Proc. NatL Acad. Sci. USA Vol. 80, pp. 1298-1299, March 1983 Botany

Hawkmoth pollination of Mirabilis longiflora (Nyctaginaceae) (hawlanoth flowers) VERNE GRANT AND KAREN A. GRANT Department of Botany, University of Texas, Austin, Texas 78712 Contributed by Verne Grant, December 17, 1982 ABSTRACT A guild composed of very-long-tubed hawkmoth flowers (nectar tubes, 9 cm or more long), belonging to different genera and families, occurs in the American Southwest. Our knowledge of the hawkmoth associates of these flowers is frag- mentary. Mirable longJ7ora, a member of the guild with a tube 10.0-10.5 cm long, was found to be visited and pollinated mainly by quinquemaculata with a proboscis 10.7-11.6 cm long in the Chiricahua Mountains of southeastern Arizona. This ex- ample fits in with four other previously reported cases. The long- tonguedMan. quinquemaculata is nowknown to be associated with five species of very long-tongued hawkmoth flowers in the South- west, and Man. rustica has been found on one of them. The flora of the American Southwest contains a small contin- gent of species bearing hawkmoth flowers with very long floral tubes, 9 cm or more long. Relatively little is known about the hawkmoth visitors and pollinators of these flowers. Mirabilis longiflora (Nyctaginaceae) with perianth tubes 7-17 cm long is a member of this contingent for which we have pollination rec- ords, which are reported herein. The observations were made on growing in the live oak-sycamore zone in Cave Creek Canyon in the Chiricahua Mountains (Cochise County, AZ). The plants were observed on Mountains, AZ). three successive days and nights, Aug. 19-21, 1961. A few spec- FIG. 1. Flower of M. longiflora (Chiricahua imens of each class of visitors were collected for iden- ularvisitors, returning night after night, and they visited a series tification and further observation. of flowers on each foraging trip. The modem names of sphingids are used throughout Three individuals of Man. quinquemaculata were captured, this paper. Invalid generic names used in older papers are trans- measured for tongue length, and examined for pollen. Their lated automatically into the modern nomenclature in summa- tongue lengths were 10.7, 10.9, and 11.6 cm. It is evident that rizing the older work. The specific names remain unchanged. with tongues of this length can probe deep into the floral tubes while hovering just above the limb or at the level of the OBSERVATIONS stamens and style. One had Mirabilis pollen grains ad- hering to its face near the tongue base. Floral Mechanism. The bushes bear numerous flowers on Other Insect Visitors. Small moths (Bulia similaris, Pha- their tops and sides. The flowers are closed by day. Flower buds laenidae) also were attracted to the flowers at night and tried to open in the late afternoon before dusk and the flowers remain get nectar with their short tongues. Such incidental moth vis- open all night and into the next morning. At night they are very itors could transfer some pollen. fragrant. The flowers contain no nectar during the day, includ- At dusk carpenter bees (Xylocopa sp.) were stealing nectar ing the late afternoon, a trace of nectar appears at dusk, and the from the base of the perianth tube. The carpenter bees landed main nectar flow comes later when it is fully dark. on the limb, worked their way down to the tube base, where The flowers are white and salverform with a long, slender, they made a neat slit 7-8 cm long, and got the traces of nectar gently curved perianth tube oriented upward (Fig. 1). The tube present at this time of day. Because the carpenter bees contact measured 10. 0-10.5 cm long in our population. It was 2 mm in the anthers and stigma in their visits and were observed to have diameter at the orifice and 1.5 mm in diameter below. The tube Mirabilis pollen on their venter, they must bring about some is surmounted by a spreading limb 3 cm in diameter. pollen transfer. The five violet stamens and single violet style are exserted Honeybees (Apis mellifera) were seen collecting pollen dur- 1.5-2.0 cm above the limb, the style being extended beyond the ing the late afternoon and early morning when the flowers are stamens. The round anthers contain large smooth yellow pollen open and the pollen is exposed. Pollen-collecting honeybees also grains. The stigma bears numerous small tack-shaped hairs. could effect some pollination. Hawkmoth Visitors. At night the flowers were visited by in- Bees working by early morning or late afternoon, or both, are dividuals of Manduca quinquemaculata (), which a ubiquitous feature on nocturnal hawkmoth flowers. hovered over the flowers with their extended tongues inserted Other Mirabilis Species. Previously published pollination down the perianth tubes. In their normal feeding position their records of Mirabilis in western North America deal with the faces contact the anthers and stigma. The hawkmoths were reg- smaller-flowered and brightly colored species M. froebelii and 1298 Downloaded by guest on September 24, 2021 Botany: Grant and Grant Proc. Natl. Acad. Sci. USA 80 (1983) 1299 Table 1. Long-tongued hawkmoth species previously recorded on very-long-tubed hawkmoth flowers in the American Southwest Flower Species Tube length,'cm. Location Hawkmoth Ref. Oenothera brachycarpa 12.2 West Texas Man. quinquemaculata 12 Oenothera taraxacoides 13.2 New Mexico Man. quinquemaculata 12 Acleisanthes longiflora' 9-17 West Texas Man. quinquemaculata 12 and Man. rustica Datura meteloides 10.5-11.0 Southeastern Man. quinquemaculata 15 Arizona

