Population and community ecology (original) (raw)

In this chapter we discuss the best methodological tools for visually and statistically comparing predictions of the metabolic theory of ecology to data. 2 Visualizing empirical data to determine whether it is of roughly the correct general form is accomplished by log-transforming both axes for sizerelated patterns, and log-transforming the y-axis and plotting it against the inverse of temperature for temperature-based patterns. Visualizing these relationships while controlling for the infl uence of other variables can be accomplished by plotting the partial residuals of multiple regressions. 3 Fitting relationships of the same general form as the theory is generally best accomplished using ordinary least-squares-based regression on logtransformed data while accounting for phylogenetic non-independence of species using phylogenetic general linear models. When multiple factors are included this should be done using multiple regression , not by fi tting relationships to residuals. Maximum likelihood methods should be used for fi tting frequency distributions. 4 Fitted parameters can be compared to theoretical predictions using confi dence intervals or likelihoodbased comparisons. 5 Whether or not empirical data are consistent with the general functional form of the model can be assessed using goodness-of-fi t tests and comparisons to the fi t of alternative models with different functional forms. 6 Care should be taken when interpreting statistical analyses of general theories to remember that the goal of science is to develop models of reality that can both capture the general underlying patterns or processes and also incorporate the important biological details. Excessive emphasis on rejecting existing models without providing alternatives is of limited use.