non-native plant species are absent from the forested reference sites, but abundant within the amendment treatments and historically reclaimed sites. This pattern is consistent with the literature and monitoring results from other post-mine reclamation units [6,18]. The historically reclaimed reference sites represent an approach to reclamation that used non-native, introduced annual cereals, forage, and turf species. The historically reclaimed sites are best described as a cultural meadow, dominated by Festuca longifolia, Galium triflorum, Lotus corniculatus, Melilotus albus, Phleum pratense and Trifolium hybridum, with a dearth of woody shrubs and trees. This community composition is consistent with other legacy approaches to reclamation that employed non-native, legume-based seed mixes [35].

Within the amendment treatments, non-native alpha diversity was initially greater in the nutrient treatments (2016 and 2017; Control + Peat + Fertilizer and Control + Biosolids + Peat), but by 2019 there was no significant difference in non-native alpha diversity between these nutrient treatments (Fertilizer and Biosolids) and the Control and Control + Oats treatment. These results indicate that regardless of the reclamation treatments employed in this study, non-native species will readily establish, and have a similar alpha diversity to the historically reclaimed reference sites after four years.

Similar to non-native species, graminoids were absent from the upland forested reference sites, and the graminoid community explains much of the variance between the forested reference sites and the historically reclaimed and amendment treatments. Graminoids are more abundant within the amendment treatment and historically reclaimed sites compared to the upland forested reference sites, which is consistent with the literature [19]. While graminoids are present within the lowland forested reference sites, the community composition differs from the community present with reclaimed areas.

Graminoids represent three of four species used within the initial seed mix applied to the amendment treatments. Graminoid seed was employed to provide initial erosion control within the amendment treatment units, while reducing the competitive interactions associated with the non-native legume species [35]. The results indicate that graminoid alpha diversity is not significantly different between amendment treatments, but the presence and higher abundance of graminoids is associated with treatments containing Fertilizer, Biosolids, and Oats, but not Peat. In year 1, graminoid alpha diversity was limited in the Control and Control + Oats treatment units, which have greater erosion. In contrast, the Control + Oats + Fertilizer also resulted in significantly greater graminoid alpha diversity in year 1. The results indicate that applications of nutrients through fertilizer or biosolids may increase graminoid alpha diversity and abundance within reclamation units. If erosion control with graminoids is desired in year 1 of reclamation, an amendment with fertilizer should be considered.

In years 2–4, graminoid alpha diversity was not significantly different between the amendment treatments. However, graminoids are associated with higher pH, P and Mn, and the Control + Fertilizer, Control + Oats + Fertilizer, and Control + Biosolids treatments. What remains unclear, is the competitive effect of this graminoid community on the planted trees and shrubs, and long-term trajectory of this site toward the forested reference sites. Continued monitoring, and comparison of the amendment treatment units with the historically reclaimed sites will help to answer questions around the influence of the non-native herbaceous legume and graminoid-based approaches to reclamation on woody species establishment and trajectory toward a forested reference site.

When examining the woody plant community, it is clear that woody species have established and are associated with all amendment treatments. Regardless of the soil amendment treatment employed, woody plant seedlings established in all experimental units, but alpha diversity is indistinguishable between treatments. The exception to this is lowbush blueberry, which had limited establishment within all experimental units, regardless of the amendment applied. The blueberry seedlings that did survive appear stressed, with bright red foliage. The literature indicates that peat and other organics are commonly incorporated in soil to increase OM when growing blueberry in suboptimal soils [36], such as the high pH and low organic matter conditions of the till. The optimum pH for blueberry, a member of the Ericaceae or heath family, is typically 4.0–5.2 [37]. Given the relatively high pH of the soil of the Control + Peat-based amendments (7.7–8) compared with the pH of the lowland and upland reference sites (4.2 and 3.9 respectfully), the death or limited growth (e.g., a few small branches, with red foliage) of blueberry transplants in the mine till was the result. The likelihood of other ericaceous shrub species, which are abundant within the mature forested reference sites, establishing within the inherently high pH mine till should be explored further.

Within the amendment treatment units, there was little to no natural establishment of woody species by seed rain with the exception of four Populus balsamifera seedlings and two Picea glauca seedlings. This may suggest the conditions within experimental units are not conducive to woody species seed rain establishment, or that reclamation units are receiving limited seed rain of woody species. If seed rain and spontaneous succession of woody shrubs and trees within the amendment treatments continues to be limited, then further interventions by planting native shrubs and trees that are representative of native forest biodiversity within the reference sites should be considered [6]. This may work to achieve a woody plant community composition within reclamation units that better reflects forest biodiversity and structure within the forested reference community, and also facilitate the development of a surface organic layer [19].

It is important that the trajectory of this woody plant community within treatments is closely monitored moving forward, and the progress toward the forested reference sites is considered. This includes the need to monitor more specific metrics such as canopy cover, leader length, stem diameter, and vegetation structure [9] and examine for significant differences in these metrics between treatments. Monitoring these metrics may provide a better indication than alpha diversity of the amendment treatment that is best suited to support the woody plant community. Currently, the analysis herein was not able to detect an association of the woody plant community with any particular amendment treatment, only that the absence or presence of a non-dominant woody plant community distinguishes all amendment treatments and historically reclaimed sites from the forested reference sites.

