Armstrong, R. A. Princeton University, raa@splash.princeton.edu
Pascual, M. A. Center of Marine Biotechnology,
Cavender-Bares, K. K. Massachusetts Institute of Technology,
Chisholm, S. w. Massachusetts Institute of Technologh,

 
BEYOND SIMPLE POWER LAWS: A NEW STATISTICAL METHOD FOR FITTING GENERAL DISTRIBUTIONS TO SIZE-SPECTRAL DATA FROM PLANKTON COMMUNITIES
 
The ability to characterize plankton size distributions is icreasingly recognized as essential to modeling the response of plankton commuities to environmental change. The size distribution of plankton has often been characterized as power-law, but statistical methods for estimating the parameters of underlying distributions have been slow to evolve. Here we show that a maximum likelihood estimator based on Poisson statistics can be used to estimate these parameters. This method overcomes the problems associated with regression techniques based on the Normal distribution and facilitates the fitting of distributions that depart from a simple power law. We apply this method to size spectra of bacteria and phytoplankton sampled using a flow cytometer, contrasting our results with results obtained from Normal regression, where zeroes in size categories cannot be tolerated. In particular, we show that this method solves the problem of parameter estimation when data has been gathered using different settings for large and small particles on a flow cytometer.
 
Day: Monday, Feb. 1
Time: 03:45 - 04:00pm
Location: Eldorado Hotel
 
Code: SS44MO0345E