Pollination by Lemurs and Marsupials:An Archaic Coevolutionary System
Robert W. Sussman and Peter H. Raven
The earliest insect-pollinated angio-sperms seem to have been visited andpollinated predominantly by insects (1).Angiosperms in which the flowers pro-duce large quantities of nectar can becross-pollinated effectively only by ani-mal visitors with high energetic require-ments (2). We therefore presume thatangiosperm flowers with these character-istics evolved only after the appearanceof small, nectar-feeding birds, bats, andnonflying mammals that fed at least sea-sonally on nectar and pollen. Pollinationsystems involving hummingbirds andhawkmoths, which are more specializedgroups of flower visitors with high ener-
Birds. The phylogenetic relationships,distribution, and history of perchingbirds suggest that they did not originatebefore the end of the Cretaceous (5), al-though some suboscine families mightconceivably antedate the Paleocene.Most, if not all, modem families ofperching birds are Neogene or morerecent in origin. Bock's (6) analysis ofthe evolution of the Hawaiian honey-creepers (Drepanididae), which heshows could have taken place within amillion years, is an important indicationof the rate with which frequent or evenobligate nectarivory could have origi-nated in geological time. The history of
Summary. Existing geographical and ecological relationships between bats, non-flying mammals, and birds that visit flowers for food suggest novel interpretations oftheir evolutionary history.
getic requirements, seem to be evenmore recent in origin. The hypothesisthat the flowers of the earliest angio-sperms did not produce large quantitiesof nectar and were not visited by verte-brates is strengthened by the almost totalabsence of such pollination systemsamong living angiosperms that arethought to be relatively unspecialized:woody Ranales, unspecialized Ham-amelidiflorae, and generalized mono-cotyledonous plants (3).
Vertebrate Pollination Systems:
An Overview
Birds, bats, and nonflying mammalsare the groups of vertebrates most in-timately associated with the flowers ofangiosperms at present. Although otherkinds of vertebrates, such as lizards (4),also visit nectar-rich flowers on occa-sion, it is the three groups mentionedabove that provide the great majority offlower visitors, and therefore are of thegreatest historical and functional signifi-cance. Each will now be reviewed in ahistorical and geographical context.
SCIENCE, VOL. 200, 19 MAY 1978
hummingbirds is particularly enigmatic,but if they are related to swifts, theyprobably evolved no earlier than the Eo-cene (7). Nectarivory in birds is a traitthat has originated many times indepen-dently, and there is no concrete evidencethat it was widespread until the Miocene,even though some existing flower-visit-ing groups of birds, such as the suboscinePhilepittidae of Madagascar, seem cer-tainly to be much older.On a geographical basis, nectarivory
is, as Lein (8) pointed out, well repre-sented throughout the tropics and South-ern Hemisphere and virtually absent inthe Palearctic region. Its presence inNorth America is mainly accounted forby the geologically recent invasions bythe hummingbirds, doubtless of SouthAmerican origin.
Bats. Although the fossil record ofbats is poor (9), it is generally assumedthat the modern attributes of bats had al-ready been obtained by the end of theEocene (10). The existence of a well-pre-served Early Eocene fossil that com-bines to some extent the characteristicsof the two living suborders of bats (11) isin accordance with this hypothesis. Bats,
therefore, have probably contributed tothe diversification of angiosperms forless than 50 million years, and presum-ably were absent for the first 75 millionyears of the evolution of the group (12).Megachiroptera are known to have ap-peared in Europe by the Oligocene (13,14). It therefore seems reasonable to as-sume that certain members of this sub-order may have been feeding regularlyboth on flower parts and on fruits (14)and contributing to the diversification offlowering plants by the end of the Eo-cene.By the close of the Eocene and the be-
ginning of the Oligocene, many modernfamilies of Microchiroptera appeared inthe fossil record in Europe and NorthAmerica (13). The large family Phyllos-tomatidae, the leaf-nosed bats, is knownfrom the Late Miocene (15). This sug-gests that the subfamily Glossophaginae,one of the most specialized of flower-vis-iting bats and the most numerous, beingwidespread over the warmer parts of theNew World, is very unlikely to be olderthan the Miocene (14). Members of an-other subfamily, the West Indian Phyllo-nycterinae, likewise feed frequently onflowers and presumably are Miocene inorigin also. Most of the Phyllostoma-tidae, including these two specializedsubfamilies, eat a varied diet includinginsects, fruit, and flowers (16). At leastsome glossophagine bats also obtain aportion of their nitrogen from pollen (17),and they certainly are not strict nectar-feeders. Individual species do not seemto be restricted to particular plants intheir flower-visiting habits [for example,see (18)].On a geographical basis, glossopha-
gine bats are common and intimately as-sociated with the flowers of plants ofmany families throughout the tropics ofthe New World (19). In Africa, the small,specialized pteropids that are the mostreliable flower visitors are largely con-fined to the forested area in the west, towhich the only African representativeof the Macroglossinae, Megaloglossuswoermanni, is likewise confined (20). Inthe more widespread savannah regionsof Africa, fruit-eating and flower-visit-ing bats, including species of Eidolon,Epomophorus, and Rousettus, are fairlycommon locally and seasonally, but theirrole in the pollination of flowering plantsis less well understood than that of batsin the New World or West Africa (20,21).
In Madagascar, there are only three
R. Sussman is an assistant professor in the De-partment of Anthropology, Washington University,St. Louis, Missouri 63130, and P. Raven is directorof the Missouri Botanical Garden, St. Louis 63110.
species of fruit-eating or flower-visitingbats: Pteropus rufus, Eidolon helvum,and Rousettus madagascariensis (22).Pteropus rufus is found throughout theisland and is mainly frugivorous; it visitsflowers occasionally, but is regarded asdestructive to Ceiba flowers in planta-tions in Southeast Asia (23). Eidolon hel-vum is found in rain forests in easternMadagascar. Its diet and foraging behav-ior are similar to those of Pteropus rufus(24). Rousettus madagascariensis, anendemic species, is restricted to a smallportion of the rain forest in the easternparts of the island (20, 22).
