Congratulations to UC Merced F3 fellow, Yulissa T. Perez Rojas for successfully defending her PhD in Environmental Systems this summer. Her dissertation, “The Scale Integration Challenge: How Moisture Variability and Osmotic Stress Create Systematic Biases from Laboratory to Continental Carbon Predictions,” tackles one of the greatest uncertainties in climate science: soil carbon projections.
Advised by Professor Teamrat Ghezzehei, Yulissa’s research bridges soil physics and mathematical modeling to deepen our understanding of biogeochemical processes. Beyond her fellowship, she has been a dedicated leader in the Environmental Systems Graduate Group, Graduate Student Association, and Monarch Center showing.
Congratulations, Dr. Perez Rojas!
Dissertation Abstract:
Earth system models exhibit order-of-magnitude disagreements in soil carbon projections, representing the largest uncertainty in climate predictions. This dissertation reveals a fundamental yet overlooked source of this uncertainty: the tyranny of small scales in biogeochemical modeling. Through three interconnected studies spanning microseconds to continental scales, demonstrate how seemingly minor short-term processes create systematic biases that cascade through temporal and spatial scales to fundamentally compromise global carbon cycle predictions. Continental-scale analysis of hundreds of AmeriFlux stations reveals pervasive bias in soil respiration models from using average environmental conditions, with sensors showing nearly 100% underestimation compared to actual variability. Next, rigorous analysis of soil water retention-based respiration models demonstrates how mechanistic approaches capture nonlinear moisture effects without temporal averaging, with water-limited conditions exhibiting the strongest nonlinear responses. Lastly, laboratory experiments reveal incoming water chemistry biases results by over 100%, with osmotic microenvironments controlling microbial energetics within minutes. The results reveal that microscopic processes operating on millisecond timescales ultimately control century-scale climate predictions, highlighting the need for models that preserve nonlinear responses across temporal and spatial scales.