Sun Gro Horticulture - Avenir, Alberta

Avenir Bog

The Avenir study area is a newly extracted forested continental bog with one peat extraction field and an undisturbed natural site serving as a reference. The bog was cleared and ditched starting in 2019, with vacuum extraction activities commencing in late summer 2020. It is located approximately 210 km north of Edmonton (54.992349, -112.412725) (Figure 5.). The mean annual temperature is 1.6 °C and average precipitation is 443 mm per year.

The vegetation in the adjacent undisturbed site is predominantly bog and poor fen vegetation dominated by Picea mariana, Sphagnum spp., and intermittent Larix laricina and feathermoss species. The soils are typic mesisols sitting on parent material composed of clay-rich moraine and organic deposits. The peat depth ranges from 0.75 m to 2.65 m.

Regular ditch maintenance to facilitate drainage and allow for extraction activities occurred during the spring, summer, and fall months of 2020 and 2021. The extracted field at Avenir is referred to as "Young" to differentiate this location in the study.

Sun Gro Horticulture - Seba Beach, Alberta

Seba Beach Peatland Complex

The Seba Beach facility has been active since 1975, with vacuum harvesting ongoing since 1980. The study area is located approximately 100 km west of Edmonton ( 53.480001, -114.861784) and has a mean annual temperature is 3.0 °C and average precipitation of 508 mm per year (Figure 5.). Prior to development, the peatland was classified as a forested continental bog.

There are multiple extraction and restored sites occurring simultaneously at this facility:

• One natural site (undisturbed, used as a reference) (Figure 14.)

• Two peat extraction sites (Figure 10., Figure 11.)

  • Field 16 - "Mature" site; active peat extraction

  • Field 12 - "Complete" site; peat no longer extracted but restoration has not commenced

• Two restored sites (Figure 12. , Figure 13.)

  • Field 16 - "Unsaturated" site; one perimeter ditch is not blocked, soil is not saturated to the surface

  • Field 5 - "Saturated" site; all ditches are blocked, soil is saturated to the surface

The vegetation in the adjacent undisturbed site is typical of bog and poor fen peatlands. The dominant vegetation at the Seba Beach study area is Picea mariana and Sphagnum spp., with pockets of Larix laricina and feathermoss species. The soils are typic mesisols and have parent material composed of clay-rich moraine and organic deposits. The peat depth ranges from 0.75 m to over 4 m across the study area.

Ditches within and between extracted sites were continuously cleared during active extraction to facilitate drainage and allow for peat extraction activities. At Field 12, extraction operations ceased in fall 2020 and ditches were blocked in summer 2021 in preparation for future restoration efforts. Extraction operations are ongoing in Field 16.

Ditches within the restored area were blocked and the water table was allowed to rise in Field 5 during summer of 2018, and in a portion of Field 16 in the fall of 2019. Due to its close proximity to the extracted area in Field 16, the perimeter ditch surrounding the north section of the Field 16 restoration site was not blocked. This has resulted in a partially lowered water table and unsaturated soil conditions. Restoration sites were revegetated with moss propagules using the moss layer transfer technique from an adjacent donor site, amended with phosphate fertilizer, and protected with straw mulch in the fall of 2019 and spring of 2020. Vegetation at both restoration sites is dominated by Eriophorum spp., Carex spp., Sphagnum spp., and Polytricum strictum.

Sampling Procedure and Experimental Design

A systematic, stratified sampling design was used to conduct transects within undisturbed, extracted, and restored peat fields varying in age and restoration technique (Figure 15. ). A random design was not selected due to site operational constraints based on location accessibility and required equipment access that would disrupt the study. A systematic design was selected to cover as much area as possible within these constraints and to encourage measuring the natural variability within each extraction treatment.

Sites were spaced approximately 20 m apart in natural, "complete" extracted, and restored locations, with a randomly assigned transect starting point and direction where operationally possible (Figure 15.). Each site was comprised of 3 locations spaced 5 meters apart (Figure 15.). In the "young" and "mature" extraction areas, sites were spread evenly along the length and width of the site ditch to cover as much area as possible; sites were placed 1.5 m in from the ditch edge (Figure 15.). Sites were selected along the ditch because ongoing peat extraction with heavy equipment prevented site placement away from ditches. A 30 m temperature and soil moisture transect across the width of the peat field (encompassing the ditch edge, centre of the field, and the following ditch edge) was conducted prior to the study and showed negligible differences in temperature and soil moisture within the top 10 cm. Thus, the site locations along the ditch edge were determined to be adequately representative of the peat field surface.

Measurement Protocols

Plant root simulator (PRS) probes measuring NO3-N, NH4-N, P, Ca, Fe, and Al were used to assess the relative rate of nutrients and major ions available for root and microbial uptake per surface area over a known time in the top 5-10 cm of the peat surface at both peatlands (Harris et al., 2020; Munir et al., 2017) (Figure 26.).

