Butterflies and Plants: a Study in Coevolution Paul R. Ehrlich; Peter

Butterflies and Plants: A Study in Coevolution

Paul R. Ehrlich; Peter H. Raven

Evolution, Vol. 18, No. 4. (Dec., 1964), pp. 586-608.

Stable URL: http://links.jstor.org/sici?sici=0014-3820%28196412%2918%3A4%3C586%3ABAPASI%3E2.0.CO%3B2-8

Evolution is currently published by Society for the Study of Evolution.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/ssevol.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org Wed Apr 18 12:34:09 2007 BUTTERFLIES AND PLANTS: A STUDY IN COEVOLUTIOK1

PAULR. EHRLICHAND PETERH. RAVEN Depovtnlei?t of Biological Sciei?ce~,Stan fovd Unive~sity,Stat? fovd, Califori?ia

Accepted June 15, 1964

One of the least understood aspects of plants with the hope of answering the fol- population biology is community evolu- lowing general questions: tion-the evolutionary interactions found 1. Without recourse to long-term esperi- among different liinds or organisms where mentation on single systems, what can be exchange of genetic information among the learned about the coevolutionary responses kinds is assumed to be minimal or absent. of ecologically intimate organisms? Studies of community evolution have, in general, tended to be narrow in scope and 2. Are predictive generalities about com- to ignore the reciprocal aspects of these munity evolution attainable? interactions. Indeed, one group of orga- 3. In the absence of a fossil record can nisms is all too often viewed'as a kind of the patterns discovered aid in separating physical constant. In an extreme example the rate and time components of evolution- a parasitologist might not consider the ary change in either or both groups? evolutionary history and responses of hosts, 4. Do studies of coevolution provide a while a specialist in vertebrates might as- reasonable starting point for the under- sume species of vertebrate parasites to be standing of community evolution in general? invariate entities. This view~ointis one factor in the general lack of progress toward FACT^^^ DET~R~ININGFOODcHOICE the understanding of organic diversifica- tion. Before proceeding to a consideration of One approach to what we like to the relationships between butterfly groups call coevolution is the examination of pat- and their food plants throughout the world, terns of interaction between two major it is necessary briefly to consider some of groups of organisms with a close and evi- the factors that determine the choice of dent ecological relationship, such as plants food plants in this group and in ph~toph- and herbivores. The considerable amount agous insects in general. Any group of of information available about butterflies phytophagous animals must draw its food and their food plants make them particu- supply from those plants that are available larly suitable for these investigations. in its iFographica1 and ecological range Further, recent detailed investigations have (Dethier, 1954). For instance, the butter- provided a relatively firm basis for state- flies are primarily a tropical group, and merits about the phenetic relationships of therefore there is a relatively greater utili- the various higher groups of Papilionoidea zation of primarily tropical than of tern- (Ehrlich, 1958, and unpubl.). It should, Perate families of plants. The choice of however, be remembered that we are con- oviposition site by the imago is also im- sidering the butterflies as a model. They portant. Man~.adult butterflies and moths are only one of the many groups of herbiv- lay their eggs on certain food plants with orous organisms coevolving with plants. In great precision as stressed by Merz (1959), this paper, we shall investigate the relation- but On the hand, urnis- ship between butterflies and their food takes" have been recorded (e.g., Reming- ton, 1952; Dethier, 1959). In such cases, larvae have either to find an appropriate lThis work has been supported in part by ~ ~ science~ ~ i~ ~ ~G~~~~~ ~ ~d GB-~~~~ ~ plantt li or perish.~ ~ There is an obvious selec- (Ehrlich) and GB-141 (Raven). tive advantage in oviposition on suitable EVOLUTION18: 586-608. December, 1964 586 BUTTERFLIES AND PLANTS 587 plants, but inappropriate choices can be These substances are repellent to most overcome by movement of the larvae. Fur- insects and may often be decisive in pat- thermore, larvae feeding on herbs often terns of food plant selection (Thorsteinson, consume the entire plant, and then must 1960). It has further been demonstrated move even if the adult originally made an that the chemical compounds that repel appropriate choice. most animals can serve as trigger sub- Larval choice therefore plays an impor- stances that induce the uptake of nutrients tant role in food plant relationships. An by members of certain oligophagous groups excellent review of a long series of experi- (Dethier, 1941, 1954; Thorsteinson, 1953, ments pertinent to this subject has recently 1960). Presence of such repellent com- been presented by Merz (1959); much of pounds may be correlated with the presence the following is based on his account. The of the nutrients. Both odor and taste seem condition of a given larva often has an to be important. effect on what foods it will or will not The chemical composition of plants often accept. In addition, many structural and changes with age, exposure to sunlight, or mechanical characteristics of plants modify other environmental factors (Merz, 1959; these relationships, mostly by limiting the Fliick, 1963), and this may be critical for acceptability of those plants in which they phytophagous insects (Dethier, 1954). For occur. For example, Merz (1959) found example, insects that feed on Umbelliferae that larvae of Lasiocampa quercus, a moth prefer the old leaves, which appear to us that normally feeds along the edge of less odorous than the young ones. Some leaves, could not eat the sharply toothed insects that feed on alkaloid-rich species of leaves of holly (Ilex, Aquifoliaceae). When Papaver (Papaveraceae) prefer the young these same leaves were cut so that untoothed leaves, which are relatively poor in alka- margins were presented, the larvae ate loids. Diurnal chemical cycles, influenced them voraciously. In other cases, larvae by exposure of the plant to sunlight, may eat the young, soft leaves of plants but not be of prime importance in determining the the old, tough leaves of the same plants. habits of night-feeding groups, such as Many Lycaenidae feed on flowers, and Argynnini. these butterflies may be unable to utilize Merz (1959, p. 159) has given a particu- the tough foliage of the same plants. larly interesting case of chemical repellents iXumerous similar examples could be given, at the specific level. The larvae of the but it must be borne in mind that chemical moth Euchelia jacobaeae feed on many factors are operative in the same plants species of Senecio (Compositae), but not that present mechanical difficulties to lar- on the densely glandular-hairy S. viscosus. vae (Thorsteinson, 1960), and actually When the glandular substance was dis- may be more important. solved in methyl alcohol, the larvae ate S. Chemical factors are of great general viscosus. When the same substance was importance in determining larval food painted on the leaves of other normally choice. In the first place, potential food acceptable species of Senecio, these were sources are probably all nutritionally un- refused. In an extensive study of the food balanced to some extent (Gordon, 1961). plants of Plebejus icarioides (Lycaeninae) , The exploitation of a particular plant as Downey (1961, 1962) showed that larvae a source of food thus involves metabolic would feed on any species of Lupinus adjustments on the part of an insect. These (Leguminosae) in captivity, but popula- render the insect relatively inefficient in tions in the held normally utilized only one utilizing other sources of food and tend to or a few of the possible range of Lupinus restrict its choice of food plants. Secondly, species growing locally. This work suggests many plants are characterized by the the subtle interaction of ecological, chemi- presence of secondary metabolic substances. cal, and mechanical factors that doubtless 588 PAUL R. EHRLICH AND PETER H. RAVEN characterizes most natural situations. Re- Lycaenidae about tied for third. Difficult lationships with predators (Brower, 1958), as this diversity is to estimate, it is clear parasites (Downey, 1962), or, at least that Papilionidae are a more heterogeneous in the case of Lycaenidae, ants (Downey, group of organisms than any of the other 1962), may further modify patterns of families, whereas, considering the number food plant choice. of species and genera included, Lycaenidae Despite all of these modifying factors, are remarkably uniform. A rough idea of there is a general and long-recognized pat- the phenetic relationships of the major tern running through the food plants of groups of butterflies is given by Ehrlich various groups of butterflies, and it is this (1958). pattern with which we shall be concerned. With food plant records from between 46 It certainly should not be inferred from and 60% of all butterfly genera (table I), anything that follows that all members of it seems highly unlikely that future dis- a family or genus of plants are equally coveries will necessitate extensive revisions acceptable to a given butterfly (for ex- of the conclusions drawn in this paper. The ample. see Remington, 1952). We have food plants of Riodininae are very poorly placed our main emphasis on positive known, however, and it will be interesting records, especially at the level of plant to have more information about them and species and genera. records for other outstanding "unknowns" such as Styx and Pseudopontia. It is, how- ever, difficult to imagine any additional food plant record that would seriously dis- The butterflies comprise a single super- tort the patterns outlined here. family of Lepidoptera, the Papilionoidea. In comparison with many other superfamilies of insects they are uniform morphologically and behaviorally. Table 1 gives a rough Sources of Information idea of the taxonomic diversity of this The food plant information abstracted in superfamily. this paper is derived principally from two Papilionoidea are divided into five fam- sources. First, we have examined all the ilies. Two of these, Nymphalidae and extensive and scattered literature that we Lycaenidae, contain at least three-quarters could uncover with a library search and of the genera and species; it is uncertain through the recommendations of various which family is the larger. Two smaller lepidopterists. Particularly helpful have families, Pieridae and Papilionidae, include been the volumes of Barrett and Burns virtually all remaining butterflies. Pieridae, (1951), Corbet and Pendlebury (1956), although containing many fewer genera and Costa Lima (1936), Ehrlich and Ehrlich species than either Lycaenidae or Nym- (1961), Lee (1958), Seitz (1906-1927), phalidae, form a prominent part of the van Son (1949, 1955), Wiltshire (1957), butterfly fauna in many parts of the world, and Wynter-Blyth (1957), as well as the making up in number of individuals what Journal of the Entomological Society of they lack in number of kinds. Papilionidae South Africa, the Journal of the Lepidop- are a group about half the size of Pieridae, terists' Society (formerly Lepidopterists' but gain prominence through the large size News), and the Journal of Research on the of the included forms. The tiny family Lepido ptera. Libytheidae, closely related to Nymphali- Our second major source of information dae, is obscure to everyone except butterfly has been provided by the following scien- taxonomists. tists, who have aided us in this ambitious The Papilionidae lead the butterflies in undertaking not only by sending unpub- morphological diversity. Nymphalidae prob- lished data and reprints of their works, but ably take second place, with Pieridae and by helping to evaluate the validity of cer- BUTTERFLIES ATqD PLANTS 589

Approxitnate number of Taxon Distribution Genera* Specie5

Papilionidae 24(22) 575-700 Baroniinae 1(1) 1 Central Mexico Parnassiinae 8(8) 45-55 Holarctic and Oriental; greatest diversity, Asia Papilioninae 15(13) 480-640 \krorldwide; mainly tropical. Greatest diversity, Old World tropics

Pieridae 58(40) 950-1,150 Coliadinae 11(8) 225-250 Cosmopolitan; greatest diversity tropics outside of Africa Pierinae 43 (29) 650-750 Cosmopolitan ; greatest diversity tropics Dismorphiinae 3 (3) 80-120 Primarily Keotropical; one small Palearctic genus Pseudopontiinae l(0) 1 \Vest equatorial Africa

