The ‘non-Euclidean’ distance
measure was used as it closely approximated geographic
distances between sample sites for when sea levels were
lower (i.e. when more of the ‘Sundaland’ land mass was
exposed) and therefore represents our hypothetical pathway
for historic land-restricted dispersal by B. dorsalis s.
l.. The strength of the association for either approach
was determined by linear regression analysis using the
program SPSS v17.0.