LOCAL δ18O-ELEVATION LAPSE RATES MODELED FROM UPSTREAM BASIN CHARACTERISTICS ACROSS WESTERN USA: IMPLICATIONS FOR PALEOELEVATION ESTIMATES OF BASIN AND RANGE, ROCKY MOUNTAINS

Stable isotopes of oxygen are often employed for paleoaltimetry studies, but the relationship between δ¹⁸O and elevation is complicated, particularly within continental interiors. To address these complexities, we compiled mean δ¹⁸O values from ~1,500 modern surface water samples across the Basin and Range (BR) and Rocky Mountains/Colorado Plateau (RMCP) to model the δ¹⁸O-elevation “lapse rate” for the interior continental western United States. For each sample site, we derive upstreams watersheds from digital elevation models (DEMs) and extract watershed-specific climatic and physiographic characteristics from PRISM 30-year normals to develop three multivariate models that consider the δ¹⁸O-elevation relationship in the context of upstream watershed climate variables. Each model considers a different set of basin-wide elevation statistics (mean, minimum, or both) as well as basinwide humidity, insolation and temperature characteristics chosen via model selection packages in R. We applied these models to develop 600 “local” δ¹⁸O-elevation lapse rates across the western United States. Tukey range tests for the resulting lapse rates suggest that, although models considering upstream climate characteristics predict δ¹⁸O variability better than those only considering characteristics at the point of collection, there are only minimal differences in the distribution of lapse rates derived from each model for each given region. 99.9% confidence intervals around the mean δ¹⁸O-elevation lapse rate for BR (-0.93‰ per km to -0.79‰ per km) is considerably shallower than the rates defined by previous studies from precipitation (e.g., Lechler and Niemi, 2011), while the same confidence interval for RMCP (-0.042‰ per km to 0.080‰ per km) suggests a potential mean lapse rate of 0‰ per km when the lapse rate is derived from upstream watershed characteristics. Thus, within the RMCP, modern surface water δ¹⁸O lapse rates suggest that geologic proxies of δ¹⁸O may fail to capture substantive elevation differences within deep geologic time. Moreover, the magnitude of paleoaltitude for the BR may be greater than previously estimated. Our results also suggest that, within continental interiors, lapse rates will be variable on local and regional scales, which may need to be considered for paleoaltimetric study - along with temporal variability not accounted for in this analysis.