Dormancy strategies in a random environment: Comparing structured and unstructured models

Hatching strategies are studied in a model of population dynamics in a rand om environment. The strategies are the fraction of dormant eggs that hatch each year, A projection matrix with random elements describes the populatio n dynamics. The model extends the density-independent model of Dan Cohen an d the density-dependent model of Stephen Ellner by adding structure to the egg bank, such that a fraction of the dormant eggs are buried and unavailab le for hatching each year. Differences between the structured and unstructu red models are analysed by deriving a small variance approximation for the population's long-term growth rate and relating the growth rate to the egg bank structure and the environmental parameters. Tuljapurkar and Istock (19 93) claimed that structured and unstructured models give very different pre dictions, especially when dormancy is high. We show that Tuljapurkar and Is tock's conclusions are the result of the particular dormancy type in their model, in which the maximum duration of dormancy is 2 years, rather than th e addition of population structure. Using parameters estimated from field a nd experimental studies of the freshwater copepod Diaptomus sanguineus in B ullhead Pond, RI, the density-dependent model correctly predicts that the p opulation should maintain a long-term pool of diapausing eggs, whereas the density-independent model makes the incorrect prediction that all eggs shou ld hatch at their first opportunity. If the per capita burial and emergence rates are low, as may be the case for many plant populations, the ESS opti mal hatching strategy is nearly that of an unstructured model with the mort ality of non-hatchers increased by the burial rate.