In tropical Asia there are 11 genera ofpteropid bats, two of which-Eonycterisand Macroglossus-obtain most of theirfood from flowers (25); in New Guineathere are eight genera, three of which-Macroglossus, Syconycteris, and Melo-nycteris-are flower bats (23, 26). InAustralia, a similar, but depauperate,fauna is found in the tropical extremenorth. Three species of Pteropus occursouth to the forests of New South Wales,as does Syconycteris australis, anotherblossom-feeder (27). The occurrence ofplant-feeding bats in Australia presum-ably dates from the Late Miocene ormore recently (28). Another macroglos-sine genus, Notopteris, is endemic inwestern Polynesia.
Flower-visiting bats in the Old World,therefore, seem to be well represented intropical Asia through New Guinea andwestern Polynesia, fairly well represent-ed in West Africa, and less well repre-sented in East Africa and tropical to sub-tropical Australia. Flower-visiting batsare also poorly represented, and bywidespread genera only, in Madagascar.This pattern suggests no great antiquityin the region.Although bat-pollination is not repre-
sented among the more archaic of the liv-ing angiosperms, it, together with bird-pollination, seems to be of early originand establishment in Myrtaceae (espe-cially the more advanced subfamily Myr-toideae). Increased stamen number andflower size in Myrtaceae, linked withbird-pollination by Stebbins (29) andCarlquist (30), are often regarded asprimitive. It is probable that such polli-nation systems have existed in Myr-taceae since Paleogene time, but theywere probably not characteristic of theearliest members of this family.The small Paleotropical family Son-
neratiaceae (two genera, seven species)appears to be almost exclusively bat-pol-linated. Its large flowers and numerousstamens presumably evolved in relationto this system of pollination. Largely be-cause of these features, this family is of-
732
ten regarded as the most primitive of theMyrtales. Such a phylogenetic positionmust be doubted, however, unless it canbe confirmed by independent evidence.Myrtales and the family Myrtaceae itselforiginated before the close of the Cre-taceous (28, 31), whereas the excellentfossil record of Sonneratiaceae extendsback only to the Lower Miocene (31, 32),approximately the period when bats mayhave begun to visit flowers regularly forfood in the Old World.Nonflying mammals. A recent review
(33) documents the extent of our knowl-edge of the interactions between non-flying mammals and plants. In it, Rourkeand Wiens review a number of observa-tions of such mammals as rats and squir-rels visiting flowers, but also point outsome relationships that appear muchmore significant. They argue com-pellingly that the inflorescences andflowers of certain members of Pro-teaceae and Myrtaceae in Australia areadapted to pollination by small marsu-pials, a suggestion first made by Porsch(34). They also show that rodents regu-larly visit and may pollinate the in-florescences of certain species of Pro-teaceae in South Africa and perhaps alsoin Australia [see also (35)]. In addition,as they and earlier authors pointed out,the so-called honeypossum, Tarsipesspencerae, seems clearly to haveevolved in relation to such a source offood.Many primates feed on flowers or
parts of flowers at times, but the effecton the flowers is usually destructive [forexample, see (36)]. Cebus monkeys maydrink water or nectar, or both, from treesof Ochroma pyramidale without destroy-ing the flowers (37), and might partici-pate in their pollination. This is, how-ever, the only well-documented case ofan anthropoid which might be acting asan important pollinating agent. In Mada-gascar, however, the relation betweenlemurs and flowering plants appears tobe a significant one. All of the diurnalMalagasy lemurs are mainly vegetarianand many spend a small proportion oftime feeding on flowers and are destruc-tive to them. In contrast, the followingsix species of nocturnal lemurs havebeen observed to feed regularly on flow-ers during at least a portion of the year,and are therefore of particular interest interms of their possible significance inpollination systems.
1) Lemur mongoz [average weight,2530 grams (38)]. It was reported (39)that 84 percent of the observed feedingof this species during the dry season wason nectar licked from the flowers of Fer-nandoa madagascariensis (Kigelianthe)
and Combretum phaneropetalum, bothred-flowered and normally thought to bebird-pollinated; the flowers of the in-troduced Ceiba pentandra (40); and theextrafloral nectaries of the native Huracrepitans. The animals moved rapidlyfrom inflorescence to inflorescence andvisited many trees in a single night.
2) Lepilemur mustelinus [weight, 600to 900 g (38)]. During the driest part ofthe year, flowers ofAlluaudia ascendensand A. procera constituted a major por-tion of the diet; the entire flower waseaten so the animals were evidently noteffective as pollinators (41).
3) Microcebus murinus [50 to 150 g(38)]. Martin (42) reported that this smalllemur ate the flowers ofBrexia madagas-cariensis, Rubus roridus, Uapaca sp.,and Vaccinium emirnense during the dryseason, allowing the petals to fall to theground. In contrast, one of us (43), mak-ing observations in the dry season atAmpijoroa, observed the animals lickingnectar from the flowers of the introducedCeiba pentandra without damaging theflowers.
4) Microcebus coquereli [about 400 g(38)]. Petter (44) reported this species tofeed on flowers, fruits, gums, and in-sects.
5) Cheirogaleus medius [150 to 400 g(38)]. During the first part of the rainyseason, the main part of the diet appearsto be flowers, although the animals alsoeat insects. An important source of nec-tar was a native species, Delonix flori-bunda (44, 45).
