Landscape and Urban Planning 89 (2009) 37-48 Contents lists available at Science Direct Landscape and Urban Planning ,', ".':'.'::""" ,. .i : . :":.:,:.:.".,:,''..,:, .' journal homepage: www.elsevier.com/locate/landurbplan Past and projected rural land conversion in the US at state, regional, and national levels b Eric M. White a,*, Anita T. Morzillo , Ralph J. Aliga a Pacific Northwest Research Station. USDA Forest Service, C01vallis, OR 97331, United States b Western Ecology Division, National Healtlt and Environmental Effects Research Laboratory, US Environmental Protection Agency, Corvallis, OR 97333, United States ARTICLE INFO ABSTRACT Article history: The developed land area of the US increased by 14,2 million hectares between 1982 and 2003. Along Received 26 March 2008 with a projected US population increase to more than 360 million individuals by 2030 is an expected Received in revised form 15 July 2008 continuation of expanding rural land development. Related to population growth, rural land develop- Accepted 25 September 2008 ment and the associated loss of rural open space are expected to have a number of social, economic, Available online 14 November 2008 and ecological implications. Togain greater insight into land development patterns, we used USCensus Bureau and National Resources Inventory data to quantify per-housing-unit rates of land development Keywords: during recent decades and to model future land development to 2030 for states and regions in the Land use Urbanization US.Based on these data, 0.50 ha of additional land were developed for each additional housing unit Modeling in the US.The numbers of hectares of newly developed land per additional housing unit were great- Projections est in the South Central and Great Plains regions and least in the Pacific Coast and Rocky Mountain Resources Planning Act regions of the country. Combining population projections and trends in people per housing unit with development indices, we projected that developed area in the US will increase by 22 million hectares between 2003 and 2030, with the greatest absolute increases projected to occur in the Southeast and South Central regions of the US.We used sensitivity analysis to examine the impacts of changes in popu- lation migration patterns and per housing unit development patterns on increases in projected developed area. Published by Elsevier B.V. 1. Introduction Sturtevant and Cleland, 2007), can have deleterious effects on water quality (Atasoy et al., 2006; Pijanowski et al., 2002b; Tang et al., The area of developed land in the United States (US) increased 2005), and possibly can reduce the propensity for forest manage- by more than 48% between 1982 and 2003 (USDA NRCS, 2007), ment and timber harvest (Kline et al., 2004; Munn et al., 2002; Wear During this time, approximately 680,000 ha of rural land were con- et al., 1999). verted to developed uses annually (USDA NRCS, 2007). Net loss Land-use conversion is expected to continue in the coming and fragmentation of rural lands have many potential implica- decades, In the contiguous US,Stein et al. (2005) projected that res- tions for the goods, services, and functions of natural resources idential development will affect an additional 18 million hectares provided by such landscapes (e.g., Alberti et al., 2003; Arnold and of currently rural private land within forested watersheds by 2030. Gibbons, 1996; Collins et al., 2000), the species that use them and More specifically, Alig and Plantinga (2004) projected that 10 mil- their habitat (e.g., Faeth et al., 2005; McKinney, 2002; Riley et al., lion hectares of forested land will be converted to developed uses 2003), and the ability of invasive species to establish themselves between 1997 and 2030. Nowak and Walton (2005) projected that (e.g., Holway, 2005; Lambropoulos et al., 1999; Yates et al., 2004), 5.3%(118,000 km-) of non-urban forest land will be "subsumed" by Past research also has suggested that the addition of homes and urban growth by 2050, with the greatest impacts in the Southern other structures into rural landscapes can increase the probabil- and Eastern forests. ity of wildland fire and complicate fire management efforts (e.g. Researchers have employed a number of metrics and proj ection Berry and Hesseln, 2004; Cardille et al., 2001; Gebert et al., 2007; models for defining and quantifying the development of rural lands, based on available datasets, The USDA Natural Resources Conser- vation Service (USDA NRCS) National Resources Inventory (NRI) (USDA NRCS,2000, 2007) data have been used as a basis for numer- * Corresponding author. Tel.: +1 541 750 7422; fax: +1 541 750 7329. E-mail addresses:[email protected] (E.M.White), [email protected] ous land-cover and land-use change studies (e.g., Alig et al., 2003, (A.T. Morzillo), [email protected] (R. J. Alig). 