Vol. 69: 17-23, 1991 l MARINE ECOLOGY PROGRESS SERIES Mar. Ecol. Prog. Ser. Published January 10 Use of null models to assess patterns of size class migrations in the copepod Zausodes arenicolus Susan K. Service*, Keith Walters* Department of Biology, University of South Florida, Tampa, Florida 33620, USA ABSTRACT: Size-specific patterns in the migration (emergence) of a sediment-associated harpacticoid copepod were examined in a subtidal seagrass habitat using a null model approach. Size frequency distributions of all Zausodes arenicolus Wilson, 1932 and of females, males and copepodites collected on 2 dates in the water column and sediment were analyzed. Migrating and non-migrating size frequency distributions for all Z. arenicolus were different only in November 1983. The migration of significantly larger females in November 1983 influenced the size distribution for all migrating 2. arenicolus and was not the result of random emergence from the sediment. Size distributions of migrating and non- migrating males and copepodites were not significantly different and could not be distinguished from null distributions on either date. Evidence for size influencing migratory behavior of Z. arenicolus is weak, but the non-random migration of larger females on one date requires further investigation. INTRODUCTION The size of an individual often influences migratory behavior in marine organisms, especially Crustacea (Sulkin 1984, Epifanio 1988). Typically, the effects of body size on migration are inferred from differences in water column vertical distributions of different sized individuals (Huntley & Brooks 1982, Rothlisberg 1982, Williams & Fragopoulu 1985). Size influences mating success in several arthropod taxa (Ward 1983, Law- rence 1987, Grad & Maly 1988), and increased mating contacts is proposed as one advantage of vertical mi- gration for benthic species (Preece 1971, Williams & Bynum 1972),including harpacticoid copepods (Bell et al. 1988). Size-selective predation also can influence migrations of freshwater insects (Allan 1978) and crus- taceans (Newman & Waters 1984, Malmqvist & Sjo- strom 1987), and marine zooplankters (Robertson & Howard 1978, Fancett & Kimmerer 1985). Based on observed relationships between body size and migra- tion, a number of deterministic hypotheses, for exam- ple differential predation, have been proposed to ex- plain size-related differences in migration (Huntley & Brooks 1982, Ward 1983, Lampert 1987). An alternative to concluding that size differences in Present address: University of Georgia Marine Institute, Sapelo Island, Georgia 31327, USA O Inter-Research/Printed in Germany migration result from deterministic factors would be to suggest that observed distributions simply result from non-deterministic or random factors. The role of non- deterministic factors in shaping observed patterns can be examined using a null model. Although the term null model was coined by Colwell & Winkler (1983), equivalent models have been in use for some time (see Williams 1947).While most hypotheses assume a deter- ministic cause, null model hypotheses investigate the contribution of stochastic factors in shaping observed patterns. Null models can serve as controls for non- experimental data by evaluating the role of chance in structuring patterns (Strong 1980). Procedures for constructing a null model will vary considerably from study to study. Fundamentally, null models are used to generate randomly a new data set that is compared statistically with data actually col- lected. The newly constructed data set is generated by Monte Carlo simulations, a series of random draws with or without replacement, from a previously identified source pool that usually consists of a combination of all data collected. Numbers drawn from a source pool in simulations can be classified by variables such as species names, flowering times, or body sizes. Harvey et al. (1983) and Nitecki & Hoffman (1987) contain reviews and examples of the use of null models. Our study applies a null model approach to analyze behavioral differences in migration among size classes