Adirectionalcline in Mouriri Guianensis (Me Lastom at Ace Ae)

ADIRECTIONALCLINE IN MOURIRI GUIANENSIS (ME LASTOM AT ACE AE)

Thomas Morley ( ;t)

Abstract of specialization of the most important variable, the ovary, can be clearly identi Morphological variation in Mouriri guia- fied. The overall pattern of distribution nensis is described and analyzed throughout its range in Brazil and adjacent regions. Featu was briefly described previously (Morley, res that vary are ovary size, locule and ovule 1975, 1976); the present paper is a detai number, shape and smoothness of the leaf blade led report. and petiole length. The largest ovaries with the most ovules occur in west central Amazonia; intermediate sizes and numbers are widespread MATERIAL AND METHODS but reach the coast only between Marajó and Ceará; and the smallest ovaries with the fewest The study was carried out with locules and ovules are coastal or nearcoastal from Delta Amacuro in Venezuela to Marajó. pressed specimens borrowed from many Small ovaries also occur in coastal Alagoas and herbaria, to whose curators I am grateful. at Rio de Janeiro. Ovaries with the fewest locu The most instructive characters are those les and ovules are believed to be the most of the unripened ovary, and therefoie specialized, the result of evolution toward only flowering material was of value in decreased waste of ovules, since the fruits of all members are few-seeded. Leaf characters most cases. It was necessary that speci correlate statistically with ovule numbers. mens have a considerable excess of flo Possible origen of the distribution pattern of wers for the dissections to be made wi the species is compared in terms of present thout harm but fortunately only a few rainfall patterns and in terms of Pleistocene climatic change with associated forest refuges. collections had to be omitted for this It is concluded that both phenomena were reason. In a few critical fruiting collec probably influential. High specialization appe tions leaf measurements alone were recor ars to have accompanied isolation, for reasons ded. The omission of most fruiting that are unclear. Because the plants from Delta collections from Fig. 10 does not materi Amacuro to Marajó are the most specialized they may once have been more isolated than ally affect the distribution shown, partly now. because most collections of the species are of flowering plants, partly because by INTRODUCTION good fortune most of the fruiting collec tions are from localities where flowering Mouriri guianensis is a small some material has also been taken or from times shrub-like tree of the neotropics. places nearby. No important range exten Its distribution lies across Brazil and sions are omitted. Coverage of the collec extends to Venezuela on the north and to tions I believe to be adequate for the Rio de Janeiro on the south (Fig. 10). purposes of this paper, but there are Certain characteristics of the plants are many gaps which will probably be filled highly variable over this range; the varia in the future. tion is examined here to see to what The relevant floral features are tho extent it may correlate with the geogra se of ovary size and shape (Fig. 3-9) and phic distribution. An unusual feature of the inner components that affect those the present example is that the direction characters. The diameter of the dried

