Skip to main content
. 2011 Jan 31;6(1):e16513. doi: 10.1371/journal.pone.0016513

Table 2. Variation in African ancestry modeled by historical and geographical variables using multiple linear regression.

All island
Adjusted R2 = 0.075
β estimate S.E. t P
(Intercept) −0.5676 0.02727 −20.81 <10−4
Distance to sugar mills (km) −0.00195 0.00075 −2.59 0.0097
Distance to coast (km) −0.00646 0.00145 −4.46 <10−4
Molasses production (m3/km) −0.00141 4.0 · 10−5 −4.42 <10−4

African ancestry is log10-transformed to assume a linear fit and satisfy model assumptions (e.g. normality of errors). Separate models for the whole island, the East region and the other 5 regions combined have been constructed. For a detailed description of the models and the variables, see Text S1 and Tables S3.