Symphalidae 325-400 4,800-6,200 (ca. 202) Ithomiinae 30-40(10) 300-400 Keotropical, Telle~voAustralian Danainae 10-12(10) 140-200 Cosmopolitan; greatest diversity Old World tropics Satyrinae 120-150 1,200-1,500 Cosmopolitan; greatest diversity extratropical (ca. 70) Morphinae 23-26(12) 180-250 Indomalayan and h'eotropical Charaxinae 8-lO(8) 300-400 Tropicopolitan, few temperate Calinaginae 1(1) 1 Oriental Nymphalinae 125-150 2,500-3,000 Cosmopolitan (ca. 85) Acraeinae 8(6) 225-275 Tropical; greatest diversity, Africa

Libytheidae 1(1) 10 Cosmopolitan

Lycaenidae 325-425 5,800-7,200 (ca. 167) Riodininae 75-125 (17) 800-1,200 Tropical, ftm Nearctic and Palearctic. Metropolis, Neo- tropical Styginae 1(0) 1 Peruvian Andes Lycaeninae 25@300 5,000-6,000 Cosmopolitan; greatest diversit), Old \Vorld tropics (ca. 150)

Total 730-930 12,000-15,000 (ca. 432)

"Number in parentheses indicates number of genera for which food plant records are available. tain published records and commenting on Iwase (Japan), T. W. Langer (Denmark), other aspects of the work. The cooperation C. D. MacNeill (USA), D. P. Murray of these people has been truly extraordi- (England), G. Sevastopulo (Africa), Ta- nary, and we are particularly indebted to kashi Shirozu (Japan), E. M. Shull them: Remauldo F. d'A11neida (Brazil), (India), Henri Stempffer (France), V. G. Peter Bellinger (USA), C. M. de Biezanko L, van Someren (Africa), G. van Son (Brazil), L. P. Brower (USA), C. A. (Africa). Clarke (England), H. K. Clench (USA), John C. Downey (Southern Illinois Uni- J. A. Comstock (USA), C. G. C. Dickson versity), Gordon H. Orians (University of (Africa), J. C. Downey (USA), Maria Washington), T. A. Geissman (University Etcheverry (Chile), K. J. Hayward (Ar- of California, Los Angeles), and Robert F. gentina), T. G. Howarth (England), Taro Thorne (Rancho Santa Ana Botanic Gar- 590 PAUL R. EHRLICH .AND PETER H. RAVEN den) have been so kind as to read the entire ily on broad, repeatedly verified patterns manuscript. Their advice has been in- of relationship. valuable. To our knowledge, the data assembled The Food Plants of Butterflies here represent the most extensive body of information ever assembled on the inter- In this section, we will first outline the actions between a major group of herbivo- main patterns of food plant choice for each rous animals and their food plants. family, and then discuss what bearing these patterns have on our interpretation of re- lationships within the various butterfly families. It is necessary to give the data Extreme care has been taken in associat- in considerable detail, as no comprehensive ing insects with particular food plants, as survey on a world basis is available else- the literature is replete with errors and un- where. verified records. In evaluating records, Papi1ionidae.-There are three subfami- preference has been given to those which lies. Baronia brevicornis, the only species are concerned with the entire life cycle of of Baroniinae, occurs in Mexico and feeds a particular insect on a wild plant. Labora- on Acacia (Leguminosae; Vazquez and tory experiments and records from culti- Perez, 1961). In Parnassiinae, all five vated plants demonstrate only potentiali- genera of Zerynthiini (RIunroe, 1960; ties, not necessarily natural associations. RIunroe and Ehrlich, 1960) feed on Aristo- In the laboratory, larvae may be starved or lochiaceae, as does Archon (Parnassiini). plants abnormal. In the wild, larvae are Hypermnestra (Parnassiini) is recorded often misidentified, especially if not reared from Zygophyllum (Zygophyllaceae). Par- to maturity (cf. Brower, 195813). Even nassius feeds on Crassulaceae and herba- more serious is the lack of precise plant ceous Saxifragaceae, two closely related identifications, or their identification in families, with one small group on Fumaria- the vernacular only, which almost inevi- ceae. In view of the discussion below, it tably leads to confusion (Jorgensen, 1932). is of interest that Zygophyllaceae are close Any serious student of phytophagous ani- relatives of Rutaceae, and Fumariaceae are mals should preserve adequate herbarium rich in alkaloids similar to those of woody specimens of the plants with which he is Ranales (Hegnauer, 1963). concerned (cf. Remington, 1952, p. 62) ; The third and last subfamily, Papilioni- only by doing this can the records be nae, is cosmopolitan but best developed in verified. Despite the extremely erratic the Old World, and consists of three tribes: oviposition behavior often shown by but- Troidini, Graphiini, and Papilionini. The terflies (Dethier, 1959), oviposition records eight genera of Troidini feed mostly on have all too frequently been accepted as Aristolochiaceae, with individual species of being equivalent to food plant records. Parides also recorded from Rutaceae, Meni- Finally, nomenclatural difficulties, includ- spermaceae, Nepenthaceae, and Piperaceae. ing changes in name and careless misspell- Parides (Atrophaneura) daemonius is re- ings (e.g., "Oleaceae" for Olacaceae) , have ported to feed on Osteomeles (Rosaceae), given rise to serious errors. In the litera- and P. (A.) antenor, a Malagasy butterfly ture on butterfly food plants, errors have that is the only representative of its tribe often been compounded when copied from in the Ethiopian region, feeds on Combre- one source to another, and they are difficult tum (Combretaceae). Both of these last- to trace back to their origins. All of these mentioned records need confirmation. At problems make quantitative comparisons least two species of Battus have been re- unreasonable. We therefore have been ex- corded from Rutaceae in addition to the ceedingly conservative about accepting usual Aristolochiaceae. Records available records, and focused our attention primar- for five of the seven genera of Graphiini BUTTERFLIES AND PLANTS 591

(Eztrytides, Graphium, Lamproptera, Pro- ceae, Rutaceae (Ptelea), and Salicaceae tographium, Teinopalpus) are mostly from (Brower, 1958b). Annonaceae, Hernandiaceae, Lauraceae, In Papilionidae, Munroe and Ehrlich Magnoliaceae, and Winteraceae. This is (1960) have argued that the red-tubercu- clearly a closely allied group of plant late, Aristolochiaceae-feeding larvae of families referable to the woody Ranales. Papilioninae-Troidini and Parnassiinae- In addition, some species of Graphiurn feed Zerynthiini, plus ilrchon (Parnassiinae- on Rutaceae, and others both on Apocyna- Parnassini) are so similar, and the likelihood ceae (Landolphia) and Annonaceae (one of their converging on Aristolochiaceae so of the latter also on Sphedamnocarpus, remote, that these probably represent the Malpighiaceae). Several species of Eury- remnants of the stock from which the rest tides feed on Vitex (Verbenaceae) and one of Papilioninae and Parnassiinae were de- on Jacobinia (Acanthaceae) . Eurytides rived. Viewed in this context other food lysithous feeds both on Annonaceae and on plants of these groups are secondary. The Jacobinia, and E, helios both on Vitex and two remaining tribes of Papilioninae (Papil- Magnoliaceae. The bitypic Palearctic ionini, Graphiini) are above all~associated Iphiclides departs from the usual pattern with the group of dicotyledons known as for the group in feeding on a number of the woody Ranales. This is a diverse Rosaceae-Pomoideae. assemblage of plant families showing The third and last tribe, Papilionini, con- many unspecialized characteristics. Thorne sists only of the enormous cosmopolitan (1963) has used the food plant relation- genus Papilio. Two of the five sections ships of Papilionidae as a whole to support recognized by Munroe (1960; 11, IV) are his suggestion of affinity between Aristo- primarily on Rutaceae, with occasional lochiaceae and Annonaceae (one of thc. records from Canellaceae, Lauraceae, and woody Ranales). He appears to have Piperaceae. Members of the circumboreal established the existence of this similarity Papilio nzachaon group are not only on on morphological evidence. Likewise, simi- Rutaceae but also on Umbelliferae and lar alkaloids are shared by Aristolochiaceae Artemisia (Compositae). The African P. and woody Ranales (Hegnauer, 1963; Al- demodocus, in another group, is known to ston and Turner, 1963, p. 170). Recently, feed on Rutaceae and Umbelliferae, as Vazquez and Perez (1961) described the well as Pseudospondias (Anacardiaceae), life cycle of Baronia brevicornis, the only Ptaeroxylon (Meliaceae) ,and Hippobrornus member of the third subfamily of the (Sapindaceae). Another African species, P. group. Baronia feeds on Acacia (Legumi- dardanus, is recorded from Rutaceae and nosae) and has tuberculate larvae like those also from Xymalos (Flacourtiaceae; Dick- of the forms that feed on Aristolochiaceae. son, pers. comm.). The Asian and Austra- Considering its moiphological distinctness, lian P. demoleus is mostly on Rutaceae but and in accordance with the scheme of re- also locally on Salvia (Labiatae) and lationships presented by Munroe and Psoralea (Leguminosae). The other three Ehrlich (1960, p. 175), it appears likely sections (Munroe's I, 111, V) are primarily that Baronia represents a phylogenetic line associated with Annonaceae, Canellaceae, which diverged early from that leading to Hernandiaceae, Lauraceae, and Magnolia- the rest of Papilionidae. It may thus be ceae, with a few records from Berberidaceae, the only member of the family neither Malvaceae ( Thespea) , and Rutaceae. The feeding on Aristolochiaceae nor descended North American temperate Papilio glaucus from forms that did. group (sect. 11) feeds not only on Lauraceae Following this reasoning, we suggest that and Magnoliaceae like its more southern the original transition to Aristolochiaceae relatives but also on Aceraceae, Betulaceae, opened a new adaptive zone for Papilioni- Oleaceae, Platanaceae, Rhamnaceae, Rosa- dae. Their further spread and the multi- 592 PAUL R. EHRLICH ASD PETER H. RAVEN plication of species was accompanied by nothing is known of the biology of the the exploitation of other presumably chemi- monobasic West African Pseudopontiinae. cally similar plant groups, such as woody Of the remaining three, Dismorphiinae, in- Ranales, in areas where Aristolochiaceae cluding the Neotropical Disrnorphia and were poorly represented, like Africa today. Pseudopieris and the Palearctic Leptidia, The site of greatest diversity for both are recorded only on Leguminosae. Larval Aristolochiaceae and Papilionidae is Asia. food plants are known for 7 of the 11 It is likely tdat the major diversification genera of Coliadinae. Catopsilia, Phoebis, of Papilionidae (involving differentiation Anteos, Eurema, and Colias are mostly into Parnassiinae and Papilioninae) tool; associated with Leguminosae, but there are place after the evolution of Aristolochiaceae. a few records from Sapindaceae, Guttiferae, When this might have been is entirely Euphorbiaceae, Simarubaceae, Oxalidaceae, uncertain, despite the unfounded specula- Salicaceae, Ericaceae, and Gentianaceae tions of Forbes (1958). (the last three with northern and montane Another interesting problem is posed by species of Colias). On the other hand. the many representatives of Papilionini and three genera are associated with non-legumi- Graphiini feeding on Rutaceae, in addition nous plants: Gonepteryx with Rhamnus to woody Ranales. Rutaceae are morpho- (Rhamnaceae), Nathalis with Compositae, logically very different from woody Ranales, and Kricogonia with Guaiacurrz (Zygophyl- and have not been closely associated with laceae) . Nonetheless, Leguminosae are de- them taxonomically. Recently, however, cidedly the most important food plants of Hegnauer (1963) has pointed out that Coliadinae. some Rutaceae possess the alkaloids wide- Pierinae, the third subfamily, are di- spread in woody Ranales, in addition to an vided into two tribes, Pierini (36 genera) unusually rich repertoire of other alkaloids. and Euchloini (7). In Euchloini, the tem- Earlier, Dethier (1941) showed the simi- perate Anthocharis, Euchloz, Zegris, and larity between the attractant essential oils Hesperocharis (also from Phrygilanthus. in Rutaceae and Umbelliferae. Some Loranthaceae) feed on Cruciferae, the Rutaceae-feeding groups of Papilio seem tropical Pinacopteryx and Hebemoia on to have shifted to Umbelliferae, especially Capparidaceae. For Pierini, 14 of the 23 outside of the tropics. Dethier also im- genera for which we know something of the plicated some of the similar-scented species food plants (namely, Appias, Ascia, Bele- of Arternisia (Compositae), another plant nois, Ceporis, Colotis, Dixeia, Elodina, group fed on by at least one species of Eronia, Zxias, Leptosia, Pareronia, Pieris, Papilio. Although species of Papilio link Prioneris, and Tatochila) are primarily on these groups of plants, none is known to Capparidaceae in the tropics and subtropics feed on Burseraceae, Cneoraceae, Simaru- and on Cruciferae in temperate regions. baceae, or Zygophyllaceae, families thought Some have occasionally been reported to to be related to Rutaceae but not known feed on Resedaceae, Salvadoraceae, and to contain alkaloids or coumarins (Price, Tropaeolaceae. There are also a very few 1963). On the other hand, the record of scattered records from other plants, includ- Papilio denzodocus on Ptaeroxylon (Melia- ing one or two from Leguminosae. The ceae), in addition to numerous Rutaceae, basis for selecting Capparidaceae, Crucif- would seem to indicate a promising plant erae, Resedaceae, Salvadoraceae, and Tro- to investigate for the alkaloids suspected paeolaceae is relatively easy to comprehend, (Price, 1963) in Meliaceae. since all of these plants are known to con- Pieridae.-Our discussion of Pieridae is tain mustard oil glucosides (thioglucosides) based taxonomically on the generic review and the associated enzyme myrosinase of Klots (1933) as modified by Ehrlich which acts in the hydrolysis of glucosides (1958). There are four subfamilies. but to release mustard oils (Alston and Turner. BUTTERFLIES AND PLANTS 593

