ABSTRACT
Zeros play a unique role in ecological and environmental data and often need special modeling consideration to accommodate them. The purpose of this chapter is to elaborate this challenge. Two illustrative examples are the 0 resulting from no precipitation at a location (an environmental variable) and the 0 resulting from species absence at a location (an ecological variable).
212Modeling choices we discuss include Tobit models, zero-inflated models along with hurdle models, and ordinal and nominal categorical data models. We consider these choices in both the univariate and multivariate settings. We formalize these models and discuss their properties, fitting, and associated inference. Contexts can be spatial, temporal, and spatio-temporal.