6) Phaner furcifer [about 450 g (38)].This species utilized both Crateva gre-veana and Adansonia sp. as a source ofnectar at the start of the wet season, lick-ing the flowers and moving from clusterto cluster. Phanerfurcifer is also knownto feed on gums, insects, and fruits (46).Of the four remaining species of noc-
turnal lemurs, one-Daubentonia mada-gascariensis-has a very specializeddiet, and the feeding habits of the re-maining three-Avahi laniger, Cheiro-galeus major, and Allocebus trichotis-have not been studied. In general, how-ever, it can be asserted that nocturnal le-murs do play a significant role in the pol-lination of certain plant species in Mada-gascar.Another group of unspecialized noc-
turnal primates, the bush babies of themainland of Africa, has been reported tovisit the flowers of Adansonia for nectarand probably other food material as wellwithout destroying them (47). The spe-cies involved, Galago crassicaudatus[weight, 1000 to 1250 g (38)], has alsobeen reported to feed on fruit, insects,and gum (48). A second species of Ga-
lago, G. senegalensis [weight, 225 to 300g (38)], was reported by Doyle (48) tovisit different flowering trees in succes-sion; whether they were actually seekinginsects, as he assumed, remains to bedemonstrated.
In summary, certain marsupials inAustralia, rodents in Australia and SouthAfrica, lemurs in Madagascar, and ga-lagos in continental Africa regularly visitnectar-rich flowers and lap their nectaras an important source of food, espe-cially during the dry season. Almost allof these animals are nocturnal, and near-ly all feed in part on insects, fruits, and,in some cases, gum. In general, they donot appear to compete directly withflower-visiting birds, which are exclu-sively diurnal, for food (49). The livingmarsupials of South America do not in-clude forms that regularly visit flowers oreat fruits and that might conceivably beimplicated in pollination systems (50). Itwould be of great interest to know moreabout food habits of the single nocturnalNew World primate, Aotus, which niightfeed on and pollinate flowers.
Is There a Class of Flowers Adaptedto Pollination by Nonflying Mammals?
This question, first raised in a compre-hensive manner by Porsch (34), has beenanswered in the affirmative, with newevidence, by Rourke and Wiens [(33);see also (35)]. Whether, as Porsch sup-posed, certain Myrtaceae and Pro-teaceae of Australia fall into such a class,remains to be proved. Nevertheless, theexistence of certain species with unusu-ally large, strong flowers or in-florescences, relatively few flowers or in-florescences per plant, strong floralodors, and copious nectar, in areaswhere bat-pollination is absent or at bestsporadic, does point in this direction.For Madagascar, another region
where plants that bear flowers with thesecharacteristics exist, and where pollina-tion of flowers by bats is evidently veryrare, Jumelle and Perrier de la Bathie(51) reported that lemurs regularly vis-ited the nectar-rich flowers of Sym-phonia nectarifera, eating the leatherypetals and drinking the nectar. Certainlylemurs may be regular visitors to thelargest-flowered of the approximately 16species of Symphonia found in Mada-gascar (52), and might reasonably bethought to have participated in their evo-lution. Porsch (53) considers Symphoniain general to be bat-pollinated, butagrees with Perrier de la Bathie aboutthe probability of lemur pollination inthese species. He also suggests that19 MAY 1978
some Bombacaceae and Lecythidaceaemay be adapted for pollination by non-flying mammals.Other Mascarene plants that have
what appear to be suitable character-istics and that have been observed to bevisited by lemurs eating portions of theflowers or lapping the nectar includeAdansonia, Brexia madagascariensis,Crateva greveana, Delonix floribunda,Rubus roridus, Uapaca sp., and Vacci-nium emirnense. Lemurs have also beenobserved to feed on nectar from the ex-trafloral nectaries of the pantropicalHura crepitans and to visit and probablyefficiently pollinate the flowers of the in-troduced Ceiba pentandra. At the flow-ers of the normally bird-pollinated andred-flowered Combretum phaneropeta-lum and Fernandoa madagascariensisthe visits of lemurs were undoubtedlysecondary. Bats very often visit normal-ly bird-pollinated plants, such as Eryth-rina (54) and Spathodea (55) for nectar.Reports of lemurs devouring the entireflower ofAlluaudia should be confirmed,for there has been a tendency to viewtheir activities as mainly destructive tothe plants, and this has not been con-firmed by many recent observations.
In any event, in Madagascar, an islandwhere flower-visiting birds are frequentbut flower-visiting bats are rare, thereappears to be a strong circumstantialcase for the evolution of certain plantswith floral characteristics adapted to reg-ular visitation by and consequent polli-nation by lemurs. Just as in temperateAustralia, there is no reason to supposethat fruit-eating and flower-visiting batswere ever present in greater numbersthan at present, and therefore it seemswarranted to view the association be-tween certain species of plants and le-murs as one that is archaic, rather thanrecently derived. Such an endemic groupas Sarcolaenaceae, with some eight gen-era and 40 species of trees and shrubswith few, large, strong, presumably nec-tar-rich flowers, and large pollen shed intetrads (56), might conceivably be polli-nated by lemurs and might have beenpollinated by them since Paleogenetimes.
Contemporary Relationships
Birds, bats, and nonflying mammalsvisit and pollinate flowers regularly atthe present time. Pollination systems in-volving birds are well developedthroughout the tropics and the temperateregions of the Southern Hemisphere;most of the flowers birds visit are bright-ly colored, usually red, and odorless
(57). Systems involving bats and non-flying mammals usually include flowersthat are dull-colored and odorous; theyare almost invariably nocturnal, whereassystems involving birds are always diur-nal, as pointed out by Fenton and Flem-ing (49). With the exception of the spe-cialized flowers closely adapted for hum-mingbirds (and the analogous ones vis-ited by hawkmoths), many of the flowersvisited by birds are also visited by batsand nonflying mammals in regions whereflower-visiting members of these groupsare found. Aside from morphological ad-aptations that presumably evolved toprotect the ovules of the plants con-cerned (58), the flowers of such plantsare generally open, with copious nectar.