1963, p. 284-288). In an early series of (Loranthaceae) in South America, and food choice experiments, Verschaeffelt Delias, a large Indo-Malaysian genus, from (1910) found that larvae of Pieris rapae "Loranthus" (Loranthaceae) and Exocarpus and P. brassicae would feed on Capparida- of the closely related Santalaceae, with D. ceae, Cruciferae, Resedaceae, and Tro- aglaja on rVauclea (Rubiaceae) . Aporia, a paeolaceae, as well as another family which large genus of temperate regions of the Old contains mustard oils but upon which World, has several species on Bevberis Pierinae are not known to feed in nature: (Berberidaceae), and one on woody Rosa- Moringaceae. Verschaeffelt also found that ceae. Pereute, South American, feeds on these larvae would eat flour, starch, or even Ocotea (Lauraceae; Jorgensen, 1932). filter paper if it was smeared with juice Tiliaceae, and Polygonaceae. Two very expressed from Bunias (Cruciferae) , and peculiar genera of the Delias group are the Thorsteinson (1953, 1960) showed that the monotypic Mexican Eucheira, which feeds larvae would eat other kinds of leaves on woody hard-leaved Ericaceae, and the treated with sinigrin or sinalbin (two com- bitypic western North American iveophasia, mon mustard oil glucosides) if the leaves which feeds on various genera of Pinaceae. were not too tough and did not contain It is very interesting that Cepora, which other kinds of repellents. Very few butter- falls into the Delias group morphologically, flies outside Pierdinae feed on these plants. feeds on Capparis like many other Pierini. but there is one example in Lycaeninae. In Finally, the large, taxonomically isolated. addition, there are at least two records of Ethiopian Mylothris feeds on Loranthaceae Phoebis (Coliadinae) from Capparidaceae and Santalaceae (Osyris) , with M. bernice and Cruciferae. Numerous groups of in- rubricosta on Polygonurrz (Polygonaceae) . sects other than butterflies are character- It is difficult to understand the reasons istically associated with this same series of for large groups of Pierinae being asso- plant families (Fraenkel, 1959). ciated both with plants that possess mus- It is not so easy to interpret scattered tard oils and with Loranthaceae-Santala- records of these pierine genera feeding on ceae; neither morphological nor biochemical other plant families : Belenois raffrayi on evidence has been adduced to link these Rhus (Anacardiaceae) ; Nepheronia argyia two groups of plants. Perhaps the Lo- on Capparidaceae, but also Cassipourea ranthaceae-feeders represent an old offshoot (Rhizophoraceae) and Hippocratea (Hip- of Pierinae; in any case it would appear pocrateaceae), with N. thalassina reported that the main diversification of this group only from Hippocratea; ilscia monuste on occurred after it became associated with Rhamnaceae and C'assia (Leguminosae) ,as Capparidaceae-Cruciferae. well as Capparidaceae; and Tatochila auto- Nynzpha1idae.-This enormous family is dice on Cestrum (Solanaceae) and also divided into eight subfamilies which will be Medicago (Leguminosae) . Several species discussed one by one in the succeeding of ilppias have been reported from differ- paragraphs. ent genera of Euphorbiaceae, whereas Ithomiinae are primarily American, and others feed both on Capparidaceae and there feed only on Solanaceae (many Euphorbiaceae, but this probably can be genera). The Indo-Malaysian Tellervo, explained somewhat more simply, since only Old World representative of the group, mustard oils have been reported in some which is segregated as a distinct tribe genera of Euphorbiaceae (Alston and Tellervini, has been recorded from Aristo- Turner, 1963, p. 285). lochia (Aristolochiaceae) . The identity of The remaining genera of Pierini fall the plant was inferred from the fact that mostly into what has been called the Delias papilionid larvae normally associated with group. Of these, Catasticta and Archonias Aristolochia were found on it with Tellervo. have been recorded from Phrygilanthus Solanaceae are rich in alkaloids (as are 594 PAUL R. EHRLICH AND PETER H. RAVEK

