Disentangling trait-based mortality in species with decoupled size and age

1. Size and age are fundamental organismal traits, and typically, both are good predictors of mortality. For many species, however, size and age predict mortality in ontogenetically opposing directions. Specifically, mortality due to predation is often more intense on smaller individuals whereas mortality due to senescence impacts, by definition, on older individuals. 2. When size-based and age-based mortality are independent in this manner, modelling mortality in both traits is often necessary. Classical approaches, such as Leslie or Lefkovitch matrices, usually require the model to infer the state of one trait from the state of the other, for example by assuming that explicitly modelled age (or stage) class structure provides implicit information on underlying size-class structure, as is the case in many species. 3. However, the assumption that one trait informs on the other is challenged when size and age are decoupled, as often occurs in invertebrates, amphibians, fish, reptiles and plants. In these cases, age-structured models may perform poorly at capturing size-based mortality, and vice versa. 4. We offer a solution to this dilemma, relaxing the assumption that class structure in one trait is inferable from class structure in another trait. Using empirical data from a reef fish, Sparisoma viride (Scaridae), we demonstrate how an individual-based model (IBM) can be implemented to model mortality as explicit, independent and simultaneous functions of individual size and age - an approach that mimics the effects of mortality in many wild populations. By validating this 'multitrait IBM' against three independent lines of empirical data, we determine that the approach produces more convincing predictions of size-class structure, longevity and post-settlement mortality for S. viride than do the trait-independent or single-trait mortality models tested. 5. Multitrait IBMs also allow trait-based mortality to be modelled either additively or multiplicatively, and individual variability in growth rates can be accommodated. Consequently, we propose that the approach may be useful in fields that may benefit from disentangling, or investigating interactions among, size-based and age-based demographic processes, including comparative demography (e.g. life-history consequences of resource patchiness) and conservation biology (e.g. impacts of invasive predators on size structure but not life span of natives).