|
Article Excerpt
Introduction
Environmental geography has historically neglected the study of cities in favour of locales where the human-environmental interface is most exposed, in particular, spaces where resources are extracted or harvested from the earth (Slocombe 1993; Robbins et al. 2001). This anti-urban bias historically influenced the field of ecology as well, albeit for slightly different reasons; ecologists have been sceptical about the usefulness of studying ecological systems that are so polluted and altered by human intervention (Luck and Wu 2002). Cities have thus been framed as supremely un-natural (Pickett et al. 1997), and remained the purview of 'purely' social scientists, notably urban geographers, planners and sociologists.
In recent years, the urban-nature dualism has begun to break down as both environmental geographers and ecologists become interested in the unique contexts that cities pose for environmental research. Part of this shift is epistemological, influenced by years of geographers challenging the notion that nature is everything non-human (Smith and O'Keefe 1980). Part of the shift is related to policy concerns as the world's population continues to urbanize at rapid rates. Regardless of the specific reasons, it is clear that environmental geographers and ecologists are concerning themselves with cities as never before (Botkin and Beveridge 1997; McIntyre et al. 2000; Zipperer et al. 2000; Robbins et al. 2001).
While openness to thinking about urban ecosystems has increased in recent years, there are still fundamental gaps in our knowledge about human-environmental interactions in urban settings. Those studies that have occurred tend to examine one ecological component (i.e., soil, vegetation, air quality) in a city or conduct studies analyzing biophysical flows across a city, while paying relatively little attention to human components (McIntyre et al. 2000; Pickett et al. 2001; Alberti 2005). The relationship between urban form and land cover, e.g., contains major voids, particularly for new forms of urbanization (Alberti 2005; Miller and Small 2003). These gaps are unfortunate not only because of high rates of global urbanization, but also because the expansion of urban land uses has tremendous localized impacts on the populations of those cities. A better understanding of the interface between urban pattern and land cover could specifically assist the development of policy that ameliorates attendant problems, and generally assist the goal of building an 'ecological city' (White 2002).
This article attempts to address some of these gaps by examining the relationship between land cover and urban pattern in the greater Toronto area (GTA). In particular, measures of development pattern that reflect specific aspects of density, grain and function, as well as socioeconomic characteristics are analyzed to systematically explore the effects of urban pattern on land cover heterogeneity. A simple vegetation index, namely the Normalized Difference Vegetation Index (NVDI), is extracted from a 1999 satellite image and used to represent land cover because the focus of the article is on disentangling the influential components of urban pattern. (1) Three questions are addressed through this analysis: Can variations in NDVI be explained by distance to city centre alone? What is the relationship between NDVI and different urban land uses? What other aspects of urban pattern are correlated with NDVI? The article concludes with a discussion of future research pathways and the environmental planning implications of the analysis.
Urban Pattern and Ecological Conditions
In response to a growing interest in the ecology of urban areas, McDonnell and Pickett (1990) suggest an urban-rural gradient approach to studying urban ecological conditions. Built factors are thought to be equal to or supersede biophysical factors governing ecological conditions in urban landscapes, so a gradient of urban intensity provides a way to study the dominant influence on urban ecological conditions. In particular, urban-rural gradients represent a unique situation where 'experimental' conditions with varying levels of urbanization exist that researchers are otherwise unable to create (McDonnell and Pickett 1990). Most urban-rural gradient studies delineate one or more linear transects across a metropolitan area, using a moving window approach when relying on remotely sensed data (Luck and Wu 2002) or discrete plots along the gradient for field data collection (McDonnell et al. 1997). Such an approach has now been used to examine land use (Luck and Wu 2002; Zhang et al. 2004), vegetation and other land covers (Medley et al. 1995), species diversity (Blair 1996; Sukopp 1998), microclimate variation (Miller and Small 2003), soil processes and pollution (Pouyant and McDonnell 1991; Pouyant et al. 1995) and water quality (Wear et al. 1998).
While gradient studies have provided a foundation for understanding urban ecological conditions, the approach implicitly assumes that cities are monocentric, with outward gradients of decreasing density. However, many metropolitan areas are polycentric entities sprawling in fractal or spider-like configurations (Batty and Xie 1996; De Keersmaecker et al. 2003; Longley and Mesev 2000; White and Engelen 1993), indicating that the gradient approach many not capture all of the relevant built dimensions. Alternatively, Alberti (1999) identifies a total of four urban development pattern components that likely interact with ecological factors at the metropolitan scale: density, grain, connectivity and form. Density refers to the population, employment or building density variation that is often assumed to occur along urban-rural gradients. Grain represents the diversity of functional uses, while connectivity refers to the circulation of people and goods across a landscape. Alberti defines form as the level of centralized or decentralized development in a metropolitan area. (2)
To date, only a few studies have evaluated the influence of urban pattern on ecological conditions beyond simple density or urban-rural gradient measures. Stone (2004) looked at the relationship between impervious land cover and street pattern in addition to lot density and several site level variables. Wilson et al. (2003) found that variations in urban vegetation cover were statistically different between zoning classes in the Indianapolis metropolitan area, but they did not examine the configuration of those classes nor reasons for between class variations. Hope et al. (2003) investigated the impact of distance from city centre on plant diversity, but determined that current and past land use, housing age and household income were better correlated with diversity in their Phoenix metropolitan study area.
There is also growing evidence that socioeconomic factors play a key role in urban ecosystem interactions. In the Chicago area, Iverson and Cook (2000) found household income highly related to land cover composition. A Phoenix-based study focusing on residential neighbourhood parks indicated that even urban parkland's plant diversity is a result of a 'luxury effect' reflecting neighbourhood socioeconomic conditions (Martin et al. 2004). These results lead Kinzig et al. (2005) to suggest that inclusion of socio-economic variables into traditional urban-rural gradient studies can improve explanations of intra-urban biodiversity patterns.
In a major North American metropolitan area such as Toronto, we would expect that a simple distance to city centre measure would not be a good predictor of land cover or broader ecological heterogeneity given the complexity of urban pattern present and previous findings from other regions. Instead, more specific aspects of urban development pattern, coupled with local socioeconomic characteristics, are likely the primary drivers of ecological variation in such settings. The following section describes the methods used...
|