In contrast, forb alpha diversity increased as both peat and a nutrient source (fertilizer or biosolids) were added as soil amendments. The analysis indicates that variability in the composition of the forb community explains a large proportion of the variance between the amendment treatments. As soil roughness, coarse wood, organic matter, Mg, Ca, Zn Fe increased, so did the abundance of forbs. The cumulative additions of fertilizer, biosolids and peat increased alpha diversity of non-native species, graminoids and forbs within the amendment treatments. This is not consistent with other studies, which saw fertilization have no effect on species richness, but potentially increased vegetative cover, when applied to a peat mineral soil mix growth medium [18]. Additionally, it was observed that reclamation units are likely to have more graminoids and forbs than forested reference sites [19]. In this study, alpha diversity within the reference sites and amendment treatments is similar; however, the vascular plant community within the amendment treatments is not indicative of the diversity found in the forested reference sites. Other researchers have reported similar findings when comparing plant communities within reclamation units to forests [6,18].

The results indicate that the amendment treatments and historically reclaimed reference sites do not resemble the upland and lowland forested reference sites. The plant community present within the amendment treatment sites is best described as early successional, with the presence of non-native legumes dominating the historically reclaimed sites (e.g., Lotus corniculatus, Trifolium spp.). This is consistent with reclamation units monitored from the oil sands region in Alberta, Canada [18,35].

The Control and Control + Oats treatments have a lower abundance of forb, graminoid and non-native species (relative to other amendment treatments), but also lower pH and increased erosion and deposition. While the community composition may be most similar to the reference sites (according to the ordination), this could simply be the result of the generally poor establishment of vascular plant community within this treatment. Where there is a need for erosion control within reclamation units, the Control and Control + Oats treatment does not represent a useful treatment, which is compounded by what technicians noted as a poor performance of the woody plant community (i.e., stunted, chlorotic, or dead).

Fertilizer and biosolids-based treatments result in a vascular plant community, excluding woody species that is more similar to the reference sites than the peat-based treatments. However, there is also a higher instance of non-native species in fertilizer-based treatments in year 1. Non-native species continue to be a challenge for reclamation practitioners, but the establishment of a woody shrub community and canopy structure may work to reduce the instance of non-native and early successional species. Fertilizer-based treatments have fewer non-native species than the peat-based treatments, and promoted the establishment of a graminoid community in year 1.

The peat-based treatments (Control + Peat, Control + Peat + Fertilizer, and Control + Biosolids + Peat) also developed a distinct plant community. The peat-based amendment treatment appears to result in a plant community with greater alpha diversity of forbs, and non-native species, although this variation between treatments on plant functional group alpha diversity and plant community composition was not significant in 2019. It has been found that capping the overburden with peat, compared to incorporating the peat, will lead to an increase in the rate of invasion by plant species [35]. The non-native community within the peat-based treatments has a high abundance of non-native legume species, captured within the non-natives functional group that have established through seed rain. These are species that are known to hinder woody plant establishment within reclamation units [35], and are not a desirable component of the community composition in reclamation units. The results indicate that plant community composition within the peat-based treatments is least similar to the reference sites, and the success and abundance of the woody plant community within the peat-based treatments should continue to be monitored.

A key uncertainty that remains, and was not addressed herein, is the amendment treatment soil conditions that best resemble the reference site conditions. The forested sites are associated with greater OM, N, Zn Fe and P. The literature indicates that the most important factor in promoting the development of a plant community is the organic matter content of overburden/growth medium during closure [35], while OM and N supply are important factors that may limit tree growth following mining [38]. The peat-based amended sites, for example, are associated with increased roughness, greater OM, TC, Ca, Cu, Fe, Mg, N, Zn, and lower pH. The peat-based amendments have greater OM, C and N, and are most similar to the lowland reference sites in this regard. However, it is unclear whether the peat-based treatments can be considered to be more functionally restored soil than the other amendment treatments. This question was out of scope for this article, but will continue to be investigated by the authors moving forward, as soil conditions play an integral role in reclamation success [39].

This research is working to answer a long-term set of questions about how to restore plant communities on mine-impacted landscapes. Pinno and Hawkes [6] suggest that it may take more than 20 years before notable differences in plant communities are detected within reclamation units. The results of this study indicate that the advanced successional state of the mature forest environment makes for an unrealistic point of comparison with the early successional plant community within reclamation units. What remains clear from this study, is that while plant community structure can be differentiated between the amendment treatments, the amendment treatment plant communities do not yet resemble the forested reference sites. Moving forward, including reference sites that have undergone more severe disturbance, both natural (e.g., wildfire) and anthropogenic (forest harvest), should be included. Including additional reference sites, and those along the successional trajectory from early- to mid-seral stages should be identified and monitored as reference sites for this post-mine environment. Future monitoring must continue to examine the competitive interactions and success of the woody plant community across all treatments, with an emphasis on the peat and fertilizer-based treatments. In addition to this, future work examining methods of altering pH, soil roughness, coarse wood and OM should be examined, as these variables may support the establishment of seed rain and spontaneous succession, and an organic layer and a plant community that more closely resembles the forested baseline reference sites.