( *) Botany Department, University of Minnesota, St. Paul, Minnesota, USA 55108. ovary can be used as one parameter; When the leaves are folded lengthwise no however, outside measurements of the reliable angle can be obtained. When all tapering and often distorted ovaries are are loose in a pocket an average angle can not always reliable. The ovary size is a of course be obtained but there is no reflection of the number of ovules within, certainty that the lowest and uppermost and to a lesser extent of the number and forms are present. Averages range from thickness of the partitions. Consequently 790 to 160O; the smallest and largest the numbers of locules and ovules have individual angles measured were 61° and been mapped as the major variables of the 180°. The basal angle is shown in Fig. 10 ovary. as a continuous variable. Material permitting, three to five Petiole lenght is also variable and flowers were dissected for each of the 71 has been mapped. The length ranges from collections examined and occasionally 1 to 5 mm, and the averages of the collec more. Limitations of material precluded tions vary from 1.5 to 4 mm. In this ins larger samples. In most cases the range of tance a two-state presentation seemed all variation encountered in a given collec that could be justified; a bar on the dia tion was not great with these small sam gram in Fig. 10 indicates a petiole average ple numbers, leading me to believe that greater than 2.5 mm. the results are reasonably representative. The prominence of the lateral The consistency of counts of different nerves in dried leaves is also greater in so collections from the same locality and the me areas than others and is mapped, consistency of the patterns shown in Fig. although sometimes highly variable and 10 support the opinion that the results difficult to measure consistently, only a are adequate for use. visual estimate being practical. The dried leaf surface was rated on a scale of 1—5, The leaves also vary with geography. from veiny to smooth, using a standard Blade shape is the most consistent varia veiny collection (Morley 1164) and a ble that is readily measured, and the smooth surface as reference points. In width of the basal angle is the most con this case also only a two-state presenta sistent character of the shape. As the an tion is given in Fig. 10. A bar on the gle becomes narrower, the leaves tend to diagram indicates a smoothness rating of become more elliptic and less ovate, but greater than 3. this does not always hold. The possibility that the leaves on a particular specimen The shoots of these plants are of do not represent a norm for that plant is two types, leader shoots and dorsiventral a problem that cannot be avoided. ones, and leaves of the two are dissimilar. The relatively few leaders, which grow Leaf variation on the same speci upward with great vigor, bear unusually men is often considerable. The leaf-bea short leaves which are often cordate at ring twigs grow in spurts or flushes, with base, while the much more common resting periods between. The lowest lea dorsiventral shoots produce the typical ves of each growth flush tend to have the leaves of the plant. Only the typical widest included basal angles, the upper leaves are referred to in this paper. the narrowest (Figs. 1„ 2). The difference may be very slight. When there is but a single pair of leaves on a twig, its leaves RESULTS are most often like those of the lowest leaves or sometimes the middle leaves of Examination of the distribution twigs with several pairs. The average angle map (Fig. 10) shows the following points is used here to express this leaf character. of significance: 1. Locule number. northeast tip of Marajó. C The coastal and near-coastal A. This varies from 5 (rarely 6) to collections in Alagoas and Rio one. Plants with 5 and 4 locules de Janeiro have relatively low are very^ widespread. locule numbers. The 2.6 that B. The smallest locule numbers occurs on the upper Rio Bran occur from coastal Delta Amacu- co is exceptional, by far the ro to the south side of Marajó. smallest number in the interior. All those ovaries with 2.5 or 2. Ovule number. fewer locules occur here, and none here have more than 3 ex A. This correlates with locule num cept for the collection at the ber, with some variation. Five

