plants under a physically defined set of management practices (Soil Survey Staff, 1951). Soil productivity, then, includes two aspects: the inherent productivity of soil and its response to managed inputs. Crop yield has been considered the best indicator of soil productivity as it integrates the inherent and managed components of soil productivity. However, using crop yields as a measure of soil productivity is difficult since yield data are limited, are spatially and temporally variable, and depend on the management and level of technology used (Pierce, 1991). Well-documented yield records are scarce and usually available only from farmers' long-term records, crop yield surveys, and plot experiments. Farmers' yield records are field-based, which actually represent averages of a number of soil map units and combined fields with varying management histories. As fields continue to increase in size, field yields increasingly average across management units, making identification of specific soil yield potentials more difficult. Even when available, a well-documented yield history will not necessarily be indicative of future effects of soil degradation processes. Therefore, alternatives to evaluate land and soil productivity have been intensively pursued.
Land evaluation systems and soil productivity ratings have been developed in lieu of crop yields as a measure of potential productivity (Huddleston, 1984; Olson, 1974; Riquier, 1974; Stewart, 1968; Wagenet et al., 1991). Nix (1968) recognized three approaches. The most common is the analogue or transfer-by-analogy approach and is based on land and soil classifications. A second approach, the site-factor approach; seeks to relate key parameters to biological productivity within a given environment where yield is described by a multiple regression equation. A third approach, called the system-analysis-and-simulation approach, is concerned with resolution of a complex system into simple component processes that are synthesized into a mathematical model of the whole system. The simulation of crop yields utilizing soil databases is gaining in use in land evaluation (Wagenet et al., 1991) and productivity assessment (Williams and Renard, 1985). Each of these approaches has been used not only to assess soil productivity but also the effects of soil degradation, most commonly by accelerated soil erosion, on soil productivity (see Pierce, 1991, for a review). These evaluation techniques will become increasingly important in the design and evaluation of sustainable management systems, as discussed later.
Regardless of the approach, however, land evaluation procedures or indices relate to the potential of land (soil) to produce food and fibre, and hence must correlate to crop yield. This must take into consideration, however, that weather and inputs are major contributors to crop yield, in addition to soil. Farmers consider the most productive and, therefore, valuable soils to be those that produce the consistently highest yields. While important in the equation, crop yields alone are no longer sufficient as a measure of productivity since the costs of production (economic, social, and environmental) increasingly alter the value of production.