Pollination systems involving bats andthose which involve nonflying mammalsappear to have a reciprocal geographicaldistribution. Bat-pollination is commonand well developed throughout tropicalAmerica, Asia, and northern Austra-lasia, reasonably frequent in West Af-rica, less so in East Africa, and poorlydeveloped on Madagascar. Flower-vis-iting and fruit-eating bats migrate intotemperate regions seasonally, at timesreaching temperate Australia and SouthAfrica and the southern borders of theUnited States and of the Palearctic re-gion. Those systems that appear to in-volve nonflying mammals, on the otherhand, are evidently present in temperateSouth Africa, temperate Australia, andMadagascar-all regions where flower-visiting bats are rare, seasonal, or ab-sent.
Evolutionary Relationships
The only extensive areas where non-flying vertebrate pollinators seem to becommon are where there is a limitedplant-visiting bat fauna. Can we assumetherefore that competitive interactionsamong bats, prosimians, and phalangeridmarsupials may have occurred through-out their evolutionary history? Is thereany evidence that bats may have re-placed arboreal tropical forest-dwellingmarsupials and prosimians in certain re-gions during the Tertiary?Those vertebrate taxa that are most
frequent in the tropical forest canopyand most likely to be utilizing resourcessimilar to those used by prosimians andarboreal phalangerids are bats, birds,and tree squirrels. All bats are nocturnaland most birds are diurnal. Althoughmost mammals that dwell in tropical for-ests are nocturnal, anthropoid primatesand tree squirrels, except flying squirrelsand Aotus, are diurnal (59). Bats and
birds avoid competition for many of theirresources (especially insects and otherprey and nectar) simply by these dif-ferences in activity pattern. It seems
likely, however, that bats and nocturnalprosimians and marsupials have beenmajor or potential competitors through-out much of Tertiary time. The success
of bats in tropical Asia, Africa, andAmerica and the relative lack of success
of prosimians and arboreal marsupials inthese areas may be directly related.The great eutherian and marsupial ra-
diations on different continents in theearly Tertiary coincided with the radia-tion of many modern groups of angio-sperms with relatively specialized flow-ers (60, 61). A major mammalian radia-tion, including the origin and diversifica-tion of insectivore-like primates or
primate-like insectivores, the plesiada-poids (62-66; 67, pp. 415-433; 68, 69),occurred in Europe and North Americaduring the Paleocene. A similar diversifi-cation of polyprotodont marsupials oc-
curred at the same time in South Ameri-ca (61, 70). Also in South America, some
extinct groups of smaller marsupialsseem to have been omnivorous (for ex-
ample, Palaeothentinae, Abderitinae,Polydolopidae, and Caroloameghinii-dae), and some of these groups con-
verged markedly with living Australianphalangeroids, with some primategroups such as the fossil Carpolestidaefrom North America, and with living andfossil lemurs from the Old World (50).Some extinct marsupials in the Cre-
taceous and Paleogene of South Americamay therefore have been flower-visitorsand important in the pollination of plantsat those times. None of these animalshave persisted to the present. Recall thatthe flower-visiting bats of South Americaare believed to have originated by theMiocene: did they outcompete and re-
place earlier flower-visiting marsupialson plants that produced large flowerswith copious nectar? These plants may
have developed adaptations in thecourse of evolution that would have ex-
cluded the marsupials and favored vis-itation by the wider-ranging, highly spe-
cialized glossophagine bats. The batswould undoubtedly have brought about a
greater degree of outcrossing in suchplants (2, 71). A similar pattern charac-terizes the evolution of pollination bybirds in many plant groups. At any rate,this mammalian radiation included theinvasion of numerous plesiadapoids andother mammals into arboreal mixed-feeding adaptive zones (60, 61, 65, 70,72-74).A number of these early Tertiary
mammals have dental morphology which734
suggests convergent feeding adaptations.The tiny, mouse-sized Picrodontidaewere originally considered to be bats re-
lated to Phyllonycteris (75), a fruit- andnectar-feeding phyllostomatid bat, andthere is still some controversy as to theirrelationships (67, pp. 415-433; 76). Mostauthors suggest, however, that they are
primates and merely convergent to thephyllostomatid bats (65, 77). Szalay (71,p. 16) states: "Their peculiar, phyllo-stomatid bat-like dental adaptation mighthave been the result of selection for juicyfruit or nectar feeding diet" (78). Thecarpolestids, another mouse-sized familyof plesiadapoids, share dental featureswith certain multituberculates, with an
extinct group of New World marsupials,and with some living Australian marsu-
pials (65; 67, pp. 415-433; 72, 79). Theliving genus of marsupial of the same
size as the carpolestids, Burramys, feedson seeds, fruit, insects, and nectar (33,80). The dentitions of the Paleocene andEocene genus Phenacolemur closely re-
semble those of the marsupial sugar glid-er Petaurus (67, pp. 415-433; 68) thatfeeds on insects, insect larvae, smallbirds, buds, flowers, nectar, sap, andfruit juices (24). Kay and Cartmill (69),after a detailed study of the dentition of anumber of Paleocene paromomyids, con-
cluded that the genera Palaechthon andPlesiolestes were insectivorous but tooka small amount of fruit, nectar, or gum;and thatParomomys, Torrejonia, and es-
pecially Phenacolemur fed predomi-nantly on fruit, gum, or nectar. They alsofound that in the plesiadapid Chiro-myoides, the shearing features of the mo-lars appear somewhat reduced, suggest-ing a tendency toward fruit-eating. Thus,at least by the end of the Paleocene, allfour families of plesiadapoids (Picrodon-tidae, Carpolestidae, Paromomyidae,and Plesiadapidae) contained plant-vis-iting genera.
Concerning these convergent adapta-tions, Szalay (81, p. 33) states:
A previously not fully exploited primary foodsource of fruits and leaves, etc., becameprobably more and more utilized, sometimeduring the second half of the Cretaceous, byseveral groups of therians.... One of thesegroups was undoubtedly the early prosimians.In addition to the primates, mixodectids, apa-temyids, condylarths, the ptilodontid multi-tuberculates, the eutherian plagiomenids, andprobably the Cretaceous marsupial Glasbiuswere also occupants of the frugivore-herbi-vore-omnivore adaptive zone for small mam-mals.
