Proc. Natl. Acad. Sci. USA Vol. 89, pp. 2927-2929, April 1992 Evolution Multiple occurrences of mutualism in the yucca moth lineage (coevolution/mutualism/pollination) OLLE PELLMYR*t AND JOHN N. THOMPSONt *Department of Biological Sciences, University of Cincinnati, Cincinnati, OH 45221; and tDepartments of Botany and Zoology, Washington State University, Pullman, WA 99164 Communicated by Peter H. Raven, January 3, 1992 (received for review July 2, 1991) ABSTRACT The complex mutualism between yuccas and sidered to be the result of long-term coevolution between the the moths that pollinate their flowers is regarded as one of the moths and their yucca hosts (1, 12-15). most obvious cases of coevolution. Studies of related genera The 16 recognized species of Greya, which are endemic to show that at least two of the critical behavioral and life history western North America (16), are the sister group offive small traits suggested to have resulted from coevolved mutualism in genera that include the yucca moths (Fig. 1; 17-19). Greya yucca moths are plesiomorphic to the family. Another trait, species are highly host-specific on members of the families oviposition into flowers, has evolved repeatedly within the Saxifragaceae and Umbelliferae, whereas yucca moths and family. One species with these traits, Greya politeUa, feeds on allies feed exclusively on the Agavaceae. The Agavaceae and pollinates plants of a different family, but pollination feeders include seed parasites, some of which also pollinate occurs through a different component of the oviposition be- their hosts, as well as stem borers and leaf miners (5, 19). havior than in the yucca moths. Major differences compared Similarly, Greya species include seed parasites, stem borers, with yucca moths and their hosts are that G. politeUa only and leaffeeders (16, 20), and as we demonstrate here, at least passively pollinates its host and that copolators often con- one seed-parasitic species is a major pollinator of its host. tribute to pollination. This analysis suggests that evolution of Greya politella feeds almost exclusively on Lithophragma mutualism between yuccas and yucca moths may have required spp. (Saxifragaceae), the only exception being utilization of few behavioral and life history changes in the moths. The truly the closely related Heuchera grossularfifolia along the Clear- coevolved features ofthis interaction appear to be the evolution water River in Idaho (16). All life stages are strongly asso- ofactive pollination by the moths, the associated morphological ciated with the host plant: the adults appear only during the structures in the moths for carrying pollen, and the exclusion flowering period of the host, take nectar from flowers, and of copollinators by yuccas. usually mate on the host. Eggs are deposited inside the ovary. Larvae feed initially on developing seeds and, after diapause, Many mutualisms between species are thought to have arisen on vegetative parts. Pupation occurs in a cocoon on the host. over evolutionary time from antagonistic interactions (1, 2). At least seven ofthe nine species ofLithophragma (21) are Among the most striking mutualisms suggesting antagonistic utilized as hosts by G. politella in different parts of its origins are the coevolved interactions between plants and geographic range, but local populations are usually limited to insects that are both pollinators as adults and seed parasites one host (16). In southeastern Washington, the exclusive host as larvae. These include yuccas and yucca moths, figs and fig is Lithophragma parviflorum (Fig. 2). Pollination by G. wasps, and globeflowers and globeflower flies (3-11). These politella can potentially occur during either nectaring or associations are often cited as the classic textbook cases of oviposition. Both sexes take nectar by extending the pro- mutualism (1, 12, 13), but the phylogenetic origins of all of boscis past the anthers and stigmas down to the nectary atop these interactions have been the ovary. During oviposition, which is always preceded by unknown. Recent work, how- nectaring (n = 130 ovipositions), a female struggles to push ever, on the systematics and ecology ofthe Prodoxidae-the her family including the yucca moths-allows for the first time an abdomen down the narrow floral tube, extending a evaluation of which traits in the moths may be novel to the membranous eighth segment that normally rests inside the coevolved obligate mutualism between yucca moths and their long seventh segment, to cut into the ovary and to deposit one hosts. Here we present evidence for Greya politella (Wal- to nine (mode 2-3; median 3; n flowers = 36) eggs. In the singham) (Lepidoptera: a process, the abdomen touches both the anthers and the Prodoxidae), close relative of the stigma (Fig. 2), and the eighth segment of the abdomen is yucca moths, indicating that pollination mutualism has often completely coated with pollen. evolved more than once via seed parasitism in this moth Efficacy ofnectaring and oviposition in