SMITHSONIAN CONTRIBUTIONS TO BOTANY NUMBER 13 Swollen-Thorn Acacias of Central America Daniel H. Janzen SMITHSONIAN INSTITUTION PRESS City of Washington 1974 ABSTRACT Janzen, Daniel H. Swollen-Thorn Acacias of Central America. Smithsonian Contributions to Botany, number 13, 131 pages, 119 figures, 1974.-This nomen- clatural, taxonomic, and ecological treatment of 1 1 Central American obligate ant-acacias (Acacia allenii, A. chiapensis, A. collinsii, A. cookii, A. cornigera, A. gentlei, A. globulifera, A hindsii, A nzayana, A. melanoceras, and A. sphaero- cephala) and one quasi-obligate ant-acacia (Acacia ruddiae) is based on ex- tensive field study from 1963 to 1972 and on herbarium specimens where of use. 7 he population boundaries of all species are mapped and described with respect to ecological parameters. Morphological variation, details of the interaction with the ants, and acacia reproductive biology are presented for most species. OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded in the Institution?s annual report, Srnithsonian Year. SI PRESS NUMBER 4817. SERIES COVER DESIGN: Leaf clearing from the Katsura tree Cercidiphyllurn japonicum Siebold and Zuccarini. Library of Congress Cataloging in Publication Data Janzen, Daniel H: Swollen-thorn acacias of Central America. (Smithsonian contributions to botany, no. 13) Bibliography: p. 1. Acacia. 2. Ants-Central America. 3. Botany-Central America-Ecology. I. Title. 11. QKl.S27 no. 13. [QK495.L52] 581?.08s [583?.321] 73-4401 Series: Smithsonian Institution. Smithsonian contributions to botany, no. 13. For sale by the Superintendent of Documents, US. Government Printing O5ce Washington, D.C. 20402 . Price $2.35 (paper cover) Contents Page Introduction ................................................... 1 General Biology ................................................. 2 General Comments on Swollen-Thorn Acacia Taxonomy .............. 6 Materials, Methods, and Terminology ............................... 14 General Morphological Traits .................................. 15 Taxonomic Information .............. ....................... 25 A . Key to Mature Central American Swollen-Thorn Acacias, Based on All Traits ........................................... 28 B . Key to Mature Central American Swollen-Thorn Acacias, Based on Vegetative Traits .................................. 29 Acacia cookii Safford ................. ................... 29 Acacia ruddiae, new species ..................................... 34 Acacia melanoceras Buerling ..................................... 42 Acacia allenii, new species .......................................... 53 Acacia chiapensis Safford ....................................... 61 Acacia globulifera Safford ........................................ 67 Acacia sphaerocephala Schlectendal and Chamisso ..................... 73 Acacia cornigera (L.) Willdenow .................................. 80 Acacia mayana Lundell ........................................... 93 Acacia gentlei Standley ....................................... 95 Acacia collinsii Safford ......................................... 99 Acacia hindsii Bentham ........................................ 113 Literature Cited ................................................. 130 Swollen-Thorn Acacias of Central America Daniel H. Janzen Introduction The swollen-thorn acacia group (Acacia species) is one of the two most diverse and widespread of the neotropical myrmecophytes. These acacias are the characteristic ant-plants in the drier lowlands (?tierra caliente?) of Central America, just as the genus Cecropia (Aloraceae) is the most conspicuous and diverse myrmecophyte of wetter habitats in Central and South America. This monograph on the taxonomy and natural history of neotropical swollen-thorn acacias is of- fered at this time for several reasons. (1) The neotropical swollen-thorn acacias do not appear to constitute a tight phjletic unit, yet they share many adaptive ecological and morphological traits. They are thus outstanding examples of evolutionary con- vergence. The most noteworthy of the traits they have in common is that they are protected from herbivores by colonies of ants in the genus Pseudo- myrmex, rather than by chemical traits as are most plants (Janzen, 1966a, 1967b; Rehr et al., 1973). (2) The ants can be readily removed, just as though one were able to easily remove the alkaloids and cyanogenic glycosides from the leaves of ordinary acacias such as Acacia farnesiana. This ease of ex- perimentally disarming the plant means that the protective role of the obligate acacia-ants has been easy to study and provides an excellent system for understanding the coevolution of complex interac- tions between higher organisms. For further and more detailed comparative studies among the swol- Daniel H. Janzen, Department of Zoology, University of Mich- igan, Ann Arbor, Michigan 48105. len-thorn acacias, a stable nomenclature is essential at this time. (3) Detailed data on populations of na- tive wild Central American plants are noticeably absent from the literature, A better concept is needed of the population structures of a wide va- riety of tropical plants of different life forms and in different habitats in order to generate specific hy- potheses that can lead to an understanding of com- plex tropical communities. (4) The lowland deciduous forests of Central America have nearly disappeared. The population structures of the spe- cies remaining in these habitats are already greatly altered. The swollen-thorn acacias are clearly in this category. I would like to summarize the ecological behavior of swollen-thorn acacias before their pat- terns are so far altered that we can no longer recog nize the conditions that produced the genotypes we find in today?s cattle pastures, roadsides, and abandoned fields. Most wild plants in high-yield agricultural areas must be studied under conditions not unlike studying elephants and deer in a zoo. It is hoped that the incompleteness of the data and hypotheses offered here will encourage field biologists to look harder at why plants do what they do in natural habitats as well as in those badly altered by agriculture. ACKNOWLEDGMENTS.-~ am greatly indebted to Dr. Velva E. Rudd of the National Museum of Natural History for her very competent assistance and advice throughout this study. Her patience with the bumblings of an entomologist in the world of taxonomic botany immensely facilitated the un- raveling of the nomenclatorial spaghetti surround- ing the swollen-thorn acacias. Personnel in various 1 SMITHSONIAX CONTRIBUTIOXS TO BOTANY herbaria have kindly allowed me to examine types and at times to borrow specimens. A special acknowledgment is due Karen Janzen, who participated in the laborious task of recording field data and voucher specimens. This study was supported by National Science Foundation Grant GB-5206, and is in part a by- product of NSF Grants GB-91, 1428, 7805, and 7819. Substantial portions of this study were com- pleted while I was serving in Costa Rica as a faculty member for the Organization for Tropical Studies in Costa Rica (1965-1971). The paper is Contribu- tion No. 1360 from the Department of Entomology, The University of Kansas, Lawrence, Kansas. General Biology The 12 species of swollen-thorn acacias in aggre- gate range from the Tropic of Cancer on both coasts of Mexico only as far south as northwestern Colombia. They are generally found at elevations below 1200 m (Figure 1). One Argentinian acacia, Acacia cauen (Molina) Molina, has been suggested as a swollen-thorn acacia (Fiebrig, 1909; Chodat and Vischer, 1920); however, Wheeler (1 942), Havrylenko (in litt.), and Sliillink (in litt.) have convinced me that the ants living in the thorns were only fortuitous invaders. My own unpublished observations of A. cazlen in Chile give no hint of myrmecophytism with this plant. Greenhouse A. cazlen (from Argentinian seeds) shows thorn and leaf development like that of normal acacias and therefore will not be discussed further. The African swollen-thorn acacias, or ?whistle-thorn? acacias (e.g., A. seyal, A. drepanolobium, A. formicarum, A. bussei), appear to have a mutualistic interaction with ants of the pantropical myrmicine genus Crematogaster (Hocking, 1970; Monod and Schmitt, 1968; Bequaert, 1922) . This is a clear case of mild convergence on the Central American sys- tem and is excluded from generalizations made in this paper. This restricted view should not, how- ever, be taken to suggest that hypotheses based on the much better understood Central American system are irrelevant to the African system. The pantropical and subtropical mimosoid le- gume genus Acacia contains at least 750 species, though the number may be reduced substantially if the synonymy found in this study is representative of the remainder of the genus. Plants of the genus Acacia are primarily in areas with less than 3 me- ters of annual rainfall and with a well-developed dry season of 3-11 months duration. Throughout Africa and the neotropics, those species with foliar nectaries (probably all species) have a loose associ- ation with ants. Many genera of arboricolous ants (e.g., Crematogaster, Pseudomyrmex, Camponotus) visit these nectaries and in the process probably in- crease the intensity of their foraging activity on these plants compared with adjacent plants lacking nectaries. This, accompanied by similar benefits from parasitic Hymenoptera attracted to the nec- taries, has probably been the selective force leading to the evolution of these nectaries (cf. Janzen, 1966a; Bequaert, 1922). In a few cases, however, the ant and acacia inter- action has evolved to a much higher level of com- plexity, that of obligatory mutualism (Janzen, 1966a). For background information, a brief resume of the interaction in Central America is given in the following paragraphs. More details can be ob- tained from Janzen?s (1967b) lengthy description of Acacia cornigera and Pseudomyrmex ferruginea in lowland eastern Mexico and in peripheral papers on ants and acacias (Janzen, 1967a, 1969a, 1969b, 1973; Beulig and Janzen, 1969). A colony of obligate acacia-ants lives in the swol- len, stipular thorns of one to several individual shrubs or small trees that have been given the sub- generic common name of ?swollen-thorn acacia? (Janzen, 1967a). New World obligate acacia-ants are a behaviorally distinct group (at least 13 species: P. belti, P. ferruginea, P. nigrocincta, P. venefica, P. satanica, etc.-Janzen, MS) within the exclusively New World, and mostly tropical, genus Pseudo- myrmex (150-200 species). Most species of Pseudo- myrmex have small colonies in dead hollow stems above the ground in habitats ranging from brushy pastures to well-developed forest. The exceptions, besides obligate acacia-ants, are those with an obli- gate mutualistic interaction with the neotropical trees Trip 1 ar is (Pol ygonaceae) and Tach iga lia (Leguminosae). I have never found any of the many species of obligate acacia-ants as a viable colony living elsewhere than in a swollen-thorn acacia. Those rare cases where swollen-thorn acacia pop- lations are unoccupied by obligate acacia-ants are discussed in later parts of this paper. New World swollen-thorn acacias differ morpho- logically and behaviorally from other acacias in XUMBER 13 3 FIGURE 1 .-General distribution of swollen-thorn acacias in Central America, based on herbarium records, field observations, and inferences from general vegetation. Contemporary distributions are usually continuous throughout the areas in black, owing to human destruction of the pri- mary forest that previously isolated the swollen-thorn acacia subpopulations along river edges and in other kinds of disturbed vegetation. that they have: (1) large and partly hollow stipular vested by the ants-these are the protein, lipid, and thorns that the ants clean out through an entrance major carbohydrate sources of the ant larvae; (3) hole cut in one thorn tip while the thorn is still greatly enlarged foliar (petiolar) nectaries that are green and relatively soft; (2) modified leaflet tips continually secreting the nectar that constitutes the called Beltian bodies (see Rickson, 1968, 1969, 1971, primary carbohydrate and moisture source of the for micromorphological details), which are har- adult ants; (4) year round leaf production and 4 SMITHSONIAN CONTRIBUTIONS TO BOTANY retention by most individuals, yielding a relatively continuous food source for the ants; and (5) lost most of the chemical and structural traits that pro- tect other acacias from most herbivores (cf. Rehr et al., 1973). Swollen-thorn acacias differ ecologically from other acacias in that in natural habitats they cannot survive to reproductive maturity without the patrolling activities of a large colony of obligate acacia-ants. The ants remove herbivores (primary insects) and intrusive vegetation by biting and stinging the invading pait of the plant (as do their relatives, Pachysinza, living in Barteria trees in Africa-Janzen, 1972). By analogy, the ants sub- stitute for the chemical and physical abilities of most acacias to deal with their predators and com- petitors. The one exception to the above generaliza- tions, A. rziddiae, will be discussed in its own species account. The species of obligate acacia-ants are not specific to a swollen-thorn acacia species, but rather to its life form. For example, A. collinsii has at least eight species of obligate acacia-ants living in it over its range from Mexico to Colombia, and all of these ants also live in other swollen-thorn acacias. In nature, one ant colony may be found occupying two adjacent individuals of two different swollen- thorn acacia species. The only species of swollen- thorn acacia that are occupied by only one species of ant are A. melanoceras (occupied by Pseudo- inyrmex sntanica) and A. allenii (occupied by Pseudonzy~mex ferruginea) . They appear to be this way because the climate or resources will not sup- port other ant species, rather than because the plants are unsuitable for other obligate acacia-ants. Both of these species of acacias will support other obligate acacia-ants in the greenhouse. Swollen-thorn acacias grow in natural and man- made disturbed vegetation-both primary and secondary succession. A. cornigera, A. hindsii, and A. collinsii have very successfully invaded brushy pastures, roadsides, and fallow fields, and may con- stitute a major part of the standing biomass in some l-5-year-old fallow regeneration. On the other hand, A. allenii, A. melanoceras, and A. ruddiae are apparently in danger of extinction because they do not maintain a breeding population except in very lightly disturbed, or undisturbed, primary rain forest. In contrast to most other acacias, swollen-thorn acacias originally inhabited naturally disturbed sites in forests-deciduous to evergreen-with 2-5 m annual rainfall. Other acacia species usually live in drier sites. The ability to compete with rapidly growing vegetation in wetter habitats is the direct result of the behavior of the occupant acacia-ants. In pruning vines out of the swollen-thorn acacia crown and in killing intruding branches from neighboring trees, they place the acacia in a highly insolated microhabitat (this is, incidentally, a form of allelopathy by the acacia-Janzen, 1969a). Sec- ond, since the growing points of the acacia are very heavily patrolled (protected) by ants, they lack the usual structural toughness associated with protec- tion from herbivorous insects that is found in other acacia shoot tips. This tenderness is associated with a swollen-thorn acacia rate of vertical height incre- ment high enough to make the acacia an effective competitor with other plants in wet sites. Ironically, this high growth rate would probably not be possi- ble in drier sites, and in drier sites the acacia would not be able to support the ant colony owing to loss of leaves during the dry season. From the wettest to the driest sites the species are: melanoceras, allenii, ruddine, cookii, gentlei, chiapensis, globulifera, hindsii, mayana, sphaeyocephala, cornigera, and collinsii. This progression also ranges from those with the smallest geographic ranges to the largest, except for A. mayana and A. sphaerocephala, which have minute ranges. It is noteworthy that some swollen-thorn acacias have the largest leaves of any acacia species. This may be useful in the cloudy weather which is more probable in the wetter sites than drier habitats; it is also ?allowed? by the generally reduced problem of foliar water loss in wetter habitats. In the driest swollen-thorn acacia habitats, many of the larger leaves are dropped during the dry season, and then axillary clusters of very small leaves provide nectar and Beltian bodies for the ant colony. Very few swollen-thorns and their associated large leaves are produced during the dry season. Even though they occupy wetter habitats than other acacias in general, swollen-thorn acacias dis- appear when annual rainfall exceeds about 4 m (e.g., the Puerto Viejo region of northeastern Costa Rica) and the dry seaeon is less than two months? duration. The reason for this probably centers on the decreasing number of sunny hours per day as rainfall increases. When days are overcast in rain- forest areas, the amount of light reaching small NUMBER 13 5 shrubs on the forest floor is very low, and it may be that the swollen-thorn acacia has been unable to convert its primitively heliophile physiology to that of total shade tolerance. Their upper elevational limit is about 800 to 1200 m, depending on the moisture and tempera- ture conditions at upper levels. The wetter and colder an elevational transect, the quicker the swollen-thorn acacias disappear from the vegetation with increasing altitude. Cold weather does not kill the ants on the acacias but restricts patrolling activities to the point where the acacia is either covered over by the rising vegetation, defoliated severely, or both. All species are found at sea level in at least some part of their range; their upper elevation limits vary widely, with A. globulifera, A. sphaerocephala, and A. cornigera attaining the greatest heights. In contemporary communities (disturbed by man), swollen-thorn acacias reproduce very com- monly by stump and root suckers and only rarely by seedlings. In naturally disturbed sites (where clearing of the vegetation is not periodically re- peated), however, the rare seedlings are the usual means of population recruitment. Though a large seed crop is produced annually, most seeds are killed by bruchid beetles, against which the ants offer no defense (Janzen, 196713, 1969~). There is a very high mortality rate of seedlings until the trees become occupied by an ant colony. Sucker shoots (often from lateral roots) are often near a large obligate acacia-ant colony, which simply incorpo- rates the sucker into the colony?s territory; this is especially noticeable with the obligate acacia-ant species that have multiple queens and thus un- limited colony size (e.g., Pseudomyrmex venefica in A. hindsii in western Mexico north of the Rio Balsas basin; Janzen, 1973). An unoccupied seedling must wait 9 months or longer for a founding queen ant to grow a young colony and by then the seed- !ing is often well below the surrounding vegetation and therefore at a severe competitive disadvantage. Sucker shoots produce their first seed crop about 2 years earlier than do seedlings. Adults of either seedling or sucker shoot origin generally live 15-35 years. Swollen-thorn acacias flower during the dry sea- son, as do so many other deciduous forest woody plants (Janzen, 1967c), and are pollinated by rela- tively unspecific pollen-collecting bees. In the greenhouse, swollen-thorn acacias never set fruit though they flower copiously. I inEer from this that they are obligatory outcrossers. The bees are occasionally attacked by a worker ant, but these bees are persistent in getting at the flowers, and the ant?s aggressiveness probably does little more than maximize cross-pollination by forcing bees to visit several trees. The small pods remain on the receptacle until the following dry season, growing large and maturing several months be- fore the next crop of flowers. There may be mature pods on the tree with the flowers, but these are from the previous year except in the species in very wet sites, such as A. allenii, A. ruddiae, and A. melanoceras. The seeds are distributed by birds that either eat the pod entirely (only A. cornigera and A. sphaerocephala) or eat the cluster of seeds imbedded in a sweet moist pulp that is yellow or white pulp. Swollen-thorn acacias appear to be the only neotropical acacias with bird dispersed seeds (Janzen, 1969b). If it were not for bird dispersal, there would be virtually no survival of seeds be- cause the bruchid beetles will kill virtually every seed that remains on the tree (Janzen, 1969c, 1971). Swollen-thorn acacias have a representative flowering phenology. At any given site, nearly all mature plants bear open inflorescences for 4-8 weeks at about the same time each year. This flowering period may start anytime during the first 2 months of the dry season, depending on the spe- cies and the area. For a given species and area the timing is the same from year to year. A single species may show quite extreme variation from site to site in both the absolute dates of flowering and the dates relative to onset of wet and dry season. In general, however, at least 80 percent of the sub- populations of swollen-thorn acacias flower during the dry season. In addition to the major peak of flowering shown by each species at each site, there are almost invari- ably some individual acacias that are out of phase with the remainder of their population. These are usually damaged, growing on locally peculiar drain- age types (creek bottom or hill crest) or senescent. They are often helpful in taxonomically identifying the population, but are very misleading in under- standing the morphology, ecology, and behavior of the population. Botanists have a tendency to collect these flowering or fruiting plants in preference to sterile individuals. This leads to erroneous records 6 SMITHSONIAN CONTRIBUTIONS TO BOTANY of flowering times. These individuals often set little or no seed, which gives a false impression of seed- crop size. Those seeds that are produced are some- times only mildly attacked by bruchids, giving the false impression that seed mortality is trivial for the population (see Janzen, 1968, for a discussion of this point among plants in general). These plants also may bear a reduced number of inflorescences per axil and inflorescence-bearing branch, and these branches are often sparser and smaller than on healthy acacias. Swollen-thorn acacias have very thoroughly oc- cupied most of the Central American habitat that looks as though it should support them; however, they have not yet spread into the semiarid areas of South America that are presently occupied by the species of plants that co-inhabit their Central American habitats. Their absence from the decidu- ous and semideciduous forests of Venezuela, Brazil, and Colombia is particularly striking. It seems rea- sonable to expect that these areas are not inimical to swollen-thorn acacias, but rather that they simply have not yet gotten there. A. collinsii in Colombia, however, must surely be a seed source for the re- mainder of the country. If swollen-thorn acacias were not such an economic pest in brushy Central American pastures, the obvious transplantation ex- periment would be well worthwhile. Swollen-thorn acacias have not been generally successful at colonizing islands. They occur on off- shore islands such as those in the Gulfo de Nicoya, Costa Rica, in the Gulfo de Fonseca, Honduras, and off the eastern coast of Yucatan, Mexico. The only far island they have reached is Isla Providencia, Colombia, in the Caribbean (about 125 miles off the coast of Nicaragua). A. collinsii is here occupied by one or two species of obligate acacia-ants. In colonizing islands, the acacia must already be there when the founding ant queen arrives; however, the acacia probably has little chance of surviving to reproduction unless occupied by an ant colony. It will take a high rate of simultaneous colonization by both ants and acacias for establishment. A. cornigera, introduced as an ornamental into Cuba (and Florida) without ants, has not escaped to form a noticeable population. While this may be in part due to a lack of pollinators, the absence of obligate acacia-ants may also contribute. Since agricultural man came on the scene (at least since 1200-900 B.c.-Coe et al., 1967), the population structure of most species of swollen- thorn acacia has probably been grossly altered. The majority of species have greatly increased their local density, absolute population sizes, and thoroughness of geographic coverage by invading fields, pastures, and roadsides which now connect the natural dis- turbance sites that occur along rivers, streams, and lakes. This means that the populations now ob- served are operating with a genotype evolved in slightly to grossly different habitats, and the local barriers to gene flow (climax forest on hills and ridges) have in great part been removed. Present-day populations of swollen-thorn acacias show distinct clinal variation in traits such as leaflet size, thorn size and shape, life form, seed-pod morphology, and Beltian body traits. Rather than being solely the result of extreme sensitivity to small linear environmental changes, these may be the consequence of the recently established genetic continuity between previously isolated populations on different drainage systems. Humans have probably not severely altered the gross geographic range of swollen-thorn acacias by their activities. They have, however, served as efficient long-range-dispersal agents. Fruits of A. cornigera are eaten by Central American indians and, with modern bus transport being common- place, seedlings are sometimes found near bus stops several hundred miles from a known acacia popula- tion. Further, cattle are often trucked long distances in Central America, and cattle eat the pods of A. cornigera and others. There are pockets of A. cornigera around cattle corrals scattered through the northern part of the Yucatan Peninsula; the natural vegetation at these sites is normally occu- pied by A. collinsii (though A. cornigera appears native to local wet sites in Yucatan). General Comments on Swollen-Thorn Acacia Taxonomy A close examination of the taxonomic status of swollen-thorn acacias is warranted by the fact that (1) as many as 25 percent of the swollen-thorn acacias in some herbaria are incorrectly determined, (2) there are at least 32 synonyms for the 10 previ- ously described species, (3) swollen-thorn acacias have never been treated with monographic thor- oughness (though there are short discussions in Safford, 1910, 1914, 1915, 1922; Rudd, 1964; NUMBER 13 7 Schenck, 1913; Janzen, 1967b), and (4) their spe- cies-level taxonomy has become so confused that it is currently nearly impossible to generate a detailed discussion of the mutualistic interaction in a com- parative and evolutionary framework. Examination of the distributions of the type 10- calities on the range maps of each species of swollen- thorn acacia leads one to suspect what is shown by examination of the types, namely that most syno- nyms are the result of the taxonomist possessing a herbarium sheet from two widely separated portions of a species population but having no information on the clinal changes across the connecting part of the population. Further taxonomic difficulties were generated by the greatly differing thorns that may be found on a single plant. There is no evidence that we are dealing with many sibling species but rather that there is extreme geographic variation of several conspicuous vegetative traits. Finally, the more conservative (reliable) vegetative traits of swollen-thorn acacias have not generally been used as taxonomic characters. The early parts of this study emphasized the location and delimitation of populations of swollen- thorn acacias in Central America. Once these pop- ulations were understood geographically and morphologically, then the type specimens of swol- len-thorn acacias were associated with them in the herbaria. Since virtually all the variation encoun- tered in this study was clinal, and since the clines showed little or no concordance with the subpopula- tions of a given species (being contiguous almost without exception), there is no basis for the designa- tion of infraspecific Latin names. Fortunately, the breaks between species are clearly delimited by many morphological, behavioral, and ecological traits in the numerous areas of sympatry. The species discussed here are clearly morpho- species, but all circumstantial and ecological evi- dence indicates that they are independent evolutionary units that do not normally exchange genetic information. The various species grow side by side over much of their ranges with almost no obvious hybridization, though ant-acacias may in- trogress with nonant-acacias on occasion (e.g., A. hindsii X A. pennatula (Schlect. and Cham.) Benth. and A. hindsii X A. cochliacantha Humb. and Bonpl. in western Mexico). Those species that have no range overlap with other swollen-thorn acacias are presumed to be distinct species because of the P BRACTEOLE PULVINUS FIGURE 2.-Swollen-thorn acacia terminology. many morphological and ecological differences be- tween them and other acacia populations. The total morphological variation within the populations of some of the widely distributed spe- cies (e.g., A. cornigera, A. collinsii, A. hindsii) is much greater than that within pairs of closely related species (e.g., A. allenii and A. melnnoceras, A. nzayana and A. cornigera). This, plus the im- possibility of designating what might be called a ?species-level? trait, prevents a uniform criterion for the designation of ?species.? Swollen-thorn acacias are readily determined to species in the field, either in vegetative or sexual phases. In fact, it is easier in the field than from herbarium specimens. Because of this, only oc- casional voucher specimens were collected in this study. Since swollen-thorn acacias are generally recognizable even from a fast-moving car, continu- ous distribution records along roads were often obtained in this study and are plotted on the distribution maps as continuous data rather than as ?spot localities? as are usually taken from her- barium specimens. In most parts of Central America, the total absence of swollen-thorn acacias for more than 5 km along a road or watercourse means that they are absent for reasons other than persistent attempts by humans to remove them. Their absence within areas of secondary vegetation of a few hectares in extent may be due to nonhuman FIGURE 3.-Representative leaves subtended by type A swollen thorns: A, B, Acacia cookii; c, A. hindsii; D, A. gentlei; E, A. ruddiae; F, A. globulifera; G, A. rnelanoceras; H, A. collinsii; I, A. cornigera. (All photographs taken to same scale; the leaf of A. cookii in upper left-hand corner is 40 cm long.) FIGURE 4.-Representative leaves subtended by type A swollen thorns: A, Acacia allenzi; B, A. maynna; c, A. cliiapensis; D, A. sphaevocephala; E, A. chiapetzsis. (Scale as in Figure 3.) 10 SMITHSONIAN CONTRIBUTIONS TO BOTANY TABLE 1.-Meristic traits of two representative fully developed leaves subtended by type A thorns on lateral branches of occupied adult swollen-thorn acacias (mean for large samples are not presented since this would be a study in itself; these two leaf exemplars are judged to be representative by my own obser- vations) - - - Total Total Total Length X X X Total Number of Number of Species and thorn length (mm) allenii (55+56) . . . . . . . , allenii (50 + 49) , , . , , . , , hindsii (41 +42) . . . . . . . . hindsii (36+38) . , . . , , . . cookii (51 + 48) . . . . . . , , cookii (67 + 71) , , , , , , , . mayana (51 +48) . . . , , . . collinsii (46 + 48) . . . . . . . collinsii (42+43) , . . . . , . . gentlei (48 + 48) . . . . . . , gentlei (38+39) . . . . . . . . . globulifera (43+44) . . , . . globulifera (39+33) . . . . . cornigera (45 +47) , , , . . , . cornigera (43 + 42) . . . . . . . ruddiae (no thorn) . . . . . . ruddiae (no thorn) . . . . . . chiapensis (48+43) , . . . , . chiapensis (79+75) , , . , . , sphaerocephala (63+63) sphaerocephala (54+52) , melanoceras (36+34) . . . . melanoceras (32 + 45) , , . . - leaf number number longest length length width pinnules nectaries nectaries ength pinnae pinnules pinna pinnae pinnules pinnules on longest on rachis on petiole (mm) (mm) (mm) (mm) (mm) pinna 383 84 1170 98 70.7 10.53 3.08 47 19 4 40 1 178 150 404 367 225 101 101 100 106 252 228 110 81 205 258 288 190 155 149 255 237 38 42 42 96 58 20 20 22 14 16 48 52 20 12 38 74 74 42 26 24 48 46 1?630 1924 1912 12242 7013 1049 588 632 784 988 4007 4080 848 536 3200 8180 7644 2955 952 1072 2134 1592 121 47 42 69 87 100 42 47 65 58 62 62 54 55 67 62 63 48 54 53 49 35 81.1 36.3 35.6 61.5 77.5 72.5 33.6 36.6 55.0 52.1 49.2 46.9 43.2 50.5 61.4 47.6 46.7 39.2 36.7 46.1 38.9 28.8 10.30 2.94 63 4.46 1.58 56 4.17 1.26 52 4.04 0.60 140 6.82 1 .oo 136 10.90 1.93 66 6.39 1.89 34 8.18 2.22 38 8.45 1.98 62 8.53 1.93 64 5.51 1 .oo 89 5.14 1.00 96 7.09 1.61 52 7.26 1.82 48 5.77 1.18 88 2.87 0.80 144 2.83 0.77 130 4.46 1 .oo 88 6.34 1 .go 46 6.28 1.76 52 4.90 1.12 54 5.46 1.10 42 28 21 21 54 28 9 1 0 4 8 14 1 4 20 37 35 0 0 0 25 24 a ?3 0 0 0 1 1 4 1 2 4 2 1 1 2 0 4 14 1 I 18 21 8 factors, but more likely is due to repeated cutting immediately followed by burning. Relative changes in density are especially susceptible to human ac- tivity. It is therefore very difficult to determine what climates and/or habitats are ?best? for swollen-thorn acacias. Cutting, burning, and graz- ing patterns up to at least 20 years previous may grossly influence both density and presence of swollen-thorn acacias. There are even cases of heavy insecticide use, where the ants, and apparently most other insects, have been killed and the unoccupied acacias are relatively healthy (e.g., A. hindsii in the Leon area of western Nicaragua, along cotton field edges). Some of the herbarium specimens collected during this study have been deposited in the National Museum of Natural History of the Smithsonian Institution (US), Gray Herbarium (Harvard) (GH), New York Botanical Garden (NY), University of California at Berkeley (UC), Missouri Botanical Garden (Mo), Field Museum of Natural History (F), University of Michigan (MICH), and University of Mexico (MEXA). The majority have been deposited as intact series at the Field Museum. Despite their spectacular association with ants, swollen-thorn acacias have been poorly collected and generally receive little attention in regional floras. They are not famous for medicinal uses, which is not surprising since the ants are their ?chemical? defense system. Collectors consistently avoid them (or more properly, avoid their ants). Those parts of the plants that are collected tend to be old foliage and branches, which are poorly patrolled by the ants and often unrepresentative of the foliage of the plant as a whole. The branches with well-developed swollen thorns, especially those with very large thorns on short laterals, are almost entirely missing from herbaria, presumably because they do not press easily and usually add a large number of ants to the plant press. This produces the illusion, when examining herbarium material, that the thorns of swollen-thorn acacias are gen- NUMBER 13 11 FIGURE 5.--Shoot tip from end of an elongating healthy branch: A, Acacia coo& frorn the Cardenas, Tabasco, Mexico, region; B, A. allenii from the Rincon, Osa Peninsula, Punta- renas Province, Costa Rica, region. erally small, monomorphic, and without much geographic or intraplant variation. Wheeler (1942) reviewed the generally undistin- guished presence of swollen-thorn acacias in the literature prior to 1930. Acacia cornigera was the first to be described (Linnaeus, 1753) as Mimosa cornigera, but since it was based on a European greenhouse specimen, the ant interaction was not discussed at that time. The last species to be de- scribed in the current literature, A. mayana in 1937, is the rarest and most restricted of the swollen- thorn acacias. The two species described in the present paper (A. allenii, A. ruddiae) had been collected earlier, but the herbarium specimens were misidentified. Herbarium specimens are often misleading in re- spect to the ant-acacia interaction. There is gen- erally no way of knowing if the specimen is from a heavily occupied acacia, one that has recently lost its ant colony, or one that was never occupied. The presence of ant-entrance holes in the thorns are not reliable indicators of occupation because founding queens may have made them and because, once made, they last as long as the thorn (2 or more years). Occupied acacias (those with an ant colony large enough to patrol the tree) have fully devel- oped lateral branches, short lateral branches with very large thorns, inflorescence-bearing branches, etc. Unoccupied ones have stunted branches with short internodes, smaller leaves, fewer inflorescences at times only partial damage and shading; directly, they are not a product of the ants. From 60 to 95 percent of the mature swollen- thorn acacias of each species examined in this study were occupied by obligate acacia-ants. The excep- tions are A. ruddiae, which is never occupied by obligate acacia-ants, and A. sphaerocephala. Where the latter species grows on the dunes south of Vera- cruz, Veracruz, Mexico, it is unoccupied within range of heavy wind and salt spray on leading dunes, but tenanted by Crernatogaster. Those plants tenanted by other ants (they often use the hollow thorns for nest sites as they would use hollow branches on other trees) are unoccupied from the plant?s viewpoint. The statements about vegetative characters in this paper are drawn from healthy and occupied acacias. Numerically, the majority of any swollen-thorn acacia population in repeatedly disturbed sites is made up of unoccupied seedlings and sucker shoots, but these are almost never no- ticed or collected because they are less than 50 cm tall and below the general weed canopy (mature acacias are usually emergent). These small plants are often much more difficult to identify to species than are the adults, owing to their convergence in leaf and thorn characteristics. The swollen-thorn acacias did not escape from Britton and Rose?s (1928) attempts to split the genus Acacia into more manageable subunits, The genera Tauroceras and Myrmecodendron were proposed but are not retained in the present study because they represent only one of the many ways the 12 species could be arbitrarily grouped, and thus add little to the understanding or ease of manipulation of the morphological affinities within the swollen- thorn acacias. Perhaps of more importance in subgeneric taxon- omy, the evolutionary affinities of the 12 species are not clear in most cases, and their speciation may have involved some very interesting complica- tions. Despite their possession in common of many unique vegetative traits-Beltian bodies, swollen thorns, etc.-the swollen-thorn acacias appear to be a polyphyletic group within the genus Acacia. Pairs or triplets of species appear to be more similar in the majority of their traits to acacia species not 12 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE &-Variation in nectary morphology on mature A. globulifera plants (and see Figure 7c): A, maximal number of petiolar nectaries, as occasionally found on old trees in tufts of leaves not subtended by swollen thorns (Honduras, site 26); B, maximal condensation of the petiolar nectary area, as characterizes leaves from type A thorns at high (cold) and dry areas (Honduras, site 28) ; c, maximal attenuation of the petiolar nectary area, as characterizes leaves from type A thorns on rapidly growing 2-4-year-old trees in riparian (wet) sites (Honduras, site 26). Acacia collinsii sometimes has nectaries similar to Figure B, and A. gentlei sometimes has nectaries similar to Figure C (and see Figure 7). involved in mutualism with ants than they are to other swollen-thorn acacias. Even the designation of subgenera based on these small groups seems to have little usefulness (e.g., Tauroceras would con- tain A. cornigera, A. sphaerocephala, and A. may an a). The possibility of a polyphyletic origin for swollen-thorn acacias is especially noticeable if only the reproductive parts are compared. Three expla- nations are available. The first, and least likely in view of dissimilarity of sexual reproductive parts, is that a single swollen-thorn acacia evolved, and by classical speciation patterns split into the 12 modern populations (species) discussed in this pa- per. This carries the implausible implication that floral and fruit morphology are as evolutionarily unconservative as the vegetative parts of the plant, and that the similarity of the flowers and fruits of swollen-thorn acacias to those of nonant-acacias is simple convergence. The second possibility is that each cluster of similar swollen-thorn acacias had an independently evolved, swollen-thorn acacia paren- tal species. Here, the vegetative similarity between the clusters would be convergence selected for by the requirements of the mutualistic interaction. This also seems unlikely since it requires that a complex interaction must have arisen several times independently. The third hypothesis, and the one that I favor, is a combination of the two above by the mechanism of introgression. This would yield a reticulate phylogeny. Starting with one species of swollen-thorn acacia, new acacia species become part of the system by introgression of the set of ant- acacia traits into their genotype. The outcome is two species of swollen-thorn acacia, one of which (the donor) is probably not much changed while the other may change its ecology considerably since it now possesses a very different, and in some ways more effective, set of defenses against herbivores and competitors. NUMBER I3 I3 FIGURE 7.-A, Representative petiolar nectary on a leaf subtended by a type A thorn of A. col- linsii from the moister end of the species? range; the number of nectaries may range from 1-4 and there never are rachis nectaries (Mexico, site 15). B, Representative petiolar and rachis nectary of A. gentlei on a leaf subtended by a type A thorn (Guatemala, site 21). c, Representa- tive petiolar and rachis nectary of A. globulifera on a leaf subtended by a type A thorn (Guate- mala, site 13). 14 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE %-A, Representative petiolar and rachis nectaries on leaf from type A thorn of A. hind- sii; the number and size of rachis nectaries is variable (Mexico, site 17). B, Representative petiolar nectary of A. cornigera; on rare occasions there are two petiolar nectaries and a rachis nectary (Mexico, site 6). c, Representative petiolar and rachis nectaries on leaf from type A thorn of A. ruddiae; nectaries vary greatly in height (Guatemala, site 8). This type of ?speciation? (note that there is no net increase in the number of species of Acacia) will almost certainly result in a range extension for the ant-acacia interaction, and perhaps lead to con- tact with, and subsequent incorporation of, other nonant-acacia species. It appears to be occurring now between A. hindsii and A. pennatula (see later discussion of hybrids). This process, coupled with the usual kind of speciation, could easily lead to the present 12 species. Materials, Methods, and Terminology A bit of jargon has evolved in the discussion of swollen-thorn acacias and needs brief review here (also see Janzen, 196713). An ?occupied? swollen- thorn acacia is one that has an ant colony large enough to patrol its surface with sufficient intensity such that the acacia will gain or remain in moderate to excellent health. An unoccupied acacia may therefore have founding queen ants in its thorns or have a large colony of an unprotective ant species tenanting its thorns. The ?basal circle? is that circular patch of bare ground directly under a swollen-thorn acacia that results from the ants attacking seedlings and sucker shoots beneath the acacia. A ?multiple-queen? ant colony is of one of the species that has egg-producing queens in many of the tree?s thorns, in contrast to having only one NUMBER 13 15 queen per colony. Multiple-queen ant colonies tend to be associated with swollen-thorn acacias that naturally occur as dense thickets, such as A. hindsii in western Mexico. The stipular thorns of Neotropical swollen-thorn acacias are not galls in any sense of the word. Inci- dentally, despite the caution displayed by Hocking (1970), the enlarged thorns of African acacias are not galls either, as shown by their full development in North American greenhouses. ??Type A? thorns are those more ordinary-appearing thorns to be found on long, lateral branches and on the central axis; ?type B? thorns are the large and often con- voluted thorns found on greatly shortened lateral branches. See Figure 2 for other morphological characteristics. For precise localities, the lists under each species should be referred to, rather than to the distribu- tion maps. Localities are listed under the state or province containing a reference town. The pro- jected ranges on the distribution maps are just that; they are based on my intuition and the gen- eral ranges of other plants from the same habitats. They are probably overall correct in general, but cannot be expected to be accurate in detail. The key collection localities are given in the distribution figures as sites or as continuous distributions along roads (solid broad lines). Where my own distribu- tion records overlap those obtained from herbarium specimens, to avoid redundancy I usually have not given the herbarium data. The site numbers are unique only within a species; for example, site 16 for A. cornigera is not the same locality as site 16 for A. hindsii. While some data are given as to the reproductive condition of the plants on the dates of observations or collection (fl = inflorescences, gp = immature fruits, mp = mature fruits), little reli- ance should be placed on these data in determining sexual timing owing to the collectors? habit of seeking out plants in flower and fruit but out of phase with the remainder of the population. All scales shown in photographs are in milli- meters unless the caption indicates otherwise. In the terminology of this paper, ?isotypes? are collections made from the same plant at the same time, while the ?holotype? is only that herbarium sheet with the type on it. ?NC? is used to signify a plant examined in the field but not collected. While Safford?s descriptions sometimes were not in Latin, they were published before 1 January 1935 and are still recognized as valid in accordance with Article 36 of the International Code of Botanical Nomen- clature (1972). General Morphological Traits The leaves of swollen-thorn acacias differ from those of other Central American acacias primarily in being larger overall and having larger leaflets; in short, the leaves extend far past the tips of the swollen type A thorns and can be easily removed from the tree without contacting the thorns. On any given tree the leaves vary strongly in size as a function of season (smaller leaves being produced during the dry season) and part of the tree (smaller leaves being subtended by type B thorns). However, if only the leaves produced during the rainy season and subtended by fully developed type A thorns on lateral branches from healthy trees are consid- ered (Figures 3 and 4, Table 1) , there are diagnos- tic interspecific differences in leaf and leaflet size and shape. Not surprisingly, those species from the driest habitats have the smallest leaves (e.g., A. collinsii, A. cornigera, A. hindsii). There may be considerable variation within a species over its range on the parameter of leaf size. In Figure 3, a leaf from A. cookii at the northern and dry end of its range in the Cardenas, Tabasco, Mexico, area (Figure 3~) may be compared with a leaf from the wet end of the range in the Puerto Barrios, Guate- mala region (Figure 3~). One species, A. chiupensis, shows strong variation at a given site as well, as earlier discussed for the Temascal, Oaxaca, Mexico, area (Janzen, 1967b); this variation (Figures 4c and 4E) is probably an expression of introgression of Acacia macracantha and A. cornigera genes into the A. chiupensis population in this area (and see Rehr et al., 1973). The shoot tips on elongating branches of swollen- thorn acacias are unprotected physically (as well as chemically) when compared with other plants. The shoot tip of A. allenii in Figure 5~ is repre- sentative of swollen-thorn acacias in being free of hairs or defensively sticky compounds and in having the newly expanding leaves quite separate and ex- posed. They are extremely brittle and easily broken, reminding one of the inner shoot of a celery plant. The shoot tips of A. cookii (Figure 5A) and A. ruddiae are more like those of nonant-acacias, with the new leaves clasped tightly around each other 16 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 9.-Ontogenetic development of nectaries of A. collinsii: A, characteristic seedling leaf produced at about the 20th node and often not subtended by swollen thorns; B, leaf subtended by type A thorn on juvenile but well occupied plant; c, leaf subtended by type A thorn on relatively short lateral branch in crown of mature acacia. and covered with a dense pile of short stiff hairs. In the northern part of its range, A. cookii is some- times unoccupied, and in this area the difference is more pronounced than in the Puerto Barrios, Guatemala, region where the shoot tips appear more like those of a regular swollen-thorn acacia. The extrafloral nectaries on the petiole and rachis of swollen-thorn acacia leaves are one of the most useful and reliable characters for species-level determinations. While extrafloral nec- tary morphology may vary strongly in detail from area to area for a few species (the most extreme case being A. globulzfeia, Figure 6), for the ma- jority of species there is little geographic variation and at any given site the morphology is quite con- stant; it may be the only way to distinguish between young plants of species that are easily separated as adults on the basis of thorn shape. Since the entire function of the extrafloral nectaries appears to be physiological-excreting sugar for the ant colony- they are not under strong pressure for adaptive ex- ternal morphologies; much as with the thorns, I suspect that numerous different kinds of external morphology may be equally functional and that the interspecific variation is primarily due to inter- specific differences in the adaptive peaks from which the different swollen-thorn acacia genotypes were derived. The extrafloral nectaries in Figures 7 and 8 and in Table 1 are representative of the gamut of interspecific variation. As a seedling grows, there is a progression of extrafloral nectary development (young sucker shoots from cut stumps tend to have NUMBER 13 17 FIGURE ~O.-A, Cross section of a 5- year-old main trunk of A. hindsii (Mexico, site 17). 8, Conspicuous lenticels on 6-month-old branch of A. ruddiae (Costa Rica, site 5). FIGURE 1 1.-Representative bark topography of 2-5-year-old main trunks of swollen-thorn aca- cias: A, Acacia cornigera (Mexico, site 26) ; B, A. gentlei (Guatemala, site 11); c, A. chiupensis (Mexico, site 9) ; D, A. hindsii (Mexico, site 15) ; E, A. ruddiae (Costa Rica, site 5) ; F, A. al- lenii (Costa Rica, site 5). 18 Species mayana , , . . , . . , . . . SMITHSONIAN CONTRIBUTIONS TO BOTANY Knobby Thorn projection shape in Thorn on inner cross surface External thorn wall section General thorn contour pubescence at base near base configuration smooth c+ none present elliptical flat V that is bent downward slight ridges smooth smooth with 2 sharp elongate projections absent round to slightly elliptical slightly in larger thorns; never recurved around stem gentlei . . . . . . . . none flat V, slightly distorted and bent irregularily in the plane of the V; thorn may be closely appressed to stem longitudinally and partly bent around it ,... ,... ..., .... globulifera . . , . melanoceras . . . cookii . . . . . , . . absent round flat V, slightly to strongly curved downward; never recurved around stem flat V, slightly curved downward in larger thorns heavily warped U, asymmetrical with one side or both twisted around the stem partly or fully; asymmetry highly regular in pattern flat V, tendency for one thorn half to be larger than the other; never recurved about stem none smoothed longitudinal ridges smooth smooth c slight ridges sometimes smooth smooth smooth none heavy present round to elliptical absent round ruddiae very light absent round to slight ellipse round chiapensis .... .... ,... none absent flat V that may be bent slightly downward in very large type B thorns flat V; never recurved around the stem flat V or U, bent upward or down- ward to various degrees; type B thorns often heavily recurved around stem, at right angles to it sphaerocephala cornigera . . . . . none none absent absent round round to slightly elliptical collinsii none usually present smooth smoothed longitudinal ridges smooth to faint smoothed longitudinal ridges round to elliptical to very elongate and thin elliptical round to flat V that may be very open and curve around the stem at various angles allenii . . . . , hindsii . . . . flat V; never recurved around the stem none present none absent elliptical to very elongate and thin flat V; never recurved around the stem NUMBER 13 on adult plant over entire range brown, nearly black; trace of red on green thorns dark brown, nearly black to ivory white; bicolor beige + ivory, brown+ ivory ivory; darker thorns with reddish cast ivory to beige to black very dark brown to very dark brown, nearly black; no trace of red on green thorns red to brown to very dark brown 19 foliar nectaries and other leaf traits essentially identical at the qualitative level to those of the adult crown). A characteristic progression is dis- played in Figure 9; in general, with increasing age of the plant, the number of nectaries increases, their size increases, the distance between the first pair of pinnae and the thorn decreases, and nectar flow increases. Swollen-thorn acacias are more ?treelike? than other Central American acacias in that they tend to have more of the growth concentrated in the central axis (see figures in Janzen, 1967b). The wood is harder than that of most fast-growing tropical trees and generally displays obvious annual growth rings (Figure 10~); the slower wood produc- tion coincides with the dry season and, in areas with almost no dry season, the rings are much more difficult to see if they are present at all. The tree bark (Figure 11; Table 5) is thin and often green- ish; presumably considerable photosynthesis occurs just under the bark. There are well-developed lenticels on young branches (Figure ~OB), and many of these persist on trunks 2-6 years old (e.g., Figures The swollen thorns on adult acacias are very distinctive at the species (and sometimes local- population) level, but the differences are difficult to quantify and incorporate in a dichotomous key. Tables 2 and 3 attempt to organize some of these differences, but for a good feeling for the differences the reader must also compare the thorn photographs associated with each species account. Thorn color is perhaps one of the most useful diagnostic species- level traits at a given site, but varies strongly with location and the plant?s history. Damage to green thorns may produce premature darkening of the llA, C, E, F). TABLE 3.- Color of swollen thorns for seasoned and undamaged types A and B thorns from living branches on occupied adult swollen-thorn acacias 1 Thorn color Usual color Color at any Color among Species allenii . , , . . . . . cornigera collinsii hindsii rnelanoceras . . gentlei .. change from one site 50 cm to among adult plant adult plants none no variation observed beige to dark beige+brown; brown; white brown+very to ivory dark brown; ivory+ brown beige to beige + brown; brown beige + ivory beige to beige + black; black; white ivory+ beige; to ivory; brown + black beige to brown none no variation observed brown to red + dark reddish; brown light brown to dark brown seedlings Color clines over entire where obvious range as adult plants in full sun have slightly browner thorns than those in full shade white to brown, usually beige ivory in dry and hot lowlands to very dark brown in wet upper elevations; ivory in dry lowlands to brown in wet lowlands white to none obvious beige to brown ivory to beige, lightest thorns in driest sites but black thorns there also; black thorns most common at low elevations, light at high elevations usually brown as adult none obvious reddish brown none obvious to brown 20 SMITHSONIAN CONTRIBUTIONS TO BOTANY TABLE 3.- Color of swollen thorns for seasoned and undamaged types A and B thorns from living branches on occupied adult swollen-thorn acacias-continued Species globulifera . . . . chiapensis ruddiae mayana , . . . . . . sphaerocephala . cookii Thorn color Usual color Color at any Color among on adult change from one site seedlings Color clines plant over 50 cm to among over entire where obvious entire range adult plant adult plants range brown to bicolor brown + brown to bicolor brown+ none obvious except that wet sites olive to ivory; brown black to bi- ivory; brown bicolor brown to white; color brown+ with white ivory and white to black ivory thorn tips ivory+ black very dark gray reddish cast to green thorns black; reddish cast to green thorns brown (seedling unknown) lark brown to none dark brown to none very dark none lark brown to none ivory: bicolor brown + beige, black or very dark brown + ivory; yellow dark red lark brown to none no variation dark brown to observed very dark gray no variation dark brown to observed black no variation as adult? observed (only (seedling one plant unknown) known from each site) brown to usually cream ivory; bicolor to beige dark brown + ivory to yellow dark brown to as adult dark red appear to have the darkest thorns none obvious none obvious none obvious darkest in wettest sites, most strongly bicolor inland and solid color near coast none obvious thorn wall. The death and drying (seasoning) of the thorn wall proceeds from tip to base, often resulting in temporarily bicolor or tricolor thorns; in some areas, however, all thorns are bicolor irre- spective of age. The thorns of all species turn lighter with age and bleaching by the sun. There is likely to be some adaptive significance to thorn color in the context of optimizing the internal thorn microclim- ate for the ants. In addition to detailed swollen- thorn morphology, the placement of the thorns on the branches varies from species to species; when looking down the long axis of a branch, the thorns either radiate out in all directions or lie in two parallel planes, one on each side of the branch. This appears to be a by-product of leaf placement, rather than being adaptive in the context of the thorns themselves. In general, the species of dry regions tend to have small leaves-and thorns-radiating outward from all sides of branches that are oiten more vertical than horizontal (e.g., A. collinsii), while those of wetter areas often have the leaves in a more or less horizontal plane from more hori- zontal branches. There are conspicuous differences among the species as to the proportion of leaves lacking swollen thorns (as in Figure 10~) but quanti- tative data are not available on this parameter. This parameter is also greatly confused by strong vari- ation in swollen-thorn production as a function of tree age and health; in general, seedlings and senes- cent trees produce the fewest swollen thorns, and leaves produced during the dry season often lack swollen thorns. Though not discussed here, there is moderate interspecific and intraspecific variation in thickness of the thorn wall, thorn-wall hardness, longitudinal fiber orientation in the thorn wall, insulating quality of the thorn wall, point of attach- ment of the leaf to the thorn, ease of thorn removal, and rates of natural thorn shedding. Healthy swollen-thorn acacias produce large num- bers of bright yellow inflorescences during a short part of the dry season. They are either elongate cylindrical (Figure 12) or spherical (Figure 13) and NUMBER 19 21 FIGURE 12.--Swollen-thorn acacias with elongate inflorescences: A, Acacia cornigera (Honduras, site 63) : B, A. collinsii (Mexico, site 15) : C, A hindsii (Mexico, site 16). composed of many individual florets (Table 4). Pollinators-pollen-gathering bees-generally move around over the surface, scraping off pollen with their legs; there is no evidence of nectar production. Those inflorescences that are pollinated are re- tained with their very small and immature pods through the following rainy season, but only a small fraction of the inflorescences are retained. Presuma- bly the plant produces a great number of inflores- cences to attract pollinators, and then sheds what it does not have sufficient energy to mature. A. cornigera and A. collinsii have produced large quantities of flowers in greenhouses at the Uni- versity of Kansas and the University of Chicago. Swollen-thorn acacias differ from all other Cen- tral American acacias (and probably from the re- mainder of the genus) in having bird-dispersed seeds (Janzen, 1969~). The seeds are enclosed in a sweet white to yellow pulp or aril, and in all species ex- cept the A. cornigera, A. sphaerocephala, and A. mayana group the pod opens to expose the pulp when ripe (Figure 14; Table 5). Presumably this 22 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 13.-Swollen-thom acacias with spherical inflorescences: A, Acaciu globulifera, florets unopened; two immature pods on receptacle on far right (Honduras, site 28); B, A. melanoceras (Canal Zone, site 6); c, A. allenii (Costa Rica, site 2). NUMBER 13 23 Species collinsii . . . , , . . . Distance Range Infl. In$ Znfl. Znfl. Range Range Number from sexual peraxil bearing General max. max. inp. Peduncle peduncle florets bracteole branchlet of axils per shapeof diam. length length length length this toinfl. length sexual sexual inflorescence (mm) (mm) (mm) (mm) (mm) infl. base (mm) (cm) branchlet branchlet elongate 4 17 14-20 12 11-13 227 10 5-14 3-6 5-8 cylinder mayana , . , . . . . . gentlei . . . . , . . globulifera . . . . . melanocerac . , . . cookii . . . , . . . . . terminally tapered cylinder cylinder elongate globular globular globular ruddiae . . , . . . . . chiapensis . . . . . sphaerocephala . hindsii cornigera , , . . . . . . . . . . . . allenii . . . . . . . . . 3.5 4.5 3.5 5.0 4.5 4.0 3.8 4.0 4.0 5.5 5.5 ? 5 5 9 6.5 6 6 6.5 6 7 7 6 6 8 9 5 5 globular globular nearly globular cylinder cylinder tapered tylinder elongate slightly globular 17.5 I, 26 22-26 23 20-25 30 25-34 46 40-50 23 20-24 21 28 26-28 29 5 4-5 5 11 ? ? 5 45 5 9 9-10 6.5 6-7 6 5.5-6.5 6 6 11 10-14 10 27 25-36 22 15-23 n w I, I1 6.5 5.5-7.0 I, I, 26 28 I1 I1 5 4-5 5 It 10 16 8 3 10 14 8 14 15 6 12 12 3 6 37 8 9 9 18 20 12 12 20 20 5 6 3 7 I, 14-18 13-17 1-4 5-12 8-12 10-15 6-12 10-12 3-6 26-38 8-18 6-10 16-20 8-12 A ,, I1 I, I1 I1 ,I 1422 1-10 11 I1 3-6 ,I 222 301 283 1003 1408 533 534 494 519 69 124 ? 54 117 537 105 198 182 124 120 340 319 153 142 1112 992 65 91 8 12 5.5 .2 2 8 5 8 10 6 12 12 1 1 18 4 3 3 5 3 10 10 13 13 4 5 2 3 I, 2-3 4-13 13 5-15 2-14 6-9 1-9 8-22 10-20 85-1 80 10-23 4-12 15-30 II I, ,I I, I1 11 2-18 ,I 10-23 I, 5-20 r, 3-15 I1 I, 2-3 3-5 1-2 1 1 1 2-5 It w n 2-5 3-5 37-86 25-46 n 8-8 ,I 35-60 34 ,I n 5-7 1-4 I, I1 20-40 I, I, 5-6 4-8 20 21 4-15 4-10 41 1 10-20 29-36 16-20 5-18 2-4 10-15 2-12 I1 I, ,, II 0 I1 II 8-1 3 11 4-1 3 I1 2-10 ,I difference reflects the ease with which birds can enter the crown and area beneath an occupied swol- len-thorn acacia, as contrasted to mammals which are much more bothered by the ants. The pods of swollen-thorn acacias are very diagnostic at the species level except within the A. cornigera group. A. collinsii displays the most extreme geographic variation in pod morphology of all the species, rang- ing from very short and linear pods with thick walls in the drier parts of Panama to much longer, curved, and thin walls in Mexico and Guatemala. All swollen-thorn acacias have very hard seeds (ex- cept A. allenii and A. melanoceras) which pass through the intestinal tract unharmed and are scari- fied in the process. Swollen-thorn acacias are sub- ject to heavy seed predation by bruchid beetles (e.g., Janzen, 1969c) except for those species in very wet habitats-A. melanoceras and A. allenii and perhaps A. ruddiae. The general pattern is that the eggs are laid on the full sized and nearly mature pod, the larvae bore in through the pod wall and into a seed, and one larva matures inside each seed. Upon emerging from the pod, the new adult beetle will oviposit on pods that remain, resulting in de- struction of that portion of the seed crop that has not been removed by dispersal agents (Janzen, 1969c.) Undamaged and scarified seeds yield almost 100 percent germination if taken from truly mature pods. With considerable experience in studying the populations in the field, all species of swollen-thorn acacias can be correctly identified from specimens 24 SMITHSONIAN CONTRIBUTIONS TO BOTANY TABLE 5.-Reproductive and other traits of swollen-thorn acacias recorded from occupied adult trees Species mayana , . . . . . . gentlei . . . . . glo bulifera melanoceras cookii , . . . . . . . . ruddiae . . , . . , . chiapensis . , . . , sphaerocephala . hindsii . . cornigera , . . . . allenii . . . . . . . . . collinsii . . . . . . . . Estimated Maximum Maximum Life Bark age at first Pod valve Pod Color of height basal form color seed repro- color when dehiscion Aril mature of adult diameter of and iuction (yrs) mature behavior color seed (m) (4 adult texture 4-6 maroon indehiscent yellow dark 45 10 single smooth 6 5 8 a 10+ 6-7 4 6 4 7 4 brown black dehiscent pale olive along both yellow green margins but cen- trally together along one margin and strongly curled black dehiscent yellow greenish dehiscent white black along one margin olive green black black ? greenish ? black trunk gray - tree green trunk rust red tree 14 12-25 single rough 8 12 dark dehiscent white brown along both margins maroon indehiscent yellow yellow red brown with waxy bloom maroon yellow red greenish black dark brown 12 15 ? ? 20 30 ? 7 18 50 dehiscent yellow along one margin yellow dehiscent white along both margins along both margins dehiscent yellow greenish beige black dark brown black black olive drab to black single olive drab trunk to reddish tree single or green to multiple light trunk tree brown single dark trunk brown tree single smooth gray- trunk green with tree longitudinal whitish cracks 5 20 single rough trunk rust tree red 1-4 15 single trunk smooth tree to gray- multiple green trunk shrub smooth la 35 single trunk gray- tree to green multiple trunk tree 6-18 20 single trunk smooth gray- tree green to rough 25 25 single trunk brown tree green to light brown, 6 15 single trunk rough tree silvery gray to brown NUMBER 13 25 species accounts. Table 6 presents the degrees of sympatry and gives some feeling for the amount of caution required to identify the species on collec- tion locality only; A. melanoceras and A. allenii are the only species that are never sympatric with other swollen-thorn acacias. While hybrids of swollen-thorn acacias with other acacias (A. globulifera, A. hindsii, A. collinsii, and A. chiapensis are the offenders in this case) may be difficult to place to species, swollen-thorn acacias apparently hybridize only very rarely (though un- derstandably they might be difficult to notice); I have seen only one, and that was a single plant of A. cornigera x A. sphaerocephala on the dunes south of the city of Veracruz, Mexico. FIGURE 14.-~, Newly dehisced fruits of A. collinsii, with the yellow pulp enclosing the seeds and hanging down beneath the valves on thin fibers; birds pick these off in flight (Costa Rica, site 71), B, Opened pod valves of A. hindsii; the pulp and seeds have been removed by birds (Mexico, site 17). from adult trees even if the localities are unknown. Individual variation owing to local ecological fac- tors, however, can make this process confusing, and knowledge of the collection site often aids tremen- dously in quick and accurate identification; the field work in this study was extensive and intensive enough so that the reader should be extremely suspicious if his identification leads him to decide that he has located a population of a given species far from the ranges listed under the individual Taxonomic Information While working with a population of swollen- thorn acacias, one should have no difficulty in identifying them with key A. Key B is provided for use in conjunction with key A, when flowers and fruit are not available. On occasion there may be some difficulty in identifying old and very incom- plete herbarium material with these keys, but that is not likely to be a significant problem inasmuch as the value of species-level determinations for such material is highly questionable anyway. When a population of swollen-thorn acacias is encountered during the rainy season, sexual material may be difficult to locate. If one searches for damaged trees, however, often an occasional inflorescence-bearing branch may be located. Since the immature infruc- tescences are retained on the tree through the rainy season, careful search for these usually succeeds in locating sexual material. It is worth noting that the inflorescence receptacle generally retains its shape while bearing immature or mature fruits (e.g., Figure 112), allowing easy use of the first couplet of key A even when fresh inflorescences are not avail- able. The physical locations of the type specimens of swollen-thorn acacias are given in Table 7, but these are generally of little use in identification. Vernacular names are of little use in correct identification of swollen-thorn acacias, inasmuch as they tend to be lumped under ?cornizuelo? or some similar word at a given site (Table S), in contrast to the other acacias which generally carry names such as ?guisache.? 26 SMITHSONIAN CONTRIBUTIONS TO BOTANY Name collinsii . . . . . . . sphaerocephala TABLE 6.-Sympatry of swollen-thorn acacias (+ indicates that they may be found growing within 200 m of each other in some part of the range and that the populations overlap) Species 1 COT. sph. may. glo. coo. hin. col. mel. all. rud. gen. chi. Type location and sheet number Synonyms Location of synonym types US 692159* yucatanensis B (presumed destroyed) NY (cotype)* GH (?type collection?)* US 571618 (?type COIL?)* F 417619 (paratype)? F 417618 (paratype)? US 943274 (potential lectotype) * GH (cotype)* K (same as US 943274)t US 677927 (isotype)* US 764673 (photo of type)* nelsonii US 399366 (type)? costaricensis B (presumed destroyed) penonomensis NY (type)* panamensis B (presumed destroyed) (presumed destroyed) ueracruzensis (presumed destroyed) US 764680 (frag. and photo)* US 764683 (frag. and photo)* US 764684 (frag.)? US 692164 (isotype). NY (isotype)* dolichocephala F 189552 (type)* cornigera sphaerocephala mayana globulifera cookii hindsii collinsii melanoceras allenii ruddiae gentlei chiapensis + ++ +o+ +oo+ +ooo+ +oo+o+ +oo+o++ ooooooo+ oooooooo+ oooo+oooo+ +ooo++ooo++ +oo++ooooo++ mayana . . . . . . . cornigera . , . . . . US 1688106 (isotype)* GH (isotype)* MICH (type)* NY (isotype). BM (no number) US 2396019 (frag. and photQ)* spadicigera US 764687 (frag. and photo)* cu bensis GH (type)* US 764633 (frag. and photo)* MO 16066, 46866 (type)? US 577752 (?type toll.")* GH (?type collection?)* NY (isotype)* nicoyensis B (presumed destroyed) K (cotyfie)t hernandexii US 692170 (type)* NUMBER 13 Acacia species mayana . . . . . . . . . . . . . globulifera . . . . . . . . . . cornigera , . . . . . . . . . , 27 Vernacular Geographic region cruzetillo Campeche, Yucatan, Mexico cachetoro ishcanal cornezuelo blanco Yucatan, Mexico cachetoro ishcanal cornizuelo lowland eastern and southern cornezuelo Mexico, generally throughout Coban, Alta Vera Paz, Guatemala Coban, Alta Vera Paz, Guatemala Central America TABLE 7.-Location of type specimens of swollen-thorn acacias-continued Name hindsii . . , . . . , , . cookii globulifera . . . . . chiapensis gentlei . . . . . . . . melanocerm . . . allenii . . . . . , . . . ruddii . . . . . , , . . . Name gladiata . . . . . . . , hirtipes standleyi oaxacanum Examined in .- - Type location and sheet number Synonyms Location of synonym types GH (isotype)* K (no number)? US 574387 (type)? US 764658 (paratype)* F 58707 (type)? F 36458 (isotype?)* US 1267507 (isotype?)* GH (istotype?). NY (isotype). US 692157 (type)? F 464247 (photo)* F 1000801 (type)* furcella interjecta rossiana campecheana turgida bursaria tepicana sinaloensis bucerophora donnelliana NY (isotyie)* US 692166 (type)* unknown MO (no number)* US (photo). B (presumed destroyed) (photo only, in Wheeler, B (presumed destroyed) GH (cotype)* US 637234 (type)* US 636818 (type)* NY (isotype)* US 764696 (frag. & photo)* F 659304 (?type?)* GH (?type?)* US 943282 (type)? 1942) K (cotype)t locality unknown multiglandulosa B (presumed destroyed) US (photo, no. 1281)* us (type)* us (type)* Type location US 764660 (frag. & photo)* MO 46838 (type)* Bonpl. ex Willd. US 764659 (frag. & photo)* US 716563 (?paratype?). US 694036 (type)* A. hindsii x A. pennatula Benth. US 637251 (type)* US 1208317 (type)* iis study; numbers refer to herbarium sheet numbers. and sheet numbers Hybrids A. hindsii x A. cochliacantha Humb. and A. hindsii x A. pennatula or A. macracantha A. hindsii x A. pennatula t Information provided by Sir George Taylor (in litt.). 28 Acacia species ....... gentlei .............. collinsii ............ SMITHSONIAN CONTRIBUTIONS TO BOTANY Vernacular Geographic region cuernitos Veracruz, Oaxaca, Mexico ishcanal blanco El Salvador cuernos de tor0 ishcanal Pacific lowlands of Guatemala ixcanal carnizuelo Central Honduras torito Salina Cruz, Oaxaca, Mexico cornezuelo Michoacan, Guerrero, Mexico gara terrera Sinaloa, Nayarit, Mexico jarretadera cockspur British Honduras red cockspur cockspur Isla Providencia, Colombia crowsfoot British Honduras San Luis Potosi, Veracruz, Chiapas, Tabasco, Mexico TABLE 8.-Vernacular names of swollen-thorn acacias in Central America as hindsii I cachitos Pacific coast of Panama melanoceras ........ sphaerocephala ..... cookii ............. 1 cockspur Guatemala subin cornezuelo cuerni tos cachitos cornezuelo guascanal Yucatan Peninsula, Mexico Chiapas, Yucatan, Mexico, Porto Bello, Panama Panama Veracruz, Mexico Guatemala Colombia A. Key to Mature Central American Swollen-Thorn Acacias, Based on All Traits 1. Inflorescence distinctly globular ................................................... 2 Inflorescence very elongate or at least distinctly longer than broad ................... 3 2. Leaves from type A thorns (on lateral branches) over 30 cm long ..... Leaves from type A thorns (on lateral branches) less than 25 cm long . 3. Pinnules less than 1 mm wide ...................................................... 6 4. Petiole deeply grooved, young thorns pubescent, nectaries b ised above surface of Petiole slightly grooved, young thorns glabrous, nectaries str outstanding conical, Pinnules 2 or more mm wide .............................. ................. 7 petiole and elongate, occupied by obligate acacia-ants ..................... .I. A. cookii Petiole with many nectaries (8-30) .................................. 3. A. melanoceras Petiole with few nectaries (0-6) ........................................... .4. A. allenii very soft .............. ......................................... .5. A. chiupensis shaped .................................................. 8 not occupied by obligate acacia-ants ...................... . . , . . , . .2. A. ruddiae 5. 6. Petiole with many nectaries (6-25) closely packed and often not in a straight line, wood Petiole with few and widel aced nectaries (1-5) in a straight line ..... .6. A. globulifera 7. Inflorescence peduncle very thick, at least half as wide as inflorescence, nectaries trough Inflorescence peduncle very less than one-fourth as wide as the inflorescence, nec- 8. Inflorescence 34 times as long as broad; secondary venation clearly visible on leaflets. ... 9 Inflorescence one and one-half times as long as broad, approaching a globular form; .................. taries round .......................................... secondary venation on leaflets not visible ........................ 7. A. sphaerocefhzla 9. Swollen-thorns round in cross section ................................... .8. A. cm'gma Swollen-thorns triangular in cross section ................................. .9. A. mayam 10. Inflorescence approaching globular form, but nearly twice as long as wide ................ ................................. ..Hybrid of A. hi&$ with other Acacia species Inflorescence very elongate, 3-8 times as long as broad ................................ 11 NUMBER 13 29 11. Seed pods less than 8 cm long, valves curling heavily upon opening ................... 12 Seed pods more than 12 cm long, curling only slightly upon opening ....... .lo. A. gentlei 12. Petiolar nectaries dome shaped in lateral view with 1 mm or greater diameter, usually 3-4 very close together in widened part of petiole and none on rachis .... .11. A. collinn'i Petiolar nectaries cylindrical in lateral view with less than 1 mm diameter, usually 1 or more on the petiole and one at junction of each pinna with rachis ....... .12. A. hiMi B. Key to Mature Central American Swollen-Thorn Acacias, Based on Vegetative Traits 1. Petiolar nectary trough shaped, much longer than broad (Figure 8~) .................. 2. Type A thorns circular or elliptical in cross section .................................. 3 3. Secondary venation on leaflets clearly visible .......................... 8. A. codgera Secondary venation on leaflets not visible ........................ .7. A. sphaerocephala 4. Type A thorns very heavily flattened, nectaries scattered on petiole and rachis, and small enough so as not to cause heavy swelling of petiole or rachis (Figure 8A) ,12. A. hindn'i Type A thorns circular or elliptical in cross section, nectaries various ................. 5 5. Nectaries restricted to petiole and broad dome shaped (Figures 7A, 9) ... .11. A. collinsii Nectaries other than above ......................................................... 6 6. From type A thorns on lateral branches, leaves 30 cm or more long . . , . From type A thorns on lateral branches, leaves less than 25 cm long . . 7. Pinnules less than 1 mm wide ...................................... Pinnules more than 2 mm wide , , . . .................................... 9 8. Petiole deeply grooved, young thorns (and sometimes old thorns) , nectaries elongate and low (Figures 18~, 18~)~ obligate acaciarant entrance holes in thorns near Petiole only slightly grooved, young thorns glabrous, nectaries conical and well raised above petiole and rachis (Figure 8c), never with obligate acacia-ant entrance holes in thorns near apex .............................................. , 2. A. ruddiae Petiole with few nectaries (0-6) (from Costa Rica) ......................... 4. A. allmii 10. Petiole with many nectaries (6-25) closely packed and often not in a straight line Petiole with 1 to few nectaries; if more than one, then linearly arranged . . , , . , , , , , , . 11 11. Extrafloral nectaries as in Figures 6 and 7c, a1 always more than one on the petiole ........................................... ................... 6. A. globulifera Extrafloral nectaries as in Figure 78, almost never more than one on the petiole, bark and thorns deep red or mahogany in color, type A thorns very slender (as little as 3 mm di- 2 Petiolar nectaries circular (domes, cups, conical) . . , , . . ....................... 4 Type A thorns triangular in cross section ................................. .9. A. mayana apex .................................................. ........... 1. A, cookii 9. Petiole with many nectaries (8-30) (from Panama and Canal Zone) . , ......................... ............ ................. 5. A. chiapensis amrter at base) 1. Acacia cookii Safford Acacia cookii Safford, 1910:677. Acacia bucerophora B. L. Robinson, 1913:502. Mymecodendron cookii (Safford) Britton and Rose, 1928:93. Acacia cookii is easily distinguished from all other swollen-thorn acacias by its tomentose, long and thin thorns that bend back around the stem and are often bilaterally asymmetrical (Figures 15, 16). A. collinsii type A thorns may also wrap around the branch that bears them, but they are very short and thick. Only the type B thorns of A. cornigera wrap around the branch that bears them. A. ruddiae thorns may be tomentose when green, but they are glabrous upon maturing. 10. A. gentlei The type specimens for both A. cookii and A. bucerophora are representative of specimens from the southern part of the population (Figure 17). The branch designated as the type of A. cookii (from site 1) lacks swollen thorns and was probably taken from the end of a long and old reproductive branch such as that in Figure 15~. Since the in- florescences on the type are still in bud, the pe- duncles are much shorter than they are when in flower (e.g., Figures I~B, 16~). The leaves of the type specimen are not from the inflorescence-bearing branch; they are of the form normally subtended by large type A thorns (cf. ends of branch in Figure 18) and were probably broken from the branches in 30 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 15.-Lateral branches from mature A. cookii (Guatemala, site 4): A, sterile branch with fully developed thorns characteristic of the southern part of the acacia?s range; B, reproductive branch with axillary inflorescences characteristic of the southern part of the acacia?s range. order to avoid pressing the ant-filled thorns. The type specimen of A. bucerophora (site 3) differs in no significant way from that of A. cookii. The single thorn of the specimen is old and rotted and there- fore not the dark brown that is characteristic of most thorns on a tree of A. cookii. The single small seed pod is immature and shrunken, giving the illusion of being thin, while in fact A. cookii pods are over 1 cm wide and very long (Figure 19). The northern and southern parts of the A. cookii population differ on a number of morphological traits; however, these differences appear to be the result of comparing the two ends of a cline, since the intermediate elevation wet-forest habitat of A. cookii is continuous from the northern to southern part of the range. This hypothesis can only be veri- fied by collections from the area between the broken distribution in Figure 17. Some of the more out- standing changes in A. cookii in progressing from the northern to the southern part of the range are: (1) shortening of the type A thorns (cf. Figures 15~ vs. 16~); (2) shortening of the sexual branchlets (cf. Figures 15~ vs. 16~); (3) increased number of in- florescences per inflorescence-bearing axil; (4) in- creased number of Beltian bodies per mature leaf; and (5) increased percentage of the acacia plants being occupied. KEY COLLECTION LOCALITIES 1. 2. 3. 4. 5. 6. Type of A. cookii: Finca Trece Aguas near Secanquim, Depto. Aka Verapaz, Guatemala (300 m) . 111-8-1907, G.P. Goll, no. 102 (US 692154), fl. Paratype of A. cookii: Secanquim, Aka Verapaz, Guate- mala (300 m) . Vl-4-1905, O.F. Cook, no. 4. Type of A. bucerophora: about Toledo, 4 mi from the coast and 5 mi from Punta Gorda, Belize Prov., British Honduras (sea level), 111-29-1907, ME. Peck, no 632. 2.8 mi W Puerto Matias de Galvez (on road to Escobas) , Depto. Izabal, Guatemala (10 m). VIII-10-1967, DHJ, no. 1524, gp. 9.8 mi W Cardenas on Mexico 180, Tabasco, Mexico (10 m). VI-17-1966, DHJ, no 515. From Km 56 sign post, 9.8 mi N turn off Tapilula to 10.8 NUMBER 13 31 FIGURE 16.-Lateral branches from mature A. cookii: A, sterile branch with fully developed thorns characteristic of the northern part of the acacia?s range (Mexico, site 7); B, reproductive branch with long sexual branchlets characteristic of the northern part of the acacia?s range (Mexico, site 6). mi S Pichucalco on Villa Hermosa-Tuxtla Gutierrez highway, Chiapas, Mexico (240 to 380 m) . VII-10-1966, DHJ, nos. 530, 537, fl. 7. N side of Chontalpa, 26 mi S Cardenas, Tabasco, Mexico (80 m). VI-17-1966, DHJ, nos. 548, 550. 8. Along river, Lancetilla Valley, near Tela, Atlantida, Hon- duras (5 m) . XII-19-1927 to 111-27-1928, P.C. Standley, no. 54627. 9. Rd. from Secanquim to Sepacuite, Depto. Alta Verapaz. Guatemala. IV-1-1902, O.F. Cook and R.F. Griggs, no. 374 (US 408103). 10. Sierra del Mico, between Los Amates and Lake Izabal, Depto. Izabal, Guatemala (650 m) . 11-27-1908, W.A. Kel- Zerman, no. 7169 (US 2442365, US 2442347), fl. 11. Between Virginia and Lago Izabal, Montana del Mico, Depto. Izabal, Guatemala (50-100 m) . IV41940, J.A. Steyermark, no. 38758 (F 1056559). 12. British Honduras. 1944 (F 1403119). 13. Big Rock, British Honduras (65 m) . XII-16-1932, W.A. Schipp, no. 1072 (F 713932). 14. Maskall, British Honduras? 1-1934, P.H. Gentle, no. 1121 (F 739552). NATURAL HISTORY Acacia cookii is a narrowly distributed species that has its highest densities at intermediate eleva- tion sites (200 to 400 m) with very mild dry seasons (e.g., sites 1,2,6). At site 6, this habitat was slightly higher than that of A. cornigera and A. chiapensis, where all three occur along the east-facing foothills of the Sierras in western Tabasco, Mexico. Occa- sional adults of A. cookii were found at lower and drier sites (sites 5,7) in the northern part of the range, but their sympatry with A. cornigera and A. chiapensis may have been the result of seed dis- persal by road-building machinery. In the southern part of the range, isolated plants (e.g., site 4) gen- erally grow in naturally disturbed sites along rivers in very wet forests. They are sympatric with A. gentlei at these sites (as at other Guatemalan sites such as sites 10 and ll), and there is no evidence 32 SMITHSONlAN CONTRIBUTIONS TO BOTANY FIGURE 17.-General postulated distribution (stippled area; type localities indicated by arrows). collection site of A. cookii A collection site of A. rnayana that they are unnatural populations. A. cookii dis- appears when the wet forest is cut in areas like site 4; this local extinction is probably due to the same causes as cited for A. allenii and A. melanoceras. In the northern part of the range, where the dry season is more severe, A. cookii appears to be able to persist in the face of repeated wilting and occasional burn- ing. It was very common as a roadside and pasture tree shrub at site 6. The hard seeds of A. cookii are attacked by un- identified bruchids to the same degree as those of swollen-thorn acacias of drier areas. The seeds are dispersed by birds; the pods dehisce incompletely and the dark seeds are buried in a white, sweet pulp. The seed crops are not unusually large, but the large number of inflorescences per axil is matched only by the axils of A. chiapensis (Figure 20~). In the southern part of the range, where the inflorescence-bearing branch is reduced to a bump of inflorescences per axil may be as high as 64 for a FIGURE 18.-Crown of mature 5-m-tall A. cookii growing in the thorn (Figure 15B)9 an average number along creek in rain forest (Guatemala, site 4). NUMBER 13 FIGURE 19.-Acuciu cookii (Guatemala): A, rachis nectaries on leaf siibtended by type A thorn (site 4); B, petiolar nectaries of leaf; c, mature seed pods (site 1, US 692155). large sample (Figures ~OB, 22). In the northern part of the range, where the sexual branchlets are long, a sample of 10 axils on a representative branch had a mean of 32 inflorescences per axil (16, 29, 29, 29, 31, 31, 35, 37, 38, 42). The pattern of development of A. cookii from seedling to adult is unknown, but I have no reason to suspect it differs in any substantial way from that of other wet-forest, swollen-thorn acacias such as A. allenii. At site 4, the mature trees had a relatively dense crown (Figure IS) perched on top of a single and straight thin trunk. Based on growth rings, in wet forest the trees die of senescence when about 15 to 25 years old. In slightly drier sites (definitely site 6 and probably sites 1 and 2) the old trees may FIGURE 20.-Axillary inflorescences: A, immature pods of A. attain the largest size of any swollen-thorn acacia, chiupensis (Mexico, site 9); 8, bud scars (with two in- frutescence peduncles projecting to 12:OO and 2:OO o'clock) except perhaps '* in Mexico and of A. cookii (Guatemala, site 4). A. ruddiae in the Costa Rican mountains. Growing 34 SMITHSONIAN CONTRIBUTIONS TO BOTANY in old landslide scars at site 6, there were trees of A. cookii up to 30 m tall, up to 20 m across the crown, and with a DBH as much as 40 to 50 cm. The ant colonies were not examined in these trees, but I suspect that they were either a multiple-queen species or they restricted their patrolling to the newer foliage. While the crowns of these acacias were very large, the actual amount of green foliage was small compared to what it would be in another species of legume of equal size during the rainy season. It is possible that the herbivore pressure is some- what reduced at this site and/or that A. cookii in this area has better chemical defenses than do most swollen-thorn acacias. I suspect this because road- side clumps of poorly occupied A. cookii at site 6 had foliage with very little herbivore damage. 1 favor this hypothesis because the A. cookii growing at site 5 was in a lowland roadside pasture where unoccupied A. cornigera was being heavily defoli- ated by insects that normally feed on other species of plants. This A. cookii was, however, completely unoccupied and almost in perfect condition. At least 5 years old, the plant was 4 m tall and had a crown similar in foliage density to the one in Figure 18. About one out of every four shoot tips had been - 160 Y I X t-1 u 2 W cl eaten off by insects, which is about the frequency usually observed during the rainy season on non- ant-acacias and occupied swollen-thorn acacias. These comments should not be construed to indi- cate that A. cookii does not have a mutualistic asso- ciation with its ants. At site 6 there was ample defoliation of unoccupied sucker shoots, in contrast to occupied ones. This indicates that the results of removing the ants would have had an effect on A. cookii much like that with A. allenii, but not as dramatic as with ,4. cornigera or A. collinsii. It is possible that A. cookii is still in the process of coevolving a tighter mutualism with its occupant ants-Pseudomyrmex ferruginea in the northern part of the range, and P. ferruginea plus an un- identified species in the southern part. Its petiolar nectaries (Figure 19~) are morphologically poorly developed-their sugar production rate may be high, however-though the nectaries on the rachis may compensate for this (Figure 19~). On some individuals at site 6 there were Beltian bodies on only the basal 10 to 30 pinnules, and even at the sites where the healthiest ant colonies occur (e.g., site 4), Beltian bodies are produced on only about half the pinnules. The swollen thorns on A. cookii in the northern part of the range are very thin- walled-birds could probably break into them. with ease-and they harden very soon after attaining full size, which allows the ant only a short time to cut an entrance hole. The very narrow thorns have a volume (Figure 21) representative of that of other swollen-thorn acacias because of their extreme length; however, these narrow thorns probably make very cramped living quarters for the ants. The largest thorns to be found (Figure 21) had a mean volume of only 1.64 cc, which is representative only of A. chiupensis, A. globulifera, A. gentlei, and A. ruddiae. 2. Acacia ruddiae, new species t-1 100 11 lo0 200 3 00 TOTAL THORN VOLUME ( cc ) FIGURE !?I.-Distribution of 131.3 cc of thorn volume among 19 very large type B thorns from one adult A. cookii (Guate- mala, site 4). While the largest to be found, these thorns had a maximum diameter in cross section of 5-7 mm. Arbol usque ad 20 m alta; ramuli novelli puberuli, glabrescentes; stipulae spinescentes inter- durn magnae, teretisculae, usque ad 10 cm longae, 9 mm latae; folia circiter 15-40 cm longa; nectaria sessilia cupulata ad juncturas pinnarum omnium et 0-2 in petiolis; pinnae (12-) 25-40-jugae, 4-5 cm longae; pinnulae 65-75-jugae, lineariae, 2-4 mm longae, 0.5-1.2 mm latae, acutae vel obtusae, NUMBER 13 v) W W V v) 30 el IL w I 3 z m 123456 A NUMBER OF PODS PER INFLORESCENCE u) cl x < IL 0 U I 1 35 7 9 11 B NUMBER OF INFLORESCENCES WITH PODS PER AXIL FIGURE 22.-Acacia cookii (Guatemala, site 4): A, distribution of 113 seed pods among 63 in- fructescences; B, distribution of the 63 infructescences among their 10 axils. There were 642 inflorescences produced by these 10 axils. puberulae; flores aurei in capitulo globoso; involu- cellum annulare in medio pedunculo; legumina (immatura) atrobrunnea, puberula, granulata gla- brescentes, oblonga, 5 cm longa, 7 mm lata, com- planata, obtusa, stipite 5 mm longo; legumina matura probabiliter usque ad 6-8 cm longa. TYPE.-?. H. Janzen 828, Costa Rica, Limon Province, ferry landing, Moin River, Moin, 13 Feb- ruary 1965 (holotype US 2639554; isotypes CAS, F, GH, MEXU, MICH, MO, NY, UC). Type tree 20 m tall with 35 cm diameter trunk 1 m above ground; gray-green bipinnate leaves mostly 15 to 40 cm long with 50 to 80 pinnae per leaf, pinnae 40 to 50 mm long with 130 to 150 pin- nules per pinnae, pinnules 2.5 to 3.0 mm long and 0.75 to 1.0 mm wide (up to 9,000 pinnules per leaf); occasional swollen-thorns on less than 10 percent of the nodes and very frequently only one side of the thorn swollen, thorns not occupied by obligate acacia-ants but by a miscellany of other ant species; no food bodies (Be1 tian bodies) on pinnules; single nectary at junction of each pinna with rachis, 0 to 2 nectaries on petiole, nectaries cylindrical up to 1 mm tall; yellow inflorescences spherical and up to 6 mm diameter, peduncle up to 10 mm long and 0.50 to 0.75 mm thick; immature fruits flat and slightly curved, probably 6 to 8 cm long at maturity. Acacia ruddiae is distinctive by never being occu- pied by obligate acacia-ants, lacking swollen thorns on most branches except the bases of some lateral branches, having better than 5,000 very small leaflets per leaf, growing in rain forest, becoming a very large tree, and by its range (Figure 23). It also differs from other Central American members of the genus Acacia by the same traits; it is undoubt- edly distinguishable from the remainder of the genus by fine details of flower, fruit, and leaf struc- ture, but it is impossible to know these differences at the present?state of Acacia taxonomy. The tree from which the type specimens (13 sheets) were collected was at least 30 years old, and perhaps much older, and was still standing in 1967. It was about 80 m south (west side of the road) of the south beach of the ferry landing on the Moin River and 200 m from the ocean beach, at Moin, Costa Rica (site 1). The branches were cut from high in the tree and were representative of the 1-2- year-old branches in the canopy. Although the speci- mens were collected during the dry season, the area was still subjected to frequent rains and the tree was only partly deciduous. It was at the end of its flow- ering season, and the full sized but immature green pods were from the same flower crop, rather than from the previous year?s flowers as is usual with acacias of drier habitats. As is usual with the canopy branches from A. ruddiae this size, there are few swollen thorns on the branches. The tree from which the type specimens were collected was ex- ceptional in two respects: First, most of the A. ruddiae population in this area has long since been destroyed by housing development. Second, the tree grew in a relatively open site and thus had strong 36 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE Pt.-General postulated distribution (heavily stippled area) and specific collection sites of A. ruddiae (all on the Caribbean side of Ccntral America). General postulated distribution (lightly stippled area) and specific collection sites of A. allenii (all on the Pacific side of Central America). NUMBER 13 87 development of lateral branches; most A. ruddiae are involved in intense intertree competition and thus are tall and slender. This acacia is named in honor of Dr. Velva E. Rudd whose unfailing assistance and encourage- ment has greatly aided in bringing this study to completion, and whose taxonomic work with Legu- minosae will always be a tremendous assistance to field biologists. Acacia ruddiae is included with the swollen-thorn FIGURE 24.-Acacia ruddiae: A, very young seedling; note nod- ules on roots (Costa Rica, site 5); B, young sapling (Costa Rica, site 5). acacias, despite the fact that it is not occupied by obligate acacia-ants, because it appears to represent an intermediate step in one of several possible evo- lutionary pathways to the evolution of swollen- thorn acacias. Its ecology is therefore of special significance in this study. A. ruddiae is the only Central American acacia, aside from hybrids of swollen-thorn acacias with nonant-acacias, that ap- pears to represent an intermediate step in the evo- lution of a New World swollen-thorn acacias. 38 SMITHSONIAN CONTRIBUTIONS TO BOTANY KEY COLLECTION LOCALITIES 1. TYPE.-Ferry landing, Moin River, Moin, Limon Prov., Costa Rica. 11-13-1965, DHJ, no. 828, fl. gp. 2. Pandora, Limon Prov., Costa Rica. VIII-26-1963, C.D. Michner, nc. 3. 16 mi N Vara Blanca on road from Heredia to Puerto Viejo (Sarapiqui Dist.), Heredia Prov., Costa Rica (320 m) . V-1-1965, DHJ, no. 819, fl. 4. 1 km W La Virgin, Sarapiqui Dist., Heredia Prov., Costa Rica (80 m) , IX-5-1971, DHJ, nc. 5. 1 km W La Lola (on Limon-Siquirres railroad), Limon Prov., Costa Rica. VIII-19-1963, DHJ, nos. 1868, 1872, 1873, 1874, 1875, 1876, 1877, 1878, 1881. 6. 11 mi N Florencia, San Carlo Dist., Alajuela Prov., Costa Rica. VII-14-1963, DHJ, no. 1883, 1885, 1866. 7. Along Changuinola River, Changuinola Valley, Bocas del Tor0 Prov., Panama. 11-29-1924, V. C. Dunlap, no. 462 (F 709739, US 1405662) (listed incorrectly as ?A. cookii? in Woodson and Schery), fl. 8. Lancetilla valley near Tela, Depto. Atlantida, Honduras. VI-6-1927 to 111-20-1928, P. C. Standley, no. 54627 (US 1408382). 9. 1.9 mi W Puerto Matias de Galvez (on road to Escobas), Izabal, Guatemala. VIII-10-1967, DHJ, no. 1582. 10. 27.4 mi S Puerto Barrios, Izabal, Guatemala. VIII-8-1967, DHJ, nc. NATURAL HISTORY Acacia ruddiae is one of the most peculiar Central American members of the genus. It lives in very wet forest habitats, which is distinctly unlike other acacias, yet does not have the obligate acacia-ants which generally allow survival of swollen-thorn acacias in wet habitats (Janzen, 1966a). In Costa Rica, where the only large populations of A. ruddiae have been encountered, the natural habitat of this tree is in primary succession along lowland rivers through rain forest (sites 4, 5), swamp edges (site 5), and in large, old landslide scars on the sides of very steep gorges (site 3). These areas receive 4 to 5 m of rain a year and have a poorly developed dry season of 1 to 2 months with occa- sional rains. Some of the sites occupied by A. mela- noceras and A. allenii receive as much rainfall, but have a more noticeable dry season. Judging from the paucity of collections, A. ruddiae apparently also exists as a very rare tree in the rain forest that ex- tends in a narrow strip up along the Caribbean coast to the Puerto Barrios area of Guatemala (sites 8 and 9), but there is no reason to believe that it does not become locally common in these forests just as it does in Costa Rica. Acacia ruddiae is by no means generally dis- tributed throughout the rain forest of the Costa Rican Caribbean lowlands. At site 4 it was common in 1- to 30-year-old primary succession on gravel river banks and grew to a 25-m-tall tree with 30 cm DBH; it had not, however, extended its popula- tion out into the surrounding old cacao groves and pastures. At site 5 there was a large population along about a mile of old river bank, and occasional large saplings in the immediately adjacent parts of a banana plantation and on swamp edges in pri- mary forest. Many of the swamp trees were old and scattered widely, in contrast with the many adjacent and even-aged A. ruddiae along the river. Extensive search throughout many other swamps and dis- turbed forest habitats within ten miles located no other A. ruddiae. At site 3, A. ruddiae occurs only FIGURE 25.-Acacia ruddiae (Costa Rica): A, representative flowering branch (site 1); B, naturally split type A thorn (site 6). NUMBER 13 39 FIGURE 26.-Representative branches of A. ruddiae: A, without swollen thorns from the crown of a large adult (Guatemala, site 9); B, from main trunk of large tree (Costa Rica, site 5). in a small portion of the total landslide area in this strongly dissected topography. It has not spread along the road into other disturbed habitats from this site. While detailed weather data are not avail- able for site 3, it may be slightly drier-hence better insolated-than other disturbed sites slightly above or below it in elevation. This is indicated by the high success of grasses in early stages of succession here, in contrast with dicot regeneration at casually cleared sites a few miles farther up or down the road. Seedlings growing in deep shade resemble those of A. melanoceras and A. allenii in that the leaves are arranged in a whorl so that in vertical view, they do not overlap. When growing on bare gravel bars, the seedlings are more bushy and asymmetrical in appearance (Figure 24~). The nodules on the rootlets of A. ruddiae seedlings are more noticeable than on any other swollen-thorn acacia (Figure 24~). The seedlings differ from those of all other swollen-thorn acacias in lacking swollen thorns. The first swollen thorns do not appear until the plant is 1 to 3 m tall and has several hundred leaves. The nectaries (Figure 8c), however, are produced as soon as on any swollen-thorn acacia. The first ones appear on the rachis just below the point where the pinnae join the rachis, and petiolar nectaries begin to appear on the 25th to about the 50th leaf. There are no Beltian bodies on A. ruddiae, and thus the only contact that the seedlings may have with ants is when occasional worker ants visit the nectaries and forage on the leaf surfaces. In this sense the young A. ruddiae are no different from nonant- acacias, except that the nectaries are larger than on nonant-acacias and may therefore produce more nectar. On saplings, the first swollen thorns are scattered along the main trunk and lateral branches. They are very thin-walled, hollow, easily crushed, and 40 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 27.-Acacin ruddiae (Guatemala, site 9): A, type B thorns on short lateral branches from main trunk; B, representative crown section of 4-m-tall sapling; c, well developed type A thorns from the crown in Figure B. occasionally only half of the stipule is enlarged. They often split at the base upon drying (Figure Z~B), and ants, especially Crematogaster, often use this as an entrance hole. As the tree grows larger, some of the lateral branches have no thorns at all (Figure %A), some branches have large swollen thorns at the base and tufts of leaves at the ends (Figures 27~, 27~), and some branches are very short and bear thorns best classified as type B (Figures 27~, c). The A. ruddiae in the Puerto Barrios area (sites 8 and 9) had most of its swollen thorns on branches such as those in Figures 26~ and 27, while in Costa Rica the branches were usually very short with many thorns (as in Figure 27~) or very long with just a few scattered thorns (Figure 28~). The lower trunk of a large, old A. ruddiae is covered with short branches with many thorns, making the tree very difficult to climb. The crowns of A. ruddiae are more diffuse than any other swollen-thorn acacia and therefore look more like most mimosaceous legumes than do swollen-thorn acacias. Looking up through a crown (e.g., Figure 28~), one can often see large amounts of sky; only the most densely leafed branches (e.g., Figure 5 1~) look like typical swollen-thorn acacia branches. This is the case despite the fact that the crowns of A. ruddiae are usually fully insolated; while the young seedlings appear to grow fairly well in heavy shade, the mature trees almost always have canopy level or emergent crowns, suggesting that A. ruddiae probably cannot mature in heavy or broken shade. NUMBER 13 41 FIGURE 28.-Acacia ruddiae (Costa Rica, site 6): A, lateral branches in crown of adult; B, looking up through crown of young adult. The trees of A. ruddiae do not bear flowers until they are in the 8 to 15 m height and 15 to 30 cm DBH class. Small clusters of yellow and bee-polli- nated inflorescences (Figure 25~) are scattered through the crown and not concentrated on main sexual branches as is common for swollen-thorn acacias. The seed pods appear to mature during the 2 to 4 months immediately following flowering, as with A. melanoceras and A. allenii, rather than waiting until the following dry season as with dry- land, swollen-thorn acacias. Adults of A. ruddiae show no more obvious insect damage than do other acacias not occupied by obligate acacia-ants. Since some have no ants living on them at all-though some worker ants forage in the foliage for nectar and probably insects-this immunity is likely due to the usual kinds of chemi- cal defenses encountered in ordinary acacias. Shoot tips of A. ruddiae are slow growing in comparison with more usual swollen-thorn acacias. Associated with this they are very tough, compact, bitter in taste, and covered with dense hair (Figure 5A). They strongly resemble those of A. macracantha and A. farnesiana. By virtue of the very well developed foliar nec- taries and enlarged thorns, A. ruddiae has far more ants foraging on its leaf surfaces than ordinary nonant-acacias. None of these ants are obligate acacia-ants, but rather those species-Pseudo- myrmex, Crematogaster, Camponotus-that live throughout the general forest canopy and that forage for dead, moribund, or live insects on foliage in general. While queens of obligate acacia-ants may occasionally be dispersed into habitats with A. ruddiae (definitely the case at sites 9 and 10 in Guatemala), I have never seen one of the character- istic type of entrance holes made by these queens in about 2,500 swollen-thorns examined carefully on 16 individual trees of A. ruddiae. To examine the use made of A. ruddiae thorns by ants, a representative 3 m tall, fully insolated sapling growing in river-edge vegetation at site 4 was completely dissected. The acacia had 798 leaves on 51 branches; far less than 20 percent of the large leaves produced by the acacia were subtended by swollen thorns. Of the 371 thorns, 302 were, or had been, tenanted by some insect (on a ?typical? swollen-thorn acacia, all thorns would have been). 42 SMITHSONIAN CONTRIBUTIONS TO BOTANY Nineteen percent of the thorns had split naturally at the base, and all of these thorns had been entered by insects. Of all the thorns that had been entered by insects, by any means, 20 percent had parts of ant colonies in them or queens attempting to start a colony. Of the remaining thorns that had been entered, 95 percent contained microlepidop- terous larvae, weevil larvae, and Orthoptera eggs. When these insects emerge, the acacia thorns are very likely to be occupied by ants. Spiders and other insects tenanted the remaining 5 percent of the thorns lacking ants. The ants represented at least 7 species in 5 genera-Solenopsis, Cremato- gaster, Camponotus, Pseudomyrmex, Paracrypto- cerus. While a detailed census is not available, cursory examination of A. ruddiae thorns in large old trees shows almost every thorn to have a part of an ant colony in it. It is extremely likely that the continual foraging of these ant species on the foliage occasionally results in less herbivore damage to the tree than if the ants were not present. If a mutant strain of A. ruddiae provided any type of solid food to the ants, the probability is very high that it would quickly lead to the evolu- tion of a strongly mutualistic interaction. It is im- possible to know if A. ruddiae represents the gradual degeneration of a once intense mutualism or is on its way to more mutualistic interaction. The latter is more likely, since the former would require a strong relaxation of selection pressure by herbivores, such as might occur if a swollen-thorn acacia were to immigrate to a small oceanic island. The evolution of a type of Beltian body may be starting to occur with A. ruddiae; some of the plants at site 4 have clear tips on the pinnules, extending into the leaf about 0.25 to 0.5 mm. If the ants were to start to chew on this for moisture or nutrients during the time of year when insect prey is scarcest, and therefore maintain larger colony sizes than on plants lacking this modified pinnule tip, the evolution of a food body might occur. This is likely to be the way that typical Beltian bodies first appeared on acacias. Plants with a larger year- round population of ants would probably be less damaged by the periodic small outbreaks of host- specific insects that probably occur on many tropical trees, and certainly occur with nonant-acacias. Interestingly, the species of ant that began this behavior would be likely to concentrate more of its foraging on the acacia and less off, and be likely to tenant an increasingly greater proportion of the thorns on the acacia. This could occur either through faster colony growth than could occur with ant species totally dependent on insect prey for growth, or through the ants moving into thorns left untenanted during periodic population lows in the other species of ants. Once the entire ant colony becomes concentrated in the acacia, it has a vested interest in the health of the acacia and coevolution of the interaction should occur at an ever increasing rate (Janzen, 1966a). As will be discussed further in respect to hybrids, this acacia appears to be a plant that would have greatly increased fitness by the gaining of the genetic information necessary for the production of Beltian bodies. This could well happen through introgression with a wet-forest acacia such as A. melanoceras at the southern end of the range of A. ruddiae, or A. gentlei or A. cookii, and the northern end of the range of A. ruddine. 3. Acacia melanoceras Buerling Acacia melanoceras Buerling, 1854:362. Acacia multiglandulosa Schenck, 1913:362. Myrmecodendron melanoceras (Buerling) Britton and Rose, 1928:93. Acacia melanoceras is easily recognized since it is the only swollen-thorn acacia in the wet lowlands of central Panama and the Canal Zone (Figure 29). Where its distribution is tangent with that of A. collinsii, about halfway through the Canal Zone, A. melanoceras is easily distinguished from that dry-forest species since A. melanoceras has a very large number of nectaries on the leaf petiole, longitudinally ridged swollen thorns, spherical in- florescences, and huge leaves, 1 have been unable to locate the type specimen of A. melanoceras and it is likely destroyed; how- ever, the type specimen of A. multiglandulosa (at Berlin, now destroyed, but represented by Field Museum photo 1281 ex Berlin) is from the same locality as that of A. melanoceras (site 1) and is probably described from the same specimen or a duplicate specimen of the same collection. The description of the type of both A. melanoceras and A. multiglandulosa agrees completely with the characteristics of the swollen-thorn acacia popula- tion whose distribution is plotted in Figure 29. The type specimen of A. multiglandulosa is a representa- NUMBER 13 43 FIGURE 29.-General postulated distribution (stippled area) and specific collections sites of A. melanoceras in Panama. tive distal end of a fertile branch from an adult acacia. Very long (20 cm) inflorescence-bearing branchlets such as the one on the type of A. multi- glandulosa are commonly seen on old trees growing along river edges that have their crowns partly shaded by taller trees. The absence of swollen thorns on a branch bearing sexual branchlets, as is the case with the type specimen, is commonplace in old crowns of A. tnelanoceras. In 1915, Safford redescribed A. melanoceras, but under the name of A. 1. 2. 3. 4. 3. multiglan dulosa. KEY COLLECTION LOCAL~TIES Types of A. melanoceras and A, multiglandulosa: Porto- bello (Porto Bello), Colon, Panama (sea level). 1824, J.G. Billberg, no. 289. Head of Gatun Valley, Canal Zone, Panama. 1914, H. Pittier, no, 6745 (US 716562, US 716560, US 716561; Saf- ford 1915, Wheeler 1942). Headwaters Rio Agua Salud, Canal Zone, Panama. Wheeler, 111-6-1923 (Wheeler, 1942). Marajal, near Colon, Canal Zone, Panama. Wheeler, VII- 26-1924 (Wheeler, 1942). Cerro Jefe, several mi N Cerro Azul, Panama, Panama. R.D. Dressier, VIII-28-1967 (pers. comm. plus thorn sample). 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 400 m upriver (east bank) from the bridge at Rio Pied- ras, Colon, Panama. VII-3-1967, DHJ, nos. 1641, 1912, 1913, 1915, 1916, gp, mp. Southwest corner of Zorro Island (in Gatun Lake), 5 km NW Nueva Providencia, Canal Zone, Panama. VII-7- 1967, DHJ, no. 1894, gp. 10 m post on Al. Ar. Trail, Barro Colorado Island, Gatun Lake, Canal Zone, Panama. DHJ, VII-7-1967, nc (I am indebted to R.D. Dressler for showing me this plant). 7 km SW Gatun Locks on road to Escobal, Canal Zone, Panama. DHJ, VII-10-1967, nc (I am indebted to M. Nauman for showing me these plants). France Field, Canal Zone, Panama (sea level). VIII-1924. D. Fairchild (US 1405611). Between France Field and Catival, Colon Prov., Panama. 1-9-1924, P.C. Standley, no. 30378 (US 1219256). Near Frijoles, Rio Agua Salud, Canal Zone, Panama. 111-6-1923, C.V. Piper, no. 5853 (US 1167694, US 1166375). Margarita Swamp, Canal Zone, Panama. VI-28-1923, O.F. Cook and R.D. Martin, no. 14 (US 1491829). Peiia Blanca Bay, Barro Colorado Island, Canal Zone, Panama. VII-3-1934, Shattuck, no. 1087 (US 2366760). Hills N of Frijoles, Canal Zone, Panama (50 m). XII-19- 1923, P.C. Standley, no. 27582, (US 1217650). Margarita Swamp near France Field, Canal Zone, Panama (sea level). 11-1925, J.B. Shropshire (US 1208884, US 1208883, US 1208885), mp. Barro Colorado Island, Canal Zone, Panama. VIII-1927, L.A. Kenoyer, no. 373 (US 1319142). 44 SMITHSONIAN CONTRIBUTIONS TO BOTANY NATURAL HISTORY Acacia melanoceras is one of the seven species of swollen-thorn acacias with small geographic ranges (A. allenii, A. mayana, A. ruddiae, A. sphaero- cephala, A. cookii, A. gentlei, the others), and like these species is restricted to comparatively wet and undisturbed forests. In 1928 Standley reported A. melanoceras as common in the forests on the At- lantic slopes of the Canal Zone; in comparison with the more widely distributed dry-land species, it is nevertheless very rare. For example, in contem- porary forested habitats there appear to be at least 100 adult plants of A. collinsii on the dry Pacific side of Panama for every one adult A. melanoceras on the Atlantic side. It is of course impossible to know to what degree this difference is due to A. collinsii becoming much more common with the development of brushy cattle pastures and waste- land. One thing may be said with certainty, how- ever, and that is that A. melanoceras disappears from any habitat subjected to disturbances that are any more catastrophic than infrequent logging. Since the Atlantic side of Panama, aside from banana farming, is infrequently logged, it is likely that the contemporary distribution of A. melano- ceras is not much different from that of pre-Colom- bian times. Every reproductively mature A. melanoceras found in this study (sites 6 to 9) was in second- growth woody vegetation that had grown better than 15 years without further disturbance. All sites were either areas that had been clear-cut long be- fore for logging or farming (e.g., Barro Colorado Island) or had been swept clean by a flooding river. Seedlings (Figure 30~) were found only in the full shade of canopies 10 to 15 m tall, and on wet to moderately well-drained soil. A. melanoceras generally becomes reproductively mature under, or in, jagged breaks in the general canopy; they are almost never fully insolated. Senescent canopies are often directly under a small hole in the general canopy (Figure 31). Neither mature plants nor seedlings are common at any site. Two adults per acre is a high density for A. melanoceras. Adults and young plants almost always occur as single FIGURE 30.-Acacia melanoceras: A, seedling from deep forest shade (scale in inches; Panama, site 6); B, normal lateral branch from 6-rn-tall plant (Canal Zone, site 9). NUMBER 13 45 FIGURE 31.-Acacia melanoceras (Panama, site 6): A, looking up through crown of 15-m-tall mature tree that is still reproducing, but only poorly; B, looking up through senescent crown of 12-m-tall tree that has lost its ant colony. shoots. The only clump of shoots encountered in this study was at site 6 on the edge of an island at Gatun Lake. A single tree-based on tree rings, at least 27 years old-had produced a large number of root suckers and seedlings covering an area about 20 by 30 m. The cluster was on the north side of a hill and situated such that the taller forest trees shaded it, yet the opening caused by the lake led to day-long indirect insolation. Wheeler (1942) illus- trates a small clump at site 4; the plant has the appearance of a cluster of stump and root sprouts from a large adult that was cut when the swamp was cleared of forest. Acacia melanoceras appears to be completely absent in repeatedly disturbed pasture, field edges, roadsides, cleared swamps, etc., in the central and northern part of the Canal Zone and adjacent areas in Panama; however, it is still present in adjacent relatively undisturbed forest, or forest that has been allowed to regenerate for a long time. Extinc- tion of A. melanoceras following human disturb- ance is probably based on the following factors: (1) A. melanoceras does not set seed until at least 8 to 10 years old; I have found heavy seed produc- tion only on very healthy old trees that are in partial shade of a tall broken forest canopy. Such a long time in reaching reproductive maturity is likely due to the shady microhabitat occupied by this acacia. (2) When adults or subadults are ex- posed to direct insolation by clearing the forest, they become very shrubby, stunted, and die after 1 to 3 years. They produce no seed during this time. Failure to grow in full sunlight appears to have two causes: First, the leaves are probably genetically programmed to operate most efficiently at low-light density and air temperature, and the opposite microclimate is probably damaging. Second, large numbers of phytophagous insects in young second- ary succession probably find the acacia to be very edible. Shoot tips are repeatedly eaten off the verti- cally elongating shoots, resulting in a bush life form which is probably very inefficient. It is not obvious, however, why the occupant Pseudomyrmex satanica, which is so effective at protecting A. melanoceras 46 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE SZ.-Acacia melanoceras: A, reproductive branch with extraordinarily heavy pod set (Canal Zone, site 7); B, reproductive branchlets with representative pod set (Panama, site 6); c, longitudinal section through immature but full-sized pod (Canal Zone, site 7). in the forest, is apparently so ineffective in more open sites. It may be that their protective abilities per insect attacking are the same in both habitats, but that the plant is less able to repair the more frequent damage in the open. (3) A. melanoceras does not sprout readily from cut stumps. The stumps of many young and mature A. melanoceras die without sucker-shoot regeneration, in strong contrast to the behavior of other swollen-thorn acacias. It may be that heavily shaded plants have insufficient energy reserves stored in their roots for reproduction by sucker shoots. (4) The seed of A. melanoceras (Figure 3%) has a soft seed coat and apparently needs to germinate immediately. Even if a seed is dispersed into early stages of succession, it would probably die of desiccation if it arrived during the dry season of one of the dry ?spells? during the wet season. (5) Being soft-seeded, A. melanoceras cannot be dispersed by the same species of birds that disperse the hard seeds of species living in drier habitats (eg., A. cornigera, A hindsii, A. collinsii); these birds (e.g., jays, derby fly-catch- ers) pass the hard seeds through their gizzard unharmed, while a seed of A. melanoceras would surely be digested. Any dispersal of these soft seeds must be via birds that regurgitate seeds (e.g., tro- gons, cotingas). These birds characteristically live in older stages of succession and primary forest and are unlikely to disperse A. melanoceras seeds out into large tracts of young second growth. (6) Since the seedlings of A. melanoceras are normally lound only in heavily shaded forest understory, it would not be surprising to find that the queens of P. satanica do not search in dry or fully insolated sites. The young acacia would then be competing with many fast-growing plants with a very high density of herbivorous insects. It is possible that the searching queens of P. ferruginea, coming from A. collinsii growing in young second growth, could occupy young A. melanoceras. A. collinsii, however, has not yet invaded much of the disturbed land left when forest suitable to A. melanoceras was cleared. This general interaction could perhaps be observed if half of an island such as Barro Colorado NUMBER 13 47 Island were clear-cut and turned into brushy pas- ture. Acacia melanoceras, A. allenii, and A. ruddiae live in the wettest sites occupied by the genus in Central America. The dry season in its habitats is indistinct and the sites generally receive 3.5 to 5.5 m of rainfall. At sites 6 and 7, the roots of A. melanoceras were under water or in mud almost all year. The cluster of fruits in Figure 32~ hung down to within a few inches of Gatun Lake. The bark of A. melanoceras is almost always damp from rain and those parts of the acacia not patrolled by ants often become covered with lichens and other epiphytes. That A. melanoceras lives in wet forest that lacks other acacia, and for the most part other mimosoid legumes, may explain in part why quite healthy but unoccupied adults are occasionally encountered in the forest. Judging from the holes in the thorns, these acacias were at one time occupied by large ant colonies, which died for unknown reasons. In contrast to the subsequent death of the tree that would have occurred were this a dry-land acacia, the abandoned acacias appear to live a number of years, bear successive seed crops, and are at times tenanted by a variety of other ants that normally live in hollow twigs in the general canopy. These unoccupied acacias suffer herbivore damage, but it is not very severe (in contrast to individuals even with a strong ant colony in insolated sites a few hundred meters away). I suspect that in leaving v) W U U m W el !A z 6 40 30 - 20 z 10 B 3 z behind the younger stages of succession and the large number of insects it harbors, especially those that normally feed on mimosoid legumes, A. melanoceras has moved into a habitat of reduced herbivore pressure. Such a strategy is likely to be effective until over evolutionary time other mi- mosoid legumes make the same move for the same reason. While A. melanoceras appears to be pollinated by small bees-Halictidae, Megachilidae, Ceratina in the Apidae-pollination is either very ineffec- tual or the plant has little reserves for maturing fruits. The seed crops on almost every tree ex- amined have been very small, ranging from 2 to 30 pods; the long flowering branchlets (Figure 32~) set only 1 or 2 pods per branchlet and often none at all. The 13-seed pod in Figure 32c is represent- ative of the number of seeds per pod; A. melano- cerns does not show the strong variation in number of seeds per pod found in the acacias of drier sites. The large pod crop on the acacia at site 7 provides several lines of evidence to suggest that the usual small seed crop is due to lack of energy reserves rather than poor pollination: (1) This tree was growing in an exceptionally good site, as is evi- denced by the large number of healthy sucker shoots, had an immense ant colony, and large amounts of lush foliage (as in Figure 30~ and contrasted with Figure 31~). There were no other adult swollen-thorn acacias within at least 200 m of this plant (down the island) and at least 500 m 123 13 57 A NUMBER OF PODS PER 6 NUMBER OF INFLORESCENCES WITH INFLORESCENCE PODS PER BRANCHLET FIGURE BB.-Acacia melanoceras (Canal Zone, site 7): A, distribution of 64 mature pods among the 49 infructescences on 10 sexual branchlets from one tree with an extraordinary seed crop; B, distribution of the 49 infructescences in Figure A among the 10 sexual branchlets. SMITHSONIAN CONTRIBUTIONS TO BOTANY across the lake. This plant set at least 900 pods in large clumps such as that in Figure 32~. ?The im- plication is that the tree is self pollinated, or else the clump of shoots originated from two seeds. (2) The total number of inflorescences produced by each sexual branchlet appeared to be no different from that of other acacias examined that had pro- duced only 0 or 1 pod per sexual branchlet. The number of sexual branchlets per major tree branch was likewise no different in any obvious way; how- ever, the number of inflorescences that produced pods, per branchlet, was much higher (Figure 3313) than is usually observed with A. melanoceras. Two sexual branchlets had 7 pods each on them, and none had only one. (3) There were even more than one pod set per inflorescence in many cases (Figure 33~). That the pods were distributed among the inflorescences in this manner, rather than being uniformly distributed (i.e., retain more inflores- cences, each with one fruit), suggests the possibility that the plant chooses out-crossed female florets as those to mature. This tree had an average of 31 inflorescence-bearing axils per sexual branchlet (=lo). Its total crop of about 18,540 inflorescences produced about 960 pods on 735 inflorescences for a total seed crop of about 10,560 seeds. There are about 10 female florets on each inflorescence, mean- ing that only 0.6 percent of the tree?s ovules matured. Associated with the small seed crop observed on most adult A. melanocel-as, it is particularly note- worthy that this species is free from predation by the bruchid beetles that take such a high toll of hard-seeded acacias. This means that virtually every FIGURE 34.-Acacia melanoceras (A-C, Canal Zone, site 7): A, type B thorns on short lateral branches from the main trunk of a mature tree; B, immature type B thorns on newly develop- ing short lateral branch from same trunk; ants are Pseudomyrmex satanica; c, maximally de- veloped cluster of type B thorns on lateral branch of same tree; D, type B thorns on lateral branch from crown of very old A. inelanoceras (Panama, site 6). NUMBER I3 49 seed dispersed by the plant is viable, in contrast to the case with hard-seeded, swollen-thorn acacias where most seeds are killed by bruchids before the seeds are dispersed. In short, the average number of viable seeds produced per adult A. melanoceras is likely to be as great as for acacias of drier sites that initially produce many more seeds. A seedling of A. melanoceras (Figure 30~) can probably germinate and grow slowly to adulthood almost anywhere in the forest, in strong contrast to the seedlings of hard-seeded, swollen-thorn acacias, which must land in a fairly young stage of succes- sion to survive. This means that a certain large frac- tion of the viable seeds produced by dry-land acacias will not make it to the stage where they have any chance at all of surviving. Therefore, the apparently smaller seed crop of A. melanoceras may in fact not mean that any fewer saplings develop per adult than is the case with dry-land, swollen-thorn acacias. Throughout its range, I found A. melanoceras to be occupied only ,by Pseudomyrmex satanica (Figure 3413). (The ?Pseudomyrmex satan? cited by Standley in 1928 is a misspelling.) Wheeler (1942) likewise had no records of other obligate acacia-ants in A. melanoceTas. Since P. satanica is restricted to A. melanoceias, the ant will probably go extinct as the remaining forest in Panama is converted to agricultural land. P. satanica is probably a ?wet forest derivative? of the widespread P. ferruginea, which occupied A. collinsii in the drier parts of Panama, among other places. As will be discussed later, the part of the P. ferruginea population oc- cupying A. allenii on the Osa Peninsula and the Golfito region in Costa Rica probably represents an intermediate form of the same type of speciation event. P. satanica is easily distinguished from P. ferruginea in Panama by the former?s black color, habit of holding the gaster curled under the body (Figure 34~), and having multiple-queen colonies. Unoccupied adults of A. melanoceras are tenanted by the usual miscellaneous array of ants found in unoccupied swollen-thorn acacias: Crematogaster, Azteca, Camponotzis, Paracryptocerus, and Pseudo- myrmex. Since, as mentioned previously, unoccu- pied A. melanoceras seems to survive fairly well in well-developed forest, these colonies may attain unusually large size for ants tenanting swollen-thorn acacias. At times an Azteca or Crematogaster colony will take over all the thorns on the tree, but the ants? foraging on the acacia foliage is not thorough enough to prevent frequent damage by herbivores. P. satanica founding queens establish themselves in the thorns of A. melanoceras seedlings at the stage shown in Figure 30~. Until the ant colony has grown large enough to patrol the young shoot, growth is not rapid enough for the plant to have more than 3 to 6 intact thorns less than two years old) at any one time. Not only is the number of founding queens in the system small as compared with dry-land, swollen-thorn acacias, but one very rarely finds searching queens on the outside of the acacia. This indicates that few queens are produced by adult colonies or that the queens have a difficult time locating the widely spaced seedlings. Both possibilities are probably true, FIGURE 35.-Dormant shoot tip and ?crown? of A. melano- ceras seedling above point where latest swollen-thorns were produced (Panama, site 6). 50 SMITHSONIAN CON'I'RIBUTIONS TO BOTANY FIGURE 36.-Acacia melanoceras (Panama, site 6): A, petiolar nectaries characteristic of seedling leaves near swollen thorns; B, maximal development of petiolar nectaries in crown of old plant; c, minimal petiolar nectary development of large leaves subtended by type A thorns on lat- eral branches of 5-m-tall subadult. Leaf production on thc seedling acacia differs from that of dry-land acacias. The seedling pro- duces 1-4 leaves subtended by swollen thorns fol- lowed by a cluster of leaves without swollen thorns (Figure 35). By the time that swollen thorns have rotted off (about 1.5 years later), the leaves have been shed and the process is repeated if an ant colony has not become established. How long the seedling can continue this process is unknown, but I have found seedlings that have gone through this process at least five times. The seedling in Figure 30~ had done it at least four times, yet was only 53 cm tall. On the very young seedlings there may be no nectaries on the leaves (Figure 35) except those TABLE 9.-Distribution of leaflets among the 9 leaves on a 40-cm-tall seedling of A. melanoceras at the time of production of its first swollen thorn (a total of 52 leaves had been produced by the seedling, and the thorn was on the 50th node) Rachis Petiole Number of Number of number length (mm) length (mm) pinnae pinnules Leaf 1 ~ 83 44 16 428 108 104 117 112 113 116 108 109 22 22 31 18 30 24 27 21 24 28 28 24 30 22 32 21 604 492 660 492 174 654 598 512 NUMBER IS 51 280 240 200 160 120 - z Y L4 4 UI cl 0 X F- u L m u *O el 40 . . . . . . .* . . t m. .. .. . *. '. *.*. . *. .*. .. . ..*. . t .. 0.- .* .* . . . . . . . .. . .* .. .' : **. . . . . t *. . . . . . I I I 1 I I 0 5 10 15 20 25 NUMBER OF NECTARIES PER LEAF FIGURE 37.-Distribution of 1256 petiolar nectaries among 24 leaves of A. melanoceras from sterile lateral branches in an adult crown (Panama, site 6). subtended by, or close to, swollen thorns (Figure 36~). As the plant matures, the number of petiolar nectaries increases through the progression shown in Figures QA, B, c. The distribution of petiolar nectaries among the leaves (Figure 30~) in a mature acacia's crown (Figure 37) displays all the fre- quencies that can be observed by watching seedlings develop, but the majority of leaves have between 5 and 20 nectaries. The leaves of A. rnelanoceras (Figure 30~) are at least three times the size of those found on dry-land acacias. The number of nectaries is higher on adults of A. melanoceras than on any other swollen-thorn acacia. On some A. melanoceras there are even nectaries on the pinnae. Owing to the shaded nature of its habitat, however, nectar production may be very low. Beltian body production by A. melanoceras is never fully developed, again probably from the relatively high cost to the plant of Beltian bodies in a habitat with low solar energy input and (per- haps) somewhat reduced herbivore loads. The first leaves on a seedling, as is usual with swollen-thorn acacias, have no Beltian bodies. By the time the leaves are developed to the level described in Table 9, there are 1 to 3 Beltian bodies per new leaf; a dry-land acacia would have been producing 10 to 20 per leaf by the time it had produced this many (52) leaves, even when growing in the shade. Even the very large and healthy acacia at site 6 was producing Beltian bodies on only about half its pinnules. On adult A. melanoceras, type B thorns appear on a small stubby branchlet (Figure 34~) which quickly elongates to 5 to 10 cm (Figure 34~). On large and healthy occupied acacias, such as the one at site 6, there may be many of these short shoots in relatively tight clusters (Figure 34c). On acacias in poorer condition, they are usually scattered along 52 SMITHSONIAN CONTRIBUTIONS TO BOTANY 140 120 E E Y E 100 W cl 1 .i. 140 ,120 s E - A 1.4 1.8 2.2 2.6 i.101181111'' TOTAL THORN VOLUME (cc) r c. W cl g 100 B 0 2 80 1.0 1.4 1.8 2.2 2.6 3 .O TOTAL THORN VOLUME (cc) FIGURE 38.-Mature A. melanoceras (Canal Zone): A, distribution of total thorn volume among the 25 largest type B thorns (site 9); B, distribution of thorn volume among the 31 largest type B thorns (site 7). the branches (Figure 34~). The volumes of A. melanoceras type B thorns (Figure 38) are not spec- tacular and are rarely more than twice that of type A thorns. The similarity of the two different sets of thorns in Figure 38 is probably representative of variation in type B thorns throughout the range of A. melanoceras. Several aspects of the biology of A. melanoceras reflect its occupation by a multiple-queen species of ant: (1) As noted, there is not a great difference in size or toughness between type A and type B thorns. While P. satanica colonies do not have the great number of queens characteristic of P. venefica and the undescribed multiple-queen Pseudomyrmex in Acacia hindsii in western Mexico (Janzen, 1973), they do have 5 to 20 or more per colony and this means that the loss of a queen-thorn to a bird or other predator is not fatal to the colony. This means that the primary selective pressure for very large and strong type B thorns has been removed. (2) When there is the rare opportunity for an A. melanoceras to reproduce vigorously with root suck- ers, as at site 6, the multiple-queen colony can ex- pand indefinitely to provide patrolling for the increased surface area of foliage. Therefore, while A. melanoceras normally occurs as single plants that could easily be patrolled by a single-queen colony, when the situation arises, the plant can greatly increase in size (and presumably seed production). The acacia is effectively polymorphic for growth form. (3) The new searching queens may on occasion be absorbed by the colony that produced them rather than go off in search of seedlings. This may result in a lower survival of seedlings than would be the case with a single-queen ant species situation. On the other hand, the presence of several queens in a young seedling are all contributing workers to the development of the young plant. (4) When the opportunity occurs for the growth of a cluster of A. melanoceras shoots occupied by one multiple- queen colony, the shoots are free of competition from immediately adjacent vegetation. As is usu- ally the case with large multiple-queen colonies, P. satanica is especially effective at killing foliage that contacts its acacia. Incidentally, as the name suggests, P. satanica is very aggressive toward a wide variety of foreign objects. These ants are perhaps the most difficult of the obligate acacia-ant species to work with, primarily because they run inside NUMBER 13 53 one?s clothing rather than just on top as is the case with the other species. As mentioned before, this means that a large cluster of A. melanoceras oc- cupied by a large colony of P. satanica tends to be in extremely good condition. 4. Acacia allenii, new species Arbol usque ad 25 m alta; ramuli novelli minute velutini, glabrescentes; stipulae spinescentes inter- dum magnae, teretiusculae, porcatae, usque ad 4 cm longae, 1 cm latae; folia 20-40 cm longa; nec- taria sessilia cupulata ad juncturas pinnarum omnium et 1-10 in petiolis; pinnae (5-) 10-20- jugae, 7-10 cm longae; pinnulae (1%) 20-30-jugae, oblongae, lineare, vel ellipticae, (3-) 5-10 mm longae, 1-4 mm latae, obtusae, glabrae; flores aurei in capitulo globoso; involucellum annulare in medio pedunculo vel inferiore, interdum basi; legumina (immatura) atrobrunnea, puberula, granulata, glabrescentes, lineares, 8 cm longa, 8 mm lata, complanata, acuta sessiles vel subsessiles. TYPE.-?. H. Janzen 1769, Costa Rica, Pun- tarenas Province, Osa Peninsula, bank of tributary of Rio Agua Buena, 3.5 mi SW of Rincon, 9 March 1967 (holotype US 2639555; isotypes F, GH, Rica: Puntarenas: Forested hills near Golfito, Golfo Dulce, P. H. Allen 5B7 (US). 3.4 mi E of Golfito, Janzen 1658 (F, MICH, MO, NY, UC, US). 6 mi S of Rincon on the "Road to the Pacific," Janzen 2000 (F, GH, MEXU, MICH, MO, NY, UC, US). Type tree 8 m tall with widely spaced branches and very dark green bipinnately compound leaves 20 to 40 cm long. Pinnules 8 to 11 mm long, 2 to 3.5 mm wide and up to 60 per pinna. Pinnae 7 to 10 cm long and up to 40 per leaf. 2 to 10 widely spaced petiolar nectaries and 1 to 2 nectaries on rachis at bases of pinnae. Swollen stipular thorns dark brown and up to 4 cm long and 1 cm wide at base, external thorn walls with low and rounded longitudinal ridges. New leaves with orange-yellow food body on tip of each pinnule. Internode length 5 to 10 cm on lateral branches. Sterile specimen. Fruits and flowers described below from other specimens. Janzen field collection number 1769. Types deposited in U. S. National Herbarium, Smithsonian Institution. This plant is named in honor of Paul H. Allen, who collected the first specimen (site 1) and did so MEXU, MICH, MO, NY, UC). PARATYPES.-cOSta much to aid plant biologists working in Central American wet forests. In his book (Allen, 1962) A. allenii is identified as A. melanoceras, a mistake easily made if one is working from herbarium speci- mens of either species. His specimen from site 1 bears immature inflorescences but was not chosen as the type because its leaves were badly damaged and very small. This type of foliage is typical of flower- bearing branches but not sterile branches. Sexual material is of almost no use for identification or reference since plants bearing fruits or flowers are extremely rare in time and space. For these reasons, a sterile plant was chosen for the type (site 6). Specimens of mature pods are not available from any A. allenii, and the only good collection of flowers (site 5, no. 2000) is from a nearly leafless tree. The tree from which the type specimen was taken was growing approximately in the center of the geographic range of A. allenii (Figure 23), and along a creek bank in lightly disturbed ?rain forest,? which seems to be the normal habitat for A. allenii. Its foliage is very representative of sterile and mature plants of the species; the plant was 4 m tall and originated as a sucker shoot from a 22 cm diameter downed trunk that was at least 25 years old. I estimate that the shoot would have flowered for the first time within the next 3 to 4 years. It was occupied by a large colony of Pseudomyrinex ferruginea. Acacia allenii is very similar (and presumably very closely related) to A. melanoceras. I feel justi- fied in distinguishing it as a separate species on the following grounds: 1. The population of A. allenii appears separated from that of A. melanoceras by at least 100 miles (and probably more) and by the Continental Di- vide. Despite careful searching, neither plant has been found in the wet and relatively undisturbed forests along the Pacific side of the mountains running through the western half of Panama. It seems that by the time one moves high enough in the mountains to find sufficient rainfall for A. allenii or A. melanoceras, it is too cold for the species. A. collinsii occupies the drier areas between the ranges of A. allensii and of A. melanoceras. In view of the long geographic distance between A. allenii and A. melanoceras, genetic exchange by either pollen or seed flow seems extremely improb- able. This is not meant to imply, however, that there would not be extensive exchange of genetic 54 SMITHSONIAN CONTRIBUTIONS TO BOTANY 10 u) -1 z V t; 25 b a m a w I 3 X m 40 30 u) 20 -1 x 4 l-l !!3 10 a w 3 X B 1 13519 15 13 5 7 9 11 1315 A NUMBER OF INFLORESCENCE-BEARING B NUMBER OF INFLORESCENCES PER AXIL AXILS PER BRANCHLET FIGURE 39.-Acacia allenii (Costa Rica, site 5): A, distribution of 242 inflorescence-bearing axils among 37 sexual branchlets from the crown of a 22-m-tall tree: B, distribution of 2255 in- florescences among the 242 axils in Figure A. FIGURE 40.-Acacia allenii (Costa Rica): A, left, type B thorns on the base of a lateral branch with terminal type A thorns: right, type B thorns on a short lateral branch from the main trunk (site 2): B, left, type A thorns on lateral branches and main trunk of 3-m sapling; right, type B thorns on short lateral branches of the same tree (site 5). NUMBER 13 55 material were this gap to be bridged in some man- ner, This seems unlikely, since both species become even more rare than they already are as their habitat comes under agricultural influence. 2. The foliage of A. allenii is more similar to that of A. melanoceras than it is to any other acacia. However, leaves from mature A. allenii are easily distinguished from those of A. melanoceras, since a leaf of A. allenii usually has only 1 to 5 petiolar nectaries and has pinnules with about twice the surface area of those of A. melanoceras. These two differences, along with the floral traits described below, are consistent throughout the population of A. allenii. 3. Inflorescences of A. melanoceras are borne on long (up to 45 cm) sexual branchlets that tend to be bunched near the ends of sterile branches; in- floresences of A. allenii are borne on short (3 to 15 cm) sexual branchlets scattered along major branches and therefore throughout the crown. The number of axils with inflorescences on a sexual branchlet (Figure 39, 42~) is about half that of A. melanoceras. While each sexual branch of A. melanoceras has 1 to 5 inflorescences per axil, each inflorescence-bearing axil on A. allenii has up to 15 or 16 with a mean at about 9 for large samples. 4. The type B swollen thorns of A. melanoceras are normally found on short branches, while those of A. allenii are much more likely to be strung along the lower parts of a branch that bears large leaves and type A thorns on its distal portion (Figure 40~). A. allenii thorns regularly have an inward projection that partly occludes the basal part of the thorn cavity. This is almost always absent from the thorns of A. melanoceras. 5. The differences between A. allenii and A. melanoceras are of the same magnitude that can be encountered over the ranges of one widely spread species such as A. collinsii, A. cornigera, or A. hindsii. However, these widespread species appear to have continuous ranges and almost always the variation is clinal in nature. A. allenii and A. melanoceras each appear to be going its own way evolutionarily, and I have no reason to suspect that their similarity will persist through time. If for no other reason, they are likely to diverge since A. melanoceras is occupied exclusively by a multi- ple-queen ant colony while A. allenii is occupied by a single-queen species. There is always the possi- bility that P. ferruginea will evolve multiple-queen behavior and thereby evolutionarily ?catch up? with P. satanica in A. melanoceras. In the mean- time, however, A. melanoceras will likely have further diverged from A. allenii. 6. When describing the ecology of these two populations, lumping them under one Latin name destroys resolution. To call them mutual ?sub- species? is to infer a common evolutionary origin, when in fact A. melanoceras may be a wet-forest derivative of the Panamanian portion of the A. collinsii population, while A. allenii may be a wet-forest derivative of the Costa Rican portion of the A. collinsii population. Even if one is in fact the result of a long-distance, overland-seed dispersal from the other?s population (the most likely origin), we still have no way of knowing which is the ?parent? species and which the ?subspecies.? KEY COLLECTION LOCALITIES All specimens are from Puntarenas Province, Costa Rica. 1. 2. 3. 4. 5. 6. Forested hills near Golfito. 111-3-1951, P. H. Allen, no. 5997 (USNH 2085794). Allen (1952) reports A. allenii as occasional in pastures around Golfito. 2.5 mi E (inland) of Dominical. DHJ, VI-25-1967, no. 1669, mp. 3.4 mi E Golfito. DHJ, VI-28-1967, no. 1658, fl. 11.0 mi E Golfito. DHJ, VI-28-1967, 11-28-1965, nc. Within 8 miles of Rincon along road to the Pacific and along highland road, Osa Peninsula. DHJ, 111-1-3-1965, no. 791, 803, 811, 812, 813; 111-10-1968, no. 2000, fl. TwE.-bank of tributary of Rio Agua Buena, 3.5 mi SW Rincon, Osa Peninsula. DHJ, 111-9-1967, no 1769. NATUML HISTORY Like A. melanoceras, A. allenii has a very nar- row geographic range (Figure 23) which probably reflects rather narrow tolerances of moisture and temperature. Its uppermost elevation record (160 m at site 2) is probably representative for the entire range, and, incidentally, is probably about the same as that for A. melanoceras on the Atlantic side of Panama. A. allenii is occasionally found in some- what drier sites than A. melanoceras (such as land- slide succession, roadsides), but otherwise it appears to have the same ecological difficulty with invading disturbed habitats as detailed for A. melanoceras. Acacia allenii provides a somewhat better test than A. melanoceras of the prediction that both species are likely to become extinct with the intro- 56 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 41 .-Sedling of A. allenii; note single swollen thorn in crown and old rotted thorn near bottom of photo (Costa Rica, site 5). duction of European-type agriculture to a particu- lar site. Allen (1952), based on experience around Golfito in the 1940s, when much primary forest was being cleared, reported that A. allenii was an oc- casional plant in brushy pastures. Today (1965- 1970) it is totally absent from the pastures in the Golfito region but may still be encountered on the forest edge where fields are being pushed back into the forest, especially along the road between Golfito and the Pan-American highway. When the forest is cleared, the stumps of A. allenii produce small shrubby bushes. Like those of A. melanoceras under the same circumstances, these pasture shrubs may live for 1 to 4 years and attain a height of 2 to 3 m-in the forest they live as long as 25 years and attain a height of 22 m. Since the ant colonies in these shrubs are quite ineffective at clearing the vegetation from around the acacia, pasture fires normally kill the acacia and its ant colony. In late stages of woody succession such as on landslides or undisturbed regeneration following logging, light ground fires kill neither the acacia nor ant colony (Janzen, 1967a). In the general area of Rincon on the Osa Penin- sula, healthy seedlings of A. allenii in the 0.5 to 2 m size class occur at a density of at least 1 per hectare in old natural secondary succession (such as along creeks and rivers) ranging from 20 to 40 years of age. These young plants are often occupied by colonies of P. ferruginea with 25 to 300 workers. A high percentage of these seedlings appear to attain reproductive maturity since the same habitats have densities of adults as high as one per 5 hectares. When an area is cleared by logging and then al- lowed to return to forest, the density may be much higher, indicating that in uniform woody succes- sion, seedling survival is higher than in the frag- mented successional stages along rain-forest water- ways and landslides. For example, the lower part of the hillside immediately north of the airstrip near Rincon was almost clear-cut in the late 1940s and then allowed to regenerate undisturbed. The density of adult or subadult A. allenii in this site is about 3 per hectare. Along logging roads, A. allenii adults or subadults may attain a density as high as 4 per linear mile of road. The seedlings (e.g., Figure 41) of A. allenii are morphologically and physiologically very similar to those of A. melanoceras. For purposes of later com- parison with dry-land, swollen-thorn acacia seed- lings, I offer the following description of a 90 cm tall A. allenii seedling (site 5). The plant was growing in deep shade in undisturbed forest on level ground and had a tap root 22 cm long and 8 mm in diameter at ground level. As do other swollen-thorn acacias, the rootlets had nodules on them. The taproot went straight down and had 8 lateral roots of 1 to 2 mm diameter. The stem had narrowed to 5 mm diameter at 10 cm above the ground. Based on leaf scars, between 93 and 110 leaves had been produced, 8 of which were on the plant. Between 10 and 14 swollen thorns had been produced, 2 of which were still intact and contained founding queens of P. ferruginea. The swollen thorns had the longitudinal shallow grooves characteristic of adult plants of both A. allenii and A. melanoceras. Their volumes were 0.8 and 0.9 cc. NUMBER 13 FIGURE 42.-Acacia allenii (Costa Rica): A, sexual branchlet (site 2); B, mature seeds: seed in lower left had split its thin seed coat and was starting to germinate before it left the pod (site 2); C, rachis nectaries on leaf subtended by a type A thorn (site 5). FIGURE 43.-Acacia allenii (Costa Rica, site 4): A, sapling that was growing in full shade; B, type A thorns in crown center of the sapling. 57 58 SMITHSONIAN CONTRIBUTIONS TO BOTANY As with A. melanoceras, the seedling produced a whorl of leaves without swollen thorns which were either followed or preceded by several leaves with swollen thorns. The leaves were nearly as large as those of an adult, and radiated outward such that they did not overlap on a horizontal plane. Free- dom from overlap was aided by the leaf having a 3 cm petiole about twice the length of the petiole on a mature leaf. The leaves ranged from 15 to 18 cm in length, had pinnae 22 to 29 mm long, and had pinnules 6 to 7 mm long and 1.5 to 2.0 mm wide. Beltian bodies were on only the basal 3 to 6 pinnules of each pinna, and were 2 to 4 mm long. These large Beltian bodies are about twice the length and 4 times the volume of those on a mature plant. There were no petiolar nectaries, but there was a single small nectary at the base of each pair of pinnae (as in Figure 42c). Once such a seedling is occupied by an active colony of ants, it will grow into a sapling in 2 to 3 years such as the one in Figure 43~. As the sapling grows, it maintains a single-stem life form and produces virtually no root sucker shoots. In very young second growth, the plant becomes very bushy owing to insects eating off the growing shoot tips (again as with A. melanoceras). Seed-bearing adults have 12 to 20 growth rings and are probably 15 to 23 years old. They may attain a DBH of 35 cm and are generally 12 to 25 m tall. The adults are almost never emergent to the general canopy (in strong contrast to dry-land species). Senescent canopies look like those figured for A. melanoceras. After the top has rotted out and the trunk fallen, the root crown sometimes produces a single sucker which may become oc- cupied and grow to adult status. The same happens if a healthy adult is cut, but if the sucker shoots are then removed, the stump usually fails to pro- duce more. It appears to have very limited food reserves (again, in contrast with dry-land acacias). As does A. melanoceras, A. allenii lives in a habitat (heavily shaded understory of wet forest) of less intense defoliation pressure as contrasted with the dry-land, swollen-thorn acacias. In an at- tempt to quantify this difference, an experiment of the type conducted extensively with A. cornigera in lowland eastern Mexico (Janzen, 1967b) was performed on the A. allenii population at site 5 (cf. similar experiment for A. collinsii for direct comparative data for a dry-land species). During the two weeks preceding March 1, 1968, I searched intensively for adult or subadult A. allenii in site 5. The 21 healthy plants located were in various kinds of vegetation and ranged from 5 to 20 m in height. None was fully insolated. On March 1, all were cut, leaving 60-cm-tall stumps in heavy to light shade. The control plot of 12 acacias was established by placing the acacia crown, with its ant colony, over or adjacent to the cut stump. The experimental plot of 9 acacias was established by dragging the crown at least 5 m from the stump and usually farther. Experimental and control plants were mixed in sequence, with the treatment determined by a flip of a coin. As anticipated, the ants-or the colony-abandoned 6 of the control acacias and reoccupied one of the experimental stumps by moving the colony 6.5 m from the old crown to the new sucker shoots from the stump. This left a final tally of 7 occupied stumps and 14 unoccupied; these adjustments occurred within the first month after the stumps were cut. On September 17, 1968, six and a half months later, a variety of parameters were recorded (Table 10). The effect of an ant colony on sucker regeneration is obvious, but it is not nearly as dramatic as were this experiment to have been conducted with a dry-land acacia during six and a half months of the rainy season. These data, and some of the more obvious reasons for this difference are discussed in the following paragraphs. The diameter of the stump of a swollen-thorn acacia is weakly correlated with the amount of reserves the plant has for the production of sucker shoots. While the stumps of the occupied acacias appear to have smaller diameters, thus biasing the experiment in the favor of unoccupied shoots, in fact they are not statistically different. Further, differences in growth between occupied and un- occupied shoots for other reasons are usually far greater than those resulting from difference in stump size. Without much larger samples, it is im- possible to segregate out the effect of stump size. The tallest sucker shoot may originate at any point on the side of the 60 cm stump. For this reason, small samples tend to have a very high variance in the total height of the plant in the early stages of sucker regeneration. That the oc- cupied and unoccupied acacias have significantly different total heights only at the 0.025 (t-test) level is a reflection of this high variance. In this NUMBER 13 59 TABLE 10.-General condition of occupied and un- occupied A. allenii after 6.5 months of sucker shoot growth from cut stumps (see text for details), Costa Rica, site 5 Measurements Stump diameter (mm) average ........... s.d. ............... median ........... range ............. Maximum height of shoots (a) average ........... s.d. ............. median ........... range ............ Length of longest shoot (cm) average ........ s.d. ............... median .......... range ............ Number of suckers per stump average ........... s.d. ............... median ........... range ............. Number of intact shoot tips per stump average .......... s.d. ............... median ........... range ............. Number of damaged shoot tips per stump average .......... s.d. ............... median ........... range ............. Number of swollen thorns per shoot average ........... s.d. ............... median ........... range ............. Number of leaves per stump average ........... s.d. ............... mediah .......... range ............. Total live weight of leaves per stump (gm) ............... Occupied Unoccupied (n=7) (n i= 14) 40 17 44 18-67 157 73 172 41-245 112 105 117 18-202 6.6 1.8 6 4-9 5.0 2.0 5 3-7 1.9 1.6 2 0-5 22 22 15 0-57 48 20 44 16-74 76 56 28 52 11-1 15 81 50 85 0-182 34 29 37 0-83 10.6 6.5 13 0-1 8 1.8 1.6 1 0-6 12.0 9.1 11 0-29 20 23 5 0-69 32 26 27 0-73 10 respect, the results are not outstandingly different from those expected in a drier habitat, except that in the dry land experiment this high variance would have disappeared within about 3 months of sucker regeneration. Much less so than with dry-land acacias, A. allenii suckers are not ?racing? to maintain a spot in the general canopy. They are already far below it and a difference of a few centimeters in height, between occupied and unoccupied, is not so im- portant as it is with an acacia whose survival to reproductive maturity depends on remaining inso- lated. With A. allenii, success in height increment is measured in the size of eventual seed crops, rather than in living or dying during a short period. Of course it is the length of the ?leading? sucker shoot that usually determines the height of the regenerating acacia, though sometimes the longest sucker originated low on the stump and this is in fact shorter than the tallest sucker. The significant difference at the 0.005 level (t-test) between the length of occupied and unoccupied suckers is the result of feeding damage by insects. If no shoot tips are damaged on the growing sucker, one or two quickly become the primary suckers and vertical elongation of the others nearly ceases. There is an occasional loss of a shoot tip on an occupied acacia (usually the moth larva, Coxina hadenoides, which appears to be host-specific on swollen-thorn acacias, cf. Janzen, 1967b). If the main sucker shoot is generally undamaged, however, a new vertically elongating branch is produced from a leaf axil high upon the main sucker shoot when the shoot tip is eaten and the ?single trunk? life form is not lost. If there is repeated loss of sucker shoot tips to in- sects, however, as is the case of unoccupied shoots, many short suckers are produced, none of which grow very tall. In striking contrast to sucker height, length of shoot, and condition of the shoot tips, the number of swollen thorns and leaves are not significantly different between the occupied and unoccupied acacias. However, the total live weight of leaves on the occupied acacias was 8 times as great as on the unoccupied acacias. All of the above data suggest that the following system is operating. On an unoccupied acacia, the insects that will feed on it encounter it at wide intervals. In between times, the plant has the 60 SMITHSONIAN CONTRIBUTIONS TO BOTANY chance to grow new sucker shoots to replace those eaten off. The long time between attacks by herbivores almost certainly reflects a low density of these insects. This in turn is probably brought about by the low productivity of the understory community and the absence of other mimosaceous legumes. This is in strong contrast to dry-land sites, where the insects arrive so frequently that most new shoots are eaten off before they have grown more than a centimeter. Phytophagous insects, however, find the A. allenii shoots often enough so that no one shoot ever becomes large enough to assume physiological dominance over the others. The result is a stump crowned with a scrawny clump of as many as 30 short shoots. Not only do suitable herbivores more rarely encounter the acacia, but when they do, they appear to feed primarily on the newest leaves and shoot tips. The old leaves are gradually nibbled at, but are rarely eaten in entirety. This results in the very great difference in leaf weight between the occupied and unoccupied acacias. A swollen thorn is edible only when very young, and thus if herbivores find the plant only occasionally, swollen thorns may accumulate almost as fast as damaged leaves. Judging from natural experiments of the types described above, unoccupied A. allenii eventually die. During the six and a half months of this experiment, however, only two (14 percent) of the unoccupied stumps died. In a dry-land experiment, the percentage would have been between 30 and 60 percent during one rainy season. Lowered herbivore pressure is partly responsible, but the obvious shade tolerance of A. allenii is another factor. Suckers of dry-land, swollen-thorn acacias in shade, such as the A. allenii stumps received, would have ceased growth almost entirely within about 3 months. In the dry areas, a major effect of the herbivores was to make the acacia lose the race for a spot in the sun. Such an interaction does not occur here. Incidentally, P. ferruginea has not lost its propensity to kill vegetation contacting the canopy FIGURE 44.4eneral postulated distribution and collection localities: A. chiapensis (stippled area): A. gentlei (cross-hatched area). NUMBER 13 61 of A. ullenii. Several of the occupied sucker shoots in the experiment grew through vine canopies and the ants literally chewed a hole through for the sucker shoot. 5. Acacia chiapensis Safford Acacia chiapensis Safford, 1915:356. Acacia chiapensis is known only from Mexico (Figure 44) and the type specimen was collected at the same site as the type of A. collinsii (site 1). A. chiupensis is distinguishable from all other swollen-thorn acacias by the combination of globu- lar inflorescences, multiple and circular petiolar nectaries, semi-indehiscent seed pods, and straight or slightly curved thorns that are circular in cross section. It is easily distinguished from A. collinsii, since the latter species has elongate inflorescences and few petiolar nectaries. The type specimen of A. chiupensis is representative of its species throughout the range except that the single type A thorn is light brown, while A. chiupensis thorns are usually dark brown to black. The immature seed pods on the type specimen are full-sized and display the distinct ridge bordering the margins of the valves, which is characteristic of A. chiupensis, and absent from other swollen-thorn acacias. KEY COLLECTION LOCALITIES (ALL IN MEXICO) 1. San Fernando, Chiapas. VII-12-1907, G. N. Collins, no. 164. 2. Shores of lake behind Presa Miguel Aleman, Temascal, Oaxaca (70 m) . XI-3-1963, DHJ, nc (occasional). 3. Temascal, Oaxaca to La Granja, Veracruz (16 m), All year, DHJ, nos. 118, 130, 140, 154, 158, 164, 165, 170, 185, 190, 193, 200, 204, 209, 214, 222, 230, 235, 242, 247, 266, 277, 281 (common in permanently wet sites such as below the dam at Temascal, Rio Chichicazapa, and Arroyo En- medio, very rare otherwise). 4. 3.4 to 8.5 mi E Papaloapan on Cd. Aleman-Loma Bonita Hgy., Oaxaca (25 m). XI41963, DHJ, nc (3 plants). 5. 19.6 mi SW Tuxtepec on Tuxtepec-Oaxaca Hgy., Oaxaca. XI41963, DHJ, nc (1 plant). 6. 26.4 mi SW Tuxtepec on Tuxtepec-Oaxaca Hgy., Oaxaca. XI-4-1963, DHJ, nc (common). 7. 0.9 mi S Chontalpa on Cardenas-Presa Malpaso Rd., Tabasco (about 80 m). VI-17-1966, DHJ, nc (3 plants). 8. 12.6 mi W Cardenas on Hgy. 180, Tabasco (30 m) . VI- 15-1966, DHJ, nc (1 plant), fl. 10. 11. 12. on Hgy. 185, Oaxaca (300-160 m). VIII-18-1967, DH], nos. 1500, 1510 (8 plants). Rio Cascabel, Oaxaca. 111-15-1934, C. D. Mell, no. 2239 (US 1642830) fl. Upper Rio Coatzacoalcos, Oaxaca. 111-8-1930, C. D. Mell, no. 600 (US 1492290), fl. Ubero, Oaxaca (30-90 m). IV-1937, L. Williams, no. 9124 (US 1741272, F 897560). NATURAL HISTORY Acacia chiupensis is the only dry-land lowland, swollen-thorn acacia that could be called rare, un- FIGURE 45.- Acacia chiafiensis (Mexico, site 3): A, emergent crown of an occupied 4-m-tall adult; B, two adults, 2-m tall, in an open field; c, looking up through crown of 5-m-tall ~- I 9. 15.1 mi S and 5.2 to 11.0 mi N turn-off to Matias Romero adult. 62 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 46.--Acacia chiupensis (Mexico, site 3): A, stump (4 inches in diameter) with multiple sucker shoots; 8, branch pattern of 1.5-m-tall tree. less it is found that the rare, moist forest species A. mayana also occurs regularly in dry sites. It is often represented by only one or two plants where found. At both places where it was common (sites 3 and 5), it was growing along or near watercourses in vegetation heavily disturbed at 2 to 4 year inter- vals by earth-moving machinery or logging; the sites were subjected to a severe dry season, but the acacias were growing in sites of locally high soil moisture. Based on the general condition of plants growing in a variety of sites, it appears that the pre-Colombian habitat of A. chiupensis was the evergreen or semideciduous forest along the edges of foothill streams running through deciduous forest. Road-building machinery appears to have been primarily responsible for movement out of this habitat into human disturbance sites. A. chiapensis has not invaded brushy pastures or up- land regeneration following clearing for cornfields or logging, in striking contrast to A. collinsii and A. cornigera, which probably occupied the same pre-Colombian habitat as A. chiupensis. Some of the ecology of A. chiupensis at site 3 was discussed briefly as a contrast with A. cornigera (Janzen, 196713) and A. globulifera is incorrectly listed there as a synonym of A. chiupensis. A more detailed discussion of A. chiupensis is offered here after having seen the plant growing at a variety of sites. Acacia chiapensis has a life form much like that of dry-land nonant-acacias such as A. farnesiana and A. macracantha (Figures 45~, B and 46~) but if given ample water will grow into a 8-m-tall tree. When repeatedly cut, a stump of A. chiupensis tends to form a large cluster of shoots from one root crown (Figure 46~), rather than also sprouting from lateral roots as the custom with A. cornigera. A. chiupensis is generally found in the general canopy or slightly emergent to it (e.g., Figure 45~); seedlings in the shade die even faster than those of A. cornigera and thus are rarely noticed. Acacia chiupensis resembles nonant-acacias in an- other respect. The seeds are surrounded by a dry pulp that is at best only slightly sweet and is prob- ably only rarely eaten by birds. Since the dehiscent pods (Figure 47c) do not open far enough for the seeds to fall out, the pod can either be wind dis- persed (perhaps up to 2 or 3 m), water dispersed, or rodent dispersed. None of these methods even ap- proaches being as effective as bird dispersal for NUMBER 13 63 FIGURE 47.-Acacia chiupensis (Mexico, site 3): A, flowering branch from a heavily shaded tree; B, flowering branch from a fully insolated tree; c, mature pods on a branch like that in figure A. rapidly invading newly created patches of young regeneration. All other swollen-thorn acacias, with the possible exceptions of A. cookii and A. ruddiue, are regularly bird dispersed. The flowers are pollinated by the wide variety of small bees that are available in deciduous forest habitats and contemporary second growth during the dry season. The inflorescences are borne in very dense clumps in the leaf axils of vegetative branches (as on A. cookii in the southern part of its range) produced in the early part of the previous rainy season. At any one site, the number of in- florescences per axil can vary strongly (Figures 47~, B) and the largest numbers are found on branches that are fully insolated and from bushes that are fully insolated. Inflorescences generally set 1 to 5 pods, but may on occasion have as many as 10. The seedling ecology of A. chiupensis has not been examined but the morphology of young bushes indicate that it is more like that of nonant- acacias than like that of swollen-thorn acacias. It appears that the first swollen thorns often do not appear until the fiftieth or even later leaf, and even in fully insolated bushes, many leaves on lengthening lateral branches are not subtended by swollen thorns. This behavior persists even in very old A. chiupensis, resulting in branch segments with many nodes that lack swollen thorns. The canopy section in Figure 45c has about the maxi- mum number of thorns in it to be found in an A. chiapensis canopy. While the large leaves of mature A. chiupensis have large numbers of petiolar nectaries (Figure 48~), sometimes making the petiole resemble that of A. melunocerus, the leaves on young plants have smaller numbers (e.g., Figure 48~), and I suspect that very young seedlings may only have one or zero petiolar nectaries per leaf. While nectar production of A. chiupensis is apparently very high, as with the other swollen-thorn acacias, Beltian body pro- duction is clearly reduced as compared with other 64 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 46.-Petiolar nectaries of A. chiupensis leaves (Mexico, site 9): A, from fully insolated old crown; B, from a young sapling. FIGURE 49.-Acacia chiupensis (Mexico, site 3): A, type A thorns from lateral branch of adult; B, thorns intermediate between type A and type B on lateral branch from crown of tree in Figure 45c; c, type B thorns from crown of same tree. NUMBER 13 130 - I a - 110 - I * W z w cl 90 - 0 I 65 dry-land, swollen-thorn acacias such as A. cornigera, A. hindsii, and A. collinsii. A. chiupensis leaves on fully mature and well-insolated plants may have Beltian bodies on only half the pinnules of a large leaf subtended by a swollen thorn. There is strong variation in the degree of herbivore damage to occupied A. chiupensis-all occupied plants censused have been occupied by P. ferruginea or by a smaller undescribed species of obligate acacia-ant. Even plants with the largest ant colonies to be found on A. chiapensis usually have most shoot tips eaten, as with an unoccupied A. cornigera. The result is that plants of A. chiupensis are generally very shrubby and stunted in appear- ance. The cause appears to be that with the reduced number of Beltian bodies and reduced number of thorns, almost none of which have a large volume or could be regarded as type B thorns (Figures 49, 50), the ant colony never grows very large. Since the colony may be old, its workers may be very aggressive, but there is too much acacia canopy to patrol effectively. A 2-m-tall A. chiapensis such as the one on the left in Figure 3a in Janzen (1967b) may have a maximum colony size of about 2,000 to 4,000 worker P. ferruginea, while an adjacent A. cornigera of equal size and with an equal number of leaves will have between 5,000 and 10,000 worker ants. Younger plants of A. chiapensis character- istically have a colony size reduced to the same degree, with the result that they are slow to gain effective patrolling by ants, and thus have to be on thoroughly cleared ground in order to mature. When roads or dams are built, A. chiapensis be- comes established on the bare soil where there is no competition from fast-growing suckers from old stumps of woody plants (as would have been the case in areas cleared by fire or machete). At site 6, there was a nearly pure stand of 3- to 6-m-tall A. chiapensis covering a gravel island of about 2 acres in the middle of a river; a newly bulldozed site obviously simulates such a situation for primary succession. If an A. chiupensis is growing within 1 to 2 m of an A. cornigera, the latter species always contains the queen and the A. chiupensis serves as an auxiliary unit to the main colony. This means that during a dry season sufficiently severe to cause the A. chiupensis to drop its leaves, the ant colony is very likely to abandon it and retreat into the acacia with the queen. All of these observations suggest 150 1 t 0 b 0 a a i c I I I I I 0.4 0.6 0.8 1.0 1.2 1.4 TOTAL THORN VOLUME (cc) FIGURE 50.-Distribution of 89 cc of thorn volume among 100 type A thorns from an adult occupied A. chiapensis (Mexico, site 3). that A. chiupensis is a marginal host for obligate acacia-ants, but that it is still dependent on the ants. However, it may not be as dependent as other dry-land, swollen-thorn acacias since its leaves con- tain cyanogenic glycosides that are also present in nonant-acacias (A. farnesiana) but absent from A. cornigera (Rehr et al., 1973). Not all A. chiupensis, however, demonstrate strong susceptibility to insect attack. At site 3, and only at this site, it appears that A. chiupensis may be hybridizing with nonant-acacias such as A. macra- cantha. Numerous plants occur there that are in- termediate in thorn, leaf, and branching morphol- ogy between these two species. A gradient may be recognized between the ?typical? A. chiupensis life form-dark green large leaves with large pinnules, swollen thorns subtending more than half of the large leaves on lengthening branches, new swollen thorns soft for several days allowing easy entry and with soft inner parenchyma allowing easy cleaning, fast-growing shoot tips, upright-cylindrical life form with a few major branches from a single trunk, and severe herbivore damage to uppermost shoot tips-and a life form like the shrubs in Fig- ures 45~ and 45~ which tend on all the above char- acters to be much more like A. macracantha. A. mucracantha is common in the area and shows the opposite traits of those listed above for A. chiapen- sis. It is particularly obvious that the forms inter- mediate between A. chiapensis and A. macracantha 66 SMITHSONIAN CONTRIBUTIONS TO BOTANY are more damaged by herbivores than ordinary nonant-acacias, but they are very weakly occupied, if at all. The lack of damage appears to be due to chemical and morphological defenses, since these only very mildly damaged bushes are often imme- diately adjacent to heavily damaged more typical A. chiapensis. The large stump of one of these intermediate acacias in Figure 47~ is particularly instructive in understanding the poor development of an ant colony in the apparent hybrids. It has produced at least 50 small sucker shoots since it was cut, but only five swollen thorns; sucker shoots of ?typical? A. chiapensis or A. cornigera from an equal-sized stump would have had a swollen thorn subtending almost every new leaf (e.g., Figure 53 in Janzen, 1967b). Not only are there few swollen thorns, but they harden rapidly, and the queen of P. ferru- ginea has a very difficult time hollowing out enough of the inside of the thorn for its own body, to say nothing of enoligh space for brood and workers. Many of the new leaves on a stump like this have no Beltian bodies and the nectaries are very poorly developed, again in strong contrast to A. chiapensis or A. cornigera new sucker shoots. Not only are the suckers slow growing, but dominance is poorly de- veloped with the result that many shoots develop simultaneously. The dense shrub resulting eventu- ally matures in completely insolated sites-as might be expected of dry-land nonant-acacias-but ap- pears incapable of competing with dense vigorous vegetation on land cleared by fire or machete. In short, there appears to be a gradient in ability to deal with the surrounding vegetation from A. macracantha to the apparent hybrid to A. chiapen- sis to A. cornigera. Along this gradient, the ant becomes progressively more important to the plant in protecting its faster growing shoot tips that have progressively fewer chemical defenses. Both A. macracantha and A. cornigera represent adaptive peaks that are successful in vegetation types of great areal extent. The hybrid and A. chiapensis Contour lines at 1000, 5000, and 10,000 feet FIGURE 5l.-Collection localities of A. globulifera; its detailed collection records are too patchy to allow an estimation of the general distribution at this time. NUMBER I3 67 represent adaptive peaks that do well under much rarer combinations of vegetative competition and rainfall. It is not meaningful to say that one is any more or less successful than the other, except in terms of the number of acres of vegetation occupied. 6. Acacia globulifera Safford Acacia globulifera Safford, 1915:360. Acacia donnelliana Safford, 1915:361. Myrmecodendrum globuliferum (Safford) Britton and Rose, Myrmecodendron donnellianum (Safford) Britton and Rose, 1928:98. 1928:93. Acacia globulifera is easily distinguished from other swollen-thorn acacias by the possession of spherical inflorescences, completely dehiscent pods with valves that curl back on themselves (as with A. hindsii), and 2 to 6 circular or oval petiolar nectaries (Figures 6, 7c). Its range (Figure 51) overlaps slightly with that of A. collinsii, A. cornigera, A. gentlei, A. cookii, and A. chiapensis, but it is very rarely encountered growing side by side with these plants. Herbarium specimens of A. globulifera are incorrectly listed as a synonym of A. chiupensis in Janzen (1967b), but a comparison of the petiolar nectaries and mature pods-barely indehiscent in A. chiapensis, fully indehiscent in A. globulifera-will easily separate them. The petiolar nectaries of A. globulifera may on occasion be confused with those of A. collinsii, but A. collinsii has elongate inflorescences. In field studies, failure to examine plants closely may result in mis- identifying A. globulifera as A. collinsii or as A. gentlei because it has a very similar life form to either of these when growing in their habitats. Safford's type specimen of A. globulifera (site 1) appears to have been taken from a fully insolated shrub growing in a locally exceptionally dry site, since its thorns are unusually light in color (beige) and short (the largest type A thorns are only 4 cm long), and its leaves only attain 8 cm in length. In the coastal areas of the Yucatan Peninsula such plants are usually found on old beach dunes; the more inland plants have thorns that are nearly black and twice as long, and have leaves up to 20 cm long. The inflorescences on the type specimen are representative of A. globulifera throughout its range except that they have peduncles only 10 mm or less in length, and in wetter areas the peduncles of A. globulifera may attain a length of 20 mm. Safford's type specimen for A. donnelliana was collected from that portion of the A. globulifera population growing between San Pedro Sula and Santa Rosa, Honduras (site 24), and is representa- tive of the plants in that area growing in very wet habitats (swamps, creek edges); its thorns are nearly black, its leaves are up to 20 cm long, its inflores- cence peduncles are up to 18 mm long, and its sexual branchlets are very long (e.g., Figure 52~). KEY COLLECTION LOCALITIES Mexico: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Type of A. globulifera: Silam (Dzilam), north coast of Yucatan. April 1895, G. F. Gaumer, no. 1909. 48 mi NE Puerto Real (Isla Aguada) on Mexico 180, Campeche. VI-22-1966, DHJ, nos. 611, 615, 617, 623 (OC- casional) , mp. 11.3 mi N intersection of Mexico 185 and 190, on Hgy. 185, Matias Romero, Oaxaca (290 m). VIII-18-1967, DHJ, no. 1502 (single plant) . 11.8 mi N intersection of Mexico 185 and 190, on Hgy. 185, Matias Romero, Oaxaca (290 m) . VIII-18-1967, DHJ, nc (occasional plants). 25.8 mi N intersection of Mexico 185 and 190, on Hgy. 185, Matias Romero, Oaxaca (250 m). VIII-18-1967, DHJ, nc (very abundant). 26.3 mi N intersection of Mexico 185 and 190, on Hgy. 185, Matias Romero, Oaxaca (240 m) . VIII-18-1967, DHJ, nc (3 plants). 31.5 mi N intersection of Mexico 185 and 190, on Hgy. 185, Matias Romero, Oaxaca (260 m). VIII-18-1967, DH], nc (1 plant). Chivelia, Oaxaca, Mexico. 111-6-1928, C. D. Me11 (US 1406953), fl. Silam, Yucatan, Mexico (sea level). IV-1895, G. F. Gaumer, no. 655 (US 1267146), fl. Silam, Yucatan, Mexico (sea level). IV-1909, G. F. Gau- mer, (US 1267508) fl. Guatemala: 11. 12. 13. 14. 15. 16. Finca San Joaquin, slopes above Rio Negro, near San Cristobal de Verapaz, Alta Verapaz (1080 m). VII-14- 1967, DHJ, nc (very abundant), mp. 20.5 mi NE Guatemala City on CA9 (980 m). VIII-8- 1967, DHJ, no. 1565 (occasional), fl., gp. 26.7 mi to 26.9 mi NE Guatemala City on CA9 (700 m) . VIII-9-1967, DHJ, nc (abundant), gp. 29.6 mi NE Guatemala City on CA9 (850 m). VIII-9- 1967, DHJ, nc (rare). 23.4 mi SW of turn off to Coban on Hgy. 9, Progreso, Dept. Progreso. VIII-5-1966, DHJ, no. 444. 16.2 mi NE Guatemala City on CA9 (870). V-14-1965, DHJ, no. 1741 (1 plant). 68 SMITHSONIAN CONTRIBUTIONS TO BOTANY 17. Sanarate, Depto. Progreso (900 m) . XI-9-1916, Popenoe, no. 720 (US 2089415, US 1319702), mp. 18. La Libertad, Chicbul, Depto. Peten. IV-29-1933, C. L. Lundell, no. 3088 (US 1635206, F 738059), fl. 19. Rio Teculutan, above Teculutan, Depto. Zacapa (250- 275 m). 1-7-1942, J. A. Steyermark (F 1191231) fl., mp. 20. Fallabon-Yaxha Rd., Depto. Peten. 111-22-1933, C. L. Lundell, no. 2214 (F 740476), mp. 21. Lake Yaxha, Depto. Peten. 111-23-1933, C. L. Lundell, no. 2208 (F 738041). British Honduras: 22. Vaca, El Cay0 Prov. 111-2-1938, P. H. Gentle, no. 2275 (F 945039), fl. 23. El Cayo, El Cayo. 111-5-13-1931, H. H. Bartlett, no. 11992 (US 1493114), fl. Honduras: 24. 25. 26. 27. 28. 29. 30. Type of A. donnelliana: San Pedro Sula, Cortes (200 m) . March 1888, C. Thieme, no. 5216. 19.4 to 19.9 mi SW San Pedro Sula on Honduras 18 (road to Santa Rosa) Depto. Cortes (160 m). VIII-2-1967, DH], no. 1682, 1688 (8 plants), fl., gp. 24.6 to 26.0 mi SW San Pedro Sula on Honduras 18 (banks of the Rio Chamelecon; road to Santa Rosa), Depto. Santa Barbara (380 to 420 m). VIII-3-1967, DHJ, no. 1604, 1608 (occasional). 28.4 mi SW Sula on Honduras 18 (road from San Pedro Sula to Santa Rosa), Depto. Copan (780 m) , VIII-3-1967, DH], no. 1630 (1 plant). 10.4 to 16.3 mi (Cucuyagua) SW Santa Rosa on Hon- duras 18 (road to Nueva Ocotepeque), Depto. Copan (980-1180 m) . VIII-3-1967, DHJ, no. 1634 (occasional), gP. Yoro, Depto. Yoro (850 m). 1-26-1934, J. B. Edwards, no. P-746 (US 1828182, F 759556), fl. Santa Rosa de Copan, between El Portillo and San Juan Opoa, Depto. Copan (1000 m). IX-23-1963, A Molina, no. 12873 (F 1622365). NATURAL HISTORY As with the other widespread swollen-thorn acacias, it is difficult to characterize the contempo- rary or pre-Colombian habitats of A. globulifera in one statement. It is commonly encountered at elevations that represent the upper elevational range of swollen-thorn acacias (e.g., sites 11, 28, 30); however, all of these higher elevation sites are characterized by being dry and dusty even during the rainy season. It is only at these sites that A. globulifera is sympatric with no other swollen-thorn acacia, though at some lowland sites A. globulifera may be the only riparian species, with A. cornigera or A. collinsii on the hillsides (e.g., site 26). At lower elevations in the wetter part of the range (e.g., sites 25, 26), it is most commonly encountered in riparian and swamp successional stages and oc- casionally on the edges of pastures and in roadside right-of-way. At sea level in the drier parts of the range (e.g., sites 2, 3-7) it is found on old sand dunes, rocky slopes, swamp edges, and badly eroded creek edges. No healthy plants have been found in habitats where they are heavily shaded, and almost all reproductive individuals have crowns that are emergent to the general canopy or are growing on nearly bare ground. One is left with the distinct impression that A. globulifera can withstand a very wide range of temperature and moisture regimens if it is moderately to heavily insolated. Since the largest individuals encountered have only been 3 m tall, and most are much shorter, this restricts the plant to young stages of succession, or to habitats with low or badly broken vegetative cover. At the higher elevation sites, A. globulifera is often asso- ciated with pines and oaks, and at sites 11 and 13 became very common-up to 20 adults per acre- in old pastures and regenerating fields formed by clearing mixed pine-oak forest. Acacia globulifera is distinctly shrubby in life form, and like the other dry-land acacias may bear a large seed crop when only 1 to 2 m tall if well insolated. If growing in full sun continuously, A. globulifera generally does not grow over 2 m tall (Figure 53~) but if slightly shaded, it has a slightly more attenuated life form (Figure 53B), and its crown does not expand much laterally until it is emergent to the general canopy or otherwise inso- lated. The length of the sexual branchlets of A. globu- lifera appears to be more directly related to the health of the plant than is the case with other swollen-thorn acacias. For example, the very long branchlets in Figure 52~ are from a 2.5-m-tall acacia in its third year of sexual reproduction and fully insolated with a large crown and occupied by a very large ant colony. The same is true of the long sexual branchlets bearing immature pods in Figures 54B and 54c. On the other hand, the very short sexual branchlets in Figure 53B are on the healthiest appearing branch in the senescent crown of a 4-m- tall acacia in a swamp. Though fully insolated, this 14-year-old acacia had lost its ant colony and was heavily damaged by herbivores. The extreme differences in number of inflorescences borne by the two branches in Figure 52 are documented in NUMBER 13 69 FIGURE 52.--Acacia globulifera (Honduras): A, elongate sexual branchlets from a very healthy adult (site 28): B, axillary inflorescences from the crown of a senescent tree (site 25). FIGURE 53.-Acacia globulifera (Hondu- ras): A, heavily insolated subadult (site 26); B, main trunk of subadult growing under intense lateral competition (site 28). 70 - - SMITHSONIAN CONTRIBUTIONS TO BOTANY 60- a 2 50- w 40 - 30- U a I W u z U v) W ci U 20- U 0 $10- z W cl FIGURE 34.-A, lateral branch of A. hindsii with large number of immature infructescences (Nica- ragua, site 85); B, long sexual branchlets of A. globulifera with heavy crop of immature in- fructescences (Guatemala, site 13) : c, the lateral branch bearing the infructescences in Figure B. 0 0 0. 0: 0 I I 5 15 25 A NUMBER OF INFLORESCENCES PER BRANCHLET - 32 U z 4 IT L9 m 230 a 4 W W U z 0 LL! a U 4' 210 5 190 2 170 E u, 150 0 32 z W ci 130 0 0 0 0 0 40 60 80 100 NUMBER OF INFLORESCENCES PER BRANCHLET FIGURE 55.-Number of inflorescences per sexual branchlet in A. globulifera (Honduras): A, crown of a senescent tree that had an inflorescence crop of about 250 (site 25): B, crown of a healthy adult that had an inflorescence crop of about 12,000 (site 28). NUMBER 13 71 Figure 55; the size of the inflorescence crop is clearly proportional to the condition of the acacia. Acacia gEobulifera with many long sexual branch- lets usually bear large seed crops (Figures 54B, C, 56). I suspect, however, this means that a large percent of the seeds are killed by bruchid beetles. As with the other dry-land acacias, the young green pods (Figure 112~) remain on the plant throughout the rainy season and do not mature until about the time the new flowers are produced during the following dry season. Mature undehisced pods of A. globulifera have not been collected and there- fore there is no information available on seeds. Since the pods curl open and yet stay on the tree for several months after all seeds are gone, I suspect that they are bird-dispersed as with A. hindsii. Seedling biology of A. globulifera has not been investigated except to note that they develop their first swollen thorns, Beltian bodies, and petiolar LO v) W u 2 w u v) W 3 s5 H U 0 CT W m 9 z: ~ 13 1 4 5 7 9 11 13 NUMBER OF PODS PER INFLORESCENCE FIGURE 56.-Distribution of 207 immature seed pods among the infructescences from a healthy adult A globulifera (Hon- duras, site 28). FIGURE 57.-Acacia globulifera: A, main trunk covered with type B thorns bleached white by the sun (Guatemala, site 12); B, central axis of subadult with downward curved type A thorns (Honduras, site 28). 72 SMITHSONIAN CONTRIBUTIONS TO BOTANY 120 110 - E E 100 I X F z W 4 w 90 2 80 a 0 z k 4 70 a t; b FIGURE 58.-Acacia globulifera: A, type B thorns (Honduras, site 26); B, left, type B thorns and on right, type A thorns (Guatemala, site 15). TOTAL THORN VOLUME (cc) FIGURE 59.-Distribution of 49 cc of thorn volume among 69 type A thorns from lateral branches in the crown of a healthy adult A. globulifera (Honduras, site 28). nectaries very early in development, like other dry- land, swollen-thorn acacias. Also as with other dry-land, swollen-thorn acacias, seedlings of A. globulifera appear to die within 1 to 2 years if they are in heavy shade. Acacia globulifera produces large numbers of swollen thorns early in its development, and con- tinues to do so on all branches. The thorns are exceptional among swollen-thorn acacias in showing extreme differences in color from one habitat to another. Adding to the complexity is the fact that A. globulifera thorns bleach exceptionally easily in direct sunlight (e.g., Figure 57~). They are com- monly bicolored as well, with lighter tips than bases. Type B thorns are somewhat larger than type A thorns (Figures 57~, 58~, 59, 60), but not spec- tacularly so. They also do not differ in morphology from type A thorns in any outstanding manner; even the largest type B thorns are usually round in cross section. Type A thorns often curve down- ward (Figure 57~) or project outward almost at right angles from the branch bearing them (Figure 58). Acacia globulifera shows greater variation in petiolar nectary morphology than any other species of swollen-thorn acacia. The major factor appears to be in the moisture of the ground where the individual plant is growing; the wetter the site, the longer the leaves. A great deal of the variation in nectary morphology can be viewed as simply the effect of lengthening or shortening the leaf at the NUMBER 13 140 - 120 - 100 - 73 6 6 z U 0 X I+ 160 -I 0 ooo. 0. n 0 0 0 1.0 2.0 3.0 TOTAL THORN VOLUME (cc) FIGURE 60.-Distribution of 48 cc of thorn volume among 31 type B thorns from the trunk of A. globulifera in Figure 57A. time of formation. The effect of this variation in nectary size and number on the amount of nectar produced per leaf is unknown. There is no evidence that the ants react differently to any of these different forms of nectary. The impact of herbivore on unoccupied plants of A. globulifera has not been examined experi- mentally, but plants that lose their ant colony through natural causes are badly damaged by in- sects in the lowlands. At the upper end of the elevational distribution of ?4. globulifera, appar- ently healthy acacias in reproductive condition are encountered that obviously have been unoccupied for at least a year. They are often tenanted by other arboreal ants in the genus Camponotus. At these high elevations, where the weather is often dry, windy, and cool, herbivorous insect populations appear to be much reduced as compared with the lowlands and this is probably the reason why un- occupied A. globulifera is relatively free of insect damage at these sites. 7. Acacia sphaerocephala Schlectendal and Chamisso Acacia sphaerocephala Schlectendal and Chamisso, 1830:594. Acacia ueracruzensis Schenck, 1913:362. Acacia dolichocephala Safford, 1915:355. Acacia sphaerocephala is easily distinguished from all other swollen-thorn acacias by having the elongate canoe-shaped petiolar nectaries found also on A. cornigera and A. mayana, but having a globular or thimble-shaped inflorescence. Sterile specimens of A. sphaerocephala are usually dis- tinguishable from A. cornigeya because the pinnules of A. sphaerocephala usually lack readily visible secondary venation (Rudd, 1964). As can be seen from its distribution (Figure Sl), A. sphaerocephala is probably a dry-land derivative of the widespread A. cornigera just as A. mayana is probably a wet- forest derivative of A. cornigera; the direction of this evolution cannot, however, be stated with certainty. Schlectendal and Chamisso?s type specimen of A. sphaerocephala has elongate internodes and mod- erate-sized thorns, indicating that it was collected a few meters inland from the beach dunes where most of the population of A. sphaerocephala now grows (site 1). While its spherical inflorescences are immature, they are representative of those found throughout the population of A. sphaerocephala except in the northern part of its range where some plants have more thimble-shaped inflores- cences. Rudd (1 964) has placed A. ueracruzensis and A. dolichocephala in synonymy with A. sphaero- cephala and I agree. The small maroon seed pods growing from a globular inflorescence, and the very thick thorns on the type of A. ueracruzensis identify it clearly as having been collected from the dense mat of A. sphaerocephala growing on the dunes immediately above the high-tide line south of the city of Veracruz (sites 2 and 6). At this site, the thorns of A. sphaerocephala may attain a diameter of 15 mm (e.g., Figure 65c) while equal-length thorns of A. cornigera are only 8 to 10 mm diameter at the base. Also, A. sphaerocephala thorns at this site (as at most sites) are bicolored with ivory to beige tips while those of A. cornigera are uniformly brown or gray. Safford?s type of A. dolichocephala, like that of A. sphaerocephala, was taken from the population just inland from the dunes and thus has as long internodes and thin thorns as are to be found on A. sphaerocephala. Its immature inflores- cences are slightly elongate but those of A. cornigera at the same site are 3 to 4 times as long. Acacia globulifera has been commonly misidenti- fied as A. sphaerocephala where the former species grows along the dunes sympatric with A. collinsii on the Yucatan Peninsula. Careful search has, how- 74 SMITHSONIAN CONTRIBUTIONS TO BOTANY 1000 R W 8. - I96 7 1 I I # FIGURE 61.4eneral postulated distribution (stippled area) and collection localities of A. sphaerocephala. ever, located no A. sphaerocephala south of Alva- 3. Type of A. dolichocephala: Along shore north of Vera- cruz, Veracruz (sea level). 1-24-1906, J. M. Greenman, rado, Veracruz, Mexico. _- KEY COLLECTION LOCALITIES (ALL IN MEXICO) no. 87. XI-10-1963, DHJ, nos. 1921, 1924 (common), mp. 4. 5.1 to 38.1 mi SE Jalapa, Veracruz, along Mexico 140. 5. 23.5 to 42.0 mi WNW Huatusco, Vcracruz, road from 1. 2. Type of A. sphaerocephala: Actopan, Veracruz (sea level). March, 1820, Schiede and Deppe, no. 684. Type of A. veracruzensis: sand dunes south of Veracruz, Veracruz (sea level). X-13-1908, Schenck, no. 916. Orizaba to Hgy. 140. XI-10-1963, DHJ, no. 1920 (com- mon) , mp. 6. Within 1-2 km of the ocean beach, from the southern city limits of Veracruz, south to Anton Lizard0 (approxi- NUMBER 13 75 mately 15 miles north of Alvarado) , Veracruz. DHJ, nos. 1922, 1925, .l926, 1927. 7. 26.2 mi N Ciudad Valles to Antiguo Morelos on Mexico 85 (100-300 m). VIII-25-1967, DHJ, nc. 8. Antiguo Morelos to where the Tropic of Cancer crosses Mexico 85 (e.g., 1.8 mi S Llera on Hgy. 85; 22 mi s Ciudad Victoria, Tamps.). VI-10-1964, DHJ, no. 505. 9. Ciudad Mante to Tampico (Mexico 80) (primarily in liver beds) and Tampico to the Tamuin area (e.g., 8 mi E Tampico on 110). 1-2-1964, DHJ, no. 504, Tamps. and SLP. 10. Antiguo Morelos to Ciudad del Maiz, Tamps. and S.L.P. (occasional populations on dry southeast facing slopes) . 11. 10.3 mi NE Ciudad Valles on Mexico 110, S.L.P. I-II- 1964, DHJ, no. 1923. 12. Tanquian, Partido de las Valles, Tamcanhuitz Dist., S.L.P. 1908, L.G. Cuevas, (US 692167), fl. 13. Hda. Buena Vista, Tamaulipas, Mexico. VI-15-1919, E.O. Wooton; (US 989707). 14. Between Tdla and Tampico. XII-1830, no. 728 (US 40123), fl. 15. Tampico, Tamaulipas (sea level). 1-1-31-1910, E. Palmer, no. 133 (US 764678), fl., mp. 16. Between San Luis Potosi and Tampico (Huasteca Re- gion), 1878, E. Palmer (US 764678), fl., mp. 17. Pueblo Viejo, 2 km S Tampico, Veracruz (sea level). V-23-31-1910, E. Palmer, no. 448a (US 463371), fl. 18. Between Rio de Santa Maria and Rio de 10s Pescados, Veracruz. X-7-1908, Schenck, no. 837 (US 764677). 19. 1 mi N Cd. Madero, Tamaulipas (sea level). 