Five samples, comprised of three anions and three cations each, were installed in both hummocks and hollows at the non-extraction sites, and in each extraction phase to assess variability between extraction phases (Figures 18. - 25.). At each probe location, a steel rod was inserted into the peat to observe rust accumulation as an indicator of reduction-oxidation conditions at the time of retrieval (Owens et al., 2008) (Figure 27.). Soil moisture at 6 cm below the ground, surface temperature, and peat temperature 10 cm below the ground were measured during installation and at the time of retrieval. Depth of rust and soil moisture were selected to estimate the aeration and redox conditions at each site. Surface temperature was measured to measure the effect of shading from the presence or absence of vegetation and soil temperature below the ground was recorded to characterize temperatures favorable for microbial activity.

Randomization Restrictions

Five samples, comprised of three anions and three cations each, were installed in both hummocks and hollows at the non-extraction sites, and in each extraction phase to assess variability between extraction phases (Figure 15). Although the initial intention was to use each sample as an independent, single experimental unit, this was not possible due to lack of randomization and invalid assumptions of independence. As a result, this experimental design has three levels of randomization restrictions:

1. Group of three PRS locations, together making 1 site (Figure 15.):

1. • This was done to increase the area sampled because variability was expected at all peatland locations.

   • Adding additional PRS locations increased the likelihood of collecting a representative sample instead of relying on one sampling point.

2. The transect of sites in each sampling area (natural, extracted, and restored), comprised of 5 samples (Figure 15.):

1. • A stratified technique was selected to help overcome accessibility restrictions; however, there is a risk of over-representing certain environmental conditions if the selected transect area was not representative of the entire area as whole. This may be especially problematic at the extracted sites because all probes were installed along a linear feature.

   • To assure a representative sample was selected, future studies could include short transects in different directions within the treatment area to assure all site conditions are sampled.

3. Blocks encompassing spatial variability:

1. • To account for spatial variability between the sampling locations, complete and incomplete blocks were used:

     1. Block 1: Incomplete block

        • Avenir Natural (Hummock and Hollow) and "Young" extracted peatlands

     2. Block 2: Complete block

        • Seba Beach Natural (Hummock and Hollow), "Mature" extracted peatland, and "Unsaturated" restored peatlands

     3. Block 3: Incomplete block

        • Seba Beach "Complete" extracted and "Saturated" restored peatlands

   • Although the current blocks will help decrease the effect of spatial variability, it would be beneficial to include an additional natural site in Block 3 in future studies to assure a control group is present in each block. This was challenging because representative natural sites near Block 3 have already been extracted or restored.

As a result of these restrictions, the statistical analyses performed for this project use a mean value for each sampling area. Therefore, there are 9 independent experimental units, characterized as follows;

• Natural Hummock: 2 units

• Natural Hollow: 2 units

• Extracted: 3 units

• Restored: 2 units

Please note: variability in climate, surficial and bedrock geology, soils, hydrologic connectivity, groundwater composition, and landscape position between peatlands within and between biogeoclimatic settings has made the selection of appropriate reference peatlands for study replication impossible. Therefore, a site-specific approach was used to assess the potential processes affecting peat chemistry in an attempt to control for the above confounding variables. This study additionally uses a split-plot design to differentiate between hummocks and hollows in both natural sites; however, analysis conducted for the purpose of this class project does not include the split-plot, and instead treats hummocks and hollows as individual treatments. This will be corrected in future analyses.

Statistical Tests

Data manipulation and statistical analyses were conducted in RStudio (v1.4.1717), using R Statistical Software (v.4.1.1; R Core team, 2021). Data was manipulated and graphed using the tidyverse R package (v1.3.1; Wickham et al., 2019).

ANOVA

Question: How do extraction activities impact the physicochemical conditions of the peat?

To answer this question, a one-way ANOVA with random blocks using mixed models with an alpha level = 0.05, was conducted for each physicochemical parameter to observe if there were any statistical differences between treatments:

• Fixed Effects:

  • Physicochemical parameters: depth of rust, soil moisture, surface temperature, below ground temperature

  • Treatments: natural hummock, natural hollow, extracted, restored

• Random Effects:

  • Blocks 1, 2, 3

Mixed models were produced using the lmerTest R package (v3.1.3; Kuznetsova et al., 2017) and were used because we are interested in both the treatment means and variance components, but are not interested in the differences between each block. Significant results were followed up with pairwise comparisons using Tukey's HSD, alpha = 0.05, using the multcomp R package (v1.4.17; Hothorn et al., 2008).

Correlation Analysis

Question: Are the observed physicochemical changes enough to influence the availability of nitrogen and phosphorus?

To answer this question, Spearman rank correlations were conducted to observe the relationship between physicochemical parameters and nutrient availability. Residual plots for each chemical compound were assessed. The data violated the assumptions of normality and homogeneity of variances and person correlation coefficients were not substantially improved with data transformations. Therefore, a non-parametric correlation test was selected.

Each physicochemical parameter (depth of rust, soil moisture, surface temperature, and below ground temperature) was correlated with the PRS nutrient availability (NH4 , NO3, P, Al, Ca, Fe) for a total of 6 comparisons for each physicochemical parameter. P-values (alpha = 0.05) associated with each correlation were adjusted for multiple inference using a sequential Bonferroni (Holm) adjustment, with 6 as the number of comparisons. Spearman rank correlations and associated p-values were produced using the Hmisc R package (v4.6.0; Harrell, 2021).