Author ORCID Identifier

0000-0002-9178-0692

Defense Date

2020

Document Type

Thesis

Degree Name

Master of Science

Department

Environmental Studies

First Advisor

Daniel J. McGarvey

Second Advisor

S. Leigh McCallister

Third Advisor

Paul Bukaveckas

Abstract

Rivers play a critical role in global carbon (C) budgets despite their comparatively small surface area. A significant portion of the terrestrial C that they receive is transformed, re-mineralized, or stored during transit to the ocean. Radiocarbon (∆14C) data show that a fraction of riverine organic C (OC) has been pre-aged in the terrestrial environment. Lateral export of carbon from these aged pools may contribute to atmospheric carbon dioxide emissions through microbial and photochemical oxidation. However, little is known about the regional climatic, anthropogenic, and landscape factors that promote the mobilization of aged OC to rivers. This study examines associations between riverine OC and river basin characteristics. It leverages data from two sources: 1) a spatially extensive collection of literature-reported ∆14C measurements and 2) the U.S. Environmental Protection Agency’s Stream-Catchment (StreamCat) database. The radiocarbon data include 95 dissolved (∆14CDOC) and 54 particulate (∆14CPOC) organic C measurements after averaging by location. We used the random forest (RF) machine learning algorithm to build independent models of ∆14CDOC (MSR = 7319.51, % var explained = 17.05) and ∆14CPOC (MSR = 11254.03, % var explained = 45.36). In both RF models, the StreamCat data were used aspredictor variables. Model validation was accomplished with a random, 75:25 split where 75% of the data were used for model building and the remaining 25% were used for testing and validation (∆14CDOC RMSE = 61.23, r = 0.71; ∆14CPOC RMSE = 39.82, r = 0.94). Key predictors of ∆14CDOC were generally climactic or land cover variables affecting terrestrial primary productivity. Key predictors of ∆14CPOC were primarily factors associated with sediment transport and erosion, but also included several indicators of anthropogenic influence. Human activities appear to be destabilizing both C pools, resulting in aged C flux to the more rapidly cycled C reservoir in rivers.

Rights

© Kaycee E. Faunce

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission

11-9-2020

Share

COinS