A new research paradigm for bivariate allometry: combining ANOVA and non-linear regression

Area of Science:

• * Biological statistics
• * Comparative allometry
• * Quantitative biology

Background:

• * Comparing allometric patterns across different subject groups is crucial for understanding biological scaling.
• * Standard statistical methods like analysis of covariance (ANCOVA) are often limited when dealing with curvilinear data.
• * Existing approaches may not adequately handle diverse error structures in biological data.

Purpose of the Study:

• * To introduce a novel statistical routine for exploring and comparing allometric variation in multiple groups.
• * To provide a flexible framework that accommodates various assumptions about random error distributions (e.g., normal, lognormal, heteroscedastic).
• * To demonstrate the application of this routine in analyzing field metabolic rates in mammals.

Main Methods:

• * Integration of analysis of variance (ANOVA) principles with non-linear regression techniques.
• * Utilizing a three-parameter power equation incorporating a categorical variable for group identification.
• * Development of a statistical model allowing for different assumptions regarding random error forms.

Main Results:

• * The novel routine successfully analyzed allometric variation in field metabolic rates of marsupial and placental mammals.
• * Allometric equations for both groups exhibited non-zero intercepts.
• * A higher allometric exponent was observed for placental mammals compared to marsupials, indicating distinct scaling patterns.

Conclusions:

• * The presented statistical routine offers a versatile and powerful tool for allometric analyses, surpassing standard ANCOVA on log-transformed data.
• * This method provides a more robust framework for comparing biological scaling across diverse groups and error assumptions.
• * The findings highlight significant differences in metabolic scaling between marsupials and placentals.