Aristolochiaceae), and are very poorly Closely related to Morphinae are the represented among the food plants of more temperate Satyrinae, an enormous butterflies as a whole. The diversification group that feeds mostly on Gramineae of the ithomiines that feed on them has (including bamboos and canes) and Cy- probably followed a pattern similar to peraceae, occasionally on Juncaceae. Pseu- that of Papilionidae on Aristolochiaceae donympha vigilans feeds on Restio (Res- and Pierinae on Capparidaceae-Cruciferae. tionaceae), a family close to Gramineae, Rtany other groups of insects feed primar- Physcaneura pione on Zingiberaceae, and ily on Solanaceae (Fraenl~el,1959). Ely~nniason Palmae. There are no records Danainae are a rather uniform cosmo- of this group from dicotyledons. Thus the politan group, obviously related to Itho- phenetically similar Morphinae-Satyrinae miinae. The danaines feed primarily and assemblage is the outstanding example in apparently interchangeably on Apocynaceae butterflies of a group associated primarily and Asclepiadaceae. In addition, there are with monocotyledons. records of Euploea, Ituna, and Lycorella on The distinctive tropicopolitan Charaxinae Moraceae and of the last occasionally on are often associated with woody Ranales Carica (Caricaceae). All of these plants (Annonaceae, Lauraceae, Monimiaceae, have milky juice. There is also a single Piperaceae), but also with such diverse record of Ituna ilione, which normally families as Anacardiaceae, Araliaceae feeds on Ficus (Moraceae), from Myopo- (Schefflera), Bombacaceae, Celastraceae, rum (Myoporaceae). Apocynaceae and Connaraceae, Convolvulaceae, Euphorbia- Asclepiadaceae form a virtual continuum ceae, Flacourtiaceae, Hippocrateaceae, Le- in their pattern of variation and can guminosae, Linaceae, Malvaceae, Meliaceae, scarcely be maintained as distinct (Safwat, Rtelianthaceae, Myrtaceae, Proteaceae, 1962). Both are noted for their abundant Rhamnaceae, Rutaceae, Salvadoraceae bitter glycosides and alkaloids (Alston and (Charaxes hansali) , Sapindaceae, Ster- Turner, 1963, p. 258), and share at least culiaceae, Tiliaceae, Ulmaceae, and Ver- some alkaloids (Price, 1963, p. 431) and benaceae. Records (largely Sevastopulo pyridines with Moraceae. Thus it appears and van Someren, pers. comm.) are avail- very likely that here too the acquisition of able for about 50 African species of the ability to feed on Apocynaceae and Charaxes, most of which are associated Asclepiadaceae has constituted for Da- with dicotyledons. At least three feed on nainae the penetration of a new adaptive grasses ( Gramineae), two of these on dicot- zone, in which they have radiated. Numer- yledons also. ous distinctive segments of other insect The Oriental Calinaga buddha, the only orders and groups likewise feed on these species of Calinaginae, feeds on Morus two plant families. (Moraceae). Eleven genera of Rtorphinae are recorded Nymphalinae are a huge cosmopolitan from a variety of monocotyledons: Bro- group with relatively few "gaps" in their meliaceae, Gramineae (mostly bamboos), pattern of variation which would permit the Rtarantaceae, Rtusaceae, Palmae, Pandana- recognition of meaningful subgroups (cf. ceae, and Zingiberaceae. In contrast, most Reuter, 1896; Chermocl;, 1950). The tribes species of Morpho feed on dicotyledons, do, however, display some significant pat- including Canellaceae, Erythroxylaceae, terns in their choice of food plants, with Hel- Lauraceae, Leguminosae, Rtenispermaceae, iconiini (Llichener, 1942) and Argynnini Rtyrtaceae, Rhamnaceae, and Sapindaceae, feeding mostly on the Passifloraceae-Fla- but M. aega feeds on bamboos (Gramineae) courtiaceae-Violaceae-Turneraceae com- and M. hercules on Musaceae. Whether plex of families, a closely related group the progenitors of Morpho fed on dicotyle- of plants also important for Acraeinae. dons or monocotyledons cannot be de- Acraeinae (see below), Heliconiini, and termined. Argynnini are closely related phenetically, BUTTERFLIES AND PLANTS 595 and their diversification may have taken Melastomaceae, Melianthaceae, Menisper- place from a common ancestor associated maceae, Myrtaceae, Oleaceae, Ranuncula- with this particular assemblage of plants. ceae (Vanessa on Delphiniurrz) , Rhamna- No biochemical basis is known for the ceae, Rosaceae, Rubiaceae, Sabiaceae, association of this series of four plant Salicaceae, Sapotaceae, Saxifragaceae, Ster- families, but we confidently predict that culiaceae, Thymeleaceae, Tiliaceae, and one eventually will be found (cf. also Vitaceae. Gibbs, 1963, p. 63). Rlelitaeini are often Some of the butterflies in this group associated with Acanthaceae, Scrophularia- feed on a very wide range of plants, ceae and their wind-pollinated derivatives and most of the families mentioned in the Plantaginaceae, and with Compositae and above list are represented by one or at Verbenaceae. Nymphalini feed on plants most a very few records. For example, of the same families as Melitaeini, but also Euptoieta claudia is known to feed on very prominently on the Ulmaceae-Urtica- Berberidaceae (P~doph~lhrrz),Crassula- ceae-Moraceae group and the Convolvula- ceae, Leguminosae, Linaceae, Menisperma- ceae, Labiatae, Portulacaceae, and Ver- ceae, Nyctaginaceae, Passifloraceae, Portu- benaceae. A single species in this group, lacaceae, Violaceae, and even Asclepiadaceae however, 1Vymphalis canace, feeds on (Cyanchum), and Precis lavinia is re- Liliaceae and Dioscoreaceae. Apaturini are corded from, among others, Acanthaceae, associated chiefly with Ulmaceae, especially Bignoniaceae, Compositae, Crassulaceae, Celtis. Cyrestini (Chersonesia, Cyrestis, Onagraceae (Ludwigia), Plantaginaceae, and Marpesia) and Gynaeciini (Gynaecia Scrophulariaceae, and Verbenaceae. and Historis, but not Callizona and Snzyrna) Acraeinae, a rather small tropical group, are often associated with Moraceae, and are often associated with Passifloraceae- Hamadryini (Ectima, Hamadryas), Dido- Flacourtiaceae-Violaceae-Turneraceae, as nini (Didonis), Ergolini (Byblia, Byblis, noted above, but also with Amaranthaceae, Ergolis, Eurytela, Mestra), Eunicini (As- Compositae, Convolvulaceae, Leguminosae, terope, Catonephele, Eunica, Myscelia), Lythraceae, Rtoraceae, Polygonaceae, Rosa- and Dynamini (Dynamine) mostly on the ceae, Sterculiaceae, Urticaceae, and Vita- related Euphorbiaceae, which, like Mora- ceae. In addition, Acraea encedon is re- ceae, have milky sap. In addition to the ported from Commelina (Commelinaceae) . Eunicini just mentioned, Diaethria (Calli- For the very diverse Nymphalidae as a core, Catagramma) , Epiphile, Haematera, whole, the following groups of plants are Pyrrhogyra, and Temenis feed almost ex- especially important: ( 1) Passifloraceae- clusively on Sapindaceae. There is no Flacourtiaceae-Violaceae-Turneraceae; (2) obvious dominant theme for the last tribe, Ulmaceae-Urticaceae-Moraceae, as well as Limenitini, but it is interesting to note that the closely related (Thorne, pers. comm.) two species of Euphaedra (Najas), a group Euphorbiaceae; (3) Acanthaceae-Scrophu- that is mostly on Sapindaceae, are on Cocos lariaceae-Plantaginaceae. The second and and other Palmae. Additional families third of these groups are represented among represented among the food plants of the food plants of other butterflies, such as Nymphalinae are: Aceraceae, Amarantha- Lycaeninae, but not abundantly. Con- ceae, Anacardiaceae, Annonaceae, Berberi- versely, as will be seen, the groups of food daceae, Betulaceae, Bignoniaceae, Bom- plants commonly represented in Lycaenidae bacaceae, Boraginaceae, Caprifoliaceae, -for example, Fagaceae, Leguminosae, Combretaceae, Corylaceae, Crassulaceae, Oleaceae, Rosaceae-are rare in Nymphal- Curcurbitaceae, Dilleniaceae, Dipterocarpa- idae. Although each of these two families ceae, Ebenaceae, Eleagnaceae, Ericaceae, of butterflies is very wide in its choice of Fagaceae, Gentianaceae, Geraniaceae, Gut- food plants, there is a distinctiveness to tiferae, Icacinaceae, Leguminosae (very the two patterns which suggests a history uncommonly), Loranthaceae, Malvaceae, of selection along different lines. 596 PAUL R. EHRLICH .-\SD PETER H. R.-\VES