3 4 5 6789

Figs. 1, 2. Leafy twigs illustrating variation in shape of leaf blade from plant to plant and from lower to upper leaves in a growth flush. Tracings. 1. Moore 437. 2. Broadway 275. Figs. 3—9. Inferior ovary and calyx of different flowers illustrating variation in size and shape of the ovary in dried flowers. Camera luci- da. 3. Ule 5915. 4. Level 118. 5. Kubitzki 75-79. 6. Weddell 2494. 7. Froes 2008. 8. Oliveira 2082. 9. Soubirou s.n. locules are usually associated G.The only other occurrences of with 15 or more ovules, never 10's and 11's below 11.5 are less than 11. The smallest ovule also coastal or nearly so, one in numbers occur only in ovaries Alagoas and two at Rio de with 1 or 2 locules. Janeiro. The exceptional colle B. The greatest number of ovules ction from the upper Rio Branco occurs in west central Amazonia. with 11.6 ovules (referred to in The largest number counted point 1D) is the inland collection was 33, for the plant with closest in number to any of an average of 29.3. The these: five of its flowers yielded numbers decrease to the north, ovule counts from 10 to 13. east, and south from here. An 3. Ovary diameter: the fewer the area of moderately high numbers ovules and locules, the smaller occurs along the Amazon in Pará the ovary. Size differences are C. High-intermediate ovule num slight but rather consistent. Thus bers of about 20 occur in wes the largest ovaries are found in tern Venezuela, southwest Ama western Amazonia and the smal zonia, two places northwest of lest ones on or near the coast Brasilia, and at one site in inner from Georgetown to Amapá. All Maranhão. None are on or near ovaries less than 1.2 mm in dia the coast. meter occur in the latter area, D. Low-intermediate numbers of none are 1.3 mm or greater, and about 12-15 are widespread, and only two measured 1.2 and extend to the coast between Ma 1.25 mm. The coastal collection rajó and Ceará. This is the only in Delta Amacuro has a small part of the coast occupied by locule and ovule number but plants of intermediate numbers. measured 1.3 mm as did the col E. A group of six similar collections lection on the south side of is found in west-central Mato Marajó. No inland collections Grosso. The similarity in locule had dry ovaries measuring less and ovule numbers is emphasi than 1.2 mm. zed by generally similar leaf cha 4. Leaf characters: these are highly racters. variable, but some correlations F. The low ovule numbers of 7-9 with distribution do exist. Near are all coastal or near-coastal, ly all the coastal plants, from occurring between Delta Amacu Delta Amacuro to Rio de Janei ro and Marajó. In this respect ro, have broad to moderately they coincide with the low locule broad basal angles, short petioles numbers mentioned above and and veiny leaves, while plants of with the small sizes mentioned the interior tend to have the below. No ovule numbers here opposite conditions. An abstract are greater than 11.3 except for factor to express these condi the collection at the northeast tions can be made for each col tip of Marajó. The 10's and lection by multiplying the avera 11 's are all from the Paramaribo ge petiole length by the inverse area. The smallest individual of the blade angle by the leaf number counted was 6, for the surface rating by 100. The coas Surinam collection with an ave tal plants have relatively low rage of 7.3. factors and the inland ones relatively high factors; when the two a single placenta in a locule. The total for are compared as groups the dif the ovary will be the product of locule ference is highly significant (F number times the average number of ovu test, P = < .001). The same re les per locule (Table I). The total num sult is reached using only the bers of 20 and 20, 16 and 15, 12 and 12, blade angles. The coastal plants and 9 and 8 are closer to each other than from Delta Amacuro to Amapá to adjoining numbers and would create have slightly lower factors than peaks in a histogram much like those those from Marajó to Ceará but actually found in Figs. 11A and 11B. The this difference is not statistically 20-20 peak shows only in 11B. In 11 A, significant. When the leaf factors only the peak at 14 is out of place. Con are tested for correlation with sequently, although the few-ovuled coas ovule numbers the result is also tal plants have developed more or less highly significant (P » < .001), recognizable features, they do not appear the high factors correlating with sufficiently distinct as groups to warrant high ovule numbers. Throughout taxonomic status. most of the interior the leaf cha From a purely morphological point racters are quite variable, but of view it is noteworthy that the placen- two centers of long petioles, low tation in certain collections with high basal angles, and smooth surfa ovule numbers (Appun 1711, Krukoff ces can be seen in Fig. 10, one 6613, Level 118, Prance & Silva 59321, along the lower Amazon and the Ule 59T5) resembles in part that of seve other in Mato Grosso. ral species of Mouriri in which the ovules 5. Very few collections have been are borne on all sides of each placenta. In made in the central Amazon ba the collections named the ovules are sin except along the Amazon often arranged in a full circle at the apex itself. Whether this presents a of the placenta, but spread apart to pass true picture or results from in on each side of it part way toward the complete collecting is unknown. base. In order to determine if the species might be composed of two or more taxo- DISCUSSION nomic groups, frequency histograms were made of the single most reliable character The overall distribution pattern is ovule number (Fig. 11). One diagram (A) one in which flowers with the largest uses only the average numbers plotted in ovaries and the most ovules occur in west Fig. 10, while the other (B) uses all indi central Amazonia while the smallest ova vidual counts. In A one sees a grouping of ries with the fewest locules and ovules are flowers with 7—9 ovules which is so dis coastal or near-coastal from Delta Amacu tinct that one might think it indicates a ro to the south side of Marajó, with two taxonomic unit (see point 2F above). Ho other localities of small ovaries in coastal wever, in B it is seen that the gap between Alagoas and at Rio de Janeiro. 9 and 11 ovules rs bridged by many indi Considering what is known of the vidual flower from plants with average phylogeny of the genus (Morley, 1953, numbers other than 10. This suggests at 1976) there can be no doubt that the least that the gap is not as sharp as the 5—loculed condition is primitive for the averages make it appear. There is another species and that the lower numbers of factor that partly explains the groupings locules and ovules represent progressive seen in A. The ovules tend to be the same reduction. The more primitive members number on each placenta; usually there is of the subgenus Mouriri, to which M. 7 -;V

