Context

All monitoring is done in some context. The approach and topics considered in this chapter do not apply to all situations, so for clarity I define a context for this chapter. Monitoring programs will almost always be interested in many responses. The example presented in this chapter has about 30 responses, a fairly typical number; some of the discussion will touch on the problems arising from various responses having different sorts of variation and design criteria. However, most of my discussion will focus on analyses of one response at a time, symbolized by Y with suitable subscripts to further identify it.

Monitoring, as used here, refers to studies intended to describe the geographic extent of a resource, general response size, and possible trends therein. Such studies are not sufficiently detailed in either time or space to support ecological modeling of underlying dynamic processes. Resources are always limited. When legal or regulatory requirements specify that inferences must apply to large areas, and budgets are limited, ecological modeling is neither feasible nor appropriate. This chapter is directed toward detection of trend in such situations, but this focus is not intended to imply that monitoring is done exclusively for the purpose of trend detection. In my experience, monitoring has always had many objectives. When trend is detected, a logical next step is to seek to characterize the dynamic process which causes it, assuming time and money are available to do that (see also Chapter 3).