Libytheidae.-This small family consists daceae), and Stalachtis from Oxpetalum of a single widespread genus, Libythea, (Asclepiadaceae : cf. Jorgensen, 1932, p. 43, which feeds almost exclusively on Celtis however, where it is suggested that associa- (Ulmaceae), but in southern Japan on tions of larvae of this group with ants may l'runus (Rosaceae). Libythea is obviously determine the food plant on which they closely related to Nymphalidae (Ehrlich, are found). The scanty food plant records 1958), as has recently been confirmed by for this group are thus sufficiently diverse a quantitative study of adult internal to suggest that further studies of food anatomy (Ehrlich, unpubl.) . plants will be of considerable interest. The Lycaenidae.-An enormous group, the most salient feature is the occurrence of family Lycaenidae may be larger even than Hameaerini and Abisariti on Myrsinaceae the Nymphalidae. Lycaenidae are in gen- and Primulaceae, two closely related fami- eral poorly known from the standpoint of lies that are fed on by very few other food plants (Downey, 1962). Our discus- butterflies. sion is based largely on the classification Lycaeninae likewise consist of three of Clench ( 1955 and pers. comm.) . Nothing tribes. Of these, Leptinii~iare African and is known of the life history of the Peruvian feed on lichens, some of them (Durbania, Styx injernalis, only member of Sty,'ainae. Durbaniopsis, and Durbaniella) even on Of the two remaining subfamilies, Riodin- the low crustose lichens that grow on rocks. inae, divided into three tribes, will be Liphyrini, almost entirely confined to the discussed first. Euselasia (Euselasiini) has Old LVorld tropics, are predaceous on been recorded from Mammea (Guttiferae) aphids, coccids, ant larvae, membracids. and three genera of Myrtaceae. The Old and jassids. There are no reliable records World Hamearini consist of three genera, of phytophagy in this group. with Dodona and Zemeros on Maesa The largest of the three tribes, Lycaenini, (Myrsinaceae) and Hanzearis on Primula presents a bewildering array of forms that (Primulaceae). The two plant families are can be separated only informally at present. very closely related, with hlyrsinaceae be- Many of these larvae are closely associated ing primarily tropical and woody, Primula- with and tended by ants, and this associa- ceae primarily temperate and herbaceous. tion may modify their food plant relation- The third and largest tribe, Riodinini, is ships (Downey, 1962; Stempffer, pers. divided into four subtribes. Abisara comm.) . For the large Plebejus group (the (Abisariti) feeds, like the Hamearini, on "blues"), we have records of the food Myrsinaceae; Theope (Theopiti) is on plants of 45 genera, and 33 of these are Theobroma (Sterculiaceae) ; and Helicopus known to feed, at least in part, on Legumi- (Helicopiti) is one of two members of the nosae. Records of special interest in this subfamily known to feed on a monocoty- group include Xacaduba on several genera ledon, in this case Montrichardia (Araceae). of hlyrsinaceae and Agriades on Primula- The remaining genera of Riodinini are in ceae; in this way they are like Hamaerini the exclusively New World Riodiniti, with and Abisariti of Riodininae. Chilades and very few records for a great many species. A7eopithecops are recorded from Rutaceae. Plant families represented are Acanthaceae, Four genera (Philotes, Scolitantides, Tali- Anacardiaceae, Aquifoliaceae, Chenopodia- cada, and Tongeia) are known to feed, at ceae, Compositae, Euphorbiaceae, Legumi- least in part, on Crassulaceae. Catachry- nosae, lloraceae, Myrtaceae, Polygonaceae, sops pandava feeds not only on Wagatea Ranunculaceae (Clematis), Rosaceae, Ruta- and Xylia (Leguminosae) but also on ceae, Sapindaceae, and Sapotaceae. De- Cycas revoluta (Cycadaceae), a cycad to serving special mention are the records of which it does harm in gardens. Hemiargus Cariomathus and Rhetus from Lorantha- ceraunus feeds on Marantaceae. Although ceae, Ll'apaea nepos from Oncidium (Orchi- most species of Jamides feed on Legumino- BUTTERFLIES AND PLANTS 597 sae, J. alecto feeds on Zingiberaceae (a The morphologically diverse South Ameri- monocotyledon). can group referred to "Theclan is also In the Strymon group (Clench in Ehrlich extraordinarily diverse in its choice of and Ehrlich, 1961, plus Strywzonidia), food plants: Bromeliaceae, Celastraceae, there is no obvious pattern, but there are Compositae, Euphorbiaceae, Leguminosae, several records of interest: Dolymorpha Liliaceae, Malpighiaceae, Malvaceae, Sapo- on Solanum (Solanaceae; Clench, unpubl.) ; taceae, Solanaceae, and Ulmaceae. In ad- Eurnaeus, with E. debora on both Dioon dition to Callophrys, already mentioned, edule (Cycadaceae) and Amaryllis (Lilia- many distinctive and in some cases large ceae) and E. atala on both Manihot (Eu- genera feed primarily on Loranthaceae and phorbiaceae; Comstock, unpubl.) and on the closely related Santalaceae: Charana, Za?nia integrifolia (Cycadaceae) ; Strymon Deudorix, Hypochrysops, Iolaus s. str. melinus, which feeds on a variety of (also often on Ximenia, Olacaceae), Ogyris, dicotyledonous plants, but also on the Pretapa, Pseudodipsas, Rathinda, and flowers of iyolina (Liliaceae) ; and Tmolus Zesius. 'It would appear that the epiphytic echion, which feeds not only on Lantana mistletoes and their relatives have consti- (Verbenaceae), Cordia (Boraginaceae), tuted an important adaptive zone for a Datura and Solanum (Solanaceae), Hyptis number of genera of Lycaeninae (as sug- (Labiatae), and Mangifera (Anacardia- gested by Clench, pers. comm.). Some ceae), but also on Ananas (Bromeliaceae). species of Iolaus are on Colocasia (Ara- The impressive pattern of food plant radia- ceae) . Olacaceae, Loranthaceae, and San- tion among the four subgenera of Callophrys talaceae are presumably closely related deserves special mention, for subg. Cal- (Hutchinson, 1959), and interestingly share lophrys and Incisalia feed mostly on angio- some acetylinic fatty acids (Sgirensen, sperms-Leguminosae, Polygonaceae, Rosa- 1963) and lipids (Shorland, 1963). Chliaria ceae, and Ericaceae-but three species of feeds on the buds and flowers of a number Incisalia have switched to conifers, feeding of genera of Orchidaceae, and Eooxylides, on Picea and Pinus (Pinaceae). A third Loxura, and Yasoda feed on Smilax, a subgenus, Mitoura, feeds primarily on an- hard-leaved member of Liliaceae, and the other group of conifers, Cupressaceae, with superficially similar Diosco~ea(Dioscorea- two species surprisingly on the pine mistle- ceae). Artipe lives inside the fruits of toes, Arceuthobium (Loranthaceae) . Fi- Punica (Punicaceae) ,and Bindahara inside nally, Callophrys (Sandia) nzacfarlandi, the the fruits of Salacia (Celastraceae). Fi- only species of its group, feeds on the nally, Aphnaezds inhabits galleries hollowed flowers of iyolina (Liliaceae) in the south- out by ants in the twigs of Acacia (Legumi- western United States. nosae), where it feeds on fungi (van Son, Lycaena and Heliophorus, closely re- pers. comm.) ! lated, feed primarily on Polygonaceae In summary, the plant families that are throughout the nearly cosmopolitan but best represented among the food plants of largely extratropical range of both groups. Lycaenini are Ericaceae, Labiatae, Poly- The theclines, narrowly defined (ShirBzu gonaceae, Rhamnaceae, and Rosaceae. and Yamamoto, 1956), have recently been Other records from families not hitherto treated by Shir6zu (1962), who has dem- mentioned are: Aizoaceae, Amarantha- onstrated that Fagaceae are the most im- ceae, Araliaceae, Betulaceae, Boraginaceae, portant food plants, with a number of Bruniaceae, Burseraceae, Caprifoliaceae, genera associated with Oleaceae. One genus Caryophyllaceae, Chenopodiaceae, Cista- (Shirozua) has become predaceous on ceae, Combretaceae, Convolvulaceae, Co- aphids. riaraceae, Cornaceae, Diapensiaceae, Dip- Among the remaining genera of Lycaeni- terocarpaceae, Ebenaceae, Eleagnaceae, nae, a few points are especially noteworthy. Gentianaceae, Geraniaceae, Hamamelida- 598 PAUL R. EHRLICH AND PETER H. RAVEN ceae, Juglandaceae, Lauraceae, Lecithyda- patterns, but as Merz (1959, p. 181) ceae, Lythraceae, Meliaceae, Melianthaceae, points out, we should nevertheless assume Myricaceae, Oxalidaceae, Pittosporaceae, that they probably do have a chemical Plantaginaceae, Plumbaginaceae, Protea- basis. ceae, Rubiaceae, Saxifragaceae, Sterculia- Now let us consider the groups of orga- ceae, Styracaceae, Symplocaceae, Theaceae, nisms utilized as food by butterfly larvae, Thymeleaceae, and Zygophyllaceae. As starting with the most unusual diets. Two before it must be borne in mind that many tribes of Lycaeninae have departed com- of these listings represent single records pletely from the usual range of foods: only; for example, the widespread Holarctic Liptenini feed on lichens, Liphyrini are Celastrina argiolus has been recorded from carnivorous. Many other Lycaeninae, how- food plants belonging to at least 14 fami- ever, are tended by ants and in some cases lies of dicotyledons. Nonetheless, it should the larvae are brought into the ant nests. be evident that the pattern is very different It would seem to be a relatively small step from that of Nymphalinae, the only sub- for such'larvae to switch and feed on the family comparable to Lycaeninae in size. ant grubs or fungi present in these nests. A number of species of the group ex- hibit well-developed cannibalism (Downey, 1962). Several Lycaenini, such as Shirozua, What generalities can be drawn from are carnivorous, and at least one species of these observed patterns? We shall ap- Aphnaeus feeds on fungi in ant galleries. proach this question from the standpoint of These transitional steps suggest the evolu- the utilization of different plant groups by tionary pathways to the most divergent of butterflies and see what light this throws butterfly larval feeding habits. on patterns of evolution in the two groups. Among those groups of butterflies that Butterflies, of course, are only one of many feed on plants, none is known to feed on phytophagous groups of organisms affect- bryophytes or on Psilopsida, Lycopsida, or ing plant evolution. Sphenopsida, nor is any known from ferns. Within the appropriate ecological frame- In fact, very few insects feed on ferns at work, our view of the immediate potentiali- all (cf. Docters van Leeuwen, 1958)) a ties of studies of phytophagy has been most surprising and as yet unexplained stated clearly and succinctly by Bourgogne fact with no evident chemical or mechanical (1951, p. 330)' who, speaking of the pat- basis. At least one genus of moths, Papai- terns of food plant choice in Lepidoptera, perna, is known to feed on ferns, however said: "Ces anomalies apparentes peuvent (Forbes, 1958). parfois dkmontrer l'existence, entre deux There are a few groups of butterflies that vCgCtaux, d'une affinitC d'ordre chimique, feed on gymnosperms. Two genera of quelquefois m&me d'une parenti systima- Lycaenidae (Catachrysops; Eurnaeus, two tique. . . ." Thus, the choices exercised by species) feed on Cycadaceae, but all three phytophagous organisms may provide ap- proximate but nevertheless useful indica- species involved also feed on angiosperms. tions of biochemical similarities among- ~J~eophasia(Pierinae) and three species of groups of plants. These do not necessarily Callophrys subg. Incisalia (Lycaeninae) indicate the plants' overall phenetic or feed on Pinaceae, while Callophrys subg. phylogenetic relationships. The same can Mitoura feeds on Cupressaceae (and also ie said of the choice of arrow poisons by on Arceuthobiurn, Loranthaceae, a mistle- primitive human groups (Alston and toe that grows on pines). It is well es- Turner, 1963, p. 293) and of patterns of tablished that Cupressaceae and Pinaceae parasitism by fungi (Saville, 1954). In are chemically quite distinct (Erdtman, many of these cases, biochemists have not 1963, p. 120). Judging from the taxonomic yet worked out the bases for the observed distance between these butterfly groups, it BUTTERFLIES AND PLANTS 599

can be assumed that butterflies feeding on appearance of dicotyledons, as yet undated gymnosperms had ancestors that fed on but surely pre-Cretaceous, must have ante- angiosperms. dated the evolutionary radiation that pro- An overwhelmingly greater number of duced the modern lines of diversification in butterfly larvae feed on dicotyledons than Lepidoptera and specifically in Papilionoi- on monocotyledons. The only two groups dea. All utilization of foods other than primarily associated with monocotyledons dicotyledons by butterfly larvae (and prob- are Satyrinae and Morphinae, closely re- ably by any Lepidoptera) is assumed to be lated subfamilies of Nymphalidae. One the result of changes from an earlier pat- genus of morphines (Mo~pho)is more tern of feeding on dicotyledons. often associated with dicotyledons, but we In general, the patterns of utilization by can think of no way to determine whether butterflies of dicotyledonous food plants this represents a switch from previous show a great many regularities. Certain monocotyledon feeding. No member of relationships are very constant: the plants Papilionidae, Pieridae, or Libytheidae is are usually fed upon by a single, phenetically known to feed on monocotyledons, but in coherent group of butterflies or several Nymphalidae and Lycaenidae numerous very closely related groups. As examples genera do so in whole or in part. Among we have the Aristolochiaceae-feeding Papil- the Nymphalidae, several species of Char- ionidae; Pierinae on Capparidaceae and axes, one of Acraea, one of A7ymphalis, and Cruciferae ; Ithomiinae on Solanaceae ; two of Euphaed~a(A'ajas) are known to Danainae on Apocynaceae and Asclepiada- feed on monocotyledons; all of these genera ceae; Acraeinae, Heliconiini, and Argyn- and some of the same species feed on dicot- nini on Passifloraceae, Flacourtiaceae, yledons also. In Lycaenidae, a number of Violaceae, and Turneraceae; and Riodini- very diverse groups, including at least two nae-Hamearini and Abisariti on Myrsina- qenera of Riodininae (Helicopis on Ara- ceae and Primulaceae. In many of these ceae, A'apasa on Orchidaceae) and 11 cases, the broad patterns observed probablj~ genera of Lycaeninae (Jamides on Zingi- support suggestions of overall phenetic beraceae; Zolaus on Araceae; Tmolus on similarity among the plants utilized and Bromeliaceae; Chliaria on Orchidaceae; among the groups of butterflies concerned. Heiizia~gus on Marantaceae; and Cal- Other clustering on the basis of food plant lophrys, Eooxylides, Eumaeus, Strymon, choice like that of Ulmaceae, Urticaceae, "Thecla," and Yasoda on Liliaceae) feed Moraceae, and Euphorbiaceae by certain on monocotyledons. Representatives of groups of Nymphalidae, probably also re- many of these genera and in some cases the flect phylogenetic relationship among the same species feed on dicotyledons also. - plants concerned. In several instances, the This pattern strongly suggests that butter- patterns of food plant choice of butterfly flies of two families have switched to groups underscore the close relationship monocotyledons from dicotyledons in a between certain sets of tropical woody and number of independent lines (probably at temperate herbaceous families elaborated least 18). by Bews (1927). Examples are Danainae, A corollary to the observations presented feeding interchangeably on Apocynaceae above is that the diversity we see in modern and Asclepiadaceae; Pierinae, on Cappari- butterflies has been elaborated against a daceae and Cruciferae; and part of Riodini- dicotyledonous background. Indeed this is nae, on RIyrsinaceae and Primulaceae. In probably true for Lepidoptera as a whole the first case, the families are generally (cf. Forbes, 1958). The dominant themes thought to be closely related. On the other in this particular coevolutionary situation hand, Hutchinson (1959) widely separated are therefore of considerable interest. We the members of the second and third pairs conclude from this relationship that the of plant families in his system, but this 600 PAUL R. EHRLICH AND PETER H. RAVEN disposition is considered inappropriate by Patterns of food plant utilization provide almost all botanists since it is based upon evidence bearing on the relationship of his primary division of flowering plants Araliaceae and Umbelliferae. Some groups into woody and herbaceous lines. In mak- of Papilioninae, normally associated with ing such decisions based on larval food Rutaceae, feed interchangeably on Umbel- plants, it must be remembered that we are liferae, or in some cases have switched en- dealing only with an indirect measure of tirely to this family. As we have seen, these biochemical similarity. For example, Pieri- two plant families are chemically similar. nae not only feed on Capparidaceae and But Araliaceae are close relatives of tTm- Cruciferae, which most botanists would belliferae (Rodriguez, 1957) despite their agree are closely related, but also on Sal- wide separation in the system of Hutchin- vadoraceae, which contain mustard oil son (1959), and are common in many re- glucosides but otherwise seem totally dif- gions where Papilioninae feed on Rutaceae. ferent from Capparidaceae and Cruciferae. Despite this, there is not a single record of More equivocal cases likewise occur. For a papilionid butterfly (indeed very few example, Pierinae feed on Tropaeolaceae. butterflies of any kind) feeding on Aralia- Not only do Tropaeolaceae share mustard ceae. An even more interesting relationship oil glucosides with Capparidaceae and Cru- hinges on the suggestion that the three sub- ciferae, they likewise have in common the families of Umbelliferae-Apioideae, Hy- rare fatty acid, erucic acid. Can we, with drocotyloideae, and Saniculoideae-may Alston and Turner (1963, p. 287), dismiss represent three phylogenetic lines, derived this as coincidence, or do these groups of independently from a group like the present- plants have more in common than is gen- day Araliaceae. All records of Papilioninae erally assumed? Finally, is there biochemi- from Umbelliferae are concerned with cal similarity between Loranthaceae-Santa- Apioideae, and indeed Dethier ( 1941) laceae and Capparidaceae-Cruciferae, both found that Umbelliferae-feeding papilionine common food plants of different groups of larvae refused HycEroiotyle (Hydrocoty- Pierinae (and of the genus Hesperocharis). loideae). This very strongly suggests that Whatever conclusions are drawn about biochemical analysis may go far in elucidat- the biochemical affinities of plants from ing relationships within the Araliaceae-Um- the habits of phytophagous or parasitic belliferae complex, and that the chemical organisms, little or no weight should be properties generally ascribed to Umbel- given to individual records. This is true liferae as a whole may be characteristic not only because of the numerous sources only of one subfamily, Apioideae. Aralia- of error enumerated earlier, but also be- ceae and Umbelliferae are known to share cause of the multiple explanations possible certain distinctive fatty acids (Alston and for such switches. For example. Atella Turner, 1963, p. 121) and acetylinic com- (Nymphalinae) feeds on Flacourtiaceae pounds (S@rensen,1963), and it might be and Salicaceae, among other plants. It is very instructive to see how these were quite possible that these two families are distributed in Umbelliferae outside of fairly closely related, despite the greatly Apioideae. reduced anemophilous flowers of the latter In further evaluating the patterns of (Thorne, pers. comm.). But to assume food plant choice in butterflies, it is im- that the few records involved indicate portant to consider those plant families, biochemical similarity between the groups especially dicotyledons, which are absent would be an unwarranted extension of the or very poorly represented. One outstand- data; it would be far simpler and safer at ing group is that partly characterized by that point to make comparative investiga- RSerz (1959, p. 169) as ('Sphingidpflanzen" tions of the biochemistry of the two plant -plants fed on by moths of the family families. Sphingidae. These include, among others, BUTTERFLIES AND PLANTS 60 1

