Evolution, 44(6), 1990, pp. 1637-1647 PREDICTING HOST RANGE EVOLUTION: COLONIZATION OF CORONILLA VARIA BY COLIAS PHILODICE (LEPIDOPTERA: PIERIDAE) DAVID N. KAROWE' Department ofBiology, The University ofMichigan, Ann Arbor, MI48109 USA Abstract.-Extensive sympatry is currently arising between the common sulfur butterfly, Colias philodice Latreille (Lepidoptera: Pieridae) and a potential leguminous host plant, Coronilla varia (L.). In laboratory trials, larval surviviorship and growth were higher on the primary host, Medicago sativa (L.), than on the nonhost C. varia. However, because females reared from C. varia were on average more fecund than females reared from M. sativa, fitness on C. varia (approximately as survivorship times fecundity) was commensurate with fitness on M. sativa. Thus, it is predicted that selection would favor oviposition on C. varia, if such behavior were to arise. In addition, significant among-family variation exists for several measures of larval performance on both C. varia and M. sativa, indicating that C. philodice can potentially respond to selection for increased performance on each species. Moreover, larval performance was sig­ nificantly positively correlated across these species, suggesting that selection for increased per­ formance on each species will facilitate, not constrain, evolution of increased performance on the other. It is concluded that behavioral rather than physiological barriers currently account for the absence ofC. philodice from C. varia and that, ifsuch barriers are overcome, C. philodice will expand its host plant range to include C. varia. Received January 17, 1989. Accepted December 15, 1989. The evolution of host plant range is of hosts, (2) the degree of within-population both theoretical and economic importance, genetic variation in fitness on potential and since present associations between insect existing hosts, and (3) the degree and sign herbivores and their host plants are largely ofgenetic correlation in fitness between po­ attributable to past host shifts and host tential and existing hosts. range expansions. Although the evolution­ Selection should favor host range expan­ ary lability ofhost plant range among her­ sion ifinitial fitness on the potential host is bivorous insects has long been recognized commensurate with fitness on existing hosts, (Dethier, 1954; Ehrlich and Raven, 1964), particularly if fitness is limited by the rate the study of host range evolution among at which suitable oviposition sites are en­ herbivorous insects remains essentially a countered (Levins and MacArthur, 1969; post hoc science. However, as argued by Jaenike, 1978; Courtney, 1984). If, on the Gould (1978, 1983), this need not be the other hand, fitness differs substantially be­ case. tween potential and existing hosts, then se­ For herbivorous insects with relatively lection shouldgenerally act to decrease pref­ immobile larvae, such as many Lepidop­ erence for, or to increase fitness on, the tera, colonization of a potential host plant poorer host (Futuyma, 1983a; see also Cas­ may be precluded by the failure offemales tillo-Chavez et al., 1988). If preference for to recognize the potential host as a suitable the better host evolves more rapidly than oviposition site. However, if at least some performance on the poorer host, then the females accept the potential host for ovi­ poorer host (whether potential or existing) position, and if such variation is heritable, should be eliminated from the herbivore's then the course oflocal host range evolution host plant range. might reasonably be anticipated from Genetic correlations in performance knowledge of (1) the initial relative fitness across hosts may strongly influence the rate of the herbivore on potential and existing at which performance on any single host evolves (Antonovics, 1976; Lande, 1980; Via and Lande, 1985). In particular, ifper­ 1 Present address: Department of Biology, Virginia Commonwealth University, Richmond, Virginia formance is negatively correlated across 23284. hosts, then selection for increased perfor- 1637 1638 DAVID N. KAROWE mance on the poorer host may be con­ MATERIALS AND METHODS strained by the associated "cost" ofreduced Study Organisms performance on the more suitable host Over approximately the last 200 years, (Rausher, 1983). Such circumstances are likely to inhibit the expansion ofhost plant Colias philodicehas incorporatednumerous introduced legumes into its host plant range range. On the other hand, ifperformance is (Klots, 1951; Tabashnik, 1983; Scott, 1986). positively correlated across diets, then se­ In Michigan, e. philodice feeds primarily lection for increased fitness on either host on the introduced European species M. sa­ will result in increased fitness on both, and