Introduction

Landscapes are large by conventional definitions (Forman and Godron, 1981, 1986; Urban et al., 1987; Turner, 1989) and data at that scale are dearly bought. Yet with the advent of ecosystem management (Christensen et al., 1996) – which implies a larger scale of reference than prior approaches to resource management – researchers and managers are increasingly faced with pursuing sampling and monitoring programs at these larger scales. A significant component of such programs should be the establishment of long-term monitoring systems designed to detect trends in resources, prioritize management needs, and gauge the success of management activities. This goal can be especially daunting in cases where the study area is especially large, where the signal to be detected is uncertain (e.g., potential responses to climatic change), or where the objects of concern are simply difficult to locate (e.g., rare species).

Here I consider some approaches that may prove useful in designing sampling and monitoring programs for landscape management. In contrast with large-scale efforts that are coarse-grained and intended as “first approximations” (Hunsaker et al., 1990), or more location- or taxon-specific methods (e.g., examples in Goldsmith, 1991), my concern here is with problems that are simultaneously fine-grained and of large extent. This is essentially a sampling problem at first, with the goal of capturing fine-grained pattern in an efficient manner. In many cases, however, even an efficient blanketing of the study area is logistically infeasible and so a second concern will be to focus sampling as powerfully as possible on a specific application or hypothesis.