Remote-sensing and statistical approaches to evaluating landscape changes associated with targeted livestock grazing
Loading...
Authors
Shane, Tracy L
Issue Date
2025
Type
Dissertation
Language
en_US
Keywords
Bayesian framework , cheatgrass , fuels reduction , remote sensing , targeted grazing , uncrewed aerial vehicle (UAV)
Alternative Title
Abstract
It is estimated that more than 20% of the Great Basin sagebrush rangelands are invaded by cheatgrass (Bromus tectorum L.). Land managers have been exploring the use of targeted livestock grazing to mitigate wildfires in cheatgrass-invaded sagebrush rangelands but the results of these trials have demonstrated variable levels of success. Uncrewed aerial vehicles (UAVs) mounted with active (e.g. LiDAR) or passive sensors (e.g. optical multispectral camera) have been used as a possible tool for detecting changes associated at the landscape spatial scale that are associated with targeted grazing treatments. In this dissertation I evaluate the effectiveness of autumn or fall-season targeted grazing in reducing cheatgrass-dominated fine fuel cover, structure, and connectivity by using field and UAV-derived measures of vegetation canopy cover and structure. Vegetation composition, canopy cover, and structural metrics acquired by these technologies are ultimately used to estimated aboveground biomass in drylands, or more specifically, the amount and connectivity of fine fuel. However, I first conducted a systematic literature review of how aboveground biomass is indirectly estimated in shrubland and savanna ecosystems worldwide using UAV-mounted technologies. I found this to be a knowledge gap. Secondly, at the field-plot scale, I used a non-parametric zero-inflated beta distribution in a Bayesian hierarchical framework to detect short-term changes in the occurrence, and percent conditional cover and mean cover of cheatgrass to fall-season targeted grazing. I compared the impacts of targeted grazing on plant and ground cover, structural and pattern metrics including plant height and patch dynamics using UAV-acquired time series. Finally, I share lessons learned during collaborative implementation of targeted fall-season grazing treatments across different sites and ranger districts in Nevada. The results of my dissertation show that in the short-term (1-4 years), targeted grazing treatments increased cheatgrass occurrence compared to grazing exclusion in the field cover dataset, however, targeted grazing did not increase either conditional cover or mean foliar cover of cheatgrass when compared to grazing exclusion. The UAV dataset corroborated this finding, as mean cover of a combined cheatgrass/litter fuels class did not change between summer pre-treatment and summer post-treatment years when a single application of targeted grazing treatment was applied in the autumn of 2023 at two sites. Targeted grazing treatments did increase the number of cover class patches and richness of cover class patches depending upon the study site, indicating that targeted grazing may be altering the continuity of the patches. The addition of consecutive years of targeted grazing treatment across study sites is warranted to understand longer-term effects. The results of this dissertation demonstrate that UAV monitoring of vegetation can provide insights on changes to spatial heterogeneity due to targeted livestock grazing or grazing exclusion that researchers would otherwise have difficulty detecting.