The tables below show a simplified view of the raw data used in this analysis. All codes in parentheses () are defined in the caption associated with Table 1. and Table 2.

The data from the above tables can be interpreted as follows:

Sampling Units

An average of three sets of measurements were used to create 5 individual sample locations (CATCH + EXP + WDEPTH) (Table 1.). There are 45 experimental units, each comprised of 3 pooled samples and characterized as follows:

• Natural: 20 units - 10 hummock and 10 hollow locations

• Extracted: 15 units, comprised of "young", "mature", and "complete" peatlands

• Restored: 10 units, comprised of "unsaturated" and "saturated" peatlands

Please note: for the purpose of this project, each sample location is considered independent even though the actual sample locations are in close proximity and are not considered independent.

Predictor Variables

The peat extraction treatment is a manipulated categorical variable comprised of three classes: natural, extracted, and restored peatlands.

Response Variables

The response variables are:

• Physicochemical properties (continuous variables, Table 1.)

  • Depth of rust (DRCOMB), depth of ice (DICOMB), soil moisture (PERVOL), surface temperature (TEMPs), and below-ground temperature (TEMPd))

• PRS nutrient supply rates (continuous variables, Table 1.)

  • Ammonium (NH4) and nitrate (NO3)

• Water chemistry and temperature values (continuous variables, Table 2.)

  • Ammonium (NH4), nitrite/nitrate (NO2_NO3), pH (pHt) - pH, electrical conductivity (ECt), and water temperature (TEMPt).

Physicochemical Conditions

Physicochemical conditions were plotted using box plots (Figure 28.) to observe any major outliers or highly skewed data. All conditions appear to be homogeneous and normally distributed with minimal outliers. Data were deemed acceptable and were not transformed. The depth of ice outliers were large, but the high variability of the natural site depth of ice suggests that these outliers may not be due to sampling error. Alternatively, logs or rocks present in the peat profile may have been mistaken for ice, leading to the shallow depth of ice outlier observed.

PRS Nutrient Supply Rates

PRS nutrient supply rates were plotted using boxplots (Figure 29.) to observe any major outliers or highly skewed data. Both ammonium and nitrate values were highly skewed, with high values observed at the extracted sites and values below or slightly above the detection limit. Data were log transformed (Figure 30.). Transformed ammonium values were more evenly distributed, and nitrate values were moderately improved (Figure 30. a & b).

Outflow Chemistry

Water chemistry concentrations were plotted using boxplots (Figure 30. & 31.) to observe any major outliers or highly skewed data. Nitrate values were below detection limits for all sites during the PRS installation period (Figure 30.b). For ammonium concentrations, there was one large outlier present in the restored site (Figure 30.a). Ammonium values were plotted again using log transformed data with the outlier removed (Figure 31). The distribution around the median was more uniform with the log transformed data. Therefore, log transformed values were selected for use in subsequent visualizations. Outflow values for natural (Avenir) and extracted (complete) were not collected because there was no water present during the sampling period.