Discussion

Photointerpretation of the satellite imagery for the 2001 survey proved to be fairly accurate in identifying oak wilt. Brush piles resulting from roguing Ashe juniper were the features most often confused with oak wilt. Improved training would probably reduce many of these errors, but may not eliminate them altogether. Juniper clearing is a common practice in the Hill Country, and, in many cases, piles were visually indistinguishable from the crowns of dead live oaks.

Oak wilt was found to be a prominent feature and a major cause of oak mortality throughout Fort Hood. No other cause of mortality came close to the level found for oak wilt. However, the survey was not designed to determine the volume or extent of the other major cause of tree mortality—fire. It was noted that in certain locations, fire was a dominant cause of mortality and probably far exceeded the extent of oak wilt as a disturbance. A detailed analysis of the comparative effects of fire and oak wilt on habitat is warranted and will be considered in future studies.

Habitat requirements and tree species composition associated with GCW populations have been addressed in previous studies (Kroll 1980). The dependency of GCW on Ashe juniper bark as a source of nesting materials is a well-described phenomenon (Kroll 1980, Pulich 1976). There appear to be preferences for the sizes and densities of junipers. Kroll (1980) characterized good habitats as those with juniper-oak ratios of 1.35 to 1, and poor habitats with ratios of 2.27 to 1. We also found trends for the selection of nesting sites at Ft. Hood. The juniper-oak ratios ranged from a high of 6.57:1 to a low of 1.66:1 in designated GCW habitat (Table: Juniper to oak ratios for each of the four sampling categories at Fort Hood.). But, our results indicate that preferred nesting sites were in areas with a juniper-oak ratio of 3.16:1. Other variables determined to be characteristic of good habitats by Kroll (1980) were older Ashe junipers at wider spacing and relatively lower densities than those of poor habitats. In a previous study conducted at Fort Hood, Dearborn and Sanchez (2001) made pairwise comparisons of 13 vegetation variables between nest locations and nearby nonuse vegetation patches. The only significant variable was canopy closure, which was greater at nesting sites than at the paired nest-free location. A stand density equivalent to 487 stems per ha for junipers and hardwoods combined was found at nesting sites, and junipers dominated hardwoods in all size classes. Nesting sites were characterized as having dense vegetation and nearly complete canopy closure dominated by junipers (Dearborn and Sanchez 2001). The nesting sites in our survey appear to have higher stand densities than those surveyed by Dearborn and Sanchez (2001), but we included smaller diameter stems in the survey protocol. The trends in both surveys are consistent. Of all habitats surveyed, our results confirm that GCWs prefer dense vegetation with high juniper densities.

Live oaks dominated sites where oak wilt occurred outside of GCW habitat. Within GCW habitat, the highest levels of live oaks also occurred in oak wilt centers, with a large decrease of live oak density at nesting sites and randomly selected habitat sites. The average diameters of live oak were larger in oak wilt centers than in healthy plots. A similar trend, although not as pronounced, was observed for the deciduous oaks. Oak wilt appears to be less likely to occur in places where the proportion of oaks is relatively low, such as GCW nesting sites. The incidence of live oak depends, for the most part, on availability of susceptible hosts, availability of inoculum, occurrence of infection courts (fresh wounds), and existence of nitidulid vectors (Appel 2001). Red oak density was fairly consistent among the different plot types, so inoculum availability in the form of fungal mats was potentially the same. The most likely explanation relates to vector behaviors in the live oak–dominated stands, but this suggestion would need to be confirmed with trapping studies. Our results also suggest that the site requirements for oaks may not coincide with sites preferred by GCW. The oak wilt threat to critical habitat may therefore be less than anticipated. A comparison of the GCW classification tree model developed in the present study for nesting sites, with the oak wilt model developed by FHTET, should be useful in determining whether oak wilt is a threat to GCW breeding habits.

Classification tree modeling proved to be a useful technique for establishing the site factors influential in determining the habitat for GCW nesting sites. When the comparison was made between designated habitats classified by the Nature Conservancy and nesting sites, the classification tree revealed that low road density was needed for ideal nesting-site locations. This agrees with other research findings that GCW prefers to have large unfragmented habitat for breeding and territory ranges (Kroll 1980, Ladd 1985, Moses 1996). There are, however, conflicting opinions on the GCW preferences for large blocks of unfragmented habitat or for sites bounded by edges of different vegetative composition (Moses 1996). In one study in Travis County (Texas), the estimated territory required per breeding pair of GCWs was 1.9 to 2.7 ha/pair (Ladd 1985). Kroll (1980) estimated ranges for breeding pairs were 4.49 – 8.48 ha/pair in a Texas state park. These estimates were noted to be larger than those from previous research, which resulted in estimates of territory sizes ranging from 0.81 ha to 2.55 ha per breeding pair (Kroll 1980). One effect of oak wilt is to fragment contiguous tree stands into treeless patches and expanding edges (Appel and others 1989). Dispersal distances for adult males averaged 223m in a study conducted at Fort Hood during 1991–96, whereas juvenile dispersal distances were greater, averaging 4,040 m (Jette and others 1998). These dispersal distances, patch expansion, and the creation of edges by oak wilt requires further research to determine how the disease relates to the GCW beyond the consequences of direct loss of trees

The effects of oak wilt on the landscape go beyond the destruction of trees. Gaps and edges are created, tree composition is changed, and woodland stand structure is altered. All of these effects may influence GCW populations and will require further analysis to confidently decide whether to manage the disease in the vicinity of GCW nesting sites. Oak wilt control activities need not be disruptive because they can be implemented when the birds are migrating. However, because oak wilt appears to fall in areas where oak densities are greater than those found in preferred GCW habitats, the simple loss of trees may not be sufficient justification to undertake expensive and disruptive oak wilt control methods.