Biogeochemical and hydrologic processes in montane landscapes provide immense societal benefits, sustaining the production of crops, livestock, and timber, sequestering vast amounts of carbon in soils and biomass, and supplying water to over half the global population. Predicting how changes in land use and climate affect the provision of montane ecosystem services requires characterizing soils and vegetation in landscapes with profound spatial complexity. Satellite imagery, predictions from existing empirical models, and field-based observations yield useful insights into the spatial organization of soils and vegetation in montane landscapes, but each of these sources of information carries uncertainty and potential bias that may lead to mistaken conclusions. Nonetheless, systematic handling of uncertainties is rare in practice. In my dissertation, I quantify land surface phenology (LSP) parameters (i.e., metrics of leaf seasonality) and soil organic carbon (SOC) inventories and their associated uncertainties in a montane landscape in Veracruz, Mexico, within a Bayesian framework. Additionally, I assess potential bias incurred in SOC inventories and dynamics (inferred from radiocarbon activity) when changes in soil volume due to land use are disregarded during sampling and calculations and compare existing and novel approaches to mitigate that bias. This research demonstrates (1) that the major uncertainties detected directly affect ecological inferences (e.g., effects of deforestation and succession on SOC cycling; differences in LSP-derived growing season length across land-use types) and (2) that systematic quantification of uncertainty is feasible and useful for prioritizing future data acquisition and model improvements in montane landscapes.
In partial fulfillment of the requirements for the PhD degree in the Graduate Program in Land and Atmospheric Science
Nate Looker, LAAS Doctoral Student Advised by Profs. Randy Kolka and Ed Nater