Results & Discussion

HIGH BULK DENSITY, SHALLOW ORGANIC LAYER, AND LOW SOIL NUTRIENTS MOST IMPORTANT ATTRIBUTES FOR WELL PAD RECOVERY 

Through a principal component analysis (PCA) of the scaled soil biophysical data (Fig 14.) we found that 49.9% of the variance between reclaimed well pads in this study area is explained by mainly bulk density and soil nutrient levels (total Nitrogen, total organic Carbon, and Carbon-Nitrogen ratio) and 19.9% of the variance is mainly explained by site age, soil organic layer depth, and soil pH. 

A top-down divisive cluster analysis (k-means) using Euclidean distance was used to assign groups to the data. Three clusters were specified to help separate the 30 well sites into priority groups and identify which attributes need to be addressed first. A trial of clustering with two groups showed approximately 20 sites in the higher priority category which is less feasible to reclaim. These priority groups can be seen as the ellipses in Figure 14. 

High priority sites are characterized by high bulk density (soil compaction), low soil nutrients, shallower organic layer depths, and a more neutral-alkaline pH, the median pH of high priority sites is 7.95 (less ideal for plant growth, Table 5). Higher soil pH is sometimes attributed to soil compaction due to decreased movement of water and oxygen and low microbial activity which would normally create more acidic soil. Similarly, the higher pH on well pads may also be caused by reduced organic matter compared to the reference condition. Natural region does not seem correlated with these sites as both foothill and boreal sites are about equally present in the cluster. In addition, although it appears high priority sites are younger than medium or low priority, we can see in Table 5 that the median ages are actually relatively similar in terms of forest successional age (at 16, 18, and 22.5 years post-reclamation).

The medium priority group includes the largest number of well pads (~15) out of the priority clusters. These sites are less compact and have slightly higher nutrient levels than the high priority group; however, they still fall on the higher compaction/lower nutrient side of those ranges. These sites have a variety of pH, organic layer depths, and ages with a majority of sites (~12) falling in the foothills natural region. 

Finally, low priority sites tend to have higher levels of soil nutrients and less soil compaction (lower bulk density) but exhibit a variety of soil organic layer depth, pH, and site age. Nine out of the 10 sites in this classification are central mixedwood boreal sites. The reason for this finding is unclear but could have several ecological explanations such as differences in plant/fungi species, reclamation procedures, microclimates, and microbial activity.  Since this study did not collect related data, this component of the findings likely requires further investigation to fully understand the underlying mechanisms. 

NON-NATIVE SPECIES COVER STRONGLY ASSOCIATED WITH RECLAIMED WELL PADS REQUIRING FURTHER INTERVENTION

A second PCA using percent cover data and basal area (Fig 15.) showed that variance between reclaimed well pads can be largely attributed to the separation between tree and shrub cover (approximately 48.4% variance explained) and herb, graminoid, and non-native cover (approximately 24.3% variance explained). There do not appear to be strong patterns of site age or natural region correlated with cover data in this PCA. Table 6 shows the median values for each priority group, and we can see that tree and shrub data are very different between high/medium and low priority sites and non-native cover strongly correlated with well pads that are exhibiting poor soil quality. 

Mapping the priority groups from the cluster analysis based on soil data, site age, and natural region onto the cover data PCA allows visualization of how vegetation data might relate to the biophysical data from Figure 14. Overall, many high and medium priority sites have low tree and shrub cover as shown by red and orange points dominating the left portion of the PCA (and as very low medians in table 6). It also appears that low priority sites are have a variety of plant cover ratios (in terms of dominance of trees, shrubs, or graminoids) but do tend to have lower non-native and herb cover. This indicates that the vegetation types that are present on these sites is not correlated with the soil data and that revegetation of high priority sites should focus on avoiding non-native and herb species rather than purposefully planting a specific life-form group. 

A direct gradient analysis using PCA further shows the correlations between soil and cover data (Fig 16.). Here, we again see that high priority sites have greater non-native and herb cover. This reflects the agronomic/weedy species likely planted on these sites as well as plants groups with traits better adapted to growing in nutrient poor and compact soils.  

A CanCor analysis of the relationship between site and cover data (Table 7) reveals that bulk density and organic layer depth are the biophysical attributes most highly associated (statistically significant, with p-values <0.001) with differences in cover data. High organic layer depth is strongly associated with treed sites and negatively associated with graminoid cover (high in sites experiencing arrested succession), meaning this is an essential attribute for the regeneration of reclaimed well pads. Bulk density is strongly associated with non-native and herb cover and negatively associated with shrub cover which shows that soil compaction is another main attribute inhibiting well pad recovery, and should be a principle soil trait, along with organic layer depth, addressed in reclamation (Table 7). Other site traits such as age (see CAN1 and CAN3), nutrients (see CAN2), and pH (see CAN3) still have associations with cover data but to a lesser extent than bulk density and organic layer depth, so soil reclamation should focus on managing the latter two attributes first.

RECOMMENDATIONS

High priority sites should be addressed first due to their high level of soil compaction and poor bio-chemical qualities. Medium priority sites with attributes close to high priority sites could then be reclaimed. Low priority sites likely do not require intervention as they do not exhibit evidence of arrested succession. 

To relieve the soil compaction of high priority sites, the soil should be lightly tilled which will prevent the loss of soil organic carbon through breakdown of soil aggregates. A light tiller/cultivator should be used to loosen the top 4-6 inches of the soil. 

After tilling, soil amendments should be incorporated to address high pH, low organic content, and low soil nutrients. Some advised options would be chicken manure, plant residue (compost or mulch), and sulfur. First, chicken manure contains organic matter which will help increase organic matter depth in the soil. It is also high in nitrogen, phosphorus, and potassium - all essential nutrients for plant growth. Chicken manure is also considered an acidic fertilizer (contains organic acids), so it help reduce the soil pH over time. Plant litter will aid in deepening the soil organic layer and increasing total organic Carbon. Finally, adding some sulfur will also help reduce pH through microbial breakdown into sulfuric acid. This increase in microbial activity will help make nutrients more bioavailable and maintain soil structure. Sulfur also doubles as a nutrient for plant growth. 

Shallowly tilling the soil to reduce soil compaction may not completely remove vegetation but can disturb it enough to reduce its competitiveness. Depending on the project budget and seed availability, it would be ideal to seed high priority sites with native herb, shrub, and tree species (or use transplants/saplings). Otherwise, prioritizing improved soil quality will aid in success of propagation attempts from the surrounding forest.

Conclusion

From this study, we've identified that organic layer depth and bulk density in combination with non-native vegetation cover are the attributes significantly contributing to the variance between reclaimed well pads and associated with their arrested succession. We now know that shallow organic layer depth and high soil compaction are correlated with inhibited regeneration post-reclamation certificate and this is also associated with low tree and shrub cover. Future reclamation should aim to prevent these attributes when reclaiming well pads, as they appear to be the greatest factor from this study causing well sites to maintain a grassland-like state rather than approaching reference forest conditions. 

In theory, eventually (especially with a natural disturbance like wildfire that would encourage natural succession) these well pads might progress towards the reference forest condition, but we don't know how long that might take. Therefore, there is value in going back to further reclaim older well pads that are in arrested succession to reduce forest fragmentation. Prioritizing reclamation activities based on site attributes facilitates cost-effective action to reach improved ecological outcomes effectively.