Onagraceae, Lythraceae, Balsaminaceae, (rich in alkaloids), Myrtaceae, Polemonia- I'itaceae, Rubiaceae, and Caprifoliaceae. ceae, Ranunculaceae (rich in alkaloids), The first two, and probably the third and and Theaceae. In addition, very few but- fourth, are generally regarded as fairly terflies feed on Centrospermae, a group closely related. Each one of the first five characterized both by its morphological and families (Rubiaceae only in part) is char- biochemical traits (summary in Alston and acterized by the abundant presence of Turner, 1963, p. 141-143, 276-279). This raphides, bundles of needlelike crystals of group includes such large families as calcium oxalate (see discussion in Gibbs, Amaranthaceae, Cactaceae, Caryophylla- 1963). In a very interesting experiment, ceae, Chenopodiaceae, Nyctaginaceae, and Merz (1959) offered mature leaves of Portulacaceae. Although no biochemical Vitis (Vitaceae) to larvae of Pterogon basis for this lack of utilization is known at proserpina (Sphingidae) . Young larvae ate present, one probably exists. For a fanlily these leaves, their pointlike bites falling such as the enormous Compositae, poorly between the clusters of raphides. Older represehted among the food plants of but- larvae, which make large slashing bites, terflies, the explanation may lie either in could not avoid the raphides and did not their chemical composition or largely extra- eat the leaves. After the raphides were dis- tropical distribution, or most likely a com- solved in very dilute hydrochloric acid, the bination of these. One prominent family of leaves were accepted by larvae of all sizes. monocotyledons that is practically unrepre- Although it cannot be proven that some sented among butterfly food plants is other chemical repellent was not removed Araceae (Helicopis, Riodininae, and spe- by this treatment, it is obvious that cies of lolaus, Lycaeninae, are exceptions). raphides offer considerable mechanical One can conclude only that at least some difficulty for phytophagous insects. A of the plant groups enumerated above have number of families of moths other than chemical or mechanical properties that Sphingidae feed on this same series of plant render them unpalatable to butterfly larvae. families (Forbes, 1958). Thus far the combination of circumstances Rubiaceae, one of the families mentioned permitting a shift into the adaptive zones above, is perhaps the most prominent fam- represented by these groups has not oc- ily that is nearly absent from the records of curred. The assumption that such a shift butterfly food plants. Probably the third is theoretically possible is strengthened by largest family of dicotyledons, with nearly the observation that nearly every one of 10,000 species, it is, like the butterflies these plant groups is fed upon by one or themselves, mostly tropical. One can only more families of moths. speculate that some chemical factor, per- haps the rich representation of alkaloids, sharply restricts the ability of butterfly A systematic evaluation of the kinds of larvae to feed on plants of this family. In plants fed upon by the larvae of certain this respect, the similarities between the subgroups of butterflies leads unambigu- alkaloids of Apocynaceae (which however ously to the conclusion that secondary have milky juice) and Rubiaceae are of plant substances play the leading role in interest. Other dicotyledonous families determining patterns of utilization. This that are very poorly represented or not seems true not only for butterflies but for represented at all among butterfly food all phytophagous groups and also for those plants include Begoniaceae, Bignoniaceae, parasitic on plants. In this context, the Boraginaceae, Celastraceae, Cornaceae, irregular distribution in plants of such Curcurbitaceae (with curcurbitacins, bitter- chemical compounds of unknown physio- tasting terpenes), Gesneriaceae, Hydro- logical function as alkaloids, quinones, phyllaceae, Loasaceae, Menispermaceae essential oils (including terpenoids) , gly- 602 PAUL R. EHRLICH AND PETER H. RAVEN cosides (including cyanogenic substances to the success of the larvae of Chlosyne and saponins), flavonoids, and even raph- harrisii. Similarly, in western Colorado, the ides (needlelike calcium oxalate crystals) is density of the small plants of Lomatiurn immediately explicable (Dethier, 1954; eastwoodiae (Umbelliferae) is an important Fraenkel, 1956, 1959 ; Lipke and Fraenkel, factor limiting population size in Papilio 1956; Thorsteinson, 1960; Gordon, 1961). indra (T. and J. Emmel, pers. comm.). In Angiosperms have, through occasional these, and many similar situations, it is mutations and recombination, produced a logical to assume that genetic variants able series of chemical compounds not directly to utilize another food plant successfully related to their basic metabolic pathways would be relatively favored. This advan- but not inimical to normal growth and de- tage would be much enhanced if genotypes velopment. Some of these compounds, by arose that permitted switching to a new chance, serve to reduce or destroy the food plant sufficiently novel biochemically palatability of the plant in which they are that it was not utilized, or little utilized, produced (Fraenkel, 1959). Such a plant, by herbivores in general. protected from the attacks of phytophagous The degree of physiological specialization animals, would in a sense have entered a acquired in genetic adjustment to feeding new adaptive zone. Evolutionary radiation on a biochemically unusual group of plants of the plants might follow, and eventually would very likely also act to limit the what began as a chance mutation or recom- choice of food available to the insect group bination might characterize an entire in the general flora (Merz, 1959, p. 187; family or group of related families. Phy- Gordon, 1961). As stressed by Brower tophagous insects, however, can evolve in (1958a), moreover, close relationships be- response to physiological obstacles, as tween insects and a narrow range of food shown by man's recent experience with plants may be promoted by the evolution commercial insecticides. Indeed, response of concealment from predators in relation to secondary plant substances and extreme to a single background. The food plant nutritional imbalances and the evolution of provides the substrate for the larvae, not resistance to insecticides seem to be in- just their food (Dethier, 1954, p. 38). timately connected (Gordon, 1961). If a After the restriction of certain groups of recombinant or mutation appeared in a insects to a narrow range of food plants, population of insects that enabled indi- the formerly repellent substances of these viduals to feed on some previously pro- plants might, for the insects in question, tected plant group, selection could carry become chemical attractants. Particularlj~ the line into a new adaptive zone. Here it interesting is the work of Thorsteinson would be free to diversify largely in the (1953), who found that certain mustard absence of competition from other phytoph- oil glucosides from Cruciferae would elicit agous animals. Thus the diversity of feeding responses from larvae that fed on plants not only may tend to augment these plants if these glucosides were smeared the diversity of phytophagous animals on other, normally unacceptable, leaves. (Hutchinson, 1959), the converse may also But if these glucosides were smeared on be true. the alkaloid-rich leaves of Lycopersicu~~z Changes in food plant choice would be (Solanaceae), the larvae still refused them. especially favored in situations where the Similarly, Sevastopulo (pers. comm.) was supply of the "preferred" plant is suffi- unable to induce the larvae of Danais ciently limited to be an important factor in chrysippus to eat anj~thingbut Asclepiada- the survival of the larvae. Such situations ceae even by smearing the leaves of other have been described by Dethier (1959), plants with the juice of Calotropis (As- who showed that the density of Aster clepiadaceae). urnbellatus (Compositae) plants was critical This illustrates clearly that the choice of BUTTERFLIES AND PLANTS 603 a particular food plant or of a spectrum of nini, which feed on the same plants as food plants may be governed by repellents Heliconiini and Acraeinae, do not plaj~any present in other plants (Thorsteinson, prominent role as models for mimetic 1960) as well as by attractants in the normal forms. This may be in part because Argyn- food plants; this fully accords with the nini are best developed in temperate re- model outlined above. It should not, how- gions, where butterflies and therefore ever, be assumed without experimental mimetic complexes are less common. It verification that a particular secondary may further be suggested that the mimicry plant substance is an attractant or feeding supposed to exist between dark female stimulant for the insects feeding on plants forms of various species of Speyeria and that contain it. Indeed, for the beetle the model Battus plzilenor may well be a Leptinotarsa, the alkaloids of the Solana- case of Miillerian rather than Batesian ceae on which it feeds serve as repellents mimicry. Indeed, the results of Brower (summary in Fraenkel, 1959, p. 1467- (1958) with the model Danais plexippus 1468). and its inimic Limenitis archippus suggest In view of these considerations, we pro- that there is in fact no sharp line between pose a comparable pattern of adaptive Batesian and Miillerian mimicry. These radiation for each of the more or less should be thought of as the extremes of a strictly limited groups of butterflies enu- continuum. Thus Speyeria females may be merated above. It is likewise probable that somewhat distasteful but can only acquire the elaboration of biochemical defenses has warning coloration when the selective bal- played a critical role in the radiation of ance is tipped by the presence of other those groups of plants characterized by distasteful forms. This follows logically unusual accessory metabolic products. from numerous experiments and observa- Further, it can be pointed out that all tions on the behavior of predators with groups of butterflies that are important in mimetic complexes, which rarely reveal an furnishing models in situations involving "either/or7' type of response (cf. Swynner- mimicry are narrowly restricted in food ton, 1919). Assuming a balance of this plant choice: Papilioninae-Troidini; Itho- sort would resolve some of the difficulties miinae ; Nymphalinae-Heliconiini ; Acraei- of interpretation concerning the two kinds nae; and Danainae. This is in accordance of mimicry (e.g., Sheppard, 1963, p. 145). with a long-standing supposition of natu- Numerous unusual feeding patterns scat- ralists and students of mimicry that the tered among butterfly families attest to the physiological shifts that enabled the butter- frequency of radiation into new groups of fly groups to feed on these plants conferred food plants. We can, however, only guess a double advantage by making the butter- at the probability of future radiation in the flies in question unpalatable. These groups new adaptive zone. For example, does of butterflies have-been selected for warn- Stalachtis susanae (Riodininae), which is ing coloration, and once established, this known to feed on Oxypetalurn campestre conspicuousness would tend to put any- (Asclepiadaceae) in Argentina, represent thing that would maintain their distasteful- the start of a new phylogenetic series of ness at a selective premium. butterflies restricted to this group of Conversely, those groups of butterflies plants? Probably not, since the examples that furnish most of the Batesian mimics- of this sort of unusual feeding habit today Papilioninae-Papilionini and Graphiini ; far exceed the number of radiations ob- Satyrinae; Nymphalinae except Heliconiini ; served in the past. Nevertheless, the close Pierinae; and Dismorphiinae-feed mostly patterns of coadaptation we have discussed on plant groups that are shared with other above must have started in a similar dissiinilar groups of butterflies. It is some- fashion. Comparable patterns can be what surprising that Nymphalinae-Argyn- found among plants with biochemical in- 604 PAUL R. EHRLICH AND PETER H. RAVEK novations, for example, Senecio viscosus, cannot accept the theoretical picture of a mentioned earlier, or the relationship of generalized group of polyphagous insects Seduln acre to other species of its genus from which specialized oligophagous forms (Merz, 1959, p. 160). were gradually derived. Just as there is no In viewing present-day patterns of food truly "panphagous7' insect (cf. Fraenkel, plant utilization, however, the historical 1959), so there is no universally acceptable aspect of the situation must not be ne- food plant; and this doubtless has always glected. A biochemical innovation might been true. This statement is based on the have had a considerable selective advantage chemical variation observed in plants and for a group of plants in the Cretaceous. the physiological variation observed in in- Such an advantage would, of course, have sects. Leguminosae are important food been in terms of the phytophagous animals plants for several groups of Lycaenidae and and parasites present in the Cretaceous, Pieridae, and woody Ranales are well repre- and not necessarily those of the present sented among the food plants of Papilioni- day. The crossing of an adaptive threshold dae and Nymphalidae; but this should not by a member of a living group of phytoph- be taken to prove that these groups of agous animals would have an entirely plants are "inert" chemically or readily different significance now than that which available to other phytophagous groups of it would ha"e had in the Cretaceous. insects. The initial radiation of butterfly For example, even though a species of taxa onto these groups may for a time have Stalachtis is able to feed on Asclepiadaceae, produced a pattern just as spectacular as, it shares the available supply of these for example, the close association between plants with representatives of numerous Troidini and Aristolochiaceae seen today. other groups of phytophagous animals. We hold that plants and phytophagous in- These have, somewhere in the course of sects have evolved in part in response to geological time, acquired the ability to feed one another, and that the stages we have on asclepiads. Further, if the phytophagous postulated have developed in a stepwise organisms switching early to milkweeds manner. became protected from predators by their As suggested by Fraenkel (1956, 1959). ingestion of distasteful plant juices, this secondary plant substances must have been initial advantage might have been overcome formed early in the history of angiosperms. in the intervening years by corresponding At the present day, many classes of organic changes in prospective predators. A species compounds are nearly or quite restricted to of bird long selected to like milkweed bugs this group of plants (for example, see Al- might find milkweed-feeding Stalachtis a ston and Turner, 1963, p. 164: Harborne, gourmet's delight. 1963, p. 360; Paris, 1963, p. 357). We As in the occupation of any adaptive suggest that some of these compounds may zone, the first organisms to enter it have a have been present in early angiosperms and tremendous advantage and are apt to have afforded them an unusual degree of pro- the opportunity to become exceedingly di- tection from the phytophagous organisms verse before evolution in other organisms of the time, relative to other contemporary sharply restricts their initial advantage. In plant groups. Behind such a biochemical short, the nature of any adaptive zone is shield the angiosperms may have developed altered by the organisms that enter it. and become structurally diverse. Such an From our vantage point in time we view assumption of the origin of angiosperms only the remnants, doubtless often disar- provides a cogent reason why one of many ranged if not completely shattered by sub- structurally modified groups of gymno- sequent events, of the great adaptive sperms would have been able to give rise radiations of the past. to the bewildering diversity of modern In view of these considerations, we angiosperms, while most other lines became BUTTERFLIES AND PLANTS 605 extinct. It seems at least as convincing to degree of plasticity of chemoreceptive re- us as do theories based on the structural sponse and the potential for physiological peculiarities of angiosperms. Although the adjustment to various plant secondary sub- chemical basis for the success of early stances in butterfly populations must in angiosperms may no longer be discernible, large measure determine their potential for it can be mentioned that woody Ranales, evolutionary radiation. Of secondary, but generally accepted as the most "primitive" still possibly major importance, are me- assemblage of living angiosperms on other chanical plant defenses, and the butterflies' grounds, are as a group characterized by responses to them. many alkaloids as well as by essential oils. With respect to the second question on Of course this might also be interpreted the generation of predictions the answer only to mean that the development of also seems clear. We cannot predict the alkaloids has permitted this group to per- results of any given interaction with pre- sist despite its many generalized features. cision-Stalachtis on Asclepiadaceae or In turn, the fantastic diversification of Neophasia on pines may or may not form modern insects has developed in large the basis for further patterns of radiation. measure as the result of a stepwise pattern On the other hand, the basis for a prob- of coevolutionary stages superimposed on abilistic statement of "Further radiation the changing pattern of angiosperm varia- unlikely" seems to have been developed. tion. With specific reference to the butter- A great many minor predictions can be flies, one is tempted in terms of present-day made, such as the probable presence of patterns to place more emphasis on the full alkaloids in Ptaeroxylon (Meliaceae) , the exploitation of diurnal feeding habits by solanaceous character of the food plants of the adults than on the penetration of any unknown larvae of Ithomiinae, and so particular biochemical barrier by the lar- forth. vae. On the other hand, phenetic relation- Although the data we have gathered per- ships suggest that Papilionoidea (with mit us to make some reasonable sequence Hesperioidea, the skippers) are representa- predictions about phylogenetic patterns tives of a line that is amply distinct from (e.g., diversification of Apocynaceae and all other living Lepidoptera (Ehrlich, Solanaceae before Danainae and Ithomiinae, 1958). Thus it is entirely possible that respectively), these predictions cannot be radiation onto a new food plant was de- tested and the relationships cannot be cisive at the time Papilionoidea first specified further in the absence of a fossil diverged, even though the feeding habits record. The reconstruction of phylogenies of the order as a whole are now much wider on the basis of this sort of information than those of the butterflies alone. The would seem an unwarranted imposition on impossibility of deciding objectively which the data, since evolutionary rate and time groups of Papilionoidea are more primitive are still inseparable. than others (cf. Ehrlich, 1958, p. 334- In response to the fourth question, it 335) relegates the task of identifying the seems to us that studies of coevolution original group of food plants for butterflies provide an excellent starting point for un- to the realm of profitless speculation. derstanding community evolution. Indeed We would like to return now to the four the seeming ease with which our conclusions general questions posed at the beginning have been extended to include the complex of this paper. First, what have we learned interactions among plants, phytophagous of the reciprocal responses of butterflies organisms, mimics, models, and predators and their food plants? The observed pat- leads us to believe that population biolo- terns clearly point to the critical importance gists should pursue similar studies of other of plant biochemistry in governing the re- systems. Many examples come to mind lationships between the two groups. The such as parasitoid-caterpillar, Plasmodium- 606 PAUL R. EHRLICH AND PETER H. RAVEN hemoglobin, tree-mycorrhizal fungus, in however, also be mentioned that the rela- which stepwise reciprocal selective response tively permissive tropical climate presum- is to be expected. Studying most of these ably allows a greater diversity of plant life systems experimentally tends to be diffi- forms and therefore secondarily of animals cult, and may be complicated by lack of (Hutchinson, 1959, p. 150). repeatability in the results. Probably our most important overall con- An approach to biology that is concerned clusion is that the importance of reciprocal with broad patterns quite possibly will lead selective responses between ecologically to a better understanding of some other closely linked organisms has been vastly problems of community ecology. For ex- underrated in considerations of the origins ample, biologists have long been interested of organic diversity. Indeed, the plant- in the reasons for the differences in species herbivore "interface" may be the major diversity between tropical and temperate zone of interaction responsible for generat- areas. An important factor in maintaining ing terrestrial organic diversity. these differences may be the sort of syner- gistic interactions between plants and her- bivores we have been discussing. The The reciprocal evolutionary relationships selective advantage of living in a tropical of butterflies and their food plants have climate is evident for insects, which are been examined on the basis of an extensive poikilothermal. Insects are much more survey of patterns of plant utilization and abundant in the tropics than elsewhere and information on factors affecting food plant doubtless constitute the major class of choice. The evolution of secondary plant herbivorous animals. The penetration of substances and the stepwise evolutionary relatively cold environments or other en- responses to these by phytophagous orga- vironments requiring diapause is probably nisms have clearly been the dominant fac- a rather recent occurrence in most insect tors in the evolution of butterflies and groups. That these environments are not other phytophagous groups. Furthermore, always readily entered is attested to by the these secondary plant substances have repeated failure of insects such as the probably been critical in the evolution of butterfly Mestra anzy-mone to survive the angiosperm subgroups and perhaps of the winter in localities at the northern fringes angiosperms themselves. The examination of their ranges where summer colonies have of broad patterns of coevolution permits been established (Ehrlich and Ehrlich, several levels of predictions and shows 1961). promise as a route to the understanding of The abundance of phytophagous insects community evolution. Little information in tropical regions would be expected to useful for the reconstruction of phylogenies accentuate the pace of evolutionary inter- is supplied. It is apparent that reciprocal selective responses have been greatly under- actions with plants. These interactions may rated as a factor in the origination of have been the major factor in promoting organic diversity. The paramount impor- the species diversity of both plants and tance of plant-herbivore interactions in animals observed in the tropics today. As generating terrestrial diversity is suggested. this diversity was being produced, it be- For instance, viewed in this framework the came arrayed in richly varied mixtures of rich diversity of tropical communities may species with relatively great distances be- be traced in large part to the hospitality tween individuals of any one plant species. of warm climates toward poikilothermal As Grant (1963, p. 420-422) has suggested, phytophagous insects. this arrangement would have the additional LITERATURECITED advantage of providing a maximum degree ALSTON,R. E., AND B. L. TURNER. 1963. Bio- of protection from epidemic outbreaks of chemical systematics. Prentice-Hall, Engle- plant diseases and plant pests. It must, wood Cliffs, N. J. BUTTERFLIES AND PLANTS 607

