Improving Measurement, Quantifying Variation, and Predicting Tree Carbon Mass in California Conifers

Carbon sequestration and storage are important areas of scientific research. Despite this importance, the current approach to carbon mass estimates consists primarily of estimating wood volume, multiplying that volume by a species average wood density value to obtain biomass, then multiplying biomass by 0.5. Alternatively biomass may be estimated from allometric equations and then converted to carbon with the same 0.5 carbon conversion value. This approach is fundamentally inaccurate. This dissertation presents three improvements in carbon mass estimation that are critical to moving the science forward. In the first chapter I explore the importance of utilizing carbon fraction measurement methods that are capable of capturing the volatile component of tree tissues. I demonstrate a novel method that captures an additional 1.10% - 2.21% carbon mass in tree tissues. This increase is equivalent to “finding” an additional 16 Pg of carbon at a global scale. In the second chapter I explore the variation in measured carbon fractions, wood densities, and carbon densities within tree boles of seven major conifer species. I determine the importance of accounting for this variation, and discover negative correlations between wood density and carbon fractions within trees suggesting that the current approach of studying these characteristics as two independent properties is potentially biased. Applying the carbon fraction measurements at the whole tree level led to increases in carbon mass estimates of between 3.6% to 10.6% compared to carbon mass estimates that used a carbon fraction of 0.5. These values point to a systematic bias in current carbon mass estimation protocols. In the third chapter I develop carbon mass prediction models for five conifer species that capture some of the variation determined in chapter two. The models developed are the first vertically integrable conifer tree models to be based on accurately measured carbon fraction data paired with wood density measurements. The models are validated with an independent database and applied to the whole tree level demonstrate that accounting correctly for carbon leads to carbon mass estimates that are between 98.4% and 109% of the carbon mass estimated using standard approaches. The data presented makes a strong argument for the need of models that accurately account for variation in and correlation of carbon fraction, and wood density