Variability in Oceanic Particle Size Distributions and Estimation of Size Class Contributions Using a Non-parametric Approach.

A dataset of nearly 400 measurements of the particle size distribution (PSD) compiled from the Pacific, Atlantic, and Arctic Oceans is used to examine variability in the magnitude and shape of the PSD, and to characterize the partitioning of particle number, cross-sectional area, and volume concentration among defined size intervals. The results indicate that the relative contributions of three size classes based upon the pico-, nano-, and microplankton size range exhibit substantial changes among measures of particle size and between oceanic environments. The single-slope power law model commonly employed to characterize the PSD in aquatic studies is demonstrated to have significant limitations in capturing the complexity of PSD shapes observed for natural particle assemblages, and in consequence poorly predicts the relative contributions of these different size intervals. We show that specific percentile diameters derived from the cumulative distributions of particle size are strongly correlated with the contributions of these three size classes, and that these non-parametric descriptors of the cumulative distribution provide superior performance for estimating their contributions while requiring no assumption of underlying PSD shape. A comparison of these predictive relationships with independent field measurements suggests that this approach is generally robust for particle assemblages representing a wide diversity of marine environments.