Controls on Alkenone Carbon Isotope Fractionation in the Modern Ocean.

Carbon isotope records from alkenone biomarkers (εp37:2) produced by haptophyte algae are frequently used for atmospheric CO2 paleobarometry, but this method has yielded inconsistent results during periods where CO2 variations are known independently. Recent syntheses of algal cultures have quantitatively demonstrated that εp37:2 indeed records CO2 information: εp37:2 increases as aqueous CO2 concentrations increase relative to carbon demand. However, interpretations of εp37:2 are complicated by irradiance, where higher irradiance yields higher εp37:2. Here we examine the roles of physiology and environment in setting εp37:2 in the ocean. We compile water‐column and sediment core‐top εp37:2 data and add new core‐top measurements, including estimates of cell sizes and growth rates of the alkenone‐producing population. In support of culture studies, we find irradiance to be a key control on εp37:2 in the modern ocean. We test a culture‐derived model of εp37:2 and find that the quantitative relationships calibrated in culture experiments can be used to predict εp37:2 in sediment samples. In water‐column samples, the model substantially overestimates εp37:2, largely resulting from higher irradiance at the depth of sample collection than the integrated light conditions under which the sampled biomass was produced and vertically mixed to the collection depth. We argue that the theory underpinning the conventional diffusive alkenone carbon isotope fractionation model, including the "b" parameter, is not supported by field data and should not be used to reconstruct past CO2 changes. Future estimates of CO2 from εp37:2 should use empirical or mechanistic models to quantitatively account for irradiance and cell size variations. Key Points: Irradiance is a primary control on alkenone carbon isotope fractionation in the natural environmentCulture‐calibrated model can predict alkenone carbon isotope fractionation in core‐top sediment samplesDiffusive model "b" value is not correlated with growth rate: diffusive model should not be used for paleobarometry [ABSTRACT FROM AUTHOR]