• diameter of dry ovary, 2 mm in this example. • average no. of locules -average no. of ovules -petioles relatively long -basal angle of leaf blade narrow -leaf surface relatively smooth

Fig. 10. Distribution of M. guianensis showing geographic variation of selected flower and leaf characters. See legend. A bar is shown when the petioles are mo re than 2.5 mm long, and when the leaf surface is smoother than the median on a smoothness scale. Included angles of leaf bases range from 79° (narrow) to 160° (wide). See Materials and methods. guianensis belongs, have 15—30 ovules in the most widespread species of the per ovary and it therefore appears reaso tribe Memecyleae and appear to be nable to suppose these numbers to be advantageous for bird dispersal (Morley, primitive for M. guianensis. Therefore 1975 p. 19). Since so few seeds are pro according to the present distribution the duced, many ovules must abort when a most primitive members of the species are large number is present before fertiliza those of western Amazonia and the most tion. One might therefore expect a reduc specialized ones are the coastal and near- tion series in. ovule number to occur in coastal ones with small ovaries and few the interests of greater efficiency. locules and ovules. The species appears However, it is not clear why this to have originated in the west and migra reduction should have been accelerated ted north, east and south from there. more in some areas than in others. One An adaptive value can be rationali can only speculate that unk vn compe zed for the more specialized forms. All titive forces were greater in tne areas of known fruits of the species are small and reduction. Because those areas are all rela contain one or two or at most three tively small at present and may have been seeds. Small few-seeded fruits are found smaller in the past it is possible that de-

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 no. of ovules

Fig. 11. Frequency histograms of ovule numbers in M. guianensis. A. Frequencies of the averages from Fig. 10 as read to the nearest whole number, e. g. a number from 13.6 to 14.5 is counted as 14. B. Frequencies of all individual counts me- de. The same number of counts could not always be made for each collection; see Materials and methods, and Results. creased population size with its modified (Fig. 12). If one excludes the plants of I evolutionary forces may have been one Alagoas and Rio de Janeiro, there is some factor in promoting the reduction. Diffe correlation of distribution with precipita-, ring dispersal agents and other altered tion to the extent that the largest ovaries environmental conditions might also have and the smallest ones both occur in areas played a part. of high rainfall, one inland, the other coastal. The intermediate sizes and num The association of leaf characters bers occupy a wide range of rainfall zones with ovule numbers (point 4 above) sug but the ones that reach the coast, from gests that leaves with long petioles, nar Para to Ceara, are mostly in a region of row basal angles, and smooth surfaces lower rainfall. The Alagoas collection may be primitive for the species. correlates with an isolated pocket of rela In seeking an explanation of the tively high rainfall and a very short dry distribution pattern of the species, we season (Prance, 1978). Only the Rio de should first consider whether it might Janeiro site appears to have no relation to have developed under present environ a zone of higher precipitation and here mental conditions. If so, the distribution the dry season is not very pronounced, of the variables in the species should probably a relevant factor. Both the Ala match present environmental zones goas and Rio de Janeiro sites are in a nar reasonably well. Rainfall is the factor row coastal strip of "moist" forest which most easily compared, although even for extends north from Santa Catarina to Rio it the information available is incomplete Grande do Norte (Mori at al, 1981).

Table I

Locule and ovule numbers in M. guianensis The average number of ovules per locule may vary from flower to flower and from collection to collection, resulting in diffe rent total numbers of ovules in ovaries that may have the same numbers of locu les. Some total numbers are closer to each other than to others: 20 and 20, 16 and 15, 12 and 12, 9 and 8. These groupings may partly account for some of the peaks seen in Fig. 11. See Results.

