Olsen, Anthony R., and H. T. Schreuder. 1997. Perspectives on large-scale natural resource surveys when cause-effect is a potential issue. Environmental and Ecological Statistics 4:167-180.
Our objective is to present a perspective on large-scale natural resource monitoring when cause-effect is a potential issue. We believe that the approach of designing a survey to meet traditional commodity production and resource state descriptive objectives is too restrictive and unnecessarily limits the ability to investigate cause-effect issues. We only consider terrestrial natural resources, focusing on forests and rangeland.
A large institutionalized program is required to establish cause-effect relationships when monitoring terrestrial resources. This is justified based on the growing concerns about our natural resources. A long-term vision of a desirable future terrestrial monitoring system, realizing that it is not clear yet what key variables should be measured, will increase the chances that decisions on current designs will ultimately lead to better systems in the future.
We propose a pronounced shift in the design applied to forest and range, specifically the National Resources Inventory (NRI), the Forest Inventory and Analysis (FIA), and the Forest Health Monitoring (FHM) programs. The designs must not only address simple status and trends estimation, but also give emphasis to identifying interesting changes occurring in the sampled populations thus facilitating identification and establishment of possible cause-effect relationships. We propose an integrated design consisting of a large-scale, long-term ongoing survey as the core design accompanied by supplemental experimental design studies or analytic survey. Continuous inventory involving annual measurement of a subset of the sample from selected populations should be implemented inventorying a population every five years (as with NHI) or every ten years (as with FIA) is insufficient. FHM, FIA, and NRI should collect a subset of variables in common. Complementarity of data collected would make it more likely to identify promising cause-effect relationships for a wider range of resource variables. At this stage we recommend focusing on the mortality of trees, shrubs, forbs, and grasses as the key indicator of forest and range health. Mortality is objectively measurable and can often be detected by remote sensing. When possible, follow-up observational studies to document cause-effect relationships should be limited to public lands because of concern of infringing on the personal rights of landowners. This may not be possible if unrepresentative populations result because of this. If studies are designed properly, we could achieve our objectives yet tie such studies to current natural resource inventory systems.