Estimation of Relative Trappabilities by Age and Development Delays of Released Blowflies

Abstract

Marked blowflies Lucilia cuprina (Wiedmann) were released as newly-emerged adults and trapped continuously over several consecutive days. Females were classified into reproductive age classes according to their cycle and stage of ovarian development. The experiment was aimed at determining the relative trappability of females in each age class and durations of developmental delays associated with limited availability of protein and oviposition sites in the field. A reasonably natural model led to a complicated multinomial likelihood function involving 15 unknown parameters. Considerable difficulties were experienced in maximizing this likelihood because of the number of parameters and some inequalities between them. Computer algorithms often failed to converge or to report a negative definite Hessian for the log-likelihood. Some estimates obtained strayed very far from realistic values. We conclude that some parameters are poorly identifiable, so that to obtain estimates with sufficiently small standard errors would require impractically large sample sizes.