tiva L. (alfalfa) and Trifolium pratense L. expansion ofthe herbivore's host plant range (red clover), and uses Melilotus alba L. is more likely. (white sweet clover) as a secondary host Throughout the eastern and midwestern (personal observation). United States, extensive sympatry is cur­ Coronilla varia L. (crown vetch) repre­ rently arising between the common sulfur sents a potential host plant for e. philodice. butterfly, Colias philodice Latreille (Lepi­ At present, e. philodice is not known to doptera: Pieridae) and a recently introduced occur on e. varia, though females do visit potential host plant, Coronilla varia L. (Fa­ this species for nectar (Karowe, 1988). In­ baceae). In southeastern Michigan, e. varia troduced into the eastern United States in often occurs in the same habitat as Medi­ 1890 (Henson, 1963), e. varia is a consid­ cago sativa, the primary host plant of e. erably more recent addition to the North philodice. The absence ofe. philodice from American flora than are M. sativa, T. pra­ e. varia in nature is apparently due to pre­ tense, and M. alba. Due to its attractiveness alighting discrimination by ovipositing fe­ as groundcover and forage (Reynolds et at, males (Karowe, 1988). The aim ofthis study 1969; Bums et at, 1969, 1972), and its abil­ is to determine whether physiological bar­ ity to escape cultivation, e. varia is becom­ riers to colonization also exist, and to use ing increasingly abundant throughout much this information to anticipate the likely ofthe eastern and midwestern United States. course of host range evolution should fe­ By 1974, nearly 30,000 acres of'C varia had males initiate oviposition on e. varia. Ac­ been planted along highway slopes in Penn­ cordingly, I ask: sylvania alone (Wheeler, 1974). Moreover, 1. How does present performance of e. with continued naturalization, sympatry philodice on e. varia compare to that on between e. varia and e. philodice is likely M. sativa? to increase since the typical habitats of e. 2. Can selection potentially act to improve varia, M. sativa, T. pratense, and M. alba performance on M. sativa and/or e. var­ (roadsides, waste places, and other dis­ ia, i.e., does e. philodice exhibit herita­ turbed areas) are very similar (Fernald, ble variation for performance on either 1950). species? 3. Will evolution ofincreased performance Parental Performance on one host facilitate or constrain evo­ To obtain a parental generation consist­ lution of increased performance on the ing of individuals whose relative perfor­ other, i.e., is fitness on varia corre­ e. mance on M. sativa and varia was known, lated, either positively or negatively, with e. 17 philodice females were caught in fitness on M. sativa? e. southern Montmorency County, Michigan, To answer these questions, I measured in July 1985 and their offspring were reared the performance of C. philodice larvae on to pupation on each plant species. Both M. M. sativa and e. varia during two succes­ sativa and e. varia grow in this habitat, but sive generations, determined whether the larvae occur only on M. sativa. sampled population contained genetic On each plant species, neonate larvae from variation for larval performance on these all 17 females were allowedto mix, andwere plant species, and determined the nature then divided randomly among ten 10-cm ofcorrelations in performance across plant plastic Petri dishes. Larvae were reared in species. an environmental chamberundera 16:8 L:D PREDICTING HOST RANGE EVOLUTION 1639 cycle and a corresponding 29: 19°C temper­ Offspring Performance ature cycle. Larvae were provided fresh food every second day or more often ifnecessary. Performance was measured for the off­ Within 2 hr after it pupated, each larva was spring of12 mated pairs from M. sativa and marked with a unique number. Ten pupae 12 (for survivorship) or 11 (for all other from each diet were sacrificed to establish measures) from C. varia. Parental pairs from a wet weight/dry weight ratio used to esti­ each diet were chosen to provide as wide a mate the dry weight ofall remaining pupae. range as possible ofpaternal, maternal, and The performance of43 larvae fed M. sativa mid-parent growth rates. and 92 fed C. varia was measured in terms Survivorship. - Fifty newly hatched off­ ofegg-to-pupa development time and pupal spring from each set ofparents were divided weight. Two-way analysis of variance was into two groups of25 and placed into plastic performedwith host and sex as main effects. Petri dishes containing fresh leaves ofeither To facilitate mating between fast- and M. sativa or C. varia. Larval density was 25 slow-growing individuals within each diet, per dish during the first instar and 5 per all pupae were refrigerated at approximately dish thereafter. Survivorship