Testing for Shifts in the Vertical Distribution of Plankton Using a Robust Kolmogorov-Smirnov Like Statistic

Ecological communities often occur along important physical gradients such as altitude, moisture, and light. The two sample Kolmogorov-Smirnov statistic can be used to test for equality of two population abundance profiles along a gradient. Standard Monte Carlo methods demonstrate the well known fact that this procedure is highly sensitive to the presence of extra-multinomial variation - that is, local population patchiness. However, the distribution of a scaled version of the Kolmogorov-Smirnov test statistic is shown to be insensitive to large departures from the simple random sampling assumption. The procedure was developed to investigate day-night shifts in the vertical distribution of zooplankton.

Key Words
Bootstrap pivoting; Compound multinomial; Cumulative distributions; Plankton profiles; Randomization tests.

Woollcott Smith is Professor, The Fox School of Business and Managment, Temple University, Philadelphia, PA 19122, and Senior Fellow, Marine Policy Center, Woods Hole Oceanographic Institution, Woods Hole, MA 02543 (E-mail: v5567e@vm.temple.edu). Andrew Beet is Research Assistant, and Andrew R. Solow is Associate Scientist, Marine Policy Center, Woods Hole Oceanographic Institution, Woods Hole, MA 02543.