Szalay (81) believes that the radiation ofthe plesiadapoids may have led to the de-cline and eventual extinction of many ofthe more primitive Cretaceous mam-
mals. A number of other authors have al-
so associated early extinctions with com-petitive exclusion related to early eu-therian radiations. However, the particu-lar patterns of Paleocene and Eoceneextinctions of some North and SouthAmerican marsupials and of certain taxaof plesiadapoids and Eocene primatesmay be, at least to some extent, relatedto the rapid evolution and radiation ofthe bats, and their exploitation of flowersand fruits as food.The North American marsupials de-
clined by the end of the Oligocene anddisappeared entirely by the end of theMiocene, as did the single line thatreached Europe (82). Although most ma-jor groups of South American marsupialspersisted into the Pliocene, the small,herbivorous marsupials such as theGroeberiidae, Abderitinae, and Palaeo-thentinae also became extinct by themiddle Miocene (61, 73), when flower-visiting bats were becoming well estab-lished. The plesiadapoids disappeared bythe Eocene, an unknown lineage givingrise to the "primates of modern aspect"(83) no later than the mid-Paleocene (64).These Eocene prosimians ranged in
size from about 60 to 1600 g (84), in thesize range from Microcebus to Lemurmongoz. In contrast to the plesiada-poids, they showed a number of modernprimate adaptations: sharp claws werereplaced by flattened nails overlying fric-tion pads; the toe and thumb were diver-gent and enlarged to produce effectivegrasping organs; and the two eyes con-verged toward the center of the face (64,85). These adaptations (at least the firsttwo) would have allowed the Eoceneprosimians far greater access to fruitsand flowers, as well as to many plant-vis-iting insects, making them much more ef-ficient at locomoting and foraging in thesmall terminal branches of bushes andtrees than were the plesiadapoids (63, 64,86). Since most small nocturnal primatesfeed on crawling insects, many of whichare captured on the ground (42, 74), andsince most are omnivorous and include alarge proportion of fruits or nectar, orboth, in their diet (especially during thedry season), it is probably this improvedability to feed in terminal branches thatwas the most important impetus for themajor adaptive shift seen in these Eo-cene primates. They then would becomesimply a more efficient version of theinsectivorous-frugivorous-nectarivorousplesiadapoids.By the Oligocene, however, these
early primates of modern aspect dis-appear worldwide, leaving only a few,mainly specialized genera of modernforms outside of Madagascar. By the lat-ter half of the Eocene, this group of
mixed-feeding prosimians gave rise tothe larger-bodied, diurnal, mainly folivo-rous-frugivorous anthropoids (63; 67, pp.415-433). If feeding in terminal brancheswas a very important, new adaptive zonefor a number of mammalian species dur-ing the Paleocene and Eocene, it is notunlikely that certain forms developedbetter adaptations to exploit this habitatthroughout these time periods. It is alsolikely that the early prosimians of theEocene were more efficient at exploitingterminal branches than were the plesia-dapoids, and it has been suggested (87,pp. 1-64) that competition between Pa-leocene primates and primitive bats mayhave been one of the initial causes of batflight. The Eocene seems to have been aperiod when bats and primates were re-fining their abilities to exploit the fruitsand flowers of angiosperms, as well asthe insects that also fed upon these re-sources. It may also have been a periodof intense competitive interaction be-tween these chiropterans and primates.As we have mentioned, the earliest
known fossil bat, Icaronycteris index,comes from early Eocene deposits inWyoming (11). Concerning this fossil,Jepsen (87, p. 12) wrote:
By the time in paleobiologic history that I. in-dex was a segment in the phylogeny of chi-ropts the whole vast galaxy of morphic, be-havioral, and ecologic characteristics that dis-tinguish bats from non-bat predecessors hadalready been achieved.... It indicates thatsome bats had already evolved almost to theirpresent grade of organization while horseswere the size of modem dogs and man's an-cestors were no larger than small monkeys.
By the early Eocene the evolution ofprimates of modern aspect was in itsvery early stages. The rapid develop-ment and distribution of bats in the Oldand New World tropics during the Eo-cene corresponds with the virtual ex-tinction of the plesiadapoids and the ini-tial development and later disappearanceof the first primates of modern aspect inEurope, North America, and presum-ably throughout the tropics. Jepsen (87,pp. 1-64) speculated that plesiadapoidand pre-bat competition led to a majoradaptive shift in chiropteran evolution.Be that as it may, it does seem possiblethat bats may well have influenced pri-mate evolution during the Eocene. Theycertainly may have been one of the majorforces contributing to the extinction ofmany kinds of prosimians by the end ofthis period.Such prosimians, and marsupials with
similar habits, have survived chieflywhere competition from flower-visitingand fruit-eating bats is limited or absentin Madagascar, South Africa, and tem-19 MAY 1978
perate Australia. The reasons for the ab-sence of such prosimians and marsupialsin the temperate regions of the NorthernHemisphere remain to be explored, al-though these regions have certainly beenan active site of evolution for many"dominant" groups of animals through-out geological history (15, 88). Survivalfor many groups of plants and animals,and we would suggest also for the uniqueearly Tertiary coevolutionary relation-ships between nonflying mammals andflowering plants, seems clearly to havebeen facilitated in the far-flung temperatelands of the Southern Hemisphere.
Jepsen (87, pp. 1-64), however, warnsthat speculations about the causes of batflight might be flights of fantasy, and wemight add that, by making speculationsabout the causes of Eocene primate ex-tinctions, we may be going out on a (ter-minal) limb.