111-1-1961, R.M. King, no. 3990 (US 2365430), fl., mp. XII-28-1963. ~~ATURAL HISTORY The distribution of Acacia sphaerocephala in- variably borders the dry edge of the A. cornigera distributions. At sites 2 and 6, it occupies the dunes and only extends inland to the wetter sites occupied by A. cornigera as scattered individuals. At sites 4 and 5, A. sphaerocephala extends the distribution of swollen-thorn acacias into that part of the state of Veracruz with the most severe dry season. All along the northern part of the range of A. sphaero- cephala, the northernmost individuals are found in semidesert vegetation while the southernmost indi- viduals are scattered among the northernmost individuals of the A. cornigera population. A. sphaerocephala is the only species of swollen-thorn acacia that inhabits newly formed dunes, and it occupies some of the driest habitats of any swollen- thorn acacia population. It is the only Central American acacia involved in the early stages of beach succession. In the Isthmus of Tehuantepec on the Pacific side of Mexico, A. collinsii is also found on occasion in extremely dry habitats, but it is not clear if the low density population is reproducing itself or maintained by seed immigra- tion from wetter sites. For purposes of discussion, the A. sphaerocephala population can be divided into two ecotypes: the dune ecotype and the inland ecotype. The dune form occurs only on new dunes in the vicinity of the city of Veracruz (sites 3 to 6), while the inland form occurs throughout the remainder of the range and does not approach the ocean as closely as does the dune form, When seeds of the dune form are grown at well-watered inland sites, the resulting shrub closely resembles the inland ecotype except for having slightly shorter internodes and slightly thicker swollen thorns. On the dune, however, it is a very different plant and worthy of special dis- cussion. On the dunes closest to the ocean, the dune form of A. sphaerocephala forms a dense pure stand from 0.2 to 1.8 m tall that generally crowns the leading edge of the dune (Figure 62~; Figure 10 in Sauer, 1967). The origin of the mat of A. sphaerocephala is seed, but most of the plants in the larger mats are from sucker shoots from roots. The branches are so intertwined that the entire mat appears as though it were one crown. The largest mats may be 30 m wide by 200 m long and contain at least 6,000 major stems. The dune ecotype of A. sphaerocephala differs from other swollen-thorn acacias in not being oc- cupied by obligate acacia-ants when it grows on new dunes. Its success as an unoccupied swollen- thorn acacia in this habitat is apparently due to the continuous strong onshore winds, which probably make life very difficult for phytophagous insects and prevent their flying to the acacias in the first place. The absence of other adjacent shrubby vege- tation in this barren habitat also removes the source for most of the insects that defoliate unoccupied swollen-thorn acacias. Plants grown at inland sites from dune ecotype seeds are heavily defoliated if unoccupied, just as are adjacent unoccupied A. cornigera (experiments conducted 1963-1964, at Cotaxtla Exp. Sta. and Temascal, cf. Janzen, 1967b, for site details). Inland ecotypes of A. sphaero- cephala interact with ants in the same manner as do A. cornigera. Failure of Pseudomyrmex ferruginea (the only obligate acacia-ant at the site) to occupy the dune 76 SMITHSONIAN CONTRIBUTIONS TO BOTANY ecotype is due to the following reasons: (1) Search- ing queens appear to rarely arrive at the dune plants, probably because the prevailing winds blow from the dunes toward the inland swollen-thorn acacias (A. cornigera and A. sphaerocephala), and because they would have to cross many feet of very hot sand to walk there. Their absence is indicated by the almost complete lack of their characteristic entrance holes in thorns of seedlings and adult plants of the dune ecotype of A. sphaerocephala. (2) The dune ecotype produces almost no Beltian bodies and the few that are produced appear to be harvested by colonies of Crematogaster atrata Mayr or C. coruina Mayr that tenant the thorns in large colonies. These ants cut entrance holes at the base of the thorns, tend mealybugs (Pseudococcidae) and aphids (Aphididae) inside the thorns, and scavenge on the acacia surface. Their foraging is almost entirely nocturnal and very patchy in cov- erage. There is no evidence that they protect the acacia; inland A. sphaerocephala dies as fast from herbivore damage when tenanted by this ant as it does when unoccupied. (3) The Crematogaster ants thoroughly harvest the petiolar nectar produced by the dune ecotype of A. sphaerocephala and would probably be very severe competitors for the young FIGURE 62.-~, pasture dotted with young 1-2-m-tall A. sphaerocephala; the columnar structure is typical of pasture plants in this area (Mexico, site 9); B, beach dune created by dense mat of A. sphaerocephala (Mexico, site 6). queens attempting to start a colony. The total absence of Crematogaster, however, from those in- land acacias that are producing Beltian bodies and can be found by many searching queens of P. ferruginea, indicates that obligate acacia-ants can easily exclude Crematogaster in more ordinary habitats. (4) During the latter part of the dry sea- son most of the dune ecotype individuals of A. sphaerocephala drop their leaves (e.g., Figure 63c), and this behavior makes colony establishment and maintenance very difficult. As dune succession proceeds, the dense mats of A. sphaerocephala are broken up, and the popula- tion consists primarily of very flat-topped scattered individuals that rarely exceed 1.5 m in height (Fig ure 63~). Once the succession has reached a height of 3 to 5 m, A. sphaerocephala disappears com- pletely, though A. cornigera may still be common. Birds apparently disperse seeds continually into these inland sites, however, because if they are cleared, both A. cornigera and the inland form of A. sphaerocephala appear in the early successional stages. In the inland habitat, these two acacias are very difficult to tell apart until they bear flowers. Seedling development in situ of the dune ecotype has not been examined closely, but leaflet develop- NUMBER 15 77 FIGURE 63.-Acacia sphaerocephala (Mexico): A, adult in cattle pasture (site 4); B, adult 1 m tall, leafless, in low vegetation several hun- dred meters behind the high-tide line (site 6); c, leafless subadult on beach dune (site 6). inent of a representative seedling from an inland plant is depicted in Figure 64. The gradual increase in leaf size and leaflet number with age is repre- sentative of dry-land, swollen-thorn acacias. It is characteristic of all the swollen-thorn acacias to have this short period of growth before developing the characteristic traits of a swollen-thorn acacia. In short, the seed probably lacks sufficient reserves to produce both a small amount of shoot and root, and swollen thorns, nectaries, and Beltian bodies. That these traits increase gradually in quantity indicates that the seedling continues to divide its new photosynthate between conventional vegetative growth and production of resources for the develop- ing ant colony. No matter in what habitat it is growing, a young 78 SMITHSONIAN CONTRIBUTIONS TO BOTANY 525+ .O4?& 28 5\ 26 % 16 i, /& 54 # 30 A34 FIGURE 64.--Schematic representation of 46-day-old A. sphaerocephala seedling: the number of pinnules for each pinna are recorded on the figure and .the first petiolar nec- tary appeared on the 7th leaf with the first swollen thorn and Beltian bodies at the 11th node. A fully mature leaf on the 50th or later node may have as many as 20 pinnae with 50-70 pinnules on each (Mexico, site 6). FIGURE 65.-Acacia sphaerocephala (Mexico): A, sexual branchlet with type B thorns (site 9); B, type B thorns (two very large thorns) and type A thorns from the plants found on beach dunes (site 6): c, fruits and type A thorns (site 6). NUMBER 13 79 100 90 80 h I = L T c K w - g 70 n f 60 t9 L w -1 cl 4 s c. 50 . . 5. . .... .. ... . b . i *. .. 0. . 0 . 0 . . . . . 0. . . . e. I I I I I I I I 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 TOTAL VOLUME PER THORN (CC) FIGURE 66.-Distribution of 30 cc of thorn volume among the 59 thorns on a 50-cm-tall A. sphae. rocephala from the mat of shrubs on beach dunes (Mexico, site 6). A. sphaerocephala starts out with a single major trunk (e.g., Figure ~ZA), as with A. cornigera. These plants, however, are generally in dry and, therefore, relatively well-insolated sites and thus rapidly spread out laterally (Figure 63~). A. cornigera in the same sites tends to grow taller and more slender. Within 2 to 4 years after first becoming occupied by obligate acacia-ants, the young A. sphaero- cephala bears flowers. It is generally one year ahead of A. cornigera growing at the same site. After flowering, A. sphaerocephala rarely increases in height much in later years. The pods (Figure 65c) are produced in dense clusters, again as with A. cornigera, but generally have very thin walls- easily crushed between one's fingers-in contrast to the stiff and tough walls of A. cornigera at the same sites. A. cornigera has a morph with thin pod walls, but it is not sympatric with A. sphaerocephala. In an effort to document the thorn volume of A. sphaerocephala, and its relationship to external thorn dimensions, the swollen-thorns of three dif- ferent plants were measured. The young seedling (Figure 66) has the very small thorns (as compared with those of the adult in Figures 67 and 68) char- acteristic of young plants; in general, the larger the thorn, the higher it was up on the plant. For the inland ecotype (Figures 65~, 69~, B), the maximum thorn volume is 3.5 to 4.5 cc and probably repre- sents the maximum possible inside thorn diameter for optimal ant brood care. The great increase in thorn volume found in the dune ecotype (Figures 65~, 69c) would be negatively adaptive for plants occupied by obligate acacia-ants, since it is pro- duced by inflating the thorn rather than increasing its length. It is very difficult for the ants to organize and care for brood in thorns where the thorn walls are more than $bout 4 to 6 mm apart. The ants of Crematogaster, however, that normally tenant these thorns are accustomed to living in large cavities in plants and build carton partitions in the thorns to keep the brood separate. One may envision this as the first possible step in the development of an 80 SMITHSONIAN CONTRIBUTIONS TO BOTANY .. I .. .? . * ..* .: .. . .. *. .. $;;:;:*? . I!: ..:i!:: :l:il . . a:!!! FIGURE 67.-Relationship between thorn length, volume, and width for the 244 .* *. : . thorns on a 2-m-tall adult A. sphaerocephala from the beach dune mat (Mexico, .. .I.. site 6): A, total thorn volume as a function of the total thorn length (sum of the :* two halves); B, total thorn volume as a function of the sum of the two halves of the thorn; c, relationship of total thorn length to the sum of the maximum width .?I:;. * ... I 5 10 15 20 25 ,b of the two halves of the thorn. c SUM 0? UAXIIJII LIDT? OF THE TWO THORN HALVES (mmj interaction in the New World as has evolved be- tween the African swollen-thorn acacias and Acacia turgida Safford in Wheeler, 142:plate 45. Acacia campecheana Schenck, 1913:361. Crematogaster. 8. Acacia cornigera (L.) Willdenow Acacia cornigera (L.) Willdenow, 1806: 1080. Mimosa cornigera L., 1753:520. Acacia cornigera var. americana Dc., 1825:460. Acacia spadicigera Schlechtendal and Chamisso, 1830:594. Acacia cubensis Schenck, 1913:360. Acacia nicoyensis Schenck, 1913:360. Acacia hernandezii Safford, 1914:358. Acacia furcella Safford, 1914:359. Tauroceras spadicigerum (Schlechtendal and Chamisso) Brit- Tauroceras cornigerum (L.) Britton and Rose, 1928:86 [ex- Acacia interjecta Schenck, 1913:361. Acacia rossiana Schenck, 1913:361. ton and Rose, 1928:85. cluding synonymy]. Acacia cornigera is the best known of the swollen- thorn acacias. It is readily distinguished from all other swollen-thorn acacias by the unique combi- nation of elongated ?canoe-shaped? petiolar nec- taries (Figure 8~), very elongated infloresc, ences (Figure 12~), thorns usually round in cross section, and completely indehiscent seed pods. Linnaeus? type specimen of A. cornigera is a sterile branch representative of the eastern Mexican portion of the total population; it was collected from a cultivated plant in Holland, presumably grown from Mexican seed (Rudd, 1964). Rudd has compared pinnules of the type with those of A. sphaerocephala and found that the pinnules of the type have the readily visible secondary venation lacking on A. sphaerocephala NUMBER 13 B r 20- E I i5. b : 81 .... -. .... ................ .......... ....... . ........ ..... . ...a ....... _..I ..... .. a. .. 15C 12: p ii C .. .... ...... ....... ............. ... .. I 1234567BD1011I2 L 2 3 4 5 6 7 0 9 10 11 TOTAL THORN VOLUME Iccl B TOTAL TLiORN YOLJME lccl *. -. -2. : .* *. FIGURE 68.-Relationship between thorn length, volume, and width for the 200 thorns from half of a 1.5-m-tall adult occupied A. sphaerocephala growing like a representative swollen-thorn acacia 200 m inland from the beach (contrast these data with those in Figure 67) (Mexico, site 6): A, total thorn volume as a function of the total thorn length (sum of the two halves); B, total thorn volume as a func- tion of the sum of the maximum width of the two halves of the thorn; c, relation- ship of total thorn length to the sum of the maximum width of the two halves of the thorn. *: .:.a. * *:. !i;.:* ... 'i,ili:i. i.1 ... ::!i!.i . . . 11: .*. IS 20 25 10 35 SUM OP MPXIMUM WIDTH OF THE MI0 THORN MLYES lmml FIGURE 69.-Acacia sphaerocephala (Mexico): A, black type A thorns on lateral branch (site 5); B, bicolor type A thorns from a lateral branch of an inland adult (site 6); c, bicolor type A slightly inflated thorns of plants growing in the dune mat (site 6). 82 SMITHSONIAN CONTRIBUTIONS TO BOTANY leaflets. Owing to the peculiar thorns of A. mayana, there is no doubt that the type does not belong to that species. The type specimen of A. spadicigera (site 1) is easily identified as A. cornigera because of its elongate inflorescences and thorns round in cross section. Its straight short thorns are representative of A. cornigera from lowland wet sites in Veracruz. Safford?s type specimen of A. hernandezii is un- mistakably from the northern wet lowlands portion of the A. cornigera population. A. cornigera is the only species of swollen-thorn acacia in the Rascon area (site 2). Safford?s leafless type specimen of A. furcella is representative of the A. cornigera popula- tion in the Lake Catemaco region (site 5). The type has the very dark brown thorns and widely spread type B thorns characteristic of highslevation, east- ern Mexican A. cornigera. The inflorescences of the types of both A. hernandezii and A. fnircella are representative of A. cornigera inflorescences. Schenck?s type of A. rossiana (site 4) is representa- tive of that portion of the A. cornigera population inland from Minatitlan, Veracruz, Mexico. I have been unable to locate his type specimen for A. campecheana; possibly it was destroyed at Berlin. He states that its leaves are like those of A. spadicigera and A. rossiana, and it was therefore most likely collected from that part of the A. cornigera population that extends along the coastal lowlands from Veracruz to Yucatan. Its name sug- gests that it was collected in Campeche, and there appear to be no swollen-thorn acacias in that area that could be confused with ?4. cornigera. Schenck described A. interjecta from material growing in the Singapore and Kew botanic gardens; the de- scription agrees with that of A. cornigera and the seed probably came from Mexico, since that appears to have been the source of most introduced A. cornigera seed. Schenck described A. nicoyensis from material representative of the southern end of the geographic range of A. cornigera (site 77). His type specimen has the white and slightly flat- tened type A thorn found on Costa Rican A. cornigera growing in very dry and insolated sites. The inflorescences and seed pods of the type of A. nicoyensis are unmistakably A. cornigera. Schenck?s type specimen of A. cubensis was taken from a cultivated plant (site 82); the white and partially recurved type B thorns indicate that the seed came from the A. cornigera population some- where north of Guatemala. A. cornigera apparently occurs only in gardens .in Cuba and its presence unlikely is due to natural introduction by birds or other means. The absence of ant entrance holes in the thorns of the type specimen of A. cubensis indi- cates that the ants have not yet been introduced to Cuba. Since A. turgida is supported only by a photo- graph (Wheeler, 1942), it probably represents a name that Safford intended to propose. I cannot find the locality (site 5) for the specimen, but it appears to have been collected from that rather distinctive portion of the A. cornigera population growing where the Pan-American highway crosses the Guatemala-Mexico border. KEY COLLECTION LOCALITIES * Mexico: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Type of A. spadicigera: La Laguna Verde, Veracruz, Mexico. March 1820, Schiede, no, 685. Type of A. hernandezii: Rascon, S.L.P., Mexico. VII-19- 22-1905, E. Palmer, no. 669. Type of A. furcella: Catamaco, Veracruz, Mexico (300 m). IV-26-1894, E.W. Nelson, no. 427. Type of A. rossiana: Santa Lucrezia, Isthmus of Tehu- antepec, Veracruz, Mexico, X-8-1906, H. Ross, no. 918. ?Type? of A. turgida: Herba Santa, Chiapas, Mexico (see plate 45 in Wheeler, Bull. Mus. Comp. Zool. Har- vard, 90: 1-262). Guatemala-Mexico border (near Ciudad Cuahtemoc) to 19.0 mi N on Hgy. 190, Chiapas (720-600-780 m) . VIII- 7-1966, DHJ, no. 499; VIII-16-1967, DHJ, nc (common to rare). 2.6 mi NE to 0.9 mi W Chiapa de Corzo on Hgy. 190, Chiapas (580-475 m) . VIII-17-1967, DHJ, nc (common to rare). 9.7 mi W Tuxtla Gutierrez to 8 mi NE Cintalapa on Hgy. Chiapas 190 (860-1090-570 m) . VIII-18-1967, DHJ, nc (common to rare to occasional). 13.5 to 9.0 mi E of intersection of Hgy. 190 and 185 (La Ventosa) , Oaxaca (80 m) . VIII-18-1967, DHJ, nc (com mon to rare). 9.9 to 23.4 mi N of intersection of Hgy. 190 and 185 (La Ventosa) , Oaxaca (210-290-250 m). VIII-18-1967, DHJ, nc (common to occasional). 33.3 mi N of intersection of Hgy. 190 and 185 (La Ven- tosa) to Acayucan, Oaxaca-Veracruz (23040 m) . VIII- 18-1967, DHJ, nc (common). * Hundreds of collection records of A. cornigera in the U.S. National Herbarium and elsewhere are included within the inclusivc localities given here. For the sake of brevity, only type localities or collections made in sites not visited in this study are cited. NUMBER 13 83 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. Near Salina Cruz, Oaxaca (sea level). X-3-1965, DHJ, nc (occasional). Coatzacoalcos to Veracruz on Hgy. 180, Veracruz (0-350- 0 m) . 111-21-1964, DHJ, no. 1935; VI-15-1966, DHJ, nc (occasional to common). 8.3 mi SE bridge over Rio Coatzacoalcos, Veracruz, to Cardenas, Tabasco, on Hgy. 180 (0-40 m) . VI-15-1966, DHJ, nc (very rare, about 1-2 plants per 2 miles). Cardenas to 37.1 mi S, Tabasco (40-100 m) . VI-17-1966, DHJ, nc (rare to common). Cardenas to Paraiso, Tabasco (40-10 m) . VI-17-1966, DHJ, nc (rare). Cardenas to Ciudad del Carmen, Tabasco-Campeche (40-0 m) , VI-22-1966, DHJ, no. 494 (rare, and only on well drained sites), fl. Campeche to Ruinas Edzna (29 mi E on Hgy. 180), Cam- peche (0-50 m) . VI-22-1966, DHJ, nc (very rare to oc- casional). Hopelchen to Ticul on Hgy. 180, Campeche-Yucatan 50-80). VI-28-1966, DHJ, nc (very rare to occasional). 12.2 mi S Pet0 (Santa Rosa) on Hgy. to Cd. Chetumal, Yucatan-Quintana Roo (50-80 m). VI-29-1966, DHJ, nc (common), fl. Cd. Chetumal to 69.6 mi NW on Hgy. to Peto, Quintana Roo (0-20 m) . VI-30-1966, DHJ, nc (common), fl. 30 mi S to Campeche on Hgy. 180, Campeche (0-80-0 m). VI-25-1966, DHJ, no. 491; VII-8-1966, DHJ, nc (rare to common). 3.2 mi E Cruz Grande to 19.1 mi NW Puerto Escondido on Hgy. 200, Guerrero-Oaxaca (50-250-70 m). VI-18- 1967, DHJ, nc (rare to common to occasional) . 9.3 mi N Playa Pie de la Cuesta to 16.5 mi SE Papanoa on Hgy. 200, Guerrero (0-20 m). VI-17-1967, DHJ, nc (very common in swamps, absent on dry hills). Alvarado to Tlacotalpan to Cd. Aleman to Las Tinajas to Paso del Toro, Veracruz (0-100-0 m). Throughout the year, DHJ, no. 1939 (common), flower February-May. Tierra Blanca to La Granja to Temascal, Veracruz- Oaxaca (20-70 m). Throughout the year, DHJ, nos. 1966, 1967, 1968, 1969, 1970, 1971, 1972, 1973, 1974, 1975, 1976, 1977, 1978, 1979, 1980, 1981 (very common). Shores of the lake behind Presa Miguel Aleman, Temas- cal, Oaxaca (70 m). X-20-1963, DHJ, no. 1937; XI-3- 1963, DHJ, nc (occasional to very common). Cd. Aleman to Loma Bonita to Rio Playa Vicente, Vera- cruz-Oaxaca-Veracruz (50-100 m) . XI-4-1963, DHJ, no. 1938 (common to very rare on red clay hills), Cd. Aleman, Veracruz to Valle Nacional, Oaxaca, on Hgy. to Oaxaca, XI-6-1963, DHJ, nc (common to occasional). 22.8 to 27.5 mi SW Conejos (turn off of Hgy. 140 to Huatusco) on Hgy. to Huatusco to Cordoba, Veracruz. XI-10-1963, DHJ, nc (common to rare). 4.3 mi E Cordoba, Veracruz on Hgy. 150, Veracruz (800-0 m) . X-16-1963, DHJ, nc (rare to very common). Hgy. 150 SE to San Ignacio de la Llave, Veracruz (60-10 m) . X-16-1963, DHJ, nc (common). Tampico to Tuxpam on Hgy. 105, Veracruz (0-100-0 m) . 1-2-1964, DHJ, nos. 1942, 1943; VI-13-1966, DHJ, nc (common to rare). Tuxpan to 1.2 mi S Villa A. Camacho on Hgy. 130, Vera- 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. cruz. 1-2-1964, DHJ, no. 1940; VI-13-1966, DHJ, nc (oc- casional to rare). Poza Rica to Nautla on Hgy. 180, Veracruz (50-0 m), 1-20-1964, DHJ, nc (occasional to rare). 6 mi SW Tamazunchale to Cd. Valles on Hgy. 85, San Luis Potosi (500-100 m) . 1-1-1964, DHJ, no. 1936; XII- 31-1963, DHJ, nc (rare to common). 36.6 mi E Rio Verde to Cd. Valles on Hgy. 86, San Luis Potosi (1140-100 m). VIII-24-1967, DHJ, nc (rare to common). Cd. Valles to 6.4 mi N Antiguo Morelos on Hgy. 85, San Luis Potosi-Tamaulipas (100-250 m) . VIII-24-1967, DHJ, nc (rare). Antiguo Morelos to 20 mi R?W Cd. del Maiz, Tamaulipas- San Luis Potosi (250-? m). XII-31-1963, DHJ, nc (com- mon on wetter hillsides) . Cd. Valles to Tampico on Hgy. 70, San Luis Potosi- Veracruz (100-10 m). VI-11-1966, DHJ, nos. 496, 497; VIII-24-1967, DHJ, nc (common to occasional). Pt. Talisman (Guatemala-Mexico border) to 3 mi N Arriaga on Hgy. 200, Chiapas (50-150 m). 1-11-1965, DHJ, no. 1047 (common). Presa Benito Juarez, near Jalapa de Marquez on Hgy. 190, Oaxaca (200 m). IX-7-1968, DHJ, nc (1 plant). Pichucalco, Chiapas. 1-1907, GN. Collins and C.B. Doyle, no. 260 (US 692169) mp. 3 mi S Pinola Las Rosas along road to Soyatitan, Venus- tiano Carranza (W of Comitan), Chiapas (1400 m). VII- 27-1965, D.E. Breedlove, no. 11367 (US 2484541), fl. Zacualpan, Veracruz (1000 m) . VII-1917, C.A. Purpus, no. 7748 (US 884536), fl., and IX-1908, H. Schenck, no. 836 (US 764665). Fortuno, Rio Coatzacoalcos. 111-1937, L. Williams (F 897476), fl. British Honduras: 47. Sta. Elena to Belize on main Hgy. (5-0 m) , VI-30-1966, DHJ, nc (rare to occasional), fl. 48. El Cayo, El Cay0 Prov. V-5-1931, H.H. Bartlett, no. 13007 (US 1493273), mp. Guatemala: 49. 50. 51. 52. 53. 54. 55. Escuintla to Taxisco along Hgy. CA-7, Depto. Escuintla- Santa Rosa (330-100 m). 1-13-1965, DHJ, nc (occa- sional) , Escuintla, Drpto. Escuintla, to Puente Talisman, Depto. San Marcos (Guatamela-Mexico border) on Hgy. CA-7 (330-50 m). VII-16-1966, DHJ, no. 466; 1-11-1965, DHJ, nos. 1045, 1046 (common to occasional), fl. Retalhuleu SW to Champerico, Depto. Retalhuleu (200- 0 m) . VIII-1-1966, DHJ, nc (occasional). Retaljuleu to 14.0 mi NE on Hgy. S, Depto. Retalhuleu (200-900 m). VII-28-1966, DHJ, no. 464 (common to rare). Escuintla to 7.4 mi NE on Hgy. 3, Depto. Escuintla (330- 880 m) . VIII-13-1967, DHJ, nc (common to occasional). 0.3 mi SW to 4.1 mi NE Los Amates on Hgy. CA-9 (150- 110 m). VIII-9-1967, DHJ, nc (common) . Finca San Joaquin, Slopes above Rio Negro, near San SMITHSONIAN CONTRIBUTIONS TO BOTANY 84 56. 57. 58. 59. Cristobal de Verapaz, Depto. Alta Verapaz (1420-1080 m) . VIII-14-1967, DHJ, nc (rare to common). 14.9 mi SE to Guatemala-Mexico border (near Ciudad Cuauhtemoc) , Depto. Huehuetenango (900-1000-720 m) . VIII-16-1967, DHJ, nc (common). Patalul, Depto. Solola. 11-14-1906, WA. Kellerman, no. 5915 (US 578841), fl. Finca Sepacuite, Depto. Alto Verapaz. IV-1905, O.F. Cook and R.F. Griggs, no. 8 (US 590344), mp. 1-2 mi N Ocos, Depto. San Marcos (sea level). III-15- 1940. ].A. Steyermark, no. 37867 (F 1035247). Honduras: 60. 12.5 to 19.5 mi NW Siguatepec on Tegucigalpa-Puerto Cortes Hgy., Depto. Comayagua (910-540 m). VII-31- 1967, DHJ, nc (comnion to very rare). 61. 0.7 mi N Rio Lindo to 18.6 mi S Puerto Cortes on Tegucigalpa-Puerto Cortes Hgy., Depto. Cortes (90-10 m) , VIII-2-1967, DHJ, nc (common to occasional). 62. San Pedro Sula to 8.1 mi SW on Hgy. to Santa Rosa, Depto. Cortes (90-130 m) . VIII-2-1967, DHJ, nc (occa- sional along streams). 63. 20.6 mi NE to 28.4 mi SW Sula on San Pedro Sula-Santa Rosa Hgy., Depto. Santa Barbara (260-780 m). VIII-3- 1967, DHJ, nc (common to rare), fl. 64. Vegas del Rio Agua, 3 km from Yoro, Depto. Yoro (1000 m), V-8-1956, A. Molina, no. 6807 (US 2412019), fl. El Salvador: 65. From 11.2 to 26.3 mi N (Depto. Cuscatlan) San Salvador on San Salvador-Nueva Ocotopeque Hgy. (420-590 m) . VIII41967, DH], nc (common to rare). 66. 6 mi E San Salvador on Interamerican Hgy., Depto. San Salvador (650 m) . 1-14-1965, DHJ, nc (occasional), 8. 67. 29.7 to 31.6 mi E La Libertad on La Libertad-La Union Hgy., Depto. La Paz (50-80 m). VIII-5-1967, DHJ, nc (rare). 68. Intersection of Hgy. from La Herradura with La Liber- tad-La Union Hgy., Depto. La Paz (20 m) . VIII-5-1967, DHJ, nc (1 plant). fig. 15.2 to 10.2 mi SE Santa Ana on Interamerican Hgy., Depto. Santa Ana (440-700 m). VIII-6-1967, DHJ, nc (common). 70. San Vicente, Depto. San Vicente (350-500 m), III-2- 1922, P.C. Standley, no. 21687 (US 137448, US 1137096), fl. 71. Ateos, Depto. La Libertad. IV-17-1922, P.C. Standley, no. 23360 (US 1139047), fl. 72. San Martin to Laguna de Ilopango, Depto. San Salvador. IV-1-1922, P.C. Standley, no. 22580 (US 1138317), fl. Nicaragua: 73. Penas Blancas (Depto. Rivas) to 7.2 mi NW on Hgy. 2 (50 m). VII-25-1967, DHJ, nc (common along rivers). 74. 3.4 mi SE Leon on Hgy. 12, Depto. Leon (20 m). VI-27- 1967, DHJ, nc (30 acre patch), fl. 75. 1.2 mi W Matagalpa to 3.6 mi N on Hgy. 3, Depto. Mata- galpa (650-880 m) . VII-28-1967, DHJ, nc (common), 8. 76. 8 mi S Esteli on Hgy. 1 (940 m) . VII-28-1967, DHJ, nc (1 plant). Costa Rica: 77. Type of A. nicoyensis: Shores of the Gulf of Nicoya, Guanacaste, Costa Rica (sea level). February 1900, Ton- duz, no. 13538. 78. 20 mi NW turnoff of Interamerican Hgy. to Puntarenas, to Pefias Blancas (border of Costa Rica and Nicaragua) on Interamerican Hgy., Guanacaste Prov. (10-100 m) . January-September, periodically in 1963-1967, DH], nc (rare to common, patchy distribution and associated with wet spots), fl., February-April. 79. Liberia to Playa COCO along Hgy. 21, Guanacaste Prov. (100-0 m). January-September, periodically in 1963-1967, DH], nc (rare to very common, primarily in wet spots), fl. and mp., in February. 80. Hda. Palo Verde, COMELCO, Bagaces, Guanacaste Prov. (sea level). IV-2-1966, DHJ, nc (occasional), 81. Filadelphia, Guanacaste Prov. 111-2-1946, Echeuerria, no. 279 (F 1442853), mp. Cuba: 82. Type of A. cubensis: North coast of Cuba. IV-21-1963, C. Wright, no. 2402. NATURAL HISTORY Acacia cornigeru (Figure 70) and A. collinsii have the greatest geographic and ecological range of all the swollen-thorn acacias. The riparian and swamp habitats that were occupied by A. cornigeru previous to human interference are in lowland areas that are ideal for slash-burn milpa agriculture. A. cornigera has probably been common in lowland agriculture fallow fields as long as Indians have lived in the Central American lowlands. Since oc- cupied A. cornigeru may reach reproductive ma- turity in 3 to 4 years, and since birds, people, and cattle distribute its seeds widely in second growth vegetation, A. cornigera populations have expanded widely into pastures, roadsides, and short-fallow fields generated by contemporary agricultural prac- tices. At many locations, A. cornigera is found in all types of disturbance sites, from creek bottoms to hilltops. Only A. sphaerocephala and A. collinsii are on occasion found in drier habitats than A. cornigera. The natural history of A. cornigeru in eastern Mexico has been described in detail (Jan- Zen, 1967b), and the comments in the remainder of this section are designed to highlight the way it differs in other parts of its range from that portion of the population. Local short-distance dispersals of A. cornigera by NUMBER IS 85 humans make it difficult to understand its pre- Colombian range. For example, at site 42 in western Mexico, a single A. cornigera was found growing by a new hydroelectric dam. The closest native population of A. cornigera is several hundred miles away in the Isthmus of Tehuantepec and along the west coast in the Acapulco region (sites 23, 24) (also an introduction?). This plant probably arrived by either road-building machinery or through humans; farmers and Indians often eat the sweet pulp around the seeds and ingest many seeds simultane- ously (Janzen, 196713). The work crews on public construction in Mexico often come from several hundred miles away. If more than one plant was introduced in the area of the dam, it is likely that viable seeds will be produced. The bruchids that normally kill most A. cornigera seeds will be absent, at least at first, and this swollen-thorn acacia will likely spread down the watercourse away from the dam. It will probably not spread around the lake since the general area of the lake is extremely dry and water levels will fluctuate greatly. The single plant was occupied by Pseudomyrmex ferruginea. The queen probably came from the population of A. collinsii about 25 miles to the south or the population of A. hindsii 50 miles to the northwest. As a second example, in the central Yucatan Penin- sula, A. cornigera is only common around small settlements, cattle corrals, and old Indian ruins (sites 18 to 21). These small sites are also usually the wetter habitats in the region (the outlying areas are heavily occupied by A. collinsii), and it is im- possible to know if the presence of A. cornigeru is due to recent and frequent introduction by hu- mans and cattle, or by natural dispersal by birds. Deliberate long-distance introduction of A. cornigera may have some interesting repercussions in the future. Introduced to the southern tip of Florida as a garden plant, A. cornigera has oc- casionally escaped to form feral populations (e.g., Oneco and Chapman Field, US 1433083 and US 764685 in the U.S. National Herbarium). These plants are severely damaged by the frosts that oc- casionally damage orange orchards, but they do flower and set fertile seeds if there are several together. Obligate acacia-ants have not been intro- duced, but it would be interesting to examine the interaction of occupied swollen-thorn acacias with the subtropical communities of southern Florida. Reproductively isolated from the Mexican parent genotype, A. cornigera might even develop a frost- resistant population in Florida. A. cornigeru has also escaped on Martinique, Guadeloupe, and Cuba in the Caribbean (US 2229384, US 1714928, US 764657). Here, with a climate not unlike that of some parts of eastern lowland Mexico, the popula- 86 SMITHSONIAN CONTRIBUTIONS TO BOTANY tions (again without their ants) have a chance to expand into the native vegetation just as many other species of introduced trees have done on Caribbean islands. The central question here is whether the herbivore pressure in the island com- munities is strong enough to require occupation by obligate acacia-ants for the survival of a large breed- ing population of A. cornigera. All introductions to the Old World tropics appear to have been to botanical gardens, and I have no information on their escape into natural habitats. Acacia cornigera and A. collinsii occupy adjacent habitats over much of their range. Always A. corni- gera is in the wetter creek bottoms and swamp edges while A. collinsii is on the drier hillsides (e.g., Guanacaste Province, Costa Rica). In the wetter parts of the range, A. hindsii and A. cornigera often occur together (e.g., Pacific foothills of Guatemala) and here, A. cornigera is the drier hillside plant with A. hindsii along the watercourses. The distinc- tiveness of these separations by habitat was prob- ably once very clear. Current agricultural practices have rendered wetter habitats drier and more inso- lated by clearing the vegetation, with the conse- quence that the formerly separated swollen-thorn acacias now grow side by side. In the southern part of its range, the A. cornigera population is much more broken up into subpopu- lations than in eastern Mexico where it is found in all sorts of disturbance and climatic regimes from sea level to 1200 m, and from areas with a 7-month dry season to areas with almost no obvious dry season. In habitats where one expects to find A. cornigera in the northern part of the range, one instead encounters A. collinsii and A. hindsii. To determine the role of competitive exclusion in this system, we need much more information on the parasites and predators (specifically, bruchid beetles) that these species have in common, and the ability of each to grow and maintain the largest possible obligate acacia-ant colony in the shortest period of time. Why A. cornigera has stopped its southward niovement in Guanacaste Province in Costa Rica is very enigmatic. A. collinsii continues south to Colombia with a nearly continuous population that occupies many habitats similar to those occupied by A. cornigera further north. It may well be that it has not stopped at all, but rather that it is in the process of slowly moving southward. The only possible ecological explanation that comes readily to mind is that the Guanacaste plants of A. cornigera are specialists at surviving the severe dry season there and therefore cannot advance slowly through the wet-forest area immediately down the coast from Puntarenas. What is needed is a long- distance dispersal of more than one seed simultane- ously from Guanacaste to the dry Pacific coast of Panama. The dependency of A. cornigera on a colony of obligate acacia-ants for survival to reproductive maturity has been well documented for lowland eastern Mexican habitats (Janzen, 1967a). Judging from the badly damaged condition of naturally un- occupied A. cornigera throughout its range, this dependency is the general case. It is clear, how- ever, that the intensity of damage to unoccupied A. cornigera varies from one habitat to another, just as it did from one season to another in the lowlands of eastern Mexico. High-elevation sites subjected to wet, cool, and windy weather (e.g., site 75) produce the healthiest appearing unoccupied A. cornigera, but even here it is doubtful if an unoccupied tree can survive until reproductive maturity unless growing in a cluster of acacias in a large and well-grazed cattle pasture. In central highland Guatemala (e.g., site 53) and on the high- land Atlantic side of Honduras (site 60), A. cornigera is occasionally ?occupied? by a large colony of Camponotus planatus which appear to compete directly for thorns with the obligate acacia- ants colony. The C. planatus patrol the acacia quite effectively and the plants are healthy in appearance. How long an acacia can be occupied by C. planatus and survive is not clear. On some occasions, a colony of C. planatus and one of obligate acacia- ants occupy the same acacia, but usually in different branches. C. planatus occurs throughout much of the range of swollen-thorn acacias but in other areas does not appear to offer any protection to the acacia. It also lives in hollow twigs in many other species of tree. Acacia cornigera has gained the same ?freedom? from herbivorous insects gained by A. hindsii in areas of intense agriculture (e.g., site 74 in El Salvador); however, it is much rarer than A. hindsii in these sites and will probably become extinct in the near future. Seedlings are almost never en- countered. Without the ants to keep the vines off, the stiff and upright young shrub of A. cornigera NUMBER IS 87 FIGURE 7l.-Acacia cornigera (Mexico, site 26): A, unoccupied seedling from heavily shaded site; B, unoccupied seedling from strongly insolated site. makes an excellent vine standard leading to its death by shading. The young plants of A. hindsii are more shade tolerant and less rigid than A. cornigera of the same age. As with A. hindsii and A. collinsii, the seedlings of A. cornigera are scattered throughout the habitat occupied by the parents. As stressed earlier (Janzen, 1967b), they attain the height of 10 to 50 cm and die after 1 to 2 years if not occupied by a colony of obligate acacia-ants. When in heavily shaded sites, the seedlings tend to produce few swollen thorns (Figure 71~), while in fully insolated sites thorn production is much greater (Figure 71~). Once occupied, a variety of different shrub life forms may develop. In dry areas, about as much wood goes into the central trunk as into branches (e.g., Figure 72~), and the shrub ends up being very bushy in appearance (Figures 73~, B). Only very rarely does A. cornigera become a large tree to FIGURE 7P.-Acacia cornigera: A, crown of healthy adult after losing its leaves during the last part of the dry season (Nica- l5 in height; such are ragua, site 75); B, crown of healthy subadult growing in found in natural disturbance sites along rivers and marshy pasture (Honduras, site 61). 88 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 73.-Acacia cornigera: A, pasture with adult (Mexico, site 23); B, type B and type A thorns in center of crown of adult (Nicaragua, site 75); c, type B thorns on main trunk of adult (Mexico, site 24). may at one time have made up a much larger proportion of the reproducing population than at present . Over most of the distribution of A. cornigera, sexual branchlets and their products are rather uniform; however, wetness of the habitat does ap- pear to influence this. In the Pacific foothills of Guatemala (site 50) the sexual branchlets may be 20 to 50 an long (Figure 74~). In contrast to the situation at this wet site, most populations of A. cornigera produce sexual branchlets 1 to 10 cm long (Figure 74c). The inflorescence itself is not influ- enced by the amount of moisture at the site, except that at some of the driest sites, some of the shortest inflorescences were found (again suggesting that A. sphaerocephala is a dry-land offshoot of A. corni- gem). The number of florets per inflorescence is poorly correlated with the length of the inflores- cence (Figure 75) at any one site but over a variety of sites shows a weak correlation (Figure 76). NUMBER 13 89 FIGURE 74.