The scope of this study was confined to tree species that are native to or common in the United States. We define a tree as any woody plant that attains a diameter at breast height of 6 or more inches. We obtained an initial list of tree species from USDA Forest Service forest inventory data, but added additional species listed elsewhere (Little 1979; Gleason and Cronquist 1991; Hitchcock and Cronquist 1976; Radford et al. 1983; Morin 1993; Bailey 1951; Silba 1986). For purposes of tabulation, all species were listed in taxonomic order (phylum, class, order, and so on). The phylogenetic classification follows that of Cronquist (1988). Latin binomial names of gymnosperms were used as they appear in Morin (1993). Nomenclature of all other tree species were extracted from published lists of various regional floras (Gleason and Cronquist 1991; Hitchcock and Cronquist 1976; Radford et al. 1983).
The results of four field studies of host suitability were included in the summary (Campbell and Sloan 1977; Gansner and Herrick 1985; Lechowicz and Jobin 1983; Maufette et al. 1983). While there are other publications that report defoliation levels on different tree species, only these studies examined a broad range of species, though they differed in the way that host preference was measured and interpreted. Campbell and Sloan (1977) and Gansner and Herrick (1985) reported suitability in terms of observed defoliation levels. By contrast, Lechowicz and Jobin (1983) and Maufette et al. (1983) reported preference in terms of an "electivity index" which was based on larval numbers found on a given species relative to the species' dominance in the stand. We categorized the results of these studies into one of three relative ranks (Table 1) using Montgomery's (1991) concepts of "susceptible," "resistant," and "immune" as guidelines.
The results of seven laboratory feeding trials were included in the summary (Edwards et al. 1981; Forbush and Fernald 1896; Miller and Hanson 1989a, b; Montgomery 1994; Mosher 1915; Nielsen 1992). Again, there are other publications that report gypsy moth performance on different tree species but they lack a sufficiently broad range of tree species to be included here. As described by Montgomery (1991), these studies differ in procedures used to assess foliage suitability, but it is possible to categorize the results into relative ranks (Table 1). For each of the Miller and Hanson (1981a, b) trials, we extracted two groups of data: one summarized information on larval survival; the other summarized pupal weights. The 1-3 ranking was applied to each of these. Two classifications were derived from Mosher (1915). One classification gives the actual ranks (1-4) that Mosher reported in a summary table. The other classification was derived by incorporating our interpretation of Mosher's narrative description of his feeding trial results into Montgomery's 1-3 ranks. The feeding trials of Edwards et al. (1981) and Forbush and Fernald (1896) are different from the other five laboratory feeding trials in that these authors noted only whether or not larvae fed on the shoot when confined with it for several days. These studies identify which species are immune to attack but do not distinguish between resistant and susceptible species. Since they resemble binomial ranks, they were scored as (+) or (-) to indicate plants fed on or not fed on, respectively.
We also included the results of one review article (Montgomery 1991) because it is the only study that reinterpreted previous feeding trials after adjusting for variations among trials. Montgomery accomplished this adjustment by expressing foliage suitability as the ratio of pupal weight of larvae reared on a host species divided by the pupal weight of larvae reared on white oak, (Quercus alba) in the same study. Thus, the continuous relative index produced was then divided into three ranks (Table 1).
We combined all of the results of all 15 studies into a single rank statistic for each species. Again, this rank follows the same categories described by Montgomery (1991). In several cases there was disagreement as to the results of different studies. We usually gave the field studies greater importance than laboratory studies in forming the summary rank. Generally, the results of Mosher (1915), Edwards et al. (1981), and Forbush and Fernald (1896) were given little significance since their methods were problematic (Montgomery 1991). No data were available for several species. In these cases we used taxonomic affinities to guide the formation of a "guess." Results obtained from congeneric species were used to extrapolate ranks to these species. For species lacking congeneric species that were used in published studies, we used information from similar species in the same family.