In a wildlife study, where a group of animals are monitored from some prespecified age to the event of death, the observed data consists of a set of failure/censoring times as well as other measured covariates. Based on the ages of the animals upon entry into the study, the observed cohort comprises partially observed durations that are possibly unbiased or left-truncated. In this talk, we will present some methodological strategies for simultaneously including multiple types of partially observed duration data into a single modelling procedure. We will discuss some recent applications of these procedures in modelling African lion and baboon mortality as well as some open statistical problems for future research.