Riparian response to the interactive effects of livestock grazing and wildfire in Northern Nevada.

摘要

In 1999-2001, a series of catastrophic wildfires burned 2.8-million acres across much of Northern Nevada. Wildfires of varying severity and extent burned though many of the State's limited riparian areas. The broad occurrence of the wildfires, coupled with an extensive inventory of pre-fire stream survey and grazing data housed by federal and state agencies, provided a unique opportunity to study the interactive effects of fire and livestock grazing on 81 independent riparian study sites across Northern Nevada. This study addresses previous discrepancies in riparian grazing strategy research and previous unexplained model variance with the re-creation of grazing histories for pasture/use areas containing each independent riparian study site. Trends in grazing variables were assessed over a 20-year duration with discriminant analysis techniques to statistically differentiate grazing strategies derived from grazing history variables. Binary logistic regression helped to predict stream survey attribute changes resulting from specific timing, duration, and rotation of livestock grazing coupled with watershed, hydrologic, and fire characteristics from previous models. Trend analyses revealed a general decrease in grazing use in pasture/use areas over the 20-year study duration. Burned linear regression models were significantly different from unburned site models for nine out of 12 grazing attributes. Burned model slopes were consistently steeper suggesting the occurrence of fire during the 1999-2001 wildfire seasons influenced the duration of growing season grazed, the number of days rested, the timing of grazing, and the number of AUMs grazed more rapidly over time when compared to unburned sites. Discriminant analyses effectively differentiated the riparian grazing strategies used in predictive models. Mann Whitney medians tests for paired data revealed significant improvement for most stream survey attributes for the burned and unburned sites data sets. Binary logistic regression revealed that the 1999-2001 wildfires and specific wildfire characteristics were not significant predictors of stream survey attribute change. Grazing timing, duration, and rotation variables were more robust predictors of attribute change. The relationships revealed in this study warrant additional modeling of the interactive effects of livestock grazing, wildfire, and natural processes for application at the landscape scale.