OK, that's helpful. I now understand that the question is when constructing a model relating mass to a transformation of metabolic rate plus its habitat is any explanatory power added by considering the site with which the habitat is associated?
In that case I'd think of the base model as being included in, or nested within the explanatory model.
There's a good thread at https://stats.stackexchange.com/questions/4717/what-is-the-difference-between-a-nested-and-a-non-nested-model on the subtle differences among experimental design.
This scenario is actually equivalent to fitting the simple model first and removing its predicted variance from the data, and then fitting the additional component of the more complex model to the residuals from the first fit (at least with least squares estimation). Ruben van Bergen
Your first decision is whether you want to work with the F-Statistic output of lm or the logs ratio of glm. Although both are able to accommodate categorical variables, you don't always get 0 \le P \le 1 when you use lm with categorical variables unless you create dummies for habitat in your case, and you end up working with a maximum likelihood estimator in glm.
Neither approach is right. Assuming you want to use ordinary least squares, you would create two models
without_sites <- lm(mass ~ logWMR + hab1 + hab2)
mass_ex_res <- without_sites$residuals
with_sites <- lm(mass_ex_res ~ logWMR + hab1 + hab2 + site)
according to the method described by Ruben van Bergen