2004; Lubowski, 2002; Polyakov and Zhang, 2008). An alternative 0169-2046/$ - see front matter. Published by Elsevier B.V. doi: 10.1016/j.landurbplan.2008.09.004 38 E.M. White et al. / Landscape and Urban Planning 89 (2009) 37-48 source of data, the US Department of Commerce (USDC) Census broad economic conditions, local factors influencing land use (e.g., Bureau data for urban area that is defined based on population and planning, zoning, land prices), and demands associated with the proximity to established urban areas, are updated every 10 years expanding residential area (DR). Land development required for and were used by Nowak and Walton (2005) to project urban area industrial sector expansion (D1) is influenced by broad economic expansion to 2050. Hobbs and Stoops (2002) used the USDC Cen- conditions and local factors influencing land use, as well as the sus Bureau definition of urban area and urban growth over time as availability of production inputs (e.g., labor, capital, intermedi- an indicator ofland development. Likewise, other researchers have ate products). Finally, the expansion of transportation networks adapted USDA Forest Service Forest Inventory and Analysis (FIA) (DT) is influenced by growth in the residential (DR), commer- data and definition of urban lands to identify the conversion offor- cial/institutional (Dc). and industry sectors (DI) (e.g., Baldwin et est land to urban uses (e.g., Harper et al., in press; Kline and Alig, al.,2007). 1999; Thompson and Thompson, 2002). Others have quantified The additional developed land area required at T1 (Dt1-t0) is losses of agricultural (e.g., Nelson, 1999) and forest land (e.g., Alig provided by both land already in some form of developed or urban and Plantinga, 2004). However, quantifying the overall loss of agri- use (Dexisting) (through in-fill development and renovation of exist- cultural or forest land also requires identifying rural lands that will ing developed infrastructure, as in urban renewal, for example) convert to other rural land uses (e.g., agriculture land replanted to and rural land that is converted to developed uses (Rconverted) forest and vice versa), as well as rural land converted to developed (Fig. 1): uses. In addition, several spatially explicit land-cover and land- use change models have been constructed using remotely sensed land-cover data and often involve cellular automata, autonomous agents, and/or neural networks (e.g., Clarke and Gaydos, 1998; Evans and Kelley, 2004; Pijanowski et al., 2002a; Zhen et al., The amount of rural land converted to developed uses 2005). However, such spatially explicit models typically are applied (Rconverted) is a reflection of the rents of those rural lands in rural to regional- or local-level analyses rather than national-level production (e.g., the rents from agriculture or forestry production) analyses. relative to the rents for developed uses and local factors influencing Land-use studies also have included demographic characteris- land-use conversion (e.g., zoning and planning and physiographic tics as a component of a land development metric. For example, conditions) (Alig et al., 2003, 2004; Lubowski, 2002; Parks and Kline (2000) quantified land developed as the number of addi- Murray, 1994). Although rural land historically has converted back tional acres of developed land per additional resident for US states and forth between rural uses (e.g., agriculture to forest and vice as an indicator of the relative efficiency of developed land use. versa), conversion to developed uses is generally a permanent alter- More broadly, Liu et al. (2003) focused on per capita resource con- ation of rural lands. sumption at the household level and how an overall decrease in In this study, residential development (DR) served both as a household size has contributed to a net increase in resource con- driver of land development by itself and as an input to the develop- sumption for a given population size. ment of related commercial/institutional (Dc) and transportation In this study, we adopted a metric that quantifies land devel- expansion (DT )(Fig. 1). Industrial development (Dl) could be related opment on a per-housing-unit basis. Our study is an improvement positively to residential development to the extent that proximate upon past research such that our metric is not complicated by popu- residential areas provide labor inputs or consume the outputs from lation decreases (as found in Kline, 2000) and extends the approach industry (e.g., energy production). Alternately, residential develop- adopted in Liu et al. (2003) to include housing units that are not cur- ment may be related inversely to industrial expansion because of rently primary residences (e.g., second homes, newly built vacant incompatibility of residential areas and industrial production (e.g., houses). To better elucidate the patterns of land development, we noise pollution). Considering the postulated important role of resi- quantified the rate of development at the housing unit level over dential expansion in driving land development and readily available a 15-year period from 1982 to 1997.