Data tables

The raw data for this project consisted of soil, cover, tree basal area, and age, for well pads and reference plots in Lower Foothills and Central Mixedwood natural regions (Table 1). Data was subsetted to only include wells as reference data was not required for this project. In addition, fern, clubmoss, and lichen data was excluded due to most or all values being zero for those variables. 

Data was split into two datasets, one with age and soil biophysical variables and a second with vegetation cover data. This aided in visualizing relationships in the data through the lens of soil characteristics and vegetation separately to better inform recommendations for reclamation. 

Abbreviations used in Tables 1 and 2 are as follows: 

LFHmean_mm: LFH soil horizon depth (organic layer)

BD_0-15cmdepth_cm: Bulk density of the soil - 0-15 cm depth

pH_0: measured pH of the soil - 0-15 cm depth

TOC_0: Total organic carbon in the soil - 0-15 cm depth

TN_0: Total nitrogen in the soil - 0-15 cm depth

CNratio_0: Carbon to nitrogen ratio of the soil - 0-15 cm depth

tph_total: Number of live and dead trees/ha

LiveBA_m2/ha: Live basal area (BA; m2/ha) for all trees combined.

DeadBA_m2/ha: Dead basal area (BA; m2/ha) for all trees combined.

Units

All soil data and site age are continuous quantitative variables with units described in Table 2. Cover data contains continuous quantitative data in the form of trees/hectare and square meters per hectare as well as percent cover of life-form plant groups (visually determined).

Age and Soil Data

Age and soil data (Table 3) also referred to as Site Data includes the site age (years since reclamation) and soil physical (bulk density), chemical (pH, total organic carbon, total nitrogen, and carbon-nitrogen ratio), and biological (organic layer depth) properties.

The youngest site in this study was 7 years post-reclamation and the oldest site was 48 years post-reclamation; however, a majority of the sites fall around 15-20 years post-reclamation (Fig 8.).  The age median was 17.5 years post-reclamation certification (Fig 8.) A limitation to assessing the effect of time since reclamation is that the distribution of the data is not even. Here, the best case scenario would include sites spanning the 7 to 48 years post reclamation range consistently rather than a high concentration of sites reclaimed ~15 years ago. This is likely due to reclamation still being a relatively new practice for oil and gas disturbance in 1966 (when the oldest well site in this study was certified reclaimed).  Although there are fewer data points for sites older than 30 years post-reclamation, they provide important insights for successional trajectories of reclaimed well pads.

It's important to acknowledge that large variances in reclamation practices are not captured in this dataset but are likely contributing to much of the variance in cover and soil data of these sites. While reclamation criteria sets the standards for reclamation work, there are many different ways to meet those requirements, and different companies and consultants can approach a reclamation project in their own way. However, this data is not easily obtained as past reclamation records can be inconsistent or nonexistent. 

Cover Data

Cover data (Table 4) includes percent cover of herbs, shrubs, graminoids, and non-native species and cover of trees via density per hectare and basal area (both live and dead). The inclusion of dead woody debris is important as a source of structure and organic material . The tree, snag, and stump measurements are more precise than vegetation cover, which was estimated visually and not always by the same people although they were trained together and calibrated through group training. 

Predictor and Response Variables 

Tree and cover data (response variables are being assessed in relation to predictor variables which include the natural region (Boreal vs. Foothill) of the well pads, site age (years since reclamation), and site biophysical data (including soil nutrients, bulk density, pH, and organic layer depth).  Predictor variables are observed not manipulated. Natural subregion is the only categorical variable, all others are continuous numeric. 

Exploratory graphics 

DISTRIBUTIONS

Preliminary assessments (boxplots, histograms) of the data distributions of both soil/age and cover data showed skewed data. Scaled boxplots (Figs 9 and 10) were used to identify which variables were already relatively normally distributed and which variables could be transformed. Age, Organic layer depth, Total organic Carbon, and Total Nitrogen were all selected as having significant outliers or skewed distributions which could benefit from transformation (Fig 9.). Bulk density, pH, and Carbon-Nitrogen ratio were not transformed due to their already satisfactory distributions. 

The scaled boxplots for cover data revealed skewed data for tree and shrub variables which represent the lack of woody species on many reclaimed well pads in arrested or slowed succession but could also reflect younger sites in early successional stages (Fig 10.). Live and dead trees per hectare, Live basal area, Dead basal area, and Shrub % cover were therefore transformed while Herb, Graminoid, and Non-native % cover were left as they were already relatively normally distributed.

DATA TRANSFOMATIONS

The following transformations were made to normalize both datasets: 

Soil and age data (Fig 11.) log transformations of total organic Carbon, site age, and organic layer depth, and 1-(1/(Total Nitrogen+0.1))+5 (Fig 12); 

Cover data (Fig 13. )log transformations of live and dead trees/hectare+0.1, dead basal area+0.1, and shrub cover+0.1, and squareroot transformation of live basal area+0.1. 

Even with log transformation, dead basal area was zero-inflated, and still represents skewed data in this dataset. This is ecologically important however, as it shows that many well pads had no trees present likely due to young age or the competitive advantage of planted agronomic species which keep the well sites in arrested succession.