Abstract:
The use of development threat as a tool in conservation prioritization is becoming
increasingly common. When used as a component of a larger planning methodology that
may include factors such as biological integrity, area, and connectivity, degree of threat
may help an organization to protect a larger amount of land that it would otherwise have
been able to. I develop two models of development: one to predict the threat of
development for Washington County, Virginia (based on structures built between ~1960
and 2006) and a second to examine how settlement patterns have changed over time
(based on structures built before 1938). Variables that are incorporated into the pre-1938
model include: elevation, slope, Euclidean distance to roads, Euclidean distance to towns,
and relative elevation. Logistic regression results in negative coefficients for all of the
predictor variables, all of which are significant except for elevation. The strongest
predictor variables were slope and relative elevation. The model had high accuracy, with
a value of 81.4% obtained for a calibration dataset. Variables included in the ~1960-2006
model include: Euclidean distance to roads, slope, relative elevation, Euclidean distance
to fire and rescue stations, and Euclidean distance to schools. Logistic regression results
for the 1960-2006 model differ from those reported for pre-1938 settlement. Distance to
towns is no longer a significant predictor, elevation is less significant, and relative
elevation is positively correlated with development potential (for locations near roads).
Negative coefficient values still exist for slope and Euclidean distance to roads, as well as
for the new variables Euclidean distance to fire and rescue stations, and Euclidean
distance to schools. The accuracy of the model was slightly less than the pre-1938 model
of development, but still a decent 76.6%. Overall, these results show a change in
development patterns over time, with people becoming less limited by distance to towns
and high elevations, and the evidence of a preference for house sites at a high relative
elevation near roads.
