Note that the data published in the 2002 State of the Nation’s Ecosystems Report as well as the 2003 and 2005 Web-Only Updates have been superseded by the 2008 Report and thus should be used with caution. For the most recent data, purchase the 2008 Report from Island Press.

The Indicator

This indicator indirectly measures the fragmentation of farmland by developed or built-up areas. Cropland interspersed with residential subdivisions raises entirely different policy and farmland management implications than cropland interspersed, for example, with patches of “natural” land cover (forest, grasslands/shrublands, or wetlands). Thus, this indicator considers fragmentation to occur when croplands and natural lands are interspersed with development.

This indicator is an index of spatial fragmentation calculated from digital land cover maps classified from remote-sensing data. Land cover data from, for example, the National Land Cover Dataset (NLCD; see technical note for the national extent indicator) will be used. This dataset has classified grid cells, or “pixels,” which represent areas measuring approximately 100 feet (30 meters) across. This index is computed by analyzing classified land cover “layers” within a raster-based geographic information system (GIS).

The fragmentation index is calculated for each pixel in the farmland landscape (i.e., either cropland or nearby “natural” lands). This value is based on the characteristics of the surrounding pixel “neighborhood.” Such neighborhoods are often created as 3 by 3 or 5 by 5 pixels arrangements. The value for the center pixel is based on the character of the surrounding 8 pixels (in a 3 x 3 square) or 24 pixels (in a 5 x 5 square).

Although the fragmentation index will be calculated for individual pixels, index values for pixels will be aggregated at the scale of one-kilometer squares. Rather than directly reporting index values (i.e., 0 to 1), three fragmentation classes will be reported based on a statistical analysis of these aggregated index values. Each one-kilometer square block will be classified as having a high, medium, or low level of fragmentation. The percentage of surface area in each fragmentation class will be reported by region.

A sensitivity analysis should be performed so that the overall results are not an artifact of the neighborhood size (e.g., such as the 9-pixel arrangement discussed above). In addition, by enlarging the size of the pixel neighborhood (such as to 5 x 5 pixel units), the method will be more sensitive to non-adjacent development.

The index will depend not only on the amount of development interspersed within the farmland landscape, but also on how this development is distributed spatially in the landscape. Thus, development could cover, for example, 20% of two farmland landscapes, but these two landscapes would have very different index values. Clustered rural residential development (e.g., conservation subdivisions), surrounded by cropland and natural areas, would result in relatively high fragmentation index values for those developed portions of the farmland landscape. More scattered, lower density rural residential development (e.g., large estates) would result in somewhat lower fragmentation index values for those developed portions of the farmland landscape. Yet if the total gross residential densities (e.g., total number of dwelling units) were equal in both development scenarios, the proportion of the farmland landscape with an elevated fragmentation index would be much greater in the scattered, low-density development scenario.

The index is sensitive to low-density development if this development can be detected using satellite data (i.e., the development must “fill” a major portion of the pixel used in order to be classified “developed” in the land cover dataset). It should be noted that this is a subject that has garnered considerable attention in the research community. An example of an alternative approach is the one promoted by the U.S. Department of Agriculture (USDA) Economic Research Service (ERS). Its approach identifies farmland that is influenced by nearby development using property values—based on the assumption that farmland priced beyond its agricultural value must be experiencing development pressure. See Development at the Urban Fringe and Beyond: Impacts on Agriculture and Rural Land, AER- 803; http://www.ers.usda.gov/publications/aer803/. Another approach would use data from the Natural Resources Inventory (NRI; USDA Natural Resources Conservation Service). Specifically, the “segments per unit” metric might be used for the appropriate land cover category and reported on a regional basis.

The Data Gap

Several indices have been developed to quantify various aspects of pattern at the patch, class, and landscape scales. Data appropriate for calculating this indicator are available from the NLCD, which was used to define the “farmland landscape” for this report. Calculating this index requires digital data and specialized software designed to analyze landscape spatial patterns. The most commonly used software for analyzing landscape spatial patterns (Fragstats) is not capable of processing the very large file sizes that would be required to calculate this index for the entire nation. It may be possible to address this analysis using a statistical sampling technique, analytical approaches relying on GIS software, or other analytical approaches; however, the details of this were not resolved in time for production of this report.