Average Average no. of no. of ovules Total ovules Total No. of per no. of No. of per no. of locules locule ovules locules locule ovules (or (or per (or (or per placentas) placenta) ovary placentas) placenta) ovary

5 6 30 4 3 12 5 5 25 3 4 12 5 4 20 3 3 9 5 3 15 2 or 1 4 8 4 5 20 2 or 1 3 6 4 4 16 ( 1 locule always contains 2 placentas) Fig. 12. Approximate annual precipitation in meters in the range of M. guianensis, redrawn from Ratisbona (1976) and Snow (1976).

A broad relatively dry zone of less thin the wet coastal strip just mentioned than 2 m of rainfall extends from sou is also correlated with rainfall to a mode theast Brazil northwest across the Ama rate degree. A close comparison of Figs. zon and partly separates the wet coastal 10 and 12 shows a very close correlation strip between Delta Amacuro and Marajó at the Delta, while southeast along the from the wet interior. The bulk of this coast the collection from sites with 2.5 m dry arm northwest of the Amazon is of rain or more tend to have fewer locules shown by Brown (1982) as having a pro and ovules than those from slightly drier nounced dry season of from 3—5 months. areas. Thus it c^n be said that there is a The distribution of the plants wi definite but imperfect relation between present rainfall patterns and distribution the plants presumably have high mobility. of the different forms of Mouriri guianen Bird migration routes could have beene sis. factor. The very high water table of most However, the degree of reduction of Delta Amacuro might lead one to sus seen in the plants from Delta Amacuro to pect a relation between this condition Amapá is very sharp and may have been and the plants with few ovules. However, produced under conditions of greater iso in French Guiana similar plants grow at lation than now appear to exist, the edges of savannas which are relatively especially considering that a similar but dry according to Granville (1982). slightly lesser reduction took place Nor does the available information apparently under nearly complete isola on soils and other environmental factors tion at Alagoas and Rio de Janeiro. The appear to coincide with the distribution refore it is possible that the observed of the species. differences are at least partly the result of past climatic changes which involved The fact that the plants along the to some extent the various Pleistocene coast have leaves with relatively short pe forest refuges that have been proposed tioles, broad basal angles, and veiny surfa as centers of forest contraction and ex ces (point 4 above) suggests a relation pansion during these changes. The subject between these forms and maritime influ is reviewed by Simpson & Haffer (1978). ences; the adaptive value of such forms if Some correlations exist: (1) the plant any is unknown. with the most ovules, in western Amazo If the species diversified under nia, is close to one of Prances (1982) re-1 present conditions, one would hypothesi fuges although not within it. (2) The ze its history about as follows: moderate Alagoas location is within or very close reduction of the ovary tended to occur as to the proposed forest dispersal centers the plants spread north, east, and south for numerous vertebrates (Muller, 1973) from the point of origin in the west, per and several genera of butterflies (Lamas, haps in relation to environmental gradi 1973, n. v., ill. in Simpson & Haffer, ents, but this is not clear. Isolation does 1978). This site is also at the southern tip not appear to have been a major factor of the Pernambuco endemic center for in this development. Plants with modera forest butterflies as shown by Brown te reduction carried to the coast between (1977, 1982). (3) The Rio de Janeiro Marajó and Ceará. However, when the location is recognized as a probable forest plants penetrated the broad dry zone refuge on the basis of neotropical bird mentioned above and reached the coast distribution (Haffer, 1974), nymphalid between Delta Amacuro and Amapá and- butterflies (Brown, 1977, 1982), nume became adapted to coastal conditions rous vertebrates (Muller, 1973), and trees there, a sufficient degree of isolation exis (Mori etal, 1981). ted in this case to permit accelerated The coastal strip from Delta Ama evolution under present conditions. Par curo to Amapá in which occur so many tial isolation or the existence of environ plants with reduced flowers finds no close mental gradients