Answers are grounded in this paper and cite their sources.
Mixture models reveal population density in ecological data, suggesting improved management of vulnerable species.
Mixture models are frequently used in ecology for the estimation of abundance. These models adopt a hierarchical structure in which the observations are dependent on both a detection probability and abundance at the survey site. Applications are typically to discrete survey count data. Analogous mixture models have not been developed for semi‐continuous sampling data, which are characterised by a large number of zero observations and non‐zero observations measured on a continuous scale. We attempt to bridge the gap between mixture modelling approaches developed for discrete counts and their application to semi‐continuous data. We use survival analysis to derive a relationship between a continuous measure of abundance and the probability of a zero observation, and incorporate this relationship into a two‐part, log‐normal hurdle model, with the biomass represented as a hierarchical model parameter. We apply the model to semi‐continuous marine sampling data collected from a bottom trawl fishery in New Zealand. Despite the simplicity of the parameterisation, the model is able to describe the observations and predict a relative biomass density layer over space. The approach allows mixture models to be applied to semi‐continuous ecological data. By allowing the population density distribution to be properly estimated, the methods presented here can inform the management of anthropogenic impacts on vulnerable species, as well as understanding distributional shifts that may arise due to climate change.
No takes yet. Share an insight, caveat, or question.
Edwards et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: