Collection and Reliability of Rust Data

Each FIA sample location (plot) consists of a permanent cluster of 10 sample points (delineated using a sampling prism with a basal area factor of 37.5 ft²/acre) to tally trees >5 inches in d.b.h. Trees <5 inches in d.b.h. are tallied on small fixed-radius plots around the prism points. For fusiform rust, any pine >1 inch in d.b.h. is considered infected if it has a rust gall on the main stem or on a living branch within 12 inches of the main stem. These galls are potentially lethal, especially if infection occurred prior to age 5. Field crews are trained in the identification and recording of fusiform rust infection and are provided with manuals illustrating rust infection and recording procedures. Details of the data collection process can be found in "Field Instructions for the Southeast" and "Forest Survey Inventory Work Plan for Alabama" (administrative documents available from USDA Forest Service, Southern Research Station, Asheville, NC, and Starkville, MS).

In order to evaluate the reliability of rust infection data, field checks of a sample of FIA plots were made. Ten plots each of the slash and loblolly forest types were visited, and slash and loblolly trees were examined for rust infection. Results were compared to those of the regular field crews. In addition, 100 slash or loblolly trees in the area surrounding each FIA plot were surveyed for infection to evaluate how well the field plots represented rust infection nearby. To accomplish this, individual trees within the 10 prism points were identified, and any intervening trees were evaluated for rust infection. Next, trees outside the cluster of prism points were selected in an outwardly expanding area from the approximate center of the plot (i.e., the cluster of prism points) and evaluated for rust infection until a 100-tree sample was obtained.

Field checks showed that FIA field crews had 98- and 95-percent accuracy for recognition of fusiform rust on slash and loblolly pines, respectively. On uninfected trees, the accuracy rate of field crews was 100 percent for slash and 98 percent for loblolly. When plot infection was >10 percent, the 100-tree survey of the surrounding area confirmed it 100 percent of the time. Seventy percent of the time (at all infection levels), the estimate of the 100-tree sample was within 14 percent of the plot infection estimate and never varied by more than 20 percent.