may permit such diffe counterpart in the proposed refuges. rentiation (Benson, 1982, Endler, 1982). However, Muller (1S73) and Ab'Saber Long distance dispersal would have (1982) indicate a large inland arboreal brought the plants to Alagoas and Rio' de dispersal center nearby that extends from Janeiro, where they were well isolated just south of the Orinoco delta through and where accelerated evolution appears most of Amapá, avoiding the coast except to have taken place. Being bird-dispersed. just northwest of Georgetown. Prance shows two smaller refuges farther inland The Alagoas and Rio de Janeiro colo in the same general area, and Brown nies remain isolated, but some of the shows a part of Amapá as a refuge. plants of the northeast coast are now If climatic change involving refuges close to plants of the interior and it is was a major factor in the modification of here that any hybridization and thus the species, then its history might have "secondary contact" would have taken been as follows: during a Pleistocene dry place. period the broad relatively dry zone of In fact it does not appear necessary less than 2 m rainfall that extends from to choose one of the above hypotheses southeast Brazil across the Amazon to the to the exclusion of the other. It seems northwest became drier and expanded, probable that both were operative. The dividing the population into four groups evidence for climatic change is strong, of different sizes-the plants to the west, but the changes would have been of those crowded along the northeast coast varying degrees and would have produced From Delta Amacuro to Amapá, and different effects in different places de those in Alagoas and Rio de Janeiro. The pending on variables such as rainfall, soil, plants trapped along the northeast coast water table, and wind patterns. In some followed the moist zones; if the present areas present isolation is already so com rainfall pattern is relevant they would plete that lessened precipitation probably have been squeezed into a small area in would not increase appreciably the likeli eastern French Guiana and nearby Amapá hood of differentiation occurring. In with possibly a second group in Guyana. other areas a moderate decrease in rainfall During the Flandrian transgression they would accentuate existing differences and would have been forced inland, very accelerate differentiations already proce likely to the approximate areas proposed eding at a slower pace under conditions as refuges by Ab'Saber, Brown, and of partial isolation or environmental gra Prance. The transgression would also have dients or both. Plants might persist along driven plants from the area of the Atlantic watercourses and in sheltered places du embayment that might otherwise have ring dry times. Thus the populations persisted along the main watercourses. might diminish but not necessarily disap For reasons that are unclear but which pear in the driest zones during dry perio may have been related to small popula ds and then multiply during wetter perio tion size, the coastal plants evolved rapi ds. Changes of this level appear sufficient dly in the direction of smaller ovaries to account for much and perhaps all of with fewer ovules; reproductive barriers the differentiation found inM. guianensis. seem not to have been formed judging by Perhaps in some areas or for other orga the series of transitional forms, but this nisms full and drastic climatic shifts has not been tested. The plants on the would be necessary to bring about diffe west remained relatively unchanged. rentiation. When climatic conditions shifted toward those of today, the plants spread ACKNOWLEDGEMENT into all suitable habitats, occupying much I am grateful to José R. Nascimento of the territory between the four restric for translating the summary. ted zones including the coast between Marajó and Ceará. Plants within the four zones tended to keep their original cha Resumo racteristics, but the migrants developed Variação morfológica em Mouriri guiarían more or less intermediate traits, possibly sit é descrita e analisada em relação a sua dis due to hybridization. tribuição no Brasil e regiões adjacentes. Carac- terísticas que variam são tamanho do ovário, Brown, K.S. Jr.