RELATIONSHIP OF LARVAL FOOD-PLANTS AND VOLTINISM PATTERNS IN TEMPERATE BUTTERFLIES*

An interesting aspect of phenology is the number o broods that a species produces in the growing season. Even among a airly uniform group like the butterflies within a restricted geographical area, voltinism patterns oi: the dierent species may vary considerably. Explanations oi: possible causes o, and ecological implications of these patterns have apparently seldom been attempted. It might be expected that voltinism patterns are often genetically determined and are regulated by some environmental clue (cL V,rig glesworth, I967), but knowing the proximate reason does not reveal the ultimate causes that brought the voltinism patterns under such control. Obviously, the presence o a food source for the larvae (and, although less well studied, for the adults as well) will have a major influence on these patterns. The exclusive use o the vernal herb Dentaria by larvae oi: the West Virginia white, Pieris virginiensis, precludes any more than one brood per season. As an adaptation to i:eeding on this ephemeral food-plant, the larvae upon entering the pupal stage undergo obligate diapause, even though the genetic mechanism for multivoltinism exists in this species (Shapiro, 97I). A more subtle explanation o voltinism patterns may emerge rom examining larval growth rates on different plants (e.g. Dowdeswell & Willcox, 96 Hovanitz & Chang, 962; Sharii & Zarea, 97o). Nutritional (including water) dierences (Soo Hoo & Fraenkel, 966; Waldbauer, 968; Feeny, 97o; J. M. Scriber and P. P. Feeny, ms. in prep.; Slansky, 974) and differences in secondary chemical content (Gupta & Thorsteinson, 96o; Nayar & Thorsteinson, 963 Feeny, I97O) of dierent plant species and o the same plant species at di6erent stages of growth in large part determine the growth rates of larvae feeding on the plants, and thus perhaps for

It might be expected that voltinism patterns are often genetically determined and are regulated by some environmental clue (cL V,rig glesworth, I967), but knowing the proximate reason does not reveal the ultimate causes that brought the voltinism patterns under such control. Obviously, the presence o a food source for the larvae (and, although less well studied, for the adults as well) will have a major influence on these patterns. The exclusive use o the vernal herb Dentaria by larvae oi: the West Virginia white, Pieris virginiensis, precludes any more than one brood per season. As an adaptation to i:eeding on this ephemeral food-plant, the larvae upon entering the pupal stage undergo obligate diapause, even though the genetic mechanism for multivoltinism exists in this species (Shapiro,97I

METHODS
In order to examine the hypothesis the number of broods produced per year by, and the larval food-plants of, eighty-six species of butterflies were tabulated for three transition zone areas in eastern and mid-western United States (Forbes,9o6;Saunders,932;Ebner,x97o). Additional food-plant records were obtained from Ehrlich and Ehrlich 96 and Shapiro (966). The voltinism patterns of selected species of these butterflies were then related to changes in the nutritional suitability of their larval food-plants.