Conclusion
Among the vertebrates, birds, bats,and nonflying mammals include speciesthat regularly visit the flowers of angio-sperms and have evolved in relation tothese habits. Flower-visiting birds arefrequent throughout the tropics and inthe south temperate zone; in the northtemperate region they are represented insome numbers only in North America,which the hummingbirds, a South Amer-ican group of problematical origin, in-vaded presumably in late Tertiary time.Many unrelated groups of birds visitflowers regularly and participate in theirpollination to greater or lesser degrees,but the oldest of these seem to date fromthe Cretaceous-Tertiary boundary, andthere is no strong evidence for wide-spread flower-visiting among birds untilthe Miocene. Although they often visitthe same species of plants, there appearsto be little evidence for competition be-tween birds, of which the flower visitorsare all diurnal, and either bats or non-flying mammals, with which they coexistwidely.
Bats seem to have originated in theEarly Tertiary and may have been vis-iting flowers as a source of food since theEocene. Fruit-eating and flower-visitingbats occur widely throughout the trop-ics. They are less well represented in Af-rica than in Asia or America, and rare inMadagascar and in temperate regionsgenerally. They may, however, visittemperate regions during the course ofseasonal migrations. Glossophagine andphyllonycterine bats, the two specializedgroups of flower-visiting Microchirop-tera, both restricted to the New World,
seem to be of Miocene origin, so that thehabit of eating fruits and visiting flowersfor food is probably older in the EasternHemisphere than in the Western Hemi-sphere.Among the nonflying mammals that
regularly visit flowers for food, the mostprominent groups are some of the pha-langeroid marsupials of Australia, ro-dents in Australia and South Africa, le-murs in Madagascar, and bush babies incontinental Africa. Where these relation-ships are well developed, they occur inareas in which flower-visiting bats areabsent or sporadic. Almost all of theserelationships involve nocturnal mam-mals, and there appears to have been asort of competitive exclusion on a grandscale involving flower-visiting bats.Although there are clearly bird-flow-
ers and bat-flowers, whether any flowershave characteristics that evolved specifi-cally in relation to nonflying mammalsremains to be seen. The kinds of largeflowers with copious nectar that are vis-ited by vertebrates in general obviouslydid not occur in the course of angio-sperm evolution until the end of the Cre-taceous, at the earliest, and thus duringthe first half of their evolution differentgroups of flowering plants were probablyvisited and pollinated chiefly by beetlesand by flies. Some were of course wind-pollinated. The kinds of nectar-rich flow-ers that evolved in certain groups of an-giosperms at about the start of the Ter-tiary doubtless facilitated outcrossing inthese plants, particularly in the kinds ofwidely separated populations that arecharacteristic of the tropics. Subse-quently throughout the Tertiary morespecialized forms of flowers, with flow-ers that structurally restrict the kinds ofvertebrates that are able to obtain nectarand are thus more efficient in directingoutcrossing, have evolved repeatedly.Hummingbirds, sunbirds, and plant-vis-iting bats have been especially importantin this context; the kinds of specializedflowers they visit have evolved from an-tecedents of many sorts, some insect-pollinated.On the basis of these relationships, we
suggest that there have been marsupialsvisiting the flowers of angiosperms forfood since uppermost Cretaceous times,but that this relationship, once wide-spread, has persisted only in temperateAustralia, where placental mammals arerecent arrivals. Fossil forms elsewherethat were similar to the flower-visitingarboreal phalangeroids of Australiamostly disappeared at about the end ofthe Paleocene in northern continents, ap-parently giving way to prosimian formssimilar in habits to them and to the living
lemurs of Madagascar. These early pri-mates also were widespread, but appar-ently replaced by bats with similar habitsand dentition from the Eocene onward inthe Old World, with the major replace-ment in South America probably delayeduntil the presumed diversification ofglossophagine bats in the Miocene. Pro-simians with this mode of life have per-sisted only in Madagascar, where fruit-eating and flower-visiting bats are andpresumably always have been rare, andto a lesser extent (Galago) in continentalAfrica. Relationships between rodentsand flowers have evolved more recentlyin areas such as South Africa and Ha-waii, where flower-visiting bats are ab-sent. Thus the coevolutionary relation-ships between marsupials and certainflowering plants in Australia as well asthose between lemurs and other flower-ing plants in Madagascar, appear to berelicts that have survived from ancienttimes, just as many archaic plants andanimals have persisted in these isolatedlands. The relationships themselves ap-pear to be "living fossils," which have agreat deal to tell us about the evolutionof both the mammals, including some ofour own antecedents, and of the flower-ing plants.
References and Notes1. H. G. Baker and P. D. Hurd, Annu. Rev. Ento-
mol. 13, 385 (1968).2. B. Heinrich and P. H. Raven, Science 176, 597
(1972).3. L. van der Pijl, Evolution 15, 44 (1961). There
seems to be no evidence to support the con-tention of J. Smart and N. F. Hughes [in InsectlPlant Relationships, H. F. van Emden, Ed.(Blackwell, Oxford, 1973), p. 151] that the in-crease in size of "flowers" of Bennettitales inthe Triassic was associated with visitation byflying vertebrates in place of insects.
4. I. Elvers, Bot. Not. 130, 231 (1977).5. A. Fedducia, Syst. Zool. 26, 19 (1976); personal
communication.6. W. J. Bock, Evolution 24, 704 (1970).7. , personal communication; S. L. Olson,
personal communication.8. M. R. Lein, Syst. Zool. 21, 135 (1972).9. B. H. Slaughter, in About Bats, B. H. Slaughter
and D. W. Walton, Eds. (Southern MethodistUniv. Press, Dallas, 1970), pp. 51-83; J. D.Smith, in Biology of Bats of the New WorldFamily Phyllostomatidae, R. J. Baker, J. K.Jones, Jr., D. C. Carter, Eds. (Texas Tech, Lub-bock, 1976), part 1, pp. 49-69.
