In some insect and crustacean populations, animals molt independently and asynchronously such that a randomly selected animal is equally likely to be anywhere in the molt cycle at any time. Then, it is possible to estimate intermolt periods from observations on whether or not tagged animals molted while at liberty or from observations on the time it takes animals brought into captivity to molt for the first time. Existing estimators have been oversimplified by assuming that intermolt period is a deterministic function of premolt size, whereas observations from experiments measuring exact intermolt period suggest a lognormal distribution. A lognormal intermolt period model is developed here for tagging data from field experiments and time-to-first-molt data from laboratory studies. The deterministic and lognormal models are applied to a field study of spiny lobster (Panulirus argus). Simulation and sensitivity analyses are used to demonstrate the unsuitability of the deterministic estimator and the properties of the more realistic lognormal estimator.

Key Words
Censored data; Growth; Mark–recapture; Maximum likelihood estimation.

Russell B. Millar is Senior Lecturer, Department of Statistics, University of Auckland, Private Bag 92019, Auckland, New Zealand. John M. Hoenig is Associate Professor, Virginia Institute of Marine Science, College of William and Mary, P.O. Box 1346, Gloucester Point, VA 23062.