Skip to main content

Table 1 Statistical model types and specifications were tailored for each data set. Generalized linear models fit all data best. Asterisks (*) indicate both square root transformation and zero-inflated models. Non-native plant models tested included model families Poisson, negative binomial, and Gaussian; with and without zero-inflation; and transformations such as square root. Native shrub richness models tested included model families Gaussian and Poisson. Shrub cover and dominant obligate seeder cover were modeled with binomial family, in which the proportion of shrub cover was divided by the transect length for the response variable.

From: Decade-Long Plant Community Responses to Shrubland Fuel Hazard Reduction

Question Response variable Model family Data
Native plants Shrub cover Binomial Proportion
Native plants Native shrub richness Poisson Species counted
Native plants Dominant obligate seeder cover (buckbrush cover) Binomial Proportion
Non-native plants Non-native annual grass density Gaussian * Count
Non-native plants Non-native plant density by harmfulness rating Gaussian * Count