RARRETT,C., AND A. N. BURNS. 1951. Butterflies taxonomy. Academic Press, London, p. 167- of Australia and New Guinea. N. H. Seward, 186. Melbourne. FORBES,W. T. M. 1958. Caterpillars as bota- REWS, J. Mr. 1927. Studies in the ecological nists. Proc. Tenth Int. Congr. Ent., 1: 313- evolution of the angiosperms. New Phytol., 317. 26: 1-21, 65-84, 129-148, 209-248, 273-294. FRAENKEL,G. 1956. Insects and plant bio- BOURGOGNE,J. 1951. Ordre des LCpidopt&res. chemistry. The specificity of food plants for Trait4 Zool., 10: 174-448. insects. Proc. 14th Int. Congr. Zool., p. 383- RROWER,JANE VAN ZANDT. 1958. Experimental 387. studies of mimicry in some North American -. 1959. The raison d'etre of secondary plant butterflies. Part I. The monarch, Danazls substances. Science, 129: 1466-1470. plezippus, and viceroy, Limenitis archippzls GIBBS, R. D. 1963. History of chemical tax- archippus. EVOLUTION,12: 3247. onomy. In Swain, T., ed., Chemical plant BROWER,L. P. 1958a. Bird predation and food taxonomy. Academic Press, London, p. 41- plant specificity in closely related procryptic 88. insects. Amer. Nat., 92: 183-187. GORDON,H. T. 1961. Nutritional factors in in- . 1958b. Larval food plant specificity in sect resistance to chemicals. Ann. Rev. butterflies of the Papilio glazlczls group. Entom., 6: 27-54. Lepidop. News, 12: 103-114. GRANT,V. 1963. The origin of adaptations. CHERMOCK,R. L. 1950. A generic revision of Columbia University Press, New York and the Limenitini of the world. Am. Midl. Nat., London. 43: 513-569. HARBORNE,J. B. 1963. Distribution of antho- CLENCH, H. K. 1955. Revised classification of cyanins in higher plants. In Swain, T., ed., the butterfly family Lycaenidae and its allies. Chemical plant taxonomy. Academic Press, Ann. Carnegie Mus., 33: 261-274. London, p. 359-388. CORBET,A. S., AND H. M. PENDLEBURY.1956. HEGNAUER,R. 1963. The taxonomic significance The butterflies of the Malay Peninsula. Ed. 2. of alkaloids. In Swain, T., ed., Chemical plant Oliver and Boyd, London. taxonomy. Academic Press, London, p. 389- COSTALIMA, A. M. DA. 1936. Terceiro Catalogo 427. dos Insectos que vivem nas plantas do Brasil. HUTCHINSON,G. E. 1959. Homage to Santa Directoria de Estatistica da Produc~ZoSec~Zo Rosalia why are there so many kinds of de Publicade, Rio de Janeiro, p. 201-231. or animals. Amer. Nat., 145-159. DETHIER, V. G. 1941. Chemical factors deter- 93: mining the choice of food plants by Papilio HUTCHINSON,J. 1959. The families of flowering larvae. Amer. Nat., 75: 61-73. plants. Ed. 2. 2 vols. Clarendon Press, Oxford. ---. 1954. Evolution of feeding preferences in JORGENSEN, P. 1932. Lepidopterologisches aus phytophagous insects. EVOLUTION,8: 33-54. Siidamerika. Deutsch. Ent. Zeitschr. Iris, -. 1959. Food-plant distribution and density Dresden, 46: 37-66. and larval dispersal as factors affecting insect KLOTS, A. B. 1933. A generic revision of the populations. Canad. Entom., 91: 581-596. Pieridae (Lepidoptera) . Entom. Amer. n.s., DOCTERSVAN LEEUWEN,W. M. 1958. Zooceci- 12: 139-242. dia. In Verdoorn, F., ed., Manual of pteridol- LEE, C. L. 1958. Butterflies. Academia Sinica ogy. Martinus Nijhoff, The Hague, p. 192- (in Chinese). 195. LIPKE, H., AND G. FRAENKEL.1956. Insect nu- DOWNEY, J. C. 1962. Host-plant relations as trition. Ann. Rev. Ent., 1: 1744. data for butterfly classification. Syst. Zool., MERZ, E. 1959. Pflanzen und Raupen. Uber 11: 150-159. einige Prinzipien der Futterwahl bei Gross- - AND W. C. FULLER. 1961. Variation in schmetterlingsraupen. Biol. Zentr., 78: 152- Plebejzls icarioides (Lycaenidae) . I. Food 188. plant specificity. J. Lepidop. Soc., 15: 3442. MICHENER, C. D. 1942. A generic revision of EHRLICH, P. R. 1958. The comparative mor- the Heliconiinae (Lepidoptera, Nymphalidae) . phology, phylogeny and higher classification of Am. Mus. Novitates, 1197: 1-8. the butterflies (Lepidoptera: Papilionoidea) . MUNROE,E. 1960. The generic classification of Univ. Kansas Sci. Bull., 39: 305-370. the Papilionidae. Canad. Ent., Suppl., 17: -AND A. H. EHRLICH. 1961. How to know 1-51. the butterflies. Wm. C. Brown, Dubuque. --- AND P. R. EHRLICH. 1960. Harmonization ERDTMAN,H. 1963. Some aspects of chemotax- of concepts of higher classification of the onomy. In Swain, T., ed., Chemical plant tax- Papilionidae. J. Lepid. Soc., 14: 169-175. onomy. Academic Press, London, p. 89-125. PARIS, R. 1963. The distribution of plant gly- FL~~cK,H. 1963. Intrinsic and extrinsic factors cosides. In Swain, T., ed., Chemical plant affecting the production of secondary plant taxonomy. Academic Press, London, p. 337- products. In Swain, T., ed., Chemical plant 358. 608 P.4UL R. EHRLICH AND PETER H. RAVEN