10. K. F. Koopman and J. K. Jones, Jr., in AboutBats, B. H. Slaughter and D. W. Walton, Eds.(Southern Methodist Univ. Press, Dallas, 1970),pp. 22-28.
11. G. L. Jepsen, Science 154, 1333 (1966).12. Evidence derived from a comparison of proteins
in different groups suggests that bats may besomewhat older; V. M. Sarich and J. E. Cronin,in Molecular Anthropology, M. Goodman andR. E. Tashian, Eds. (Plenum, New York, 1976),pp. 141-170; V. M. Sarich, personal communi-cation.
13. J. K. Jones and H. H. Genoways, in About Bats,B. H. Slaughter and D. W. Walton, Eds. (South-ern Methodist Univ. Press, Dallas, 1970), pp. 3-21.
14. H. G. Baker, in Tropical Forest Ecosystems inAfrica and South America: A Comparative Re-view, B. J. Meggers, E. S. Ayensu, W. D. Duck-worth, Eds. (Smithsonian Institution, Washing-ton, D.C., 1973), pp. 145-159.
15. P. J. Darlington, Zoogeography: The Geograph-ical Distribution ofAnimals (Wiley, New York,1957).
16. T. H. Fleming, E. T. Hopper, D. E. Wilson,Ecology 53, 555 (1972); D. G. Howell and D.Curch, Rev. Biol. Trop. 21, 281 (1974); A. L.Gardner, in Biology of Bats of the New WorldFamily Phyllostomatidae, R. J. Baker, J. K.Jones, Jr., D. C. Carter, Eds. (Texas Tech, Lub-bock, 1977), part 2, pp. 293-350.
17. D. J. Howell, Comp. Biochem. Physiol. 48A,263 (1974).
18. __, Nature (London) 270, 509 (1977).19. S. Vogel, Flora 158, 289 (1969).20. R. L. Peterson, personal communication.21. T. A. Vaughan, personal communication.22. J. Dorst, Bull. Mus. Hist. Nat. (Paris) Ser. 2 19,
306 (1947); J. E. Hill, personal communication;K. F. Koopman, personal communication.
23. L. van der Pijl, Acta Bot. Neerl. 5, 133 (1956).24. E. P. Walker, Mammals of the World (Johns
Hopkins Press, Baltimore, ed. 2, 1968).25. A. N. Start and A. G. Marshall, in Tropical
Trees, Variation, Breeding and Conservation(Academic Press, London, 1976), pp. 141-150;E. Gould, "Foraging behavior of Malaysiannectar feeding bats," Biotropica, in press.
26. M. B. Fenton, personal communication.27. W. D. L. Ride, A Guide to the Native Mammals
of Australia (Oxford, Melbourne, 1970); andpersonal communication.
28. P. H. Raven and D. I. Axelrod, Science 176,1379 (1972).
29. G. L. Stebbins, Evol. Biol. 2, 101 (1967).30. S. Carlquist, Phytomorphology 19, 332 (1969).31. J. Muller, Biol. Rev. 45, 417 (1970).32. _ , in Ancient Pacific Floras. The Pollen
Story, L. M. Cranwell, Ed. (University Press ofHawaii, Honolulu, 1964), pp. 33-42.
33. J. Rourke and D. Wiens, Ann. Mo. Bot. Gard.64, 1 (1977).
35. F. L. Carpenter, "Hooks for mammal polli-nation?", Oecologia, in press.
36. K. Glander, in World Anthropology: Socioecol-ogy and Psychology of Primates, R. H. Tuttle,Ed. (Mouton, The Hague, 1975), pp. 37-57; P.Waser, in Primate Ecology, Studies of Feedingand Ranging Behavior in Lemurs, Monkeys andApes, T. H. Cutton-Brock, Ed. (AcademicPress, London, 1977), pp. 183-222; T. Strusha-ker, in Recent Advances in Primatology, vol. 1,Behaviour, C. M. Hladik and D. J. Chivers,Eds. (Academic Press, London, in press).
37. J. R. Oppenheimer, thesis, University of Illi-nois, Urbana (1968); C. M. Hladik, personalcommunication.
38. R. F. Kay, thesis, Yale University (1973).39. R. W. Sussman and I. Tattersall, Folia Prima-
tol. 26, 270 (1976).40. H. G. Baker, Inst. Internat. Stud. Univ. Califor-
nia Berkeley Res. Ser. 9, 185 (1965).41. C. M. Hladik and P. Charles-Dominique, in
Prosimian Biology, R. D. Martin, G. A. Doyle,A. C. Walker, Eds. (Duckworth, London, 1974).
42. R. D. Martin, Z. Tierpsychol. Beih. 9, 43 (1972).43. R. W. Sussman, in Recent Advances in Prima-
tology, vol. 2, Evolution, C. M. Hladik and D.J. Chivers, Eds. (Academic Press, London, inpress).
44. J.-J. Petter, in Recent Advances in Primatology,vol. 1, Behaviour, C. M. Hladik and D. J. Chi-vers, Eds. (Academic Press, London, in press).
45. A. Hladik, personal communication.46. J.-J. Petter, A. Schilling, G. Pariente, in Lemur
Biology, I. Tattersall and R. W. Sussman, Eds.(Plenum, New York, 1975), pp. 209-218.
47. M. J. Coe and F. M. Isaac, East Afr. Wildl. J. 3,123 (1965).
48. G. A. Doyle, in Behavior of Nonhuman Pri-mates, A. M. Schrier and F. Stollnitz, Eds. (Ac-ademic Press, New York, 1974), pp. 155-353.
49. M. B. Fenton and T. H. Fleming, Biotropica 8,104 (1976).
50. L. G. Marshall, personal communication.51. H. Jumelle and C. Perrier de la Bathie, Ann.
Mus. Colon Marseilles 18, 452 (1910).52. H. Perrier de la Bathie, in Flore de Madagascar
et de Comores (Firmin-Didot, Paris, 1951), 136efamille, pp. 1-96.
