Results

Sample Location is more important for recovery composition

 than Age Class & Recovery time is long (30 years or more)

Figure 9 shows much overlap between the composition of lichen species in reference and wellpad locations, but clear differences between reference and wellpad locations of the same site. Many of the lines connecting the sites are moderate to long in length, and go in very inconsistent directions that show no clear pattern. All 3 age classes are also mixed across ordination space. Altogether, this figure indicates that there is much site-specific variation and that the disturbance on wellpads does have an impact on species composition. Once again, there is much overlap between the sample locations, with a few wellpad sites standing out as outliers in Figure 10. However, the confidence ellipses for the wellpads are larger and taller, while the reference site confidence ellipse is shorter and longer. This could be interpreted to mean that wellpads have much greater variability in species composition and reference communities are more consistent, but these results could be influenced by the wellpad site outliers.

I used perMANOVA to test for statistical differences between sample locations, age classes, and the lichen species distance matrix using 999 permutations. Both sample location and age class together were not strongly significant (p=0.059), but when I looked at only sample location, there was a slightly more significant result (p=0.021), and age class on it’s own was not significant (p=0.581). I followed this with a linear model regression and ANOVA test on each PCoA axis. The most significant variable for PCoA 1 was Sample Location (p=0.1481), though this shows weak significance. Both Age Class (p=0.6805) and both variables together (p=0.2096) showed even weaker influences on lichen species composition. For PCoA 2, sample location once again showed the strongest results (p=0.0691), though this is overall weak evidence for a significant influence of location or age class on lichen communities. A pairwise perMANOVA on age classes also demonstrated weak results, with differences between 10 and 30 years post reclamation showing the most notable associations (p=0.542). Differences between 10 and 20 years reported a p-value of 0.914, and 20 to 30 years, 0.886. 

Sample Location is more important for recovery composition

 than Age Class & Invasive Crested Wheatgrass is having some impact on Recovery

In Figure 11, sites in reference locations are grouped more closely together (indicating more consistent community composition), where sites in wellpad locations are more dispersed (indicating more variability in community composition). There is some overlap between reference and wellpad sites, but references are grouped more on the right while wellpads spread more towards the left side of the plot, demonstrating some shift in community composition between the two groups. Hespcom (Hesperostipa comata), Boutgra (Bouteloua gracilis), and the shorter vectors closest to them are strongly associated with reference sites. Tragdub (Tragopogon dubius), Pascsmi (Pascopyrum smithii), and especially Taraoff (Taraxacum officinale) are also associated with the reference locations but are likely influenced by something other than disturbance. Agrocri (Agropyron criistatum ssp. Pectiniforme) is strongly associated with the wellpad sites, making them an indicator of disturbance. In Figure 12, the confidence ellipses of all of the age classes overlap, which tells us there is likely not much change in composition between age classes. The 10-year (red) confidence ellipse is taller and aligned more with NMDS2, while the 30-year (blue) confidence ellipse is wider and aligned more with NMDS1, which may mean that as sites are aging, they are becoming more diverse or moving toward an alternative state. 

A perMANOVA on the vascular species distance matrix using 999 permutations on both age class and sample location proved to be significant (p=0.001). On its own, sample location (p=0.001) was shown to be slightly more significant than age class (p=0.002). A linear model regression and anova test on NMDS1 showed that both age class (p=2.045e-07) and sample location (p=6.658e-8) were very influential factors on community composition and cover. For NMDS2, age class was weakly significant (p=0.1379), but sample location still seemed to be important (p=1.58e-06). When I ran a pairwise perMANOVA test on the different age classes, 20 vs 10 years post reclamation (p=0.001) was the most significant, followed by 10 vs 30 years (p=0.002). 20 vs 30 years was only weakly significant (p=0.436). 

Finally, I decided to look at the relationship between lichen species richness and the vascular plant species matrix to see how the vascular plants may be affected by the lichen species, and if so, which species are having the most significant impact. To do this, I used a negative binomial model. This model returned weak results for Agrocri  (p=0.0457) and Site_ID (p=0.0322), suggesting that Crested Wheatgrass (Agrocri: Agropyron cristatum ssp. Pectiniforme), may have some limited impact, but it depends on site-specific characteristics.

Conclusions

In conclusion, disturbance seems to be impacting the recovery of lichens, graminoids, and forbs on wellpad sites. I found that the sample location (wellpad or reference) seems to have a bigger influence on recovery than age class, though it seems that it takes a longer amount of time (at least 30 or more years) for lichens, graminoids, and forbs to recover to reference conditions on wellpad sites. These differences are clearer in the graminoids and forbs than in the lichen, which seems to be more driven by site-specific variation. This makes sense as micro-habitats, micro-climates and site-specific conditions will have such a big impact on these communities' ability to be successful in any site, let alone one that has been reclaimed post-disturbance. This also means that current reclamation strategies are in some way supporting the recovery of vascular plants, but do little to support the recovery needs of lichen. Disturbance seems to be increasing diversity and heterogeneity on wellpad sites to some degree compared to reference sites that are more homogenous in composition. Invasive Crested Wheatgrass in particular seems to be taking advantage of disturbance to establish itself strongly on wellpad locations, though it may have a bigger influence on whether vascular plants can recover to reference conditions than lichens. It is important to note, however, that many of my univariate results came back with weakly significant or insignificant p-values, so further research with larger sample sizes may be needed to confirm these results. 

Overall, I would say that reclamation practices need to take into greater consideration that it takes many years for wellpads to return to reference conditions after reclamation, and these sites likely require continued monitoring and adaptive management to ensure that invasive species like Crested Wheatgrass do not outcompete native plants over time. Native grasses and forbs that are strong competitors should be considered for restoration, and site-specific conditions should be taken into account when deciding which species to plant, especially to encourage the return of important native lichen species that may only be able to tolerate a narrow set of environmental conditions. Post-disturbance grasslands management should incorporate more frequent sampling and monitoring of all key species of these sites to ensure their successful recovery, and how other factors, like soil parameters, can be returned to reference conditions.