There is a growing need to monitor vegetation communities in semi-arid systems, as these systems play a vital role in the global carbon cycle. Water limitation is the most important constraint on primary productivity in semi-arid systems, such as southeastern Idaho. This region is projected to experience increased winter precipitation due to climate change, potentially increasing its capacity for carbon sequestration. Thus, southeastern Idaho has been implicated as a possible future carbon sink. This effect will, however, likely be scale-dependent due to variations in plant growth form (e.g. grass, forb, shrub). Therefore, this project sought to classify plant communities across the 2,300 square-kilometer Idaho National Laboratory (INL), and quantify community responses to fire. We used an amalgam of eight cluster analysis methods evaluated across seven geometric and non-geometric classification evaluators to generate optimal community assignment for 333 stratified random 50m transects. Flexible beta (β= -0.25) performed the best across evaluators with a sixteen-cluster optimal solution, indicating that sixteen distinct communities were present at the INL in summer of 2017.
Results/Conclusions
This result was surprising considering that the 2008-2011 mapping effort found twenty-two distinct communities using the same methods. Since 2008, ~17% of the site burned in a series of wildfires, so we investigated fire as a possible driver of this reduction. We hypothesized that patchy invasion within burns was contributing to a regional-scale loss of species diversity and simplification of metacommunity structure at the INL, evident, in part, as a loss of plant communities. At the scale of the whole INL site unburned transects had higher β-diversity, and extrapolated richness. Indicators of unburned communities included A. tridentata, while indicators of transects burned in the past 33-years included B. tectorum, and C. viscidiflorus. Further, cover of B. tectorum increased with increased fire frequency and severity and cover of A. tridentata decreased with increased fire frequency and severity. However, shrub cover in burns did not completely disappear, for example C. viscidiflorus increased cover in response to increased fire frequency and severity. This indicates that functional group transitions may be slowed, because native shrub cover is often maintained post-fire. These findings are important for generating site-wide estimates of carbon flux across the INL landscape, because many land cover maps assume shrub cover loss in burned areas. Current spatial distributions of alliance level vegetation communities allow a nuanced view of carbon flux due to vegetation communities.