Sampling Designs for Environmental Monitoring - Introduction to Ecological Sampling - page 157

FIGURE 10.4
Example of a cluster sample of size 4. Each primary unit was defined to be a 3 × 3 block of 9
secondary units. A total of 9 × 4 = 36 sites were chosen.
10.4.2 Monitoring Designs
Although there are many spatial designs generally applicable to research
studies, and even more methods for improving or optimizing the spatial
design for a single parameter (Müller, 2007), there are relatively few general-
purpose spatial designs that perform well in monitoring studies. The primary
characteristic of the few designs that perform well in monitoring studies is
that they ensure broad spatial coverage of sites (Kenkel et al . , 1989; Schreuder
et al . , 1993; Nicholls, 1989; Munholland and Borkowski, 1996; Stevens and
Olsen, 2004; Robertson et al . , 2013). Ensuring broad spatial coverage works
well in monitoring studies because spatial variation is often one of two large
sources of variation, the other large source being temporal variation.
Of the designs that ensure good spatial coverage, three have the highest
likelihood of satisfying the objectives of a monitoring study. Other designs
can work, but they are usually more difficult to apply in a way that satisfies
all objectives. The three sample designs generally suitable for monitoring
studies are systematic, GRS, and BAS. A fourth design, generalized random
tesselation stratified (GRTS) (Stevens and Olsen, 2004), has been implemented
in many environmental studies, but the next generation of BAS is easier to
implement than the GRTS design and has better spatial coverage properties
(Robertson et al . , 2013).