-1977. Centros de evolu número de lóculos e de óvulos, forma e lisura da folha seca, e comprimento do pecíolo. Os cão, refúgios quaternários e conser maiores ovários com o maior número de óvu vação de patrimónios genéticos na re los ocorrem no centro-oeste amazônico; ta gião neotropical: padrões de diferen manhos e números intermediários estão dis ciação em Ithomiinae (Lepidoptera: tribuídos largamente mas alcançam a costa ma Nymphalidae) Acta Amazonica, 7(1): rítima somente entre a ilhai de Marajó e Cea rá; e os menores ovários com os menores nú 75-137. meros de lóculos e óvulos encontram-se na 1982. Paleoecology and regional costa ou próximo da costa entre o Delta Ama- patterns of evolution in neotropical curo na Venezuela e Marajó. Pequenos ovários também ocorrem nas costas de Alagoas e Rio forest butterflies. In: Prance, G.T. ed de Janeiro. Acredita-se que ovários com o me Biological diversification in the tropi nor número de óvulos e lóculos são os mais es cs. New York, Columbia University pecializados, resultado da evolução em direção press, p.255-308. à diminuição da perda de óvulos, já que os frutos de todos os membros possuem poucas Endler, J.A-1982. Pleistocene forest sementes. Folhas com relativamente longos refuges: fat or fancy? In: Prance, pecíolos, ângulos basais estreitos, e superfí G.T. ed. Biological diversification in cie lisa correlacionam-se com número grande the tropics. New York, Columbia Uni de óvulos e podem ser primitivos para a espé versity press, p.641-657. cie. Folhas deste tipo tipificam as plantas do interior enquanto plantas costeiras tendem a Granville, J.J.-1982. Rain forest and ter folhas com condições opostas. A evidên xeric flora refuges in Franch Guiana. cia é insuficiente para subdividir a espécie em In: Prance, G.T. ed. Biological diver dois ou mais grupos taxonómicos. A possível sification in the tropics. New York, origem do padrão observado da distribuição da espécie é discutido em termos (1) dos pa Columbia University press, p. 159- drões da precipitação atual, e(2) das mudan 181. ças climáticas do Pleistoceno e refúgios flores Haffer, J.— 1974. Avian speciation in tais associados. Conclui-se que ambos os fenô tropical South America. Cambridge, menos tenham provavelmente influenciado. Mass. Nuttall Ornithol. Club, (14) Alta especialização parece ter acompanhado isolamento, por razões que não estão claras. Lamas, M.G.— 1973. Taxonomia e evo Visto que as plantas encontradas entre o Delta lução des géneros Ituna Doubleday Amacuro e Marajó são mais especializadas, elas (Danainaee Paititia gen. nov. Thyri- talvez tenham estado mais isoladas no passado dia Hubner e Methena Doubleday do que observa-se hoje em dia. (Ithemiinae) (Lepidoptera, Nympha lidae). Doctoral Thesis. Univ. São References Paulo. Brazil. Mori, S.A.; Boom, B.M.; Prance, G.T.- 1981. Distribution patterns and Ab'Saber, A.N.-1982. The paleoclimate conservation of eastern Brazilian and paleoecology of Brazilian Ama coastal forest tree species. Brittonia, zonia. In: Prance, G.T. ed. Biological 33: 233-245. diversification in the tropics. New Morley, T. —1953. The genus Mouriri York, Columbia University press, (Melastomaceae). A sectional revi p. 41-59. sion based on anatomy and morpho Benson, W. W. - 1982. Alternative mode logy. Univ. Calif. Publ. Bot., 26 : ls for infrageneric diversification in 223-312. the humid tropics; tests with passion 1975. The South American dis vine butterflies. In: Prance, G. T. ed. tribution of the Memecyleae (Melas- Biological diversification in the tro tomataceae) in ralation to the Guiana pics. New York, Columbia University area and to the question of forest re press, p. 608-640. fuges in Amazonia. Phytologia, 31 : 279-296. Ratisbona, L.R.-1976. The climate of 1976. Memecyleae (Melastomata- Brazil. In: Schwedtfeger, W. ed. ceae). Flor. Neotrop., 15: 1-295. Climates of Central and South Muller, P.—1973. The dispersal centers of America. New York, Elsevier, p.219- terrestrail vertebrates in the Neotro 293. pical Realm. In: Junk, W. ed. Biogeo- Simpson, B.B. & Haffer, J.- 1978. Spe- graphica II. The Hague. ciation patterns in the Amazonian Prance, G.T.—1978. The origin and evolu forest biota. Ann. Rev. Ecol. Syst., tion of the Amazon Flora. Intercien- 9:497-518. cia, 3: 207-222. Snow, J.W.-1976. The climate of nor 1982. Forest refuges: evidence thern South America. In: Schwedtfe from wood angiosperms. In: Prance, ger. W. ed. Climates of Central and G.T. ed. Biological diversification in South America. New York, Elsevier, the tropics. New York, Columbia p. 295-403. University press, p. 137—157. (Aceito para publicação em 6/7/83).