53. 0. Porsch, Biol. Gen. 11, 171 (1935).54. P. H. Raven, Lloydia 40, 401 (1977).55. E. S. Ayensu, Ann. Mo. Bot. Gard. 61, 702
(1974).56. S. Cariquist, Brittonia 16, 231 (1964).57. P. H. Raven, Evolution 26, 674 (1972).58. V. Grant, ibid. 4, 179 (1950).
59. P. Charles-Dominique, in Phylogeny of the Pri-mates, W. P. Lucket and F. S. Szalay, Eds.(Plenum, New York, 1975), pp. 69-88.
60. J. A. Lillegraven, Annu. Rev. Ecol. Syst. 5, 263(1974).
61. W. A. Clemens and L. G. Marshall, FossiliumCatalogus I: Animalia 123, 6 (1976).
62. There is some controversy concerning the tax-onomic status of this group; see R. D. Martin,Philos. Trans. R. Soc. London Ser. B 264, 295(1972); Cartmill (63, 64); Szalay (65, 66); Simons(67); Gingerich (68); and Kay and Cartmill (69).
63. M. Cartmill, in The Functional and EvolutionaryBiology of Primates, R. Tuttle, Ed. (Aldine-Atherton, New York, 1972), pp. 97-122.
64. __, Science 184, 436 (1974).65. F. S. Szalay, in The Functional and Evolution-
ary Biology ofPrimates, R. Tuttle, Ed. (Aldine-Atherton, New York, 1972), pp. 3-35.
66. __, Folia Primatol. 23, 158 (1975).67. E. L. Simons, in Prosimian Biology, R. D. Mar-
tin, G. A. Doyle, A. C. Walker, Eds. (Duck-worth, London, 1974), pp. 415- 433.
68. P. D. Gingerich, in ibid., pp. 531-541.69. R. F. Kay and M. Cartmill, J. Human Evol. 6,
19 (1977).70. W. A. Clemens, Evolution 22, 1 (1968).71. B. Heinrich, Annu. Rev. Ecol. Syst. 6, 139
(1975).72. G. G. Simpson, J. Mammal. 14, 97 (1933).73. _ _, Bull. Am. Mus. Nat. Hist. 19, 1 (1948);
Bull. Mus. Comp. Zool. Harv. Univ. 139, 1(1970); B. Patterson and R. Pascual, in Evolu-tion, Mammals, and Southern Continents, A.Keast, F. C. Erk, B. Glass, Eds. (State Univ. ofNew York Press, Albany, 1972), pp. 247-309.Use and chronology of South American landmammal ages follow L. G. Marshall, R. Pascual,G. H. Curtis, R. E. Drake [Science 195, 1325(1977)].
74. G. A. Doyle, in Behavior of Nonhuman Pri-mates, A. M. Schrier and F. Stollnitz, Eds. (Ac-ademic Press, New York, 1974), pp. 155-353;M. Fogden, in Prosimian Biology, R. D. Martin,G. A. Doyle, A. C. Walker, Eds. (Duckworth,London, 1974), pp. 151-165; P. Charles-Domi-nique, in ibid., pp. 131-150; in Ecology and Be-havior of Nocturnal Primates (Columbia Univ.Press, New York, 1977).
75. W. D. Matthew, Bull. Am. Mus. Nat. Hist. 37,569 (1917).
76. J. Schwartz and L. Kristalka, Ann. CarnegieMus. 46, 55 (1977).
77. G. G. Simpson, Bull. Am. Mus. Nat. Hist. 105,411 (1955); L. Van Valen, Nature (London) 207,435 (1965); M. C. McKenna, Coll. Int. Cent.Nat. Rech. Sci. Probl. Actuels Paleontol. 163,603 (1967).
78. There is also an African Miocene megachiropte-ran, Propotto, that is so primate-like that it wasoriginally described as a lorisine by Simpson; A.C. Walker, Nature (London) 223, 647 (1969).
79. K. D. Rose, Bull. Mus. Comp. Zool. Harv.Univ. 147, 1 (1975).
80. H. Dimpel and J. H. Calaby, Victorian Nat. 89,101 (1972).
81. F. S. Szalay, Evolution 22, 19 (1968).82. M. C. McKenna, personal communication.83. E. L. Simons, Primate Evolution (Macmillan,
New York, 1972).84. W. K. Gregory, Mem. Am. Mus. Nat. Hist. n.s.
3, 49 (1920); H. J. Jerison, Evolution of the Brainand Intelligence (Academic Press, New York,1973).
85. W. E. Le Gros Clark, Antecedents of Man(Edinburgh Univ. Press, Edinburgh, 1959).
86. M. Cartmill, in Primate Locomotion, F. A. Jen-kins, Jr., Ed. (Academic Press, New York,1974), pp. 45-83.
87. G. L. Jepsen, in Biology of Bats, W. A. Wim-satt, Ed. (Academic Press, New York, 1970),vol. 1, pp. 1-64.
88. W. D. Matthew, Ann. N.Y. Acad. Sci. 24, 171(1915).
89. Support from the National Science Foundationand the biomedical research program, NationalInstitutes of Health, in a series of grants toP.H.R. and R.W.S. is gratefully acknowledged.We appreciate the critical comments of and in-formation contributed by W. Bock, J. Cracraft,A. Fedducia, M. B. Fenton, M. Goodman, B.Heinrich, J. E. Hill, A. and C. M. Hladik, M. C.McKenna, L. G. Marshall, E. Mayr, S. L. Ol-son, R. L. Peterson, R. J. Raikow, W. D. L.Ride, V. M. Sarich, R. Schmid, L. Thien, and T.A. Vaughan. We thank P. Gerder, B. Heinrich,R. F. Kay, K. F. Koopman, L. G. Marshall, M.C. McKenna, I. Tattersall, and D. Wiens for re-viewing the manuscript.
