Using a cellular automaton model to forecast the effects of urban growth on habitat pattern in southern California

Land use change is one of the most important anthropogenic factors affecting terrestrial ecosystems, causing habitat loss, fragmentation, and interactions with other components of global change, such as biological invasions of non-native species. In southern California, population growth and economic expansion are the primary drivers of land use change, and the population is expected to double in 40 years. Although directly adjacent to the region's largest metropolitan area, the Santa Monica Mountains National Recreation Area (SMMNRA) remains mostly undeveloped, with 50% of the area protected as parkland. In this study, a cellular automaton (CA) model was calibrated using historical growth patterns in the region, and used to forecast three scenarios of urban growth in the SMMNRA from 2000 to 2050, with development prohibited on slopes greater than 25%, 30%, and 60% slope. Habitat pattern and extent under these scenarios was assessed using several landscape metrics, then compared to results from a GIS overlay model developed for the same region. The CA model predicted urbanization to increase from 11% of the landscape in 2000 to 26%, 35%, and 47% in 2050, respectively, for the three slope scenarios. In 2000, the majority of vegetation constituted one large, interconnected patch. With development prohibited beyond 25% and 30% slope, this patch will become, by 2050, increasingly perforated, but should stay relatively intact. However, if growth is permitted up to 60% slope, the patch breaks apart, resulting in a shift in spatial pattern dynamics on the landscape (as reflected by other landscape metrics). General growth patterns predicted by the GIS overlay model resembled those generated by the CA, but the CA model produced more patches and edge in the landscape. Because it is temporally explicit, the CA model was able to capture non-linear, emergent behavior and a phase transition in the type of growth occurring in the landscape that was not apparent in the GIS overlay predictions.