M. multiflora with perianth tubes 3.5-4.5 cm and 2-6 cm long, dence regarding this problem. There are five species of hawk- respectively. moth in the American Southwest possessing proboscides 9 cm Cruden (1) reported the hawkmoths Sphinx chersis and Eu- or more long. Four of these have the following average tongue morpha achemon; as well as pollen-collecting bees, on M. mul- lengths (12): Man. sexta, 9.0 cm; Man. quinquemaculata, 10.3; tiflora in southwestern Utah. Cockerell (2) found Hyles lineata Man. rustica, 13.0; and cingulatus, 9.6. The fifth spe- (Sphingidae) on a "large purple mirabilis" in southern New cies, C. antaeus, has a still longer proboscis, for which Trelease Mexico, which is probably M. multiflora. (11) gives a value of 14-15 cm or more. Baker (3) made observations on. M. froebelii cultivated in Some of these species occur in northern parts of the western northern California. H. lineata is the regular nocturnal floral vis- United States as well as the Southwest, but in such cases they itor. However, hummingbirds and bees visit the flowers in the are much more abundant in the Southwest than in the North hours of dusk and morning. (12). All of them range south into Mexico. For more details on We made observations (unpublished data) on M.froebelit with their geographical distribution, see refs. 13 and 14. David Gregory on-plants growing in Rancho Santa Ana Botanic Gregory (12) did not encounter C. antaeus during extensive Garden (Claremont, southern California) in mid-August and mid- collecting at flowers in the Southwest. It could be expected to September of 1958. By night we recorded the following sphin- occur in this area (13). But on the present evidence this species gid species on its flowers: H. lineata, Man. sexta, Erynnis elo, does not seem to play a significant role in flower pollination in and Sphinx sp. the Southwestern flora. Knuth and Loew (4) cite Cockerell (2) for arecord ofH. lineata The field experience of Gregory (12) and ourselves indicates on M. longiflora. and consequently it has long been thought that that Man. sexta and Man. quinquemaculata tendto be abundant a hawkmoth pollination record exists for M. longiflora, but such at hawkmoth flower sites in the Southwest, whereas the other is not the case. Cockerell (2) refers only to a "large purple mir- two species, Man. rustica and A. cingulatus, are present but oc- abilis," which cannot be M. Iongiflora but is probably M. mul- casional. It is these four species, and particularly the first two, tiflora, as noted above. that would be expected to be associated with the very long- According to Cruden (5), M. nyctaginea in eastern North Amer- tongued hawkmoth flowers in the Southwest. ica is pollinated by noctuid moths during the nighttime and A few available pollination records for the Southwestern by bees in the afternoon and morning; it is also partially self- hawkmoth flowers with very long tubes confirm this expecta- pollinating. tion. The known cases of long-tongued maths on long-tubed Muller (6) and Knuth (7) list Herse convolvuli for M.jalapa in flowers in the Southwest are summarized in Table 1. M. Ion- Europe. Cruden (5) reports it as autogamous in Iowa. giflora with its Man. quinquemaculata visitors adds another ex- With regard to breeding system, M.froebelii, M. multiflora, ample to this emerging pattern. and M. greenei are self-incompatible (8), whereas M. nyctaginea, M. violacea, M. Grateful acknowledgment is made to Dr. Mont Cazier and the South- jalapa, and M. Iongiflora are self-compatible (5, western Research Station in Arizona for use of the facilities in 1961 and 9). to Dr. Lloyd Martin of the Los Angeles County Museum for identifi- cation of the moths. DISCUSSION 1. Cruden, R. W. (1970) BulL.Torrey Bot. Club 97, 89-91. Most hawkmoth flowers in the western United States have nec- 2. Cockerell, R. D. A. (1897) Proc. Acad. Nat. Sci. Philadelphia, 25- tar tubes of medium length, in the range 2-7 cm. M.froebelii 107. and M. fall 3. Baker, H. G. (1961) Q. Rev. Biol 36, 64-73. multiflora into this category. A small number of 4. Knuth, P. & Loew, E. (1904-1905) Handbuch der Blutenbiologie Western species of hawkmoth flowers, belonging to different (Engelmann, Leipzig, German Democratic Republic), VoL 3. genera and families, have very long nectar tubes, 9 cm or more 5. Cruden, R. W. (1973) Am. J. Bot. 60, 802-809. long. M. longiflora is amemberofthis guild, whichalso includes 6. Miller, H. (1883) The Fertilisation of Flowers, translated by particular species of Oenothera (Onagraceae), (Ran- Thompson, D'A. W. (Macmillan, London). unculaceae), Macrosiphonia (Apocynaceae), Datura (Solana- 7. Knuth, P. (1906-1909) Handbook of Flower Pollination, trans- ceae), and Acleisanthes lated by Davis, J. R. A. (Clarendon, Oxford), 3 Vols. (Nyctaginaceae). The very long-tubed 8. Baker, H. G. (1964) Evolution. 18, 507-509. species all occur in the southwestern partof the western United 9. Bogle, A. L. (1974)J. Arnold Arbor. Harv. Univ. 55, 1-37. States. 10. Gray, A. (1883) Bot. Gaz. (Chicago) 8, 295. The identity of the hawkmoth visitors to the very long-tubed 11. Trelease, W. (1883) Bot. Gaz. (Chicago) 8, 319. flowers has been a long-standing question. A century ago Gray 12. Gregory, D. P. (1963-1964) Aliso 5, 357-419. (10) raised this question with regard to , 13. Hodges, R. W (1971) Sphingoidea, The Moths of America North and Trelease (11) suggested of Mexico (Classey, London), Fasc. 21. antaeus as a possible can- 14. Schreiber, H. (1978) Dispersal Centres of Sphingidae (Lepidop- didate. tera) in the Neotropical Region (Junk, The Hague, The Nether- The extensive field studies of Gregory (12) on flower-visiting lands). hawkmoths in the western United States yielded important evi- 15. Grant, V. & Grant, K. A. (1983) Bot. Gaz..(Chicago) 144, in press. Downloaded by guest on September 24, 2021