Mortality estimates of stage-structured populations must include uncertainty in stage duration and relative abundance

I explored mortality estimation for stage-structured populations, building on previous work that applied vertical life-table methods to populations of copepods. A new Bayesian approach for estimating mortality rates accounts for uncertainties in stage duration and number counted by stage, which have not been fully incorporated into previous analyses. This method assumes that mortality is similar among similar life stages. Results using simulated data show that realistic values of the standard deviation of stage duration and number of individuals counted result in reliable mortality estimates, though with wide confidence intervals. This uncertainty obscures variation in estimated mortality between successive stages and can also obscure bias due to violation of underlying assumptions such as that of a stable stage distribution. More importantly, the uncertainty calls into question many previous mortality estimates across pairs of life stages that do not account for these sources of uncertainty. The method was applied to an introduced population of the brackish-water cyclopoid copepod Limnoithona tetraspina in the San Francisco Estuary. Despite the uncertainties, results were interpretable: mortality was highest in nauplii and lowest in adults, probably because of high vulnerability of nauplii to invertebrate predators and low vulnerability of adults to fish predation.