Controlling the influence of variation in a factor

#1
I sampled communities in different habitats along a continental margin. Depth is a strong structuring factor in the ocean. To mitigate or control the influence of the depth factor and in a comparison of habitats, would it make more sense to perform a 2-way ANOVA with with depth as random or an ANCOVA and make depth a covariate? Thanks in advance.
 

Miner

TS Contributor
#2
Typically, variables for which you want to control, but are not of specific interest are handled through blocking or as covariates, not by including them as a variable. In addition, ANOVA assumes that all variables are held at specific levels. I doubt that depth will conveniently occur at specific values. While you could bin the depths, it will still introduce error. ANCOVA would treat depth as the continuous noise variable that it actually is.