partitioning variance in the polr command

#1
I am using ordinal regression (MASS package, polr command) to analyze ranked coloration data (% piebaldism) of an iguana species based on sex, snout-vent length (SVL), and location. I'm concerned that skewed SVL between the locations (some locations have a large average SVL and others have small average SVL) is influencing the the location coefficients.

Any insight into how partitioning variance works with this type of regression analysis would be greatly appreciated.

Thanks in advance.