-Acacia cornigera: A, exceptionally long sexual branchlets (Guatemala, site 50); 8, mature pods ripped open by birds (Guatemala, site 50); c, last year?s pods and this year?s inflorescences on short sexual branchlets in the crown of an adult (Nicaragua, site 75). The indehiscent pods of A. cornigera are split open by birds (Figure 75~) and the seeds dispersed by birds that frequent second-growth vegetation (Janzen, 1967b). Over most of the range of A. cornigera, the pod walls are thick and hard to break between ones fingers. They tend to have sharp and stiff long points on the end (Figures 74B, c); however, in riparian habitats at site 26, the pod walls are very thin and even split of their own accord when drying. The plants on the adjacent dry hills have the usual thick-walled pods. This differ- ence is genetic as shown by a lack of intermediates 90 0 SMITHSONIAN CONTRIBUTIONS TO BOTANY 0 0 LT Lu 1 3 z m I 26 28 30 35 311 3?6 38 4b LENGTH OF INFLORESCENCE (mm) FIGURE 75.-Distribution of 27,262 florets among 23 inflorescences from a single adult A. corni- gem; the inflorescences averaged about 35 mm in length (excluding peduncle) and 8.9 mm in width (Mexico, site 26). and occasional plants bearing the ?wrong? type of pods for the habitat where the plant is encountered. It is interesting that this site is only about 80 miles from that portion of the A. sphaerocephala popu- lation below the city of Veracruz that also has very thin pod walls. The number of seeds in A. cornigera pods is representative for dry-land, swollen-thorn acacias (Figure 77) but nearly all the seeds are killed by bruchid beetles. The very high predation on A. cornigera seeds by these beetles may be due to the high concentration of pod crops to be found in fields and roadsides; when the plants were widely spaced in natural disturbance sites, the bruchids might have a much more difficult time locating the pod crops. The thorns of A. cornigera display every color found in swollen-thorn acacias, but there is little intertree or intratree variation at any one site. In general the thorns in very hot and dry lowlands are white, ivory, or pale gray. At high elevations they are often very dark brown or nearly black. Though the correlation is not perfect, it is tempting to interpret these differences as being adaptive in respect to the ant colony. In lowland areas, tempera- tures in fully insolated thorns can be high enough to drive out the ants, and light-colored thorns may reflect radiation and slightly lower the temperature inside the thorn. It is perhaps significant in this connection that A. coZlinsii also has very light- colored thorns in dry and hot areas and dark thorns in high and wet sites. The dark thorns of A. cornigera could aid in warming up the ant colony on cool days and thereby increase the degree of patrolling by ants. It is significant in this respect that A. sphaerocephala has very dark thorns at the one high (cool) and dry site where it occurs, in strong contrast to the light-colored thorns it bears in dry lowland areas. All of the wet-forest swollen- NUMBER 13 91 In 25 20 15 VI 8 p 10 E5 I z a FIGURE 76-Distribution of 125,773 florets among 122 inflorescences from 16 adult A. cornigera from throughout the range of A. cornigera; there was no evidence of a geographical trend in size of inflorescence or number of florets per inflorescence. ""'nnn n 4 13 12 11 10 8 6 NUMBER OF SEEDS PER PO0 FIGURE 77.-Distribution of 1436 seeds among 100 pods from an adult A. cornigeru with a normally heavy seed crop of about 500 pods (Mexico, site 26). thorn acacias have very dark brown thorns, and this may be important in warming the thorns by occasional direct insolation. Bicolored thorns (Fig- ure 78c) are somewhat puzzling in this respect. They are most prominent in the small and insolated population of A. cornigera on the west coast of Mexico; here A. cornigera grows only in swamps, with A. collinsii on the extremely dry and hot ad- jacent hillsides. The spectacular "bull's horns" swollen-thorns of A. cornigera are extremely variable in shape, but in a consistent and often clinal manner. Many sites are characterized by a particular thorn type, and many of these thorn types resemble thorns found on other species of swollen-thorn acacia (though often not sympatrically). Type A thorns of A. cornigeru (Figures 78, 79) are usually round or nearly so in cross section and form a shallow to steep "V" with the branch wearing them. When flattened so that they superficially resemble A. hindsii thorns (Figure ~OD), they are still easily distinguished by having very thick and rough walls. Type B thorns (Figures 80, 81) are highly variable in form, but all are much larger in volume than type A thorns. There is no part of the range of A. cornigera where it is consistently occupied by multiple-queen species of obligate acacia-ants, so 92 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 78.-Geographic variation in type A thorns from occupied adult A. cornigera: A, Mexico, site 33; B, Mexico, site 40; c, Mexico, site 24; D, Guatemala, site 56. FIGURE 79.-Geographic variation in type A thorns from occupied adult A. cornigera: A, Mexico, site 33; B, Mexico, site 36; c, Mexico, site 33. NUMBER 13 FIGURE 80.-Geographic variation in type B thorns from occupied adult A. cornigera: A, Mexico, site 33; B, Guatemala, site 52; c, Mexico, site 84; D, Guatemala, site 56; E, Mexico, site 35. selection for large type B thorns is omnipresent. Type B thorns of A. cornigera are the hardest to break into of any of the swollen-thorn acacias. Several are often twisted together, making it diffi- cult to even pull one off the acacia. 9. Acacia mayana Lundell Acacia mayana Lundell, 1937:210. Acacia mayana (Figure 17) is the most easily recognized of the swollen-thorn acacias, owing to the pair of bladelike longitudinal flanges extending from the thorn base to tip. The type specimen is representative of the other specimens examined, though it should be noted that there are only 6 specimens at hand of this exceptionally rare swollen- thorn acacia. The leafless isotype of the U.S. Na- tional Herbarium was collected from the upper part of the canopy of a mature plant that has been re- producing at least 2 years, since it bears both flowers and fruit. The fruits and flowers of A. A 1. 2. 3. 4. 3. mayana are similar to those of A. cornigera and A. sphaerocephala (and therefore very different from the other 9 species of swollen-thorn acacia), but the fruits differ from those of these two species by being longer, thinner, and having a long thin base or stalk before swelling out around the seeds. The inflorescences differ from those of A. cornigera and A. sphaerocephala in that they taper to a point, rather than having a blunt end. More specimens of mayana are needed for study. KEY COLLECTION LOCALITIES Type: Rio Pasion, San Diego, Peten, Guatemala. VI-10- 1935, M. Aguilar, no. 495 (US 1688106), mp. Forest S of Finca Yalpemech, near Alta Verapaz-Peten boundary, Depto. Alta Verapaz, Guatemala (100-150 m) . 111-25-1942, J. A. Steyermark, no. 45184 (F 1188721), mp. 3 mi W Cardenas, Tabasco, Mexico (20 m). VI-18-1966, UHJ, no. 458. Fortuno, Coatzacoalcos River, Veracruz, Mexico (30-50 m) . 111-1937, L. Williams, no 8425 (F 896540). Retiro, Tenosique, Tabasco, Mexico. VI-19-25-1935, E. Matuda, no. 3410 (F 1027159) (?shrub in virgin forest?). 94 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 81.4eographic variation in type B thorns from occupied adult A. cornigera: A, Mexico, site 26; B, Mexico, site 28; c, Mexico, site 26; D, Mexico, site 41. 6. Rio Icuolay, NNW Finca Cubilquitz to Quebrada Diablo, Depto. Alta Verapaz, Guatemala (300-350 m) ; 111-6-1942, 1. A. Steyermark, no. 44726 (F 1148694). NAT~JRAL HISTORY Acacia mayana is by far the rarest of the swollen- thorn acacias. If the small number of collections represents the true density and spatial distribution of the species, it may well be self-pollinated in con- trast to the other swollen-thorn acacias. The only specimen that I have been able to find (site 3) was growing in the same type of old second-growth cornfield regeneration that commonly harbors A. cornigera, but with one exception. The forest was about 15 m tall and the single A. mayana was heavily shaded, yet growing very well; an A. cornigera in the same shade regime would have been spindly with undersized thorns. It is possible that A. mayana is a species of primary forest, such as A. melanoceras and A. allenii; Matuda recorded it at site 5 as a ?shrub in virgin forest.? If this is the case, then its rarity in areas under heavy agri- cultural pressure probably means that it will be- come extinct in the lowlands. If the above hypothesis is correct, then A. mayana is a mature and wet-forest edition of A. cornigera. It is interesting in this context that A. cornigera has perhaps produced both a beach-edge species (A. sphaerocephala) and a wet mature forest species. It is tempting to view the oddly shaped thorns of A. mayana as the result of a long past hybridization with a sword-thorn acacia such as A. macracantha. When A. macracantha hybridizes with A. chiapen- sis, the result is often a ?roundish? thorn in cross section, with longtitudinal shallow-sided ridges down each side of the thorn. On the other hand, A. mayana displays no other traits suggesting a history of introgression with other acacias. Whether the ridges are the result of introgres- sion, or a novel mutation, they have a possible adaptive value. A. cornigera thorns on heavily shaded shoots have weak walls and are easily opened by birds for the ant brood inside; apparently the starving shoot does not have the energy to make the thick-walled thorns that effectively protect the ant colony in more insolated microhabitats. The longitudinal ridges on A. mayana thorns may be a way to solve the strengthening problem without SUMBER I3 95 FIGURE 8P.-Acacia mayana: A, type B thorns (the two largest), the largest thorn contained the ant colony queen; B, type A thorns split by birds in search of ant brood (Mexico, site 5). putting a lot of bulk into the thorn wall (this bulk is also adaptive as insulation in heavy insolated sites) in a heavily shaded habitat. First, it is im- portant to note that the ant colony grows slowly at a comparatively large cost to the acacia in the shaded habitat, and the loss of thorns with brood will be proportionately larger to the acacia than in an insolated site. Second, the slow replacement rate of thorns in shaded sites, observed on all forest species of swollen-thorn acacias, means that the birds have a longer time per thorn to locate the tree and go after its ants. Figure 82 shows that birds can be a threat to the ant colony in A. mayana, and it is important to note that they opened all the thorns on the shoot except the few largest ones. All specimens of A. mayana examined have been occupied by mature colonies of P. ferruginea. In view of the small sample size, however, this should not be taken to indicate that it is a suitable host for only this species of obligate acacia-ant. The colony in the specimen at site 3 was as normal as can be expected of a colony that has just lost about three-fourths of its brood and workers to a bird. While the petiolar nectaries of A. mayana are nearly twice as long as those of A. cornigera and A. sphaerocephala, it should not be assumed that the ant colony gets twice as much nectar per leaf from the shaded plant. There are no data to show a correlation between nectary size and nectar pro- duction. A total of 12 pods from herbarium sheets was examined and there was no attack by bruchids. Owing to the small sample size, and the small sizes of the crops from which the pods were taken, this should not be taken as an indication that A. mayana is free from predispersal seed predation. On the other hand, if in fact it does occur as widely scattered individuals in mature forest, it is possible that the bruchids that commonly infest A. cornigera and A. sphaerocephala cannot cope with the very low prey density that such a rare population of swollen-thorn acacias presents (as with A. melano- ceras and A. allenii). 10. Acacia gentlei Standley Acacia gentlei Standley, 1940:77. Acacia gentlei is the common acacia of wetter SMITHSONIAN CONTRIBUTIONS TO BOTANY sites in British Honduras and lowland northern Guatemala (Figure 44). It is easily distinguished from other swollen-thorn acacias by the combina- tion of elongate inflorescences, one or at the most two elongate petiolar nectaries, and a large number of very thin thorns. Where it is sympatric with A. collinsii and A. cornigera in northern British Honduras, it is easily distinguished from both by attaining heights of 3 to 8 m, while they are gen- erally 1.5 to 3 m in height when reproductive, and by having much darker green foliage that strikingly contrasts with the rust-red bark. Both are generally found in the drier sites, while A. gentlei is gen- erally in the wetter sites. Where the distribution of A. gentlei overlaps with that of A. ruddiae and A. cookii in northern Guatemalan wet forest, A. gentlei is again easily distinguished from the former by being occupied by obligate acacia-ants and by having Beltian bodies on the new foliage. It is distinguished from the latter by having oval petiolar nectaries and elongate inflorescences. Some speci- mens of A. globulifera have oval petiolar nectaries as well, but the spherical inflorescence of A. globu- lifera is very easily distinguished from the elongate one of A. gentlei. Standley?s type specimen of A. gentlei was obvi- ously collected from the upper part of the crown of a large acacia such as the one in Figure 83~, and it is representative of the population whose distri- bution is plotted in Figure 44. The thorns on the type are slightly redder than most thorns in nature from British Honduras; they look more like those common on the north side of Lake Izabal. The seed pods on the type specimen are not mature but are nearly full length (19 to 20 cm). KEY COLLECTION LOCALITIES British Honduras: 1. 2. 3. 4. 5. Type of A. gentlei: British Honduras. P. Gentle, no. 185. 13.5 mi S Sta. Elena, Corozal Prov., to 43.8 mi SE Orange Walk (on road to Belice), Belice Prov., British Honduras (10-30 m). VI-30-1966, DHJ, nos. 582, 575 (occasional to common) fl., mp. 17.9 to 36.1 mi SW Belice (on road to Cayo), Cay0 Prov., British Honduras (50-100 m). VII-3-1966, DHJ, nc (common). 2 mi S Sta. Elena, Corozal Prov., British Honduras (sea level), VII-4-1966, DHJ, nc (3 plants). Gallon Jug, British Honduras. 111-5-1938, D. A. Lan- caster, no. 31 (US 2219935) fl., mp. 6. Maskall, British Honduras. 1-31-1934, P. H. Gentle, no. 1121 (US 1635002), mp. 7. Little Cocquericot, Belize River. 111-27-1933, C. L. Lundell, no. 4406 (F 738238). 8. Valentin, El Cay0 Prov. VI-1936, C. L. Lundell, no. 6328 (F 894493). 9. Camp 6, El Cay0 Prov. 111-9-1938, P. H. Gentle, no. 2311 (F 945128). 10. Rio Grande (southern B. H.) (80 m). 111-20-1933, W. A. Schipp, no. 1142 (F 669139) ft. Guatemala: 11. El Zapotillo, 1 km E El Estor, Depto. Izabal, Guatemala (10 m). V-16-1965, DHJ, nos. 657, 659, 665, 667, 671, 690 (common), fl., mp. 12. 16.1 mi SW (Puente El Lobo) Los Amates to 29.4 mi NE (Santa Rosa) Los Amates on CA-9, Depto. Izabal, Guatemala (200 to 50 m) . VIII-9-1967, DHJ, nc (com- mon but scattered). 13. 1.0 to 9.4 mi SE Mariscos (S side Lake Izabal), Depto. Izabal, Guatemala (100 m). V-13-1965, DHJ, nc (com- mon). 14. Uaxactun, Depto. Peten, Guatemala, IV-21-1931, H. H. Bartlett, no. 12676 (US 1493227, US 1493228), fl., mp. 15. Aguacaliente, Depto. Santa Rosa. 1-25-1908, W. A. Kel- lerman, no. 7710 (F 224606) (locality doubtful). 16. La Libertad, Depto. Peten. IV-1-1933, C. L. Lundell, no. 2379 (F 738067) fl. 17. N of Sebol, between Sebol and Carrizal, Depto. Alta Verapaz (200-300 m) . IV-18-1942, J. A. Steyermark, no. 45762 (P 1180368, F 1180419). 18. S of Sayaxche, Laguna Petexbatum. (Laguna Mexico), Depto. Peten (50 m). V-3-1942, J. A. Steyermark, no. 46221 (F 1180409). 19. 2-4 mi W Livingston, Depto. Izabal. IV-16-1940, J. A. Steyermark, no. 39546 (F 1038624). 20. Fallabon-Yaxha Road, Depto. Peten. 111-1933, C. L. Lun- dell, no. 2195 (F 739642). 21. 2.8 mi W Puerto Matias de Galvez (on road to Escobas) , Depto. Izabal (10 m) , VIII-10-1967, DHJ, nc. NATURAL HISTORY While A. gentlei is clearly a plant that was originally associated with wetter forest than extreme dry-land species such as A. cornigera, A. hindsii, and A. collinsii, it has survived very well in the face of human disturbance, in contrast to A. allenii, A. melanoceras, and A. ruddiae. Wherever the for- est is cut with at least a 15 to 20 year fallow rotation period, A. gentlei becomes very common, and has even become a roadside plant in the wetter parts of British Honduras. It is very common in fallow Indian milpas in the Izabal region of Guatemala. In Guatemala, it is commonly encountered in swamps and river-edge vegetation in forest undis- turbed by man for a long time. NUMBER 13 97 FIGURE BS.-Acacia gentlei (Guatemala, site 12): A, crown of 5.m-tall emergent adult; B, type B thorns on main trunk of adult. The obvious success of A. gentlei in moving from natural disturbance sites is probably due to two factors. First, the seeds are surrounded by a white and sweet pulp that is readily picked from the semidehiscent pods by birds such as kiskadees (Ptangus sulpheratus) and saltators (Saltator spp.). These, and other birds of second-growth vegetation, are probably responsible for frequent dispersal of seeds into newly cleared sites. Second, as will be discussed later, A. gentlei is quickly occupied and heavily patrolled by obligate acacia-ants. While A. gerttlei adults in reproductive condition are only occasionally encountered in primary forest (but slightly more often than are A. allenii or A. melanoceras), their seeds are probably moderately common. A. gentlei seeds are very hard, like those of other dry-land acacias, and remain viable at least 5 years and probably much longer. The pres- ence of A. gentlei seeds in the forest litter is sug- gested as the effect of the hurricane that leveled large expanses of wet forest to the southwest of Belice, British Honduras (site 3), in the late 1950s. In the woody regeneration that followed, by 1966 A. gentlei was scattered over the area at a density of about 10 reproducing adults per acre. There had clearly been enough seed and seedlings present to do this, despite the fact that the average density of A. gentlei in this type of forest, when undis- turbed, is about 0.1 tree per acre. Reproductive details of A. gentlei have not been recorded, but it sprouts very well from cut stumps. Seedlings grow well in very dense shade but appear to require a slight break in the general canopy be- fore they can grow large enough to reproduce. Seed crops are large and therefore comparable in size with the other dry-land acacias when fully insolated and mature. Adults are difficult to age because they show the least clear growth rings of any of the swollen-thorn acacias. Based on size, they probably live up to about 20 or 25 years and rarely have a trunk diameter in excess of 15 cm. Trees of A. gentlei are usually very heavily oc- 98 SMITHSONIAN CONTRIBUTIONS TO BOTANY h fi B - 150 FIGURE 84.-Acacia gentlei: A, type A thorns from a lateral branch from the crown of a mature tree (Guatemala, site 11); B, type A thorns from the crown of a mature tree (British Honduras, site 2); c, type B thorns on the base of a lateral branch (British Honduras, site 2). X t-r t3 2 w cl 2 0: 0 X t-r 130 140 i a* a I I I I 0.5 1.0 1.5 2.0 TOTAL THORN VOLUME (cc) FIGURE 85.-Distribution of 15.4 cc of thorn volume among 14 very large type A thorns from one A. gentlei (Guatemala, site 13). cupied by P. ferruginea or one of the two unde- scribed species of obligate acacia-ants found within its range. The young single-queen colonies appear to have little trouble becoming established- though in roadside areas of British Honduras, an unusually aggressive black morph of Pseudomyrmex gracilis often takes over young acacias. While the dark green foliage of A. gentlei closely resembles that of A. chiupensis in texture, color, and taste, A. gentlei lacks the herbivore damage so characteristic of A. chiupensis. This lack of herbivore damage is apparently because the ant colonies grow very large in A. gentlei; a single-queen colony of 15,000 work- ers is not uncommon in a 3-m-tall tree. Acacia gentlei shows that thin thorns (Figure 83~, 84) alone do not lead to poor growth of an ant colony. As can be seen in Figure 85, even the very longest thorn has a very small thorn volume. Many thorns are only 1.5 to 3 mm inside diameter, barely allowing the worker ants to pass through. As a consequence there is a very small amount of brood per thorn, but the very large number of thorns per tree probably means that the total thorn volume available to the colony is about the same in equal- NUMBER 13 FIGURE 8G.-Football-sized carton nests of Azteca ants in the crowns of 12-m-tall A. gentlei (Guatemala, site 12). sized trees of A. gentlei and other swollen-thorn acacias. The other thin-thorned species, A. cookii, does not appear to have an excessive number of swollen thorns, but the slightly larger diameter of A. cookii thorns may compensate for the lower number as contrasted with A. gentlei. Large senescent trees of A. gentlei growing in undisturbed (by humans) habitats are occasionally heavily tenanted by an unidentified species of ant in the genus Azteca. This ant genus, better known for its species occupying Cecropia and Cordia, is arboreal, predaceous on insects, and very aggressive to animals that shake the foliage near the nest. In A. gentlei the ants live both in the thorns previously opened up by the obligate acacia-ants and in hang- ing carton nests (Figure 86). While they do not systematically patrol the acacia leaves, they do forage on them, which probably results in the senescent tree living longer than normal. The obli- gate acacia-ant colony generally does very poorly in a senescent acacia since the ants have to have Beltian bodies and nectar in large quantities to maintain a colony large enough to effectively patrol the acacia. 11. Acacia collinsii Safford Acacia collinsii Safford, 1910:677. Acacia yucatanensis Schenck, 1913:361. Acacia neisonii Safford, 1914:3G3. Acacia costaricensis Schenck, 1913:361. Acacia penonornenscs Safford, 1914:363. Acacia panamensis Schenck, 1913:362. Myrmecodendron collinsii (Safford) Britton and Rose, 1928: Myrmecodendron costaricense (Schenck) Britton and Rose, 92. 1928:95. Acacia collinsii is easily distinguished from all other swollen-thorn acacias by its unique broadly dome-shaped petiolar nectaries. It has elongated inflorescences and fully dehiscent pods (which it has in common only with A. hindsii). A. hindsii and A. collinsii may be distinguished at almost all localities by their thorns, since type A thorns from mature A. hindsii are always strongly flattened while type A thorns of A. collinsii are round in cross section. Of all the swollen-thorn acacias, A. collinsii and A. cornigera have the widest geo- graphical and ecological distributions (Figure 87) . Safford?s tjpe specimen of A. collinsii offers a classic example of the inadequacy of type specimens at representing an entire population. While Saf- ford?s type specimen is representative of the popu- lation of A. collinsii at site 1, the Chicoasen population of A. collinsii is quite different from A. collinsii over the remainder of its range. The type specimen appears to have been taken from a large adult tree in a partially shaded riparian site. Its leaves are as long as any found on A. collinsii (up to 16 cm long with the 10 pinnae spaced up to 3 cm apart). The leaves have the unusually large number of 5 to 6 petiolar nectaries with the distal ones being as far as 2 mm apart (this makes one suspect a bit of introgression with A. globulifera, which also occurs at this site). The sexual branch- lets are up to 8 cm long (normal maximum) and the peduncles of the mature inflorescences range up to the exceptional length of 33 mm. The peduncles are also nearly 2 mm thick, while on most A. collinsii they are 1 mm or less thick. The thin type A thorns are round in cross section but twisted around the branch; this is characteristic of the A. collinsii population only in Chiapas, Mexico. Schenck?s type of A. yucatanensis (site 2) is repre- sentative of most of the population of A. collinsii in drier habitats, and especially the Yucatan portion of the population. The small leaves are specially characteristic of A. collinsii. His type of A. costari- censis is not available (apparently destroyed in Berlin) but a paratype from Costa Rica (site 66) is clearly drawn from that portion of the A. collinsii population at dry intermediate elevations in Costa Rica. If it were useful to designate a lectotype, this 20. 8? * 0 would be the appropriate specimen. Schenck?s de- scription of A. costaricensis applies unmistakably to A. collinsii and is very different from the other Costa Rican swollen-thorn acacias (A. cornigera, A. allenii, A. ruddiae). Britton and Rose (1928) refer to the type locality of A. costaricensis as ?Lagarto, Costa Rica? but do not give the source of these data; there is a ?Lagarto? in Puntarenas Province and A. collinsii is abundant there (sites 67 and 75). Acacia panamensis is clearly based on two species of plants. The description of the thorn and leaf- bearing branch clearly applies to A. collinsii (not to A. melanoceras, the only other central Pana- manian species of swollen-thorn acacia). Schenck (1913) notes that he is not sure that the inflores- cences belong to the sterile branch; the globular inflorescences could be from A. melanoceras (un- likely owing to their very thin peduncle) or, more likely, from one of the other nonant-acacias that grow commonly in the drier parts of Panama along with A. collinsii. A final complication is that the exact locality is given as ?Aljahuela? and I cannot SUMBER 13 101 locate such a site in Panama. This site most likely represents Alajuela in Alajuela Province, Costa Rica, which was a favorite collection site for Ger- man collectors around the turn of the century. A. collinsii is the only swollen-thorn acacia in the Alajuela region (sites 68, 69, 74). Safford?s type of A. nelsonii (site 3) is representa- tive of the A. collinsii population in the area north of the Isthmus of Tehuantepec, Mexico, but differs from more southerly populations in having large pinnules (up to 11 mm long and 3 mm wide) and somewhat flattened thorns (2 to 3 times as wide as deep in cross section). Flattened thorns are, how- ever, even better developed in the A. collinsii popu- lation from central Costa Rica to Panama. The inflorescences of the type of A. nelsonii are normal, but the thorns on the same branch have been weathered gray rather than being the usual beige color of younger thorns from western Mexico. Safforcl?s type of A. penonomensis (site 76) differs from the remainder of the A. collinsii population on the dry Pacific side of Panama in that its type A thorns are not quite as flattened as are those on many A. collinsii at sites 82 to 84. The seed pods on the type of A. penonomensis are representative of those found in the drier parts of the range of A. collinsii; these pods have stiff woody valves that only curl a slight amount during dehiscion. KEY COLLECTION LOCALITIES Mexico: 1. 2. 3. 4. 5. 6. 7. Type of A. collinsii: (between San Fernandino and) Chicoasen, Chiapas, Mexico (320 m). 1-13-1907, G. N. Collins and C. B. Doyle, no. 180 (US 692159). Type of A. yucatanensis: Yucatan, Mexico. 1895, G. F. Gaumer, no. 353 (US 571618). Type of A. nelsonii: Acapulco, Guererro, Mexico (sea level). IV-30-1903, E. W. Nelson, no. 7024 (US 399366). 3.6 mi NW, border station between Mexico and Guate- mala (La Mesilla) to 8.4 mi S La Trinitaria on Hgy. 190, Chiapas (680-720-600-1060 m) . VIII-7-1966, DHJ, nos. 441, 443; VIII-17-1967, DHJ, no. 1713 (very com- mon to occasional) . 11.8 mi E (near turnoff to Villa Hermosa) Chiapa de Corzo to 13.8 mi SW Tuxtla Gutierrez on Hgy. 190, Chiapas (10904.50-1090 m). VIII-17-1967, DHJ, nos. 424, 43.5 (common to occasional). 13.8 mi SW, Tuxtla Gutierrez to 17.9 mi SW Las Cruces (turnoff to Arriaga) on Hgy. 190 (1090-530-820 m), Chiapas. VIII-18-1967, DHJ, nc (common to exceed- ingly rare). 1-2 mi N Gabriel Zamora on Hgy. 37, Michoacan (820 m) . VI-10-20-1968, DHJ, nc (very common). 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 13.4 to 23.5 mi E Zanatepec on Hgy. 190, Oaxaca (50 m) . VIII-18-1968, DHJ, nc (very rare to common), fl. 42.1 mi E Zanatepec on Hgy. 190, Oaxaca (60 m) . VIII- 18-1967, DHJ, nc (1 plant). 3 mi N Soyalo on Tuxtla Guticrrez.Villa Hermosa Hgy., Chiapas (- 1100 m) . VII-10-1966, DHJ, nc (common), fl. 10 to 14 W Tchuantepec (Hgy. 190), Oaxaca (100 m). 1-10-1965, DHJ, nos. 1845, 1950. IX-3-1965, DHJ, nos. 1779, 1780, 1783 (common). Pucrto Escondido to 3.5 mi N on road to Oaxaca, Oaxaca (10-120 m) . VI-19-1967, DHJ, nc (common to very rare) . Puerto Escondido to 18.4 mi RTV (Hgy. ZOO), Oaxaca (10-80 m). VI-19-1967, DHJ, nc (common to very rare). Hills in and around Acapulco, Guerrero (0-100 m). 1?1-18-1967, DHJ, nc (occasional) . Acapulco to 1.3 mi SE Papanoa, Guerrero (0-30 m). VI-16-1967, DHJ, nc (common on dry hillsides, ab- sent from swamps). 48 mi NE Isla Aguada (across from Puerto Real, Cam- peche) to Merida (Yucatan), on Hgy. 180 (0-30 m), VI-22-1966, DHJ, nos. 428, 431, 433, 438, 606 (occasional to very abundant.) Ruinas Edzna north to Hgy. 180 (east of Campeche), Campechc (15-40 m). VI-25-1966, DHJ, nc (rare to occasional). Becal south to Hgy. 180, Campcche (1540 m) . VI-26- 1967, DHJ, nc (rare to occasional). Merida to 14.2 mi E (Hgv. 180), Yucatan (10-20 m). VI-28-1966, DHJ, no. 455 (rare). Mum (Yucatan) to Felipe Carillo Puerto (road to Cd. Chetumal) , Quintana Roo (10-20 m). VI-29-1966, DHJ, nc (very rare). Felipe Carillo Puerto to Cd. Chetumal, Quintana Roo (20-0 m). VI-30-1966, DHJ, no. 453 (very rare to oc- casional) . 22. 9.6 mi E San Marcos on Hgy. 200, Oaxaca (50 m). VI- 18-1967, DHJ, nc (1 plant). 23. La Ventosa, 7 km E Salina Cruz, Oaxaca (sea level). VII-3-1959, K.M. King, no. 1302 (US 2362102), fl. 24. Petatlan, Montes de Oca, Guerrero. VI-17-1937, G. B. Hinton, no. 10333 (US 1805541, F 959109). 25. Tuxpena, Campcche. 11-2-1932, C. L. Lundell, no. 1279 (US 1494445), mp. 26. Sudzal, Yucatan. IV-21-1956, 0. G. Enriquez, no. 564 (US 2448079). 27. Chichen Itza, Yucatan. V-30-1929, J. Bequaert, no. 2 (US 1477467), ff., mp. 28. Isla Cozumel, Quintana Roo. XII-5-1966, V. E. Rudd, no. 2036 (US 2518023). 29. San Vicente, Chiapas. IV-16-1904, E. A. Goldman, no. 862 (US 470663), mp. 30. Berriozabal, on road to San Fernando, Chiapas (950 m). XII-18-1949, F. Aliranda, no. 5847 (US 2367591), mp. 31. Chorreadcro de Tuxtla, 5.6 mi E Chiapa de Corzo (800 ni). 11-16-1965, D. E. Breedlove, no. 9018 (F 1627187), fl. 102 SMITHSOKIAN COSTRIBUTIONS TO BOTANY 32. Chichen Itza, Yucatan. V-30-1929, J. Bequaert, no. 2 (F 710846). 33. Izamal, Yucatan. 11-21-1906, J. M. Gaurner, no. 379 (F 189844), fl. 34. Salina Cruz, Oaxaca (sea level). IV-28-1910, Orcutt, no. 3451 (F 280154), mp. British Honduras: 35. 23.6 mi S Orange Walk to Belize, Belize Prov. (10-0 m). VII-2-1966, DHJ, nc (rare to common). 36. 10.1 to 34.4 mi SIV Belize on road to El Cayo, Belize Prov. (30-100 m). VII-3-1966, DHJ, nc (rare to occa- sional). 37. Duck Rum, Pine Ridge, El Cay0 Prov. 11-17-1931, H. H. Rartlett, no. 11546 (US 1493376, US 1493375), mP* Guatemala: 38. 2.6 mi S\V Puente Rio Hondo on Hgy. CA-9 (near turnoff to Zacapa), to 1.8 mi NE Pucnte El Lob0 on Hgy. CA-9, Depto. Zacapa (200-100 m) . VIII-9-1967, UHJ, nos. 1794, 1797 (very common to occasional). 39. ?District of Peten.? IV-8-1933, C. L. Lundelf, no. 2614 (F 738050). Honduras: 40. El Espino (Sicaragua) to 0.4 mi W on Honduras- Siraragua border on Interamerican Hgy., Depto. Cho- luteca (880-930 m). VII-29-1967, DHJ, nc (4 plants). 41. 21.6 mi I$? San Marcos de Colon on Interamerican Hgy., to Choluteca, Depto. Choluteca (450-10 m). VIII-31- 1965, DHJ, 110. 1788; V-11-1965, DHJ, no. 1807; VII- 29-1967, DHJ, nc (very common). 42. Choluteca (Depto. Choluteca) to Jicaro Galan (Depto. L?alle) on Interamerican Hgy. (10-40 m) . VII-29-1967, DHJ, no. 1803 (common to rare). 43. Jicaro Galan to 3.8 mi NE (Hyg. 1) , Depto. Valle (40- 80 m). VII-29-1967, DHJ, nc (rare). 44. Comayagua to 3 mi S on Hgy. 1, Depto. Comayagua (550 m). VII-31-1967, DHJ, nc (rare to common), fl. 43, Jicaro Galan to El Amatillo (Honduras-El Salvador border) on Interamerican Hgy. (40-200 m) . 1-15-1965, DHJ, no. 1841; V-12-1965, DHJ, nc (occasional to rare), fl. 46. Juticalpa, Depto. Olancho. 111-5-16-1949, P. C. Stand- ley, no. 17580 (F 1378910), mp. 47. Rio Guayainbre, Valle Jamastran, Depto. El Paraiso (350 in). 111-13-1956, A. Molina, no, 7364 (F 1599625). 48. Danli, Dcpto. El Paraiso (700-800 m) . 11-11-23-1949, P. C. Standley, no. 16506 (F 1378924). El Salvador: 49. El Amatillo (Honduras-El Salvador border) to 7.1 mi IV (on road to La Union), Depto. La Union (200-190 in). V-12-1965, DHJ, no. 1801 (rare), fl. Xicaragua: 50. Penas Blancas (Depto. Rivas) to Jinotepe (Depto. Carazo) (Hgy. 2) (50-500 m) . V-9-1965, DHJ, no. 1822; 51. 52. 53. 54. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 1-16-1965, DHJ, nos. 1830, 1932; VII-26-1967, DHJ, nc (occasional to very common). La Virgin to San Juan del Sur (Hgy. 16), Rivas Prov. (50-0 m). V-10-1965, DHJ, nos. 1814, 1815, 1819, 1921 (common), fl, mp. Masachapa to 14.7 mi NE (Hgy. 8), Depto. Managua (0-480 m) . VII-26-1967, DHJ, nc (common to rare) . 15.2 mi SW Interamerican Hgy. (Depto. Managua) on road to Chinandega (Hgy. 12) to 5 mi SE Leon (Depto. Leon), (30-450 in). VII-27-1967, DHJ, nc (occasional to rare). Puerto Somoza to 6.4 mi SE (junction with Hgy. 12), Depto. Leon (0-30 mi S Tipitapa on Hgy. 1, Depto. Managua (50 m). V-10-1965, DHJ, no. 1812; VII-28- 1967, DHJ, nc (1 unoccupied plant). 4.2 mi N Tipitapa on Hgy. 1, Managua (75 m). VII- 28-1967, DHJ, nc (1 unoccupied plant), fl. San Benito (N of Tipitapa on Hgy. 1) to 16.7 NE (Hgy. y) (100 m). VII-28-1967, DHJ, nc (extremely common). San Bcnito (Depto. Managua) to Sebaco (Depto. Mata- galpa) on Hgy. 1 (100-420-350-500 m). VII-28-1967, DHJ, nc (very common to very rare at 420 m to com- mon) , fl. Sebaco to 1.6 mi N Matagalpa (Hgy. 3), Depto. Mata- galpa (500-730 m) . VII-28-1967, DHJ, nc (common to occasional), fl. Sebaco (Depto. Matagalpa) to Esteli (Depto. Esteli) on Hgy. 1 (500-1040-850 m). VII-28-1967, DHJ, nc (corn. mon to very rare at 1040 m to common). Esteli (Depto. Esteli) to El Espino on the Honduras- Nicaragua border (Depto. Madriz) 011 Hgy. 1 (850-530- 880 m) . 1-15-1965, DHJ, no. 1825; V-11-1965, DHJ, no. 1808, 1810; VII-29-1967, DHJ, nc (common). Laguna de Masaya, Depto. Masaya (300 m) . VII-6-1923, W. R. Maxon, no. 7739 (US 1181452), 8. Granada, Depto. Granada (250 m) . VII-2-1923, W. R. Alnxon et al., no. 7397 (US 1181308) ; IV-21-1916, J. D. Mandragou (US 764653 and CS 764655), fl; and II-12- 1903, C. F. Baker, no. 2546 (US 862954), fl. La Libertad, Depto. Chontales, V-29-VI-1-1947, P. C. Standley, no. 8826 (F 1288080). Braggman?s Bluff (Puerto Cabezas), Yulu Village, Depto. Zelaya (100 m) , (F 585006). Costa Rica: 66. Potential lectotjpe of A. costaricensis: Alajuela, Ala- juela, Costa Rica (920 m). 1896, J. D. Smith, no. 6488 (cf. Schenck, 1913). 67. 18.4 mi to 21.4 mi E Palmar Sorte, Puntarenas Prov. (90 m) . V-23-1963, DHJ, no. 1852; VI-26-1967, VIII- 23-1965, V-15-1965, 111-21-1966, VII-21-1963, DHJ, nc (occasional), in fl. 68. Escazu to 3.6 in 11? Santaigo de Puriscal (Hgy. 7), Ala- juela Prov. (1000-600 m). VII-17-1967, 11-15-1965, DHJ, no. 1862; 111-2-1965 (fl), DHJ, nc (occasional to rare). 69. Along the electric railroad from Antonio to Orotina, Alajuela Prov. (900 to 50 in). VIII-1963, DHJ, nc (oc- casional). SUMBER 19 103 70. 71. 72. i3. 74. 75. 15.0 mi NTV turnoff of Interamerican Hgy. to Punta- renas, to Peiias Blancas (Border of Costa Rica and Nicaragua), Guanacaste Prov. (10-100 m) . January- September, periodically in 1963-1967, DH], nos. 1757, 1759, 1762, 1763, 1826 (rare to exceedingly common), fl. January to April. Liberia to Playa COCO (Hgy. 21), Guanacaste Prov. (100-0 m) , January-September, periodically in 1963- 1967, DHJ, no. 1858 (occasional to very common), fl. January to April. Hda. Palo Verde (about 18 mi S. Bagaces), Guanacaste Prov. (sea level). IV-2-1966, DHJ, nc (common), fl. Canas to 6 mi NE Guanacaste Prov. (80-300 m) . IV-15- 1965, DHJ, nc (occasional), fl. Alajuela Prov. (920 m) . 111-1896, J. D. Smith, no. 6488 (US 943274), fl. Lagarto, Puntarenas Prov. (sea level). XII-1891, Ton- duz, no. 4810 (US 943276). Panama: 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 81). 90. Type of A. penonomensis: Penome, Cocle, Panama (15- 300 m). 11-23-1908 to 111-22-1908, R. S. Williams, no. 113. 4.8 mi NE Tucuman Airport on Panama-Chepo Hgy., Panama Prov. (20 m). VII-1-1967, DHJ, no. 1704 (very rare), gp. Vicinity of Madden Dam, Madden Lake (50 m). VII-3- 1967, DHJ, no. 1653 (common), gp, fl. Miraflores Locks to Paraiso and Corozal area, Canal Zone, VII-8-1967, DH] and Wheeler (1942), nc (rare). Arraijuan to 4.1 mi E on Interamerican Hgy., Canal Zone. VII-9-1967, DHJ, nc (occasional). 3.2 mi SIV La Chorrera, Panama Prov. (100 m) , VII-I- 1967, DHJ, nc (rare). 5 mi N. Bejuco (Panama Prov.) to Santiago (Veraguas Prov.) along the Interamerican Hgy. (10-50 m) . VII-1- 1967, DHJ, nos. 1698, 1700, 1702 (rare to occasional). Interamerican Hgy. (Herrara Prov.) to 3.1 mi S Pedasi (Los Santos Prov.) on Azuero Peninsula (5-25 m) . VI- 30-1967, DHJ, no. 1695 (occasional to common). Santiago (Veraguas Prov.) to David (Chiriqui Prov.) along Interamerican Hgy. (50-200 m). VII-11-1967, DHJ, nc (very rare to occasional). 2.1 mi SE Sona to Santiago, Veraguas Prov. (50 m). 1'1-29-1967, DHJ, nc (rare). Remedios (Chiriqui Prov.) to 19.9 mi StV Jorones (Veraguas Prov.) on Sona-Remedios Road. VI-29-1967, DHJ, no. 1666 (occasional). David to Pedregal, Chiriqui Prov. (0-30 m). VI-29- 1967, DHJ, nc (rare). David to 8.5 mi S Boquete, Chiriqui Prov. (30-620 m) . VI-28-1967, DHJ, nc (rare to occasional), gp. David to 2.7 mi NIV on Interamerican Hgv., Chiriqui Prov. (90 m) , VI-28-1967, DHJ, nc (rare to occasional). Along Quebrada Ocho Paso, Rio Tonosi, Tonosi, Los Santos Prov. (20 m). 11-26-1963, W. L. Stern et nl., no. 1831 (US 2490276), fl., mp. Colombia:* * Colombian localities are not included in Figure 87. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. Str-ams of Usiacuri and Pital, in mountains SIV of Baranquilla, Depto. Atlantico. V-26-1928, E. Nienzeyer (US 1496168). Malambo to Sabanagrande, Depto. Atlantico (60 m) . 1-1936, A. Dugand, nos. 117, 404 (US 1619541). Megua, Depto. .4tlantico (50 m) . 1-13-1941, A. Dugand and Jaramillo, no. 2796 (US 1799137). Usiacuri, Baranquilla, Depto. Atlantico (250 m). VII- 1927, B. Elias, no. 245 (US 1342935), fl. Arroyo Jubilado, Depto. Atlantico (50-60 m) . XII-24- 1933, A. Dugand, no. Ilia, 40u (US 1573571). Cartagena, Depto. Bolivar. 1919, B. Heriberto, no. 39 (US 1041093). Isla de Providencia (Caribbean Sea). IV-29 to V-9-1948, G. R. Proctor, no. 3430 (US 1979268), fl., nip.; I-24- 1967, S. Anderson; VIII-23 to VIII-29-1970, DHJ, nc (very common). Providence Island (Caribbean Sea). VII-19-26-1966, .I. H. Torres R., no. 238 (COL 112189). ,Tuanmina, Baranquilla Dist., Depto. Atlantico (20 m). VII-5-1963, H. S. McKee, no. 10433 (COL 98405). Valencia de Jesus, Valledupar, Magdalena (30 m). II- 14-1948, R. Roinero-Castaneda, no. 793 (COL 44486). Arroyo de Megua, Depto. Atlantico (50 m) . 