Date of Award
5-2026
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Forestry and Environmental Conservation
Committee Chair/Advisor
Thomas O'Halloran
Abstract
Diffuse solar radiation, sunlight scattered by atmospheric aerosols and/or clouds, enhances plant photosynthesis through the diffuse fertilization effect (DFE), increasing ecosystem gross primary productivity (GPP) per unit of incoming radiation. Despite this known effect, pine forests of the southeastern United States remain comparatively understudied in the diffuse light literature. This is a critical knowledge gap, as the southeastern United States sequesters more carbon than any other region of the conterminous country (approximately 176 ± 26 Tg C yr⁻¹, or roughly 56% of national forest net carbon exchange), and the region experiences some of the highest biogenic aerosol loads in North America.
This dissertation addresses this gap through four chapters utilizing eddy covariance observations from a new network of co-located mature and young tower sites, spanning two Southern pine species, longleaf pine (Pinus palustris) in coastal South Carolina and loblolly pine (Pinus taeda) in central Virginia.
Chapter 1 details the establishment of a coastal flux mesonet at Clemson University's Baruch Institute of Coastal Ecology and Forest Science at Hobcaw Barony, South Carolina. Comprising three co-located eddy covariance towers, this mesonet measures carbon, water, and energy exchanges above a tidal salt marsh, a mature mixed pine forest, and a longleaf pine restoration site. The open-access, AmeriFlux-published datasets created by this mesonet provide an observational foundation for the dissertation's subsequent research.
Chapter 2 applies interpretable machine learning, combining Support Vector Machine regression with SHapley Additive exPlanations (SHAP), to identify the drivers of daily light use efficiency (LUE) across 22 site-years from four Southern pine sites. Diffuse fraction (K_d) emerged as the single most influential predictor of growing-season LUE, surpassing leaf area index, total photosynthetically active radiation, vapor pressure deficit, and air temperature. The DFE was most pronounced in dense, closed-canopy mature loblolly stands, demonstrating that canopy architectural complexity, represented here by leaf area index, mediates the capacity of an ecosystem to exploit scattered light. Across all sites, diffuse radiation contributed up to 19% of total annual carbon assimilation.
Chapter 3 introduces the “Big-Leaf-IDLE” model, a modification of the standard “Big-Leaf” LUE framework widely used in satellite-based global productivity products, that incorporates the DFE via the empirical linear relationship between K_d and LUE. Applied to 22 site-years, the model matched or exceeded the predictive accuracy of the more complex “Two-Leaf” approach. It achieved the most parsimonious fit at all four sites, improving ΔAIC by 400–1,255 points over the baseline “Big-Leaf” model and reducing systematic GPP underestimation under highly diffuse conditions (K_d > 0.8) from up to 2.0 g C m⁻² day⁻¹ to near zero. The companion Index of Diffuse Light Efficiency (IDLE), derived directly from the fitted model parameters, offers a robust, dimensionless metric for quantifying site-level DFE sensitivity, tracking canopy structural recovery following disturbance, and comparing responses across species and stand ages.
Site-specific IDLE parameterization further projects that ongoing regional atmospheric brightening, driven by sustained reductions in anthropogenic aerosol loading since the implementation of the Clean Air Act, could reduce GPP at the mature longleaf site by approximately 20% over the next 30 years, highlighting the vulnerability of Southern pine ecosystems to changes in radiative quality.
Chapter 4 extends this analysis to ecosystem water cycling, revealing that diffuse radiation significantly stimulates both evapotranspiration (ET) and water use efficiency (WUE) in Southern pines. For equivalent incoming shortwave radiation, ET was consistently higher under diffuse than clear-sky conditions, an enhancement found to be driven primarily by increased canopy transpiration (measured via sapflow) rather than surface evaporation, consistent with stomatal stimulation by altered light spectral quality and the tight physiological coupling of CO₂ and water vapor exchange. While NASA's Priestley–Taylor JPL model outputs showed some of this ET enhancement without explicitly modelling it (R² = 0.75 for daily ET), the satellite-derived inputs used in the operational ECOSTRESS data product degraded the signal under diffuse conditions, leading to underestimations of ET during periods of high cloud cover.
Collectively, this research establishes diffuse light quality as a first-order control on Southern pine ecosystem carbon and water cycling. As atmospheric aerosol loads continue to decline across the southeastern United States, accurately representing diffuse radiation effects in ecosystem models will become increasingly important for projecting the future of one of the nation's strongest terrestrial carbon sinks. From the ground-up establishment of a coastal flux mesonet, through machine-learning attribution of diffuse light as the dominant driver of LUE and its coupled enhancement of canopy-scale transpiration, to scalable satellite-infused models, this dissertation builds an integrated case that diffuse radiation quality — and its ongoing decline — exerts a measurable, quantifiable control on the carbon and water cycling of the southeastern United States’ most important forests.
Recommended Citation
Forsythe, Jeremy, "Diffuse Radiation Controls on Southern Pine Forest Ecosystem Functioning" (2026). All Dissertations. 4270.
https://open.clemson.edu/all_dissertations/4270
Author ORCID Identifier
https://orcid.org/0000-0003-1600-8747
Included in
Ecology and Evolutionary Biology Commons, Forest Sciences Commons, Plant Sciences Commons