Bison were almost hunted to extinction in the late 1800's but they are now recovering and have recently been reintroduced to restored prairies for their ecosystem-engineering roles. The effects bison have on remnant (never-plowed) prairie vegetation are well documented. However, there is little known about how bison will affect vegetation assembly in restored prairies. Remote sensing with unmanned aerial vehicles (UAVs) is used to identify changes within plant communities with increasing frequency but the limits of this are still unknown. The goals of this study were to 1) measure the impacts of bison reintroduction and restoration age on plant communities across a chronosequence of restored and remnant tallgrass prairies, and 2) determine the limitations of multispectral imagery in identifying the plant species or any differences in site greenness. During 2014 and 2015, bison were reintroduced to half of these sites. Plant surveys were conducted at each of the sites and compared to UAV surveys to assess multispectral performance. Vegetation indices were used to determine the uniqueness of individual species’ or plant groups’ spectral signatures. The sites’ greenness was determined using NDVI averages. Linear models were used to test for the impacts of grazing and restoration age on the plant communities.
Results/Conclusions
Plant diversity decreased with restoration age in sites with bison present. However, species richness decreased as a result of restoration age, regardless of bison presence. This suggests that plant communities are changing as the sites age and bison are impacting plant species relative abundances. The use of multispectral imagery may be limited to detect these changes. We found that certain species signatures were too similar to differentiate. The limitations of multispectral imagery for our purpose may be due to the amount of spectral bands available in the sensor. Hyperspectral imagery, which has hundreds of bands, may be much more effective at identifying small differences between species. Site greenness compared between restorations shows potential to monitor landscape level differences. These indices have been shown to be correlated with other traits such as foliar nitrogen. However, more information is needed to see if this holds true with this equipment. Despite these limitations, our results show that multispectral imagery has the potential to identify some species and other traits important for restoration. However, without the ability to identify most species, shifts in the community due to factors such as restoration age and grazing will be impossible to detect.