Developing an Empirical Rangeland Cover Legend for the Intermountain West
The classification process involves arranging individuals into groups (or classes) according to their attributes. Ideally, the classification process yields a set of groups where the individuals within the groups are more similar to each other than to the individuals in other groups. Without the structure imposed by classification systems, inherently complicated disciplines like landscape ecology and biodiversity conservation would be intractable. The cognitive models that we use to perceive and comprehend the natural world are themselves a form of classification. The formal schemes that we normally associate with the term vegetation classification system are simply an elaboration of those cognitive models. In vegetation classification, we typically use measures of plant species composition, physiognomy, or overall structure to arrange individual samples or plots into groups (normal analysis¸ Kent and Coker 1992, McCune and Grace 2002).
In terms of conservation, a well-designed vegetation classification system can provide a framework for defining resources and assessing their status and trends. Carefully generalized vegetation classification systems can also provide an avenue for information transfer across multiple jurisdictions.
Vegetation classification systems can also yield academic benefits. The assumptions required by the classification process often suggest fruitful avenues for future investigation. Likewise, the empirical aspects of vegetation classification system development and testing can lead to unexpected discoveries about community structure and ecosystem function.
Our objective is to use the field data collected for the Southwest Regional GAP Analysis Program (SWReGAP) and Landfire projects to develop empirical (derived from data) vegetation classification systems for rangelands in the Great Basin and Colorado Plateau. We anticipate that this effort will lead to better maps, provide insight into how to improve existing classification systems, and perhaps even change the way that we look at rangelands in the arid West.
The following sections describe key concepts that must be addressed during the development of any vegetation classification system. We then describe and compare the existing rangeland classification schemes that provide the context for our own work, and later demonstrate how we implemented several of these systems. Subsequent sections describe the opportunities and limitations associated with the SWReGAP and Landfire datasets and the elaborate screening process needed to prepare these data for statistical analysis. Finally, we describe our model-based approach to multivariate analysis, the diagnostic techniques that we are using to identify groups, the tools that we expect to use complete our classification, and topics for future work.
USU RS/GIS Laboratory | Open Monday through Friday, 8am to 5pm | Tel: (435)797-2385 | Contact information