Survival analyses

Abstract

Survival analyses are a key tool for demographers, ecologists, and evolutionary biologists. This chapter presents the most common methods and illustrates their use for species across the Tree of Life. It discusses the challenges associated with various types of survival data, how to model species with a complex life cycle, and includes the impact of environmental factors and individual heterogeneity. It covers the analysis of ‘known-fate’ data collected in lab conditions, using the Kaplan–Meier estimator and Cox’s proportional hazard regression analysis. Alternatively, survival data collected on free-ranging populations usually involve individuals missing at certain monitoring occasions and unknown time at death. The chapter provides an overview of capture–mark–recapture (CMR) models, from single-state to multi-state and multi-event models, and their use in animal and plant demography to estimate demographic parameters while correcting for imperfect detection of individuals. It discusses various inference frameworks available to implement CMR models using a frequentist or Bayesian approach. Only humans are an exception among free-ranging populations, with the existence of several consequent databases with perfect knowledge of age and cause of death for all individuals. The chapter presents an overview of the most common models used to describe mortality patterns over age and time using human mortality data. Throughout, focus is placed on eight case studies, which involve lab organisms, free-ranging animal populations, plant populations, and human populations. Each example includes data and codes, together with step-by-step guidance to run the survival analysis.