PINHEY,E. C. G. 1949. Butterflies of Rhodesia. acids in plant lipids. In Swain, T., ed., Chemi- Rhodesia Sci. Association, Salisbury. cal plant taxonomy. Academic Press, London, PRICE,J. R. 1963. The distribution of alkaloids p. 253-303. in the Rutaceae. In Swain, T., ed., Chemical S~RENSEN,N. A. 1963. Chemical taxonomy of plant taxonomy. Academic Press, London, p. acetylinic compounds. In Swain, T., ed.. 429-452. Chemical plant taxonomy. Academic Press. KELZINGTON,C. L. 1952. The biology of nearctic London, p. 219-252. Lepidoptera I. Food plants and life-histories SWYNNERTON,C. M. F. 1919. Experiments and of Colorado Papilionoidea. Psyche, 59: 61-70. observations bearing on the explanation of REUTER, E. 1896. ~berdie Palpen der Rho- form and colouring, 1908-1913, Africa. J. paloceren. Acta Soc. Sci. Fennica, 22: i-xvi, Linn. Soc. Zool., 33: 203-385. 1-577. THORNE,R. F. 1963. Some problems and guid- RODR~CUEZ,R. L. 1957. Systematic anatomical ing principles of angiosperm phylogeny. Amer. studies on Myvrhidendron and other woody Nat., 97: 287-306. Umbellales. Univ. Calif. Publ. Bot., 29: 145- THORSTEINSON,A. J. 1953. The chemotactic 318, pls. 36-47. responses that determine host specificity in an SAFWAT,FUADM. 1962. The floral morphology oligophagous insect (Plzltella nzaczllipennis of Secamone and the evolution of the pollinat- (Curt.) Lepidoptera) . Canad. J. Zoo1 , 31: ing apparatus in hsclepiadaceae. Ann. Missouri 52-72. Bot. Gard., 49: 95-129. ---. 1960. Host selection in phytophagous in- SAVILLE,D. B. 0. 1954. The fungi as aids in sects. Ann. Rev. Ent., 5: 193-218. the taxonomy of flowering plants. Science, >ah. SON, G. 1949. The butterflies of southern 120: 583-585. Africa. Part I, Papilionidae and Pieridae. SCITZ, A. 1906-1927. The macrolepidoptera of Part I1 (1955), Danainae and Satyrinae the world. Vols. 1, 5, 9, 13. Fritz Lehman VAZQUEZG., LEONIL$,AND PEREZ R., H. 1961. Verlag and Alfred Kerner Verlag, Stuttgart. Observaciones sobre la biologia de Baronia SHEPPARD,P. M. 1963. Some genetic studies of bvevicornis Salv. (Lepidoptera: Papilionidae- Miillerian mimics in butterflies of the genus Baroniinae). An. Inst. Biol. Mex., 22: 295- Heliconizls. Zoologica, 48: 145-154, pls. 1-2. 311. SHIR~ZU,T. 1962. Evolution of the food-habits VERSCHAEF~ELT,E. 1910. The cause determining of larvae of the thecline butterflies. Ty6 to the selection of food in some herbivorous in- Ga (Trans. Lepidop. Soc. Japan), 12: 144- sects. Proc. Acad. Sci., Amsterdam, 13: 536- 162. 542. --- AND H. YAMARIOTO.1956. h generic re- WILTSHIRE, E. P. 1957. The Lepidoptera of vision and the phylogeny of the tribe Theclini Iraq. Rev. ed. Kicholas Kaye, London. (Lepidoptera: Lycaenidae) . Sieboldia, 1: 329- \~'YNTER-BLYTH,M. A. 1957. Butterflies of the 421. Indian region. Bombay Nat. Hist. Soc., SHORLAND,F. B. 1963. The distribution of fatty Bombay.