1-13-1941, A. Dugand and R. Jni-amillo, no. 2769 (COL 05500). NATURAL HISTORY Over its wide geographical and ecological range (Figure 87), A. collinsii is found in habitats ranging from moderately wet to very dry, but the majority of the population is found in habitats that appear generally too dry for other swollen-thorn acacias (except A. sphaeroccpJiala with its peculiarly local distribution). These habitats ale often too dry for agriculture on hillsides (8 month dry season), and as a result, the majoiity of the contemporary popu- lation of A. colliiisii lives in shrubby vegetation that is overgrazed by cattle and goats (Figure 88). In the moister paits of its habitat, the natural vegetation has been almost entirely destroyed to plant cornfields, but where this is done with a 5 to 25 year fallow period (e.g., sites 67 and 88), A. collinsii is a common part of the regeneration. In the very dry parts of its range, such as the east side of the Yucatan Peninsula and the Tehuantepec area, A. collinsii probably grows in vegetation simi- lar to that occupied before the advent of European agriculture except for graiing by goats. TVherever A. collinsii and othei species of swollen- thorn acacias occur in the same general area, A. collinsii is always in the driest subhabitat (and thus mixed in with several nonant-acacias). This is par- ticularly striking at a place such as site 15 where it 104 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 88.-Typical pasture habitat of A. collinsii (Nicaragua, site 60). is common on the dry ridges that extend into the swampy coastal flatland occupied by A. cornigera. Also, in the very dry lowlands in northeastern Guatemala (site 38), A. collinsii occupies the dry rocky pastures, while A. hindsii, A. gentlci, and (rarely) A. cornigera occupy the swamps and stream bottoms and banks. Since much of the Yucatan Peninsula lacks surface watercourses, it would lack swollen-thorn acacias in most of the eastern drier area if it were not for A. collinsii. While A. collinsii is apparently sympatric with either A. cornigern (e.g., site 17) or A. globziliferu (e.g., site 16), the latter two species are always in a local wet site, such as a lagoon running along the beach or the site of a cenote or well. Acacia collinsii appears rarely to be moved about by contemporary human activities, and the geo- graphic distribution recorded in Figure 87 is prob- ably little changed since Europeans began disturb- ing Central American vegetation. Like A. cornigera and A. hindsii, however, A. collinsii is probably much more common now than in pre-Colombian times, since it grows very well in the dry and shrubby pastures that have replaced much of low- land Central America's deciduous forests. The fail- ure of humans to move the seeds of A. collinsii is probably due in part to the fact that the pods dehisce so that the edible pulp is either taken by a bird or drops to the ground (in contrast to the indehiscent pods of A. cotnigera that can be easily harvested, and carried about). Also, A. collinsiz lives in habitats that are so dry that earth-moving equip- ment is less likely to pick up mud with seeds in it than is the case for either A. cornigein or A. hindsii. Finally, and perhaps most important, A. collinsii appears to occupy L irtually all the suitable habitats in the dry lowlands on both coasts of Central America. The conspicuous exceptions to this state- ment are the dry northern portions of the east and west coasts of Mexico, occupied by A. cornigera and A. sphaerocephala, and by A. hindsii, respectively; human-movement patterns in these pal ts of hlexico are not at ,all favorable to contemporary transport of A. collinsii further north than at present. The failure of A. collinsii to spread throughout the di y lowlands of Colombia is the most perplexing aspect of its distribution. There are two possibili- ties, neither of which can be discussed fully with the data at hand. It may be a relatively recent introduction from Panama or further north along the coast; old coastal ports such as Acapulco, La Libertad, Puntarenas, and Panama City are well within the range of A. collinsii. On the other hand, the twin dry seasons in Colombia may be inimical to the development of A. collinsii. Jesus Idrobo (personal communication) reports that it is most commonly found in swamps in Colombia, a habitat rarely occupied by A. collinsii in Central America. SUMBER 13 105 FIGLIRE 89.--Acncia collinsii: A, left, unoccupied sucker regeneration and on the right, occupied sucker-shoot regeneration of the same age (Costa Rica, site 70); B, occupied lateral branch end; contrast with Figure c (Costa Rica, site 67); c, unoccupied lateial branch end growing near Figure B. IVhile swollen-thorn acacias may reach immedi- ately offshore islands (eg., occupied A. collinsii on Isla Co/umel, site 28), there is only one case where the ant >< acacia interaction appears to have be- come n:iturall~~ established on an island far from the Central ;\merican coast. A. colliiisii is a com- iiion member 01 the flora of Isla Providencia (sites 9i antl 98) where it is occupied by Pseudoiugmzex ferruginen. This island is about 100 miles off the Caribbean coast of Nicaragua, and even though the acacia is completely wild, there is no guarantee that it was not introduced bl, postColombian human activities. Isla Providencia has been in contact with Europeans for at least 200 years. Even if the acacia was recently introduced, it is most likely that the ants got there of their own accord; even the most avid gardener is unlikely to bring an acacia with an ant colony on a boat from the mainland to Isla Providencia. The Isla Providencia population of A. collinsii is morphologically representative of the mainland population, with the exception that the aeed pods are the smallest 1 have seen. Sot only are they very hea\.ily ~voody (like those on the dry Pacific side of Panama) but range from 2 to 3 cm long and 4 to 8 inm thick. Almost all of the vegeta- tion on Isla Pi-ovidencia has been cut many times and is repeatedly disturbed by livestock antl cultiva- tion; however, even in equally disturbed sites on the mainland, the population density of adult A. col- liiisii is higher on this island than I have ever encountered in mainland sites of comparable land use. This high abundance is probably attributable 106 SMITHSOSIAS COSTRIBLTIOSS TO BOTAXY FILVRE SO.--Accccic~ collinsii (Panama, site 83) : A, crown of adult with clumps of type A thoins caused by an unknown disease; B, close-up of the thorn cluster from the crown. FIGURE 91.--Acacia co11imii: A, two seedlings (compound leaces) with their first pair of leaves (Costa Rica, site 70); B, sucker spiout from burned stump (Costa Rica, site 70); c, unoccupied 2-4-year-old seedlings (Panama, site 88). NUMBER 13 107 to several factors: (1) Based on two small seed samples taken in August 1969, the bruchids that kill a high percentage of the mainland seeds have not arrived on the island; each of the seeds pro- duced is therefore still alive when dispersed. (2) Based on sweep samples taken during the rainy sea- son (late August 1969) the herbivorous insect density on the island is far lower than found in any main- land lowland habitat during the rainy season; this means that the primary environmental challenge that unoccupied seedlings face is competition from other plants, (3) Since the number of plant species is very low on the island, most of the usual com- petitors with A. collinsii are absent from the habitat; in particular, vines are missing. The ants appeared to be as aggressive as their mainland counterparts and attacked vegetation contacting the A. collinsii crown as usual. The only major herbi- vore on A. collinsii that appears to have arrived on Isla Providencia is the noctuid moth (Cocina had- enoides) whose larva can protect itself against the ants and eats the shoot tips of occupied swollen- thorn acacias (Janzen, 196713). On the mainland, this larva invades the acacia often enough to cause occasional multiple branching in the shrub's de- velopment. On Isla Providencia, its damage was intense, causing nearly all the acacias to take on a shrubby life form. 'M'hile A. collinsii lives in a drier habitat than A. cornigern, it appears to have a dependence on obligate acacia-ants as great as that of A. cornigern. A brief test of this was conducted with the hilltop portion of the A. collinsii population two miles southeast of Canas, Guanacaste Province, Costa Kica (site 70). While there are three species of obligate acacia-ants in the area (P. ferruginca, P. belti, P. nigrocincta), only those plants occupied by P. belti were used in tlie experiment. The hill- top site was covered with regenerating pasture shrubs (Baziliinin, Ci,oton, Cassia, Lithen, Acacia, etc.) and about 40 adults of A. collinsii to the acre. On February 16, 1967, 52 plants of A. collinsii in the 2 to 3 m class (3 to 4 years of occupied growth) were cut off 60 cm above the ground. The acacia crown was left next to 25 of the stumps, and for 27 stumps tlie crown was dragged at least 10 m away. After the usual movements of P. belti colonies following this disturbance, 9 of the stumps in the control plot lost their colonies and none of those in the experimental plot gained one (if this ex- FIGURE 92.--Acncia collinsii: A, ten-year-old adult (Panama, site 83); B, crown of adult (Sicaragua, site 54); c, left, 7-year- old clump occupied by Pseudoniyrrnex belti; right, 7-year-old tt'w occupied by Pseudomyrmex ferruginea (Sicaxagua, 57). periment had been conducted during the rainy season, when colonies are larger and the ground is not so hot, 5 to 10 of the stumps in the experimental plot would have regained their colonies despite the 108 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 93.-Acacia collinsii (Panama): A, central crown of adult (site 78); B, crown of 10-m-tall tree (site 88) : c, senescent crown of 10-m-tall tree (site 88). distance to which they were removed). By July 19, 1967, about five months later, the mean length of the longest sucker shoot from each of the 25 stumps in the control plot was 25.1 cm (s.d. 10 cm) while the same measurement was 5.1 cm (s.d. 4 cm) in the experimental plot. These differences are highly significant (Mann-Whitney U test) and differ in no major way from the results obtained with occupied and unoccupied sucker regeneration of A. cornigera and P. ferrziginea in eastern Mexico (Janzen, 1967b). If we consider only the 16 members of the control plot that were occupied, the results are more dramatic; the mean length of the longest sucker shoot is 36.4 cm (s.d. 14 cm). The 9 unoccupied suckers of the control plot also had a mean length of 5.1 cm (s.d. 12 cm). Figure 89~ illustrates a representative shoot from each of the occupied and unoccupied stumps of this date. Figures 898 and c compare the branch ends of large lateral branches from naturally occupied and unoccupied adult acacias in a much wetter habitat in southern Costa Rica (site 67). Since nearly all the regeneration was during the dry season, this low total amount of sucker-shoot development was expected. That no stump died during the course of the experiment suggests strongly that substantial energy reserves were stored in the lower stump and root system. One year later, on July 12, 1968, only 4 of the stumps un- occupied on July 19, 1967, were still alive; all the plants that had been occupied in 1967 were still alive. By July 14, 1969, all the occupied shoots had approximately regained their original height (1967) or grown taller, and there was no trace of the 36 stumps unoccupied on July 19, 1967. Acacia collinsii has a small number of leaf and stem galls just as do other swollen-thorn acacias. At site 86 in Panama, however, the population of NUMBER 13 109 FIGURE 94.-Acacia collinsii (Costa Rica, site 68): A, adult growing in open pasture, August 1, 1963; B, same tree crown, July 1, 1967; the lack of change is due to lack of interplant competition. A. collinsii is afflicted with an outstanding disease. Almost all the acacias have thick clusters of swollen thorns scattered thiougli the crown (Figure 90). These are the consequence of some agent that causes extreme branch proliferation, yet each inter- node is extremely shoit. The number of thorns produced by the acacia is far greater than the ants can use, with the consequence that many of the thorns in the cluster have no entrance holes. The potentially negative effect of this excess of thorns on seed production has not been examined. Young seedling and sucker shoot development of A. collinsii is very similar to that of A. cornigera and A. hinclsii. The bird-dispersed seed germinates at the beginning of each rainy season (Figure 91~) and, if fully inso- lated, forms into a small bush that produces its first swollen thorn on the 12th to 17th leaf (about two months old). Suckers from stumps in recently FIGURE 95.-Acacia collinsii: A, crown of adult growing in area with very severe dry season (Nicaragua, site 54); B, crown of tree growing in area with very mild dry season (Costa Rica, site 67). burned areas do the same (Figure ~IB), but the larger the stump, the more the first foliage resembles that of the mature plant. As the months pass with- out a new colony developing in the young plants, they gain in height at the rate of about 10 cm per year and suffer very heavy damage from defoliating insects (Figure 91c). In most open habitats, A. collirisii grows into a bushy shrub once occupied (Figures 92~, B). If it is in dense regeneration, it may become a small tree that gradually loses crown size and senesces when 15 to 20 years old (Figure 93). It may retain this shape for many years (Fig- ure 94). There is considerable variation in branching pat- 110 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 96.-Acacia collinsii (Nicaragua): A, dense crown of adult caused by shoot damage by Pseudomyrmex ferruginea (site 57) : B, diffuse crown of adult occupied by Pseudomyrmex nigro- cincta (site 51). terns of A. collinsii in fully insolated habitats. In the very driest sites, where A. collinsii may almost be regarded as a member of the ?climax? forest, the plant has only a few major branches, and the new leaves each year are produced in axillary tufts or on very short new branches (Figure 95~). In the wettest sites, the canopy is uniformly dense owing to the production of a large number of small lateral branches (Figure 95~). A second type of crown variation is produced by the interaction be- tween the eastern Nicaraguan portion of the Pseudomyrmex ferrziginea population and A. col- linsii. This species of ant in this area prunes the shoot tips of the swollen-thorn acacia it occupies. The consequence is that two adjacent A. coZlinsii of the same age-one occupied by P. fer7ziginea and the other by P. belti or P. nigrocincta-will be extremely different in height (Figure 92c). If the shoot is not occupied by P. ferruginea until it is 1 to 2 m tall, it produces a ??lollypop?? life form (Figure 96). As discussed in Janzen (1969c), one possible explanation for the ants? behavior is that birds preying on ant brood cannot get into the dense crown to get the queen-thorn; in this area, A. collinsii does not have the well-developed type B thorns which are normally adequate to protect the queen froni birds. Acacia collinsii, like A. cornigera, bears its first flowers at the end of three years of occupied growth. Seeds therefore appear in the fourth dry season after the plant is occupied, meaning that A. collinsii can exist as a reproducing population in cultivated sites with only four years fallow. Its long inflores- cences are normally born in short axillary clusters (Figure 98~) but in very wet sites may be produced on long sexual branchlets (Figure 97A). There appears to be a general cline in increasing inflores- cence length from the northern to the southern end of the range of A. collinsii. The woody pods are usually born 1 to 6 per inflorescence; upon SUMBER 13 111 FICL?RE 97.--Acacia collinsii: A, exceptionally well-developed sexual branchlet (Mexico, site 4); B, type A thorn at base of leaf with three droplets of nectar from petiolar nectaries (Costa Rica, site 65); c, newly grown type B thorns and fully developed branch on tree in Figure B. dehiscing, the seeds drop partly out and are wrapped in sweet yellow pulp which is picked off by birds in flight. This dispersal mechanism is common to the entire population, but there is considerable variation as to how woody the pods are and how much they curl upon opening. In no areas do the pods curl as extremely as those of A. hindsii and A. globulijern. Type A thorns on A. collinsii are easily dis- tinguished from those of all other swollen-thorn acacias, but the differences are difficult to charac- terize. A. collinsii bears the shortest type A thorns to be found on a mature swollen-thorn acacia. They are usually round in cross section and give the impression of being ?fat? (Figures 98, 99). When flattened (Figures 98~, 99~) like A. hindsii, they still have much thicker walls than would an A. hindsii thorn of the same size. .4. collinsii probably shows the most extreme variation on type A thorn morphology of any of the swollen thorn acacias. Most of the population of A. collinsii bears thorns that are ivory to beige, but in wet or high elevation sites they are often dark brown with a maroon or reddish cast. There is usually a prominent bump on the inner wall of each thorn near the base; this may aid the ants in making use of the large volume of space at the base of the ?fatter? thorn types (e.g., Figure 98~). Type B thorns of A. collinsii (Figures 100-103) are easily as variable in shape as those of A. cornigera and A. hiiadsii. As is the case with these latter two species, there is little intrapopulation variation in this trait, but of course there is rather great variation associated with the age of the acacia. Type B thorns are generally produced at the ex- pense of the leaf they subtend (Figure 97c) and are produced very rapidly once branch elongation starts. It is common to find all the type B thorns on a short branch to be at approximately the same stage in hardening. While type B thorns may appear 112 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 98.--Geographic variation in type A thorns from occupied adult A. collinsii: A, Costa Rica, site 68; B, Nicaragua, site 61; c, D, Mexico, site 16. FIGURE 99.-Geographic variation in type A thorns from occupied adult A. collinsii: A, Mexico, site 11; i%, c, Mexico, site 16; D, Nicaragua, site 61; E, Panama, site 86. NUMBER 13 113 FIGURE 100.-Type B thorns around main trunk of A. collinsii: A, Panama, site 88; B, Costa Rica, site 67. to converge on those of A. coriiigera or A. hindsii, A. colliiisii is not sympatric with those species where this occurs. 'Cl'hile the very largest type B thorns to be found on A. collinsii (Figure 104) have ii large volume and are therefore similar to the largest on A. hiiidsii, in general both type A and type 13 thorns of A. colliiisii are small in volume when compared with swollen-thorn acacias in general. 12. Acacia hindsii Bentham Acacia hindsii Bentham, 1842:504. Acacia bwsaria Schenck, 1913:363. Amcin tepicnna Safford, 1914:366. Acacia sinaloemis Safford, 1914:365. Myrrttecodendron lzindsii (Bcntham) Britton and Rose, 1928: 91. A mature Acacia hindsii is distinguished from all other swollen-thorn acacias by its very flat and dis- tended thorns. On occasion the thorns of A. collinsii display superficially similar distension, but A. hindsiz has small diameter petiolar nectaries rather than broad dome-shaped ones as on A. collinsii. A. Iiiizdsti, A. collinsii, and A. cornigera are the three most widely distributed swollen-thorn acacias. A. hiiidsii, however, is essentially a plant on the Pacific lowlands and foothills (Figure 105). 1 was unable to examine the type specimen of A. hindsii, but A. hindsii is the only species of swollen- thorn acacia in the Nanzanillo Bay area (site l), and the t) pe description agrees closely with the swollen-thorn acacia population in that area. The northernmost record of A. collinsii (A. collinsii site 7) is from slightly to the south of site 1, but at this site it has cylindrical thorns and is not sympatric with A. hindsii. A. hindsii is not to be confused with its hybrids with nonant-acacias such as A. pennatula, e.g., A. oaxacanum, A. hirtipes, and A. 114 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 101.-Geographic variation in type B thorns of occupied adult A. collinsii: A, B, Mexico, site 4; c, Nicaragua, site 51; D, Costa Rica, site 67; E, Mexico, site 11. FIGURE 102.-Geographic variation in type B thorns of occupied adult A. collinsii: A, Costa Rica, site 67; B, Nicaragua, site 51; c, Mexico, site 4. NUMBER 13 140 - 130 - 2120 - g110 - 2 100 - 2 90- 6 X z W 4 I 0 X I- I- 0 I- 115 FIGURE 103.4eographic variation in type B thorns of occupied adult A. collinsii: A, Panama, site 82: B, Mexico, site 11: c, Guatemala, site 38. c 1- I I TOTAL THORN VOLUME (cc) 2 4 6 8 10 FIGURE 104.-Distribution of thorn volume among the largest type B thorns of A. collinsii (Costa Rica, site 67). standleyi. None of the hybrids has a full comple- ment of Beltian bodies, large nectaries or fully de- veloped swollen thorns. Shenck's description of the type of A. bursaria applies without doubt to the A. hindsii population at site 49, rather than to A. cornigera, which is the only other swollen-thorn acacia at the type locality. The thorns on the type specimen of A. bursaria are exceptionally large type A thorns, of the kind often found on large and very healthy A. hindsii in the wetter parts of Guatemala. The vegetative branch chosen for the type speci- men of A. tepicana is a representative young sucker shoot or central axis from a young seedling before it became occupied by obligate acacia-ants. The thorns are not flattened, but this is the case with most young A. hindsii. Their very light color is also representative of young A. hindsii in areas where the adults have very dark colored thorns. The peduncle with mature fruits accompanying the type specimen was taken from a representative mature A. hindsii. A. hindsii is the only swollen- thorn acacia found as far north as Acaponeta in western Mexico (site 2). Safford's type specimen for A. sinaloensis is a representative branch from the upper crown of a normal mature A. hindsii of the form common at the type locality (site 3). Type A thorns of all de- grees of flattening occur in such a crown and the 116 SMITHSONIAN CONTRIBUTIONS TO BOTANY FICIIRF. 105.-Postulated general distribution (stippled area) and collection localities of A. hindsii. type specimen has thorns representing the smaller end of the spectrum. The seed pod accompanying the type is a normal A. hiizdsii pod. A. hindsii is the only swollen-thorn acacia found as far north as Villa Union, Sinaloa, Mexico. KEY COLLECTION LOCALITIES Mexico 1. Type of A. hindsii: Shore of Manzanillo Bay, Colima, Mexico (sea level). 1836-39 (or 1841; Sir George Taylor, in litt.), R. C. Ilinds, no. 248. 2. Type of A. tepicana: Acaponeta, Nayarit, Mexico (30 m). IV-10-1910. J. N. Rose, P. C. Standley, and P. G. Russell, no. 14357. 3. Type of A. sinaloensb: Villa Union, Sinaloa, Mexico (10 m). IV-2-1910, J. N. Rose, P. C. Standley, and P. G. Rus- $ell, no. 13972. 4. Arriaga to 13.2 mi N on Hgy. 200, Chiapas (10-550 m) . 1-9-1965, DHJ, nc (common). 5. Tehuantepec to 21.2 mi E on Hgy. 190, Oaxaca (10-30 m). 1-10-1965, DHJ, no. 1041; IX-2-1965, DHJ, nc; VIII-7-1966, DHJ, no. 415 (occasional to rare). 6. Near Salina Cruz, Tehuantepec, Oaxaca (sea level). IX- l-1965, no. 1934 (occasional to common). 7. Tapilula to 3 mi E, Chiapas (700 in). VII-10-1966, DHJ, nc (occasional). 8. 12.2 mi SW El Camaron to 10.2 mi W turn into Tex- quisistlan on Hgy. 190, Oaxaca (925-735 m). 1-10-1965, DHJ, nos. 1033, 1034. 9. Puerto Escondido to 36.9 mi N on Hgy. to Oaxaca, Oax- aca (sea level to 940 m). VI-19-1967, DHJ, nc (occa- sional to ra:e) . 10. Las Cruces (just inland from Acapulco) to Puerto Es- condido on Hgy. 200, Gucrerro-Oaxaca (20-300-10 m) . VI-18-1967, DHJ, nc (common to occasional). 11. 14.0 mi S Chilpancingo to Las Cruccs, Guererro (1175-30 m) . 1-7-1965. DHJ, nos. 1031, 1035; DHJ, nc (common). 12. Playa de Pie de la Cuesta (just W of Acapulco) to 13.6 mi SE Papanoa on Hgy. 200, Guererro (sea level). VI- 17-1967, DHJ, nc (common to very rare, only along up- per edges of swamp sites). 13. 21.G mi SW Autlan to Barra de Navidad, Jalisco (450-0 m). 1-5-1965, DHJ, nos. 1026, 1027, 1028, 1029, 1030 (common to rare, only in wetter sites) . 14. Barra de Navidad to Tecoman, Jalisco-Colima (50-0 m.) 1-5-1965, DHJ, nc (rare-common) . 15. 12 mi NW Tepic to 28.4 mi SE Mazatlan on Hgy. 15, NUMBER 13 117 Nayarit-Sinaloa (700-5-70 m). IX-14-1963, DHJ, no. 1745; 1-2-1965, DHJ, no, 1011; VI-8-1967, DHJ, nc (rare to common to rare) , fl. 16. Hgy. 12 to San Blas on Hgy. 46, Nayarit (200-0 m). I- 13-1965, DHJ, nos. 1019, 1020, 1021, 1022, 1023, 1024, 1025. 17. 23.5 mi SW El Palinito to 9.0 mi SW, Concordia on Hgy. 40 (1089-10 m). 1-2-1965, DHJ, nos. 1005, 1006, 1012, 1013, 1014, 1015, 1016, 1017; VI-6-1967, DHJ, nc (com- mon to rare). 18. Palo Blanco, San Ignacio, Sinaloa (340 m). IV-18-1918, M. N. Montes and H. E. Salazar, no. 286 (US 1035197). 19. Puerto Angel, Oaxaca (sea level), IV-12-1933, C. V. Morton and E. Makrinius, no. 2614 (US 1585606), fl. 20. Cafetal Concordia, Oaxaca (400-650 m) . IV-15-1933, C. V. Morton and E. Makinius, no. 2427 (US 1585440), fl. 21. Pochutla, Oaxaca (50 m) . IV-19-1917, no. 3146, C. Con- zatti (US 763852), fl. 22. Huatulco, Oaxaca (150 m). 11-1941, B. P. Reko, no. 6010 (F 1510963), fl. 23. 13.7 mi NE La Mira to Playa Azul and to Melchor Ocampo on Hgy. 37, Michoacan (200-0 m). VI-12-1968, DHJ, nc (common). 24. Guatemala-Mexico border (near Ciudad Cuauhtemoc) to 3.6 mi N on Hgy. 190, Chiapas (720-680 m). VIII-6- 1966, DHJ no, 416; VIII-16-1967, DHJ, nc (common to rare). 25. 5.1 mi SW Cintalapa (Chiapas) to 20.9 mi W Tapana- tepec (Oaxaca) on Hgy. 190 (610-820-110 m) . VIII-18- 1967, DHJ, nc (common to rare). 25a. 23.3 mi S Pichucalco, on Villa Hermosa-Chiapa de Corzo road, Chiapas (1000 m) . VII-10-1966, DHJ, no. 420, 564. 26. 29.3-35.0 mi W Tapanatepec, Oaxaca (100-20 m) . VIII- 18-1967, DHJ, nc (occasional). 27. 6.8-4.1 m S of turn into Matias Romero, on Hgy. 185, Oaxaca (230 m) . VIII-18-1967, DHJ, nc (3 plants). 28. Guatemala-Mexico border (Pte. Talisman) to Arriaga on Hgy. 200, Chiapas. 1-10-1965, DHJ, nc (common). Guatemala: 29. 6.8 mi SW (1370 m) Guatemala City to Escuintla, Depto. Esquintla on Hgy. 2. 1-12-1965, DHJ, no. 1050; VIII-12- 1967, DHJ, nc (common), 30. Shores of Lake Amatitlan and surrounding hills (1300- 1350 m). Depto. Guatemala. VIII-11-1967, DHJ, nc (common). 31. Escuintla, Depto. Escuintla (330 m) to Guatemala- Mexico border (Pte. Talisman), Depto. San Marcos (50 m) , 1-12-1965, DHJ, nos. 1048, 1049 (rare to common). 32. Retalhuleu SW to Champerico, Depto. Retalhuleu (200- 0 m) . VIII-1-1966, DHJ, nc (occasional to rare). 33. Retalhuleu to 15.2 mi NE on Hgy. 9S, Depto. Retalhuleu (200-1000 m). VII-18-1966, DHJ, no. 400; VII-16-1966, DHJ, no. 404; VII-28-1966, DHJ, nc (common to rare). 34. 16.5 mi SE to Guatemala-Mexico border (Ciudad Cuauhtemoc), Depto. Huehuetenango (900-1000-720 m) on Interamerican Hgy. VIII-16-1967, DHJ, nc (com- mon). 35. 9.5 mi SW to 6.2 mi NE Puente El Lob0 on Hgy. 4, Depto. Zacapa (150 m). V-14-1965, DHJ, nos. 1929, 1930; V-20-1965, DHJ, no. 1931; VIII-9-1967, DHJ, nc (very rare except as clumps on river banks). 36. Guatemala-El Salvador border (25 m; Pijije) , Depto. Jutinpa to Escuintla (330 m), Depto Escuintla on Hgy. 6E. 1-13-1965, DHJ, nc (rare to common), fl. 37. Guatemala.El Salvador border (San Cristobal), Depto. Jutinpa to 15.8 mi SE Guatemala City on Hgy. 2, Depto. Guatemala (520-1250 m) , VII-7-1967, DHJ, nc (common to rare to common). 38. Tikal, Depto. Peten (100-500 m). IV-2-1922, 0. F. Cook and R. D. Martin, no. 170 (US 1084640), mp. 39. San Carlos, Depto. Retalhuleu (650 m) . 111-22-1921, Rojas, no. 241 (US 1014948), fl. 40. Santa Barbara, Depto. Solola (400 m) . VIII-1891, W. C. Shannon, no. 578 (US 943298). 41. Pancajche, Depto. Alta Verapaz (360 m). IV-5-1939, P. C. Standley, no. 70648 (F 978766), fl. 42. narranca Hondo, lower slopes Volcan de Fuego, Depto. Sacatepequez (1800 m) , XII-16-1938, P. C. Standley, no. 60262 (F 994870), gp. 43. Ocos, Depto. San Marcos (sea level). 111-15-1940, J. A. Steyermark, no. 37779 (F 105266). 44. Between Finca Pirineos and Finca Soledad, lower south- facing slopes of Volcan Santa Maria, between Santa Maria and Calahuache, Depto. Quezaltenango (1300- 1400 m). 1-5-1940, J. A. Steyermark, no. 33599 (F 1051210). 45. Near Calahuache, Depto. Quezaltenango (1020 m) . III- 1-1939, P. C. Standley, no. 67107 (F 987673), mp. 46. Rio Chiquimula, 1.5 mi NE Chiquimula, Depto. Chiqui- mula (400 in). X-21- 1939, J. A. Steyermark, no. 30123 (F 1038583). 47. Finca El Porvenir, Rio Cabus, SW slope Volcan Taju- mulco, Depto. San Marcos (1000-1300 m) . 111-12-1940, J. A. Steyermark, no. 37550 (F 1049161). 48. Along Rio Gaucalate, Depto. Escuintla (500 m). XI-28- 1938, P. C. Standley, no. 58188 (F 993751), gp. 49. Type of A. bursaria: San Felipe, Retalhuleu, Guatemala (300 m). February, 1878, Bernouilli and Cario, no. 1129 or Laguna Amatitlan, Depto. Guatemala, Guatemala (1100 m). 1890, J. D. Smith, no. 2304 (Sir George Taylor, in litt.) . Honduras: 50. 13.1 mi W San Marcos de Colon (850 m) to 11.6 mi NE Choluteca (330 m) on Interamerican Hgy., Depto. Cho- luteca. 1-14-1965, DHJ, no. 1053; V-11-1965, DHJ, nos. 1928, 1933; VII-29-1967, DHJ, nc (rare to common to rare). 51. 9.4 mi NE Choluteca on Interamerican Hgy., Depto. Choluteca (300 m) . VII-29-1967, DHJ, nc (1 plant). 52. 4.3 mi NE Jicaro Galan, Depto. Valle (70 m) to 21.6 mi S Tegucigalpa, Depto. Francisco Morazan (1000 m) on Hgy. 1. VII-30-1967, DHJ, nc (rare to common to.rare). 53. Honduras-El Salvador border (El Amatillo) to 15.5 mi E on Interamerican Hgy., Depto. Valle (50-100 m). I-15- 1965, DHJ, nc (very rare). 54. 13.1 mi W Zambrano, Depto. Francisco Morazan (830 m) to 7.8 mi SE Villa de San Antonio, Depto. Comayagua 118 SMITHSONIAN CONTRIBUTIONS TO BOTANY 55. 56. 57. 58. 59. 60. 61, 62. 63. 64. 65. 66. 67, 68 (750 m) on Tegucigalpa-Puerto Cortes Hgy. VII-31-1967, DHJ, nc (common). 7.8 mi SE to 4.7 mi N (550 m) Villa de San Antonio on Tegucigalpa-Puerto Cortes Hgy., Depto. Francisco Mora- zan. VII-31-1967, DHJ, nc (occasional to very rare). 5.3 to 6.3 mi NW Comayagua on Tegucigalpa-Puerto Cortes Hgy., Depto. Comayagua (510 m). VII-31-1967, DHJ, nc (rare). 13.2 mi NW Comayagua on Tegucigalpa-Puerto Cortes Hgy. (850 m). VII-31-1967, DHJ, nc (1 plant). 10.7 mi NW (820 m) Siguatepec, Depto. Comayagua to 16.4 mi S (50 m) San Pedro Sula, Depto. Cortes on Tegucigalpa-Puerto Cortes Hgy. VIII-2-1967, DHJ, nc (common to occasional). 4.6 mi SW (90 m) to 12.9 mi SW (150 m) San Pedro Sula, Depto. Cortes on road to Santa Rosa. VIII-2-1967, DH], nc (occasional). 24.4 mi SW (240 m, bank of Rio Chamelecon) San Pedro Sula, Depto. Cortes to 10 mi N Santa Rosa. VIII-2-1967, DHJ, nc (occasional to very rare). 4.4 mi SW Santa Rosa, Depto. Copan (940 m) , VIII-3- 1967, DHJ, nc (1 plant). 4.4 mi NE (1310 m) Nueva Ocotepeque to Honduras-El Salvador border (850 m) on Santa Rosa-San Salvador Hgy., Depto. Ocotepeque. VIII-4-1967, DHJ, nc (rare to common). Amapala, Isla Tigre, Depto. Valle (less than 250 m) . II- 14-1922, P. C. Standley, no. 20738 (US 1136566, US 1136557), fl. Siguatepec, Depto. Comayagua (1080-1400 m). II-14-27- 1928, P. C. Standley, no. 56333 (US 1409186), fl. Las Limas, Depto. Comayagua (1000 m) . IV-16-1932, J. B. Edwards, no. P-89 (F 713842), fl. Vicinity of Siguatepeque, Depto. Comayagua (1080-1400 m) . 11-14-27-1928, P. C. Standley, no. 56333 (F 581757), fl. La Lima, Depto. Cortes (30 m). IV-11-20-1947, P. C. Standley and J. Chacon, no. 7548 (F 1268590), fl. Near Lancetilla, Depto. Atlantida. VIII-10-1934, T. G. Yuncker, no. 5099 (F 749502). El Salvador: 69. 70. 71. 72. 73. 74. Honduras-El Salvador border (850 m) to 10 mi S (1170 m), Depto. Chalatenango on Santa Rosa-San Salvador Hgy. VIII-4-1967, DHJ, nc (occasional to very rare). 11.7 mi S (1130 m) Honduras-El Salvador border (Depto. Chalatenango) to 9.3 mi N San Salvador (580 m) on Santa Rosa-San Salvador Hgy. VIII-4-1967, DHJ, nc (rare to very common to rare). Intersection just E of La Libertad (sea level) to 11.8 mi N on Hgy. to Interamerican Hgy. (910 m), Depto. La Libertad. VIII-5-1967, DHJ, nc (rare to very common). La Libertad, Depto. La Libertad to La Union, Depto. La Union (sea level to 30 m) on La Libertad-Zacatecoluca- Usulutan-La Union Hgy. VIII-6-1967, DHJ, nc (common to occasional to rare), La Herradura to La Libertad-Zacatecoluca Hgy., Depto. LaPaz. VIII-6-1967, DHJ, nc (very rare) . 41.5 mi SE San Miguel to La Union on San Miguel-La n? 13. 76. 77. 78. 79. 80. 81. 82. 83. 84. Union Hgy., Depto. La Union. V-12-1965, DHJ, no. 1932; 1-15-1965, DHJ, nc (rare), La Libertad, Depto. La Libertad to El Salvador-Guate- mala border (La Hachadura) , Depto. Ahuachapan. 1-16 1965, DHJ, no. 1051 (common to rare). Sonsonate (280 m) SE to Interamerican Hgy#, Depto Sonsonate. VIII-6-1967, DHJ, nc (common to accasional). Sonsonate through Armenia to San Salvador-Guatemala Hgy. (280-920-450 m). VIII-6-1967, DHJ, nc occasional to rare). Intersection of San Salvador-Guatemala City Hey., with Sonsonate-Armenia Hgy. to Guatemala-San Salvador bor- der (San Cristobal) , Depto. Santa Ana (450-810-520 m) . VIII-6-1967, DHJ, nc (very rare to common to occa- sional) , San Martin de Laguna Ilopango, Depto. San Salvador. IV-1-1922, P. C. Standley, no. 22622 (US 1138360), fl. 2 km N7V Monte Cristo, 15 km XE San Miguel, Depto. Morazan (150 m). XII-7-1941, J. iM. Tucker, no. 486 (US), mp. Laguna de Olomega, Depto. San Miguel (75 m). II-20- 1922, P. C. Standley, no. 21027 (US 1136842). San Vicente, Depto. San Vicento (350-500 m) . III-2-11- 1022, P. C. Standley, no. 21291 (US 1137447), fl. Ahuachapan, Depto Ahuachapan (700-1100 m) . I-16- 1925, P. C. Standley and E. Padilla, no. 2497 (F 1256801). 2 km NW Monte Cristo, 15 km NE San Miguel (150 m) . XII-7-1941, J. M. Tucker, no. 486 (F 1492748). Nicaragua: 85. 3.4 mi SE Leon on Hgy. 12, Depto. Leon (20 m). VII- 27-1967, DHJ, nc (30 acre patch), gp. 86. 4.9 mi SE Leon on Hgy. 12, Depto. Leon (20 m). VII-27- 1965, DHJ, nc (1 plant). 87. Ameya (near Pto. Morazan) Depto. Chianandega (sea level). VI-19-21-1923, W. R. Maxon et a!., no. 7176 (US 1180890). NATURAL HISTORY Being widespread geographically (Figure 105) on soils and in climates very suitable to seasonal agri- culture, A. hindsii is difficult to characterize ecologically since nearly all the semideciduous and deciduous forests that it occupied before the advent of European agriculture have been destroyed. As McVaugh (1966) has stressed, when both of these forest types are cleared, A. hindsii is a common member of the shrubby regenerating vegetation. In contrast to those swollen-thorn acacias that live in natural disturbance sites in wet or rain forests (A. allenii, A. melanoceras, perhaps A. mayana), the A. hindsii population density has very likely in- creased greatly since the advent of European agri- culture, except in those areas where Indian slash- burn agriculture was common on a system with at NUMBER I3 119 least ten years fallow per field. Based on examina- tion of A. hindsii throughout its contemporary range, I suspect that its pre-Colombian ecological distribution was the following. The plant was ex- tremely common along rivers and mangrove swamps in deciduous and seinideciduous forests and formed large thickets where occupied by multiple-queen species of obligate acacia-ants (Janzen, 1967a, 1973). Scattered large trees surrounded by sucker shoots from lateral roots occurred on dry hillsides, be- coming progressively rarer the more intense and the longer the tlry season. Along the wetter margins of its range (usually higher as well) it was found only along watercourses and other natural dis- turbance sites. In the very driest parts of its range (e.g., where the Tropic of Cancer crosses western Mexico), it occurred only along watercourses and in the foothills where the slightly cooler weather compensated for a lack of rain equal to that in the well-drained parts of the coastal plain. In no places did it occur as a shrub in the semidesert conditions as A. collinsii may do on rare occasions. In short, A. hiizdsii was a plant of locally wet sites in the drier habitats of the Pacific side of Central America between Nicaragua and the Tropic of Cancer. Throughout this range it invaded wet sites, as wet as those occupied by A. gentlei, A. cookii, A. may- ana, A. globulifera on the Atlantic coast. Like these four swollen-thorn acacias, it was restricted to natural disturbance sites such as river banks in the wetter habitats. The wetter sites on the Pacific side, however, tend to experience a much more severe tlry season than do sites on the Atlantic side with roughly equal annual rainfall. That the plant had to be able to survive this dry season, and thus could move into drier sites fairly readily, had been in- strumental in preventing the very widespread popu- lation of A. hindsii from differentiating into many species. In other words, drying of the general weather would not be so likely to split that popula- tion of A. hindsii into as many small and isolated populations as it would a species widely distributed along the wetter Atlantic side of Central America. The restriction of A. hindsii to the Pacific side of Central America is very puzzling. It has been collected east of the Continental Divide at only 7 sites, and 5 of these (sites 7, 27, 35, 55, 56) could well be the result of seed dispersal via road building machinery; at these sites the population is restricted in the area within a mile of the highway. At Tikal and Lancetilla (sites 38 and 68) there has been much traffic by biologists from other parts of Cen- tral America, and the collections could well be recent introductions. A. hiizdsii is certainly not the common swollen-thorn acacia at these sites; how- ever, there were reproducing trees at each of the 5 sites that I examined, and if they are not very recent introductions, it is most enigmatic that they have not spread from these sites over much of Caribbean Central America. Even if these 7 sites represent recent introductions, it is puzzling that the many natural introductions that must have occurred in the p.ast have not resulted in established popula- tions of A. hindsii on the Atlantic side of Central America. Being bird dispersed, the seeds surely must have been carried across the Continental Di- vide. Since isolated greenhouse plants of A. hindsii never set seed despite the fact that they flower profusely, it is likely that they are self-sterile. Suc- cessful establishment across the Continental Divide would then be exceedingly difficult except with simultaneous multiple-seed dispersal as could easily occur via the mud caked on road-building equip- ment. Low-income workers on the roads also eat the sweet pulp around the seeds and may have dis- persed some seeds in this manner. The dependency of A. hindsii on obligate acacia- ants for survival to reproductive maturity appears no different from that of A. cornigera, though ex- periments have not been conducted. Naturally un- occupied plants of A. hindsii are heavily damaged by herbivorous insects except under two circum- stances: one natural and the other artificial. At the upper elevation limits of their distribution (e.g., sites 8, 37) in windy cool habitats, it appears that herbivorous insect pressure is low enough so that only during the warmest months of the year are unoccupied A. hindsii subject to severe defoliation. A. hindsii is one of the