causing pollination family. Moreover, several of the behavioral and life history was tested by measuring the seed production resulting from traits found in ovipositing and pollinating yucca moths sug- single moth visits. Seeds of L. parviflorum collected from gested to be novel are likely not to be the direct result of numerous ramets in the experimental population near Granite coevolution with yuccas. Point in southeastern Washington (Whitman Co: T13N R46E A female yucca moth (Tegeticula) oviposits into the ovary S24; see ref. 22 for site description) in 1989 were stratified in of a Yucca flower and immediately afterward actively polli- early January 1990 in troughs, and individual seedlings were nates the flower, at least if it has not been pollinated before transplanted to tree tubes as soon as size permitted. The (4). Pollination secures seeds for her developing offspring to plants were grown at ambient temperature in Pullman, WA, feed upon, but the destruction of seeds is sufficiently limited without artificial lighting. When field and greenhouse plants that, on balance, the interaction is also beneficial to the plant were at the peak offlowering (March 28-April 4), greenhouse (3-6). The sequence of behaviors followed by Tegeticula plants were brought to the source population. Individual females during oviposition and pollination is generally con- plants were set out one at a time within the natural population and were observed until a moth alighted on a flower. Each The publication costs of this article were defrayed in part by page charge plant was then transferred with its moth to a 1-m3 mesh cage payment. This article must therefore be hereby marked "advertisement" containing about 20 additional plants. The moth is extremely in accordance with 18 U.S.C. §1734 solely to indicate this fact. tolerant of disturbance while on Lithophragma flowers and 2927 2928 Evolution: Pellmyr and Thompson Proc. Natl. Acad Sci. USA 89 (1992) Lampronla 30 - - Tetragma 25 a Nectaring cl C Oviposition Greya punctiferella group ? 20- 0 Greya solenoblella group Hi 15 - .0 Greya polltella group E 10 z Meseplola 5- Tegeticula 0 LO in Un LI) n Ln in SI in LI) CN - CIN N r- IN N% - ) C -, - 7 r,N N I. t 't N 0D LD s m L L OF P Parategeticula - CN ! N r- aN r N - - N N fnv OF Number of seeds per flower Agavenema FIG. 3. Seed number in Lithophragmaparviflorum resulting from Prodoxus a single bout of nectaring or oviposition by Greya politella. Filled OF bars, nectaring only (n = 27); stippled bars, oviposition (n = 70). FIG. 1. Partially collapsed cladogram of Prodoxidae, indicating mental plants were returned to the greenhouse, and the sites of evolution of life history traits critical to the evolution of flowers were dissected 9 days later. At that time, developing pollination based on oviposition and of pollination itself. LHS, local and nondeveloping ovules were readily separated, but moth host specificity; MH, mating on host; OF, oviposition in flower; P, larvae had not yet hatched from the eggs. Nonpollinated pollinator. All genera from Mesepiola on feed on Agavaceae. The filled bars indicate gain, open bars indicate loss, and the mixed bar flowers abort shortly after anthesis, and 4 of 70 experimental indicates both states present within the genus. Data are missing for flowers in the oviposition trials aborted before they could be one genus (Tridentaforma) and three Greya species. They have been examined for eggs. The results (Fig. 3) showed that nectaring excluded in this figure. Phylogeny is based on information in refs. alone contributed very little to seed set, causing no seeds at 16-19, and life history traits are based on refs. 16, 17, 19, and all in 59% of all cases, 10 or fewer in 85%, and never more unpublished data. than 27 seeds in any observation (median and mode = 0 seeds; n = 27 flowers). Individual flowers contained 276 ± 78 can be moved without any effects. After a single nectaring or ovules (mean ± SD; n = 187). Consequently, nectaring oviposition in the flower, the moth was allowed to alight on resulted in a rare single case in 10% seed set and, predomi- its choice of flower on any other plant in the cage. Visited nantly, none at all. In contrast, oviposition resulted in 28-453 flowers were then marked and denied further visits. Experi- seeds (median = 93) in 63% of the visited flowers and failed ( ~~~~~~~~~~~~~1~~ ~ ~ ~ ..:.0 FIG. 2. Greya politella ovipositing into flower of L. parviflorum (flower shown in cross-section; composite from photographs). The flower is genetically self-incompatible (21). The actual corolla diameter is ca. 15-18 mm, and the forewing length of the moth is ca. 9-10 mm. Evolution: Pellmyr and Thompson Proc. Natl. Acad. Sci. USA 89 (1992) 2929 to result in seed production in only 37% (n = 70). Considering some of the seeds as larvae, and this damage may outweigh the exceedingly low efficiency of nectaring behavior in its contribution to pollination when copollinators are abun- pollination, in effect only ovipositing females were poten- dant.