PARENT SESSION
Oral Session - Human Influences on Landscape and Watershed Processes Chair(s): Lopez, Ricardo ¹, ¹ Landscape Ecology Branch, Las Vegas, NV
Wednesday, March 31, 2004 1:00 PM - 4:20 PM Zeus Room B

Exploring spatio-temporal dynamics of agricultural to residential land-use transitions using panel data. ^*AN, LI , BROWN, DANIEL G. , NASSAUER, JOAN I. and LOW, BOBBI , ¹ School of Natural Resources and Environment, Ann Arbor, MI, USA

ABSTRACT- The focus of much work on residential development has been to understand the influences on location choices of individuals and households, but those choices are limited by the types of developments available to them. This study addresses this issue using panel (cross-section and time series) data from two townships in Southeast Michigan. The dependent variable is a typology of exurban residential developments that characterizes aspects that we assumed to have a strong influence on resulting land-cover patterns. With spatial data over five time periods (1950s to 1990s), a total of 84 human residential parcels or subdivisions were classified using our typology for each time period based on their characteristics. The "choice" of each parcel/subdivision at each time was regressed against a set of locational, landscape, demographic, and socioeconomic variables using multinomial logit modeling, where the coefficients were allowed to be both temporally and cross-sectionally variant. We found that (1) 57%-74% of the variation in the choice of development type was explained by the fixed effect logit models, indicating existence of time-invariant coefficients but varying intercepts. (2) Different development types tended to be explained by different variables. (3) Land use changes were temporally, but not spatially, stationary. Aside from providing an exurban typology of residential developments to help explain where and when these developments come about, this study has demonstrated an approach to integrating data into a framework that accommodates temporal and spatial variation, and thus has the potential to shed important light on the mechanisms behind landscape changes.

KEY WORDS: panel data analysis, land-use and land-cover change, socioeconomic and demographic stressors, landscape dynamics, geographic information systems (GIS)