RESULTS AND DISCUSSION
That differences in larval growth rates are reflected in the voltinism patterns of butterflies is seen in the following example. Under similar temperature and humidity conditions the duration of the larval stage of the imported cabbageworm, Pieris r@ae, is about I5 to 2o days on some of its food-plants (Slansky,I974), while that of the Eastern tiger swallowtail, P@ilio glaucus, is about 35 to 45 days on some of its food-plants (J. M. Scriber, unpublished data). As a consequence, P. r@ae generally exhibits one brood more than P. glaucus in the areas where they occur together (Table ).* Further support for the hypothesis is found when the voltinism patterns of selected butterfly species are examined in relation to seasonal changes in the nutritional suitability of larval food-plants. For example, species whose larvae feed on oaks (Quercus), such as the hairstreaks 8atyrium edwardsii, 8. tiarops, and S. calanus, and the skippers Erynnis brizo, E. horatius, and E. juvenalis, usually have *Although P. glaucus is generally considered to have two broods in the areas surveyed in this study, it may be that there is only one 'true' brood. It appears that a high percentage of the pupae from the previous season emerge in the early summer, constituting the 'true' first brood, and that the remaining pupae emerge later in the summer, constituting an 'apparent' second brood (i.e., not the progeny of the first brood of that season) (Scudder, 1889; R. Lederhouse and J. M. Scriber, unpublished data).   (Table ). This may be due to the presence in the oak leaves of tannins that restrict protein utilization and hinder larval development, in many cases limiting consumption of the oliage to spring and early summer when the protein content is highest and the tannin content lowest (Feeny,97o). In addition, the leaves of many species of shrubs and trees contain substantially lower percentages of water than the leaves of many annual and perennial herbs (Way,  lower than their own tissue may be limited by the availability of water so that they exhibit slower growth rates in comparison to larvae feeding on plants that have a percentage of water equal to or greater than that of the larvae (Southwood,972; J. M. Scriber & P. P. Feeny, ms. in prep.). The fact that a number of tree and shrub eeders exhibit but one or two broods (Table ) supports this contention. Along these lines Feeny (974) has suggested that plant species that are abundant and/or persistent have apparently evolved "quantitative" defenses (low nutrient contents, tough leaves, high contents of unspecific chemicals like tannins) that act as significant ecological barriers to phytophagous insects, in contrast to plant species that are rare and/or ephemeral that have apparently evolved "qualitative" defenses (specific secondary chemicals) that act as only slight ecological barriers to adapted insects although having considerable impact as evolutionary barriers to non-adapted insects.
The majority o.f grass-and sedge-eeders (all Satyridae and several Hesperiidae), with characteristically sluggish larvae (Scudder, 889), exhibit but one brood (Table ). In view o the low moisture and high /]ber content of many grasses (Way,853;Watson,95), it is not surprising that larval development might be slow on such plants. The hibernation of many o. these satyrids and hesperiids as early instar larvae (Table I) may be an adaptation for the avoidance of nutritionally poor, mature grass plants and or the maximum utilization o the spring flush of succulent growth when moisture and nitrogen levels are high and fiber content low (Watson,95I).
Thus, the voltinism patterns ot several butterfly species are given an ecological meaning, and the hibernation o many butterfly species as eggs, larvae, and perhaps fertile adults (Scudder, I889), when 1974] 8lansky Temperate Butterflies 249 viewed as a means of utilizing tender, nutritious plant tissue in the spring, is seen as an adaptive strategy (c.f. Morse, 1971;Schoener, 1971). Of course, it is not suggested that the nutritional quality of the larval 'food-plant is of sole importance in determining voltinism patterns. For example, larvae of both the Tawny crescent, Phyciodes batesii, and the Pearl crescent, P. tharos, feed on Aster; the former exhibits one brood and the latter two or three (Table ). Larvae of two Bolaria species and of four 81eyeria species all feed on Viola; the former exhibit two to three broods, the latter one brood (  (Table I).
Obviously, the interaction of several other factors (e.g. larval size, larval mortality rates, and the presence of adult nectar sources) with the nutritive and secondary chemical contents of the larval ood-plants influences the voltinism patterns. For example, the slow growth rates of larvae of P. glaucus may only be 'permitted' because the combination of the patchy distribution and warning coloration of the larvae may reduce mortality losses from predators and parasitoids (J. M. Scriber, personal comm.). On the other hand, the rapid growth rates of larvae of P. raae may be a 'necessity' because of the nature o.f its food-plants (i.e., mostly early successional species) and because of the high mortality losses of the larvae to predators, parasitoids, and disease (Richards,94o Pimentel,96 Dempster,969;Parker,97o). The more rapid growth rates of lepidopterous larvae from 'geographical races' occurring in regions with short growing seasons in comparison to those of larvae from regions with longer growing seasons (C-oldschmidt, 94o) provides another example of the ecological importance of differences in larval growth rates. 3) What selective orces cause the complex voltinism pattern exhibited by a number of butterfly species in which part of a brood becomes dormant while the remainder continues normal development (Scudder,889 Oliver,972) ? Perhaps this may allow these species to exploit marginally favorable periods while maintaining a reserve population or the usually favorable season and/or may aid in reducing losses to temporally restricted predators and parasitoids (c.L Baltensweiler, 968 Waloff,968). 4) Why do some butterflies exhibit multivoltinism ? Perhaps this is a means of building up large populations to withstand high mortality losses in the summer, especially because o biological causes, and in the winter, especially because of physical causes, such as is seen in several pierid butterflies that start out with low population levels in the spring and become more abundant as the summer progresses (Scudder,889).
Detailed ecological studies of a taxonomically and geographically well-known group like the butterflies will help to answer such questions and will aid in making meaningful predictions about such important phenomena as the population buildup of a potential or actual pest species.

2KCKNOWLEDGEMENTS
The author wishes to thank J. Mark Scriber and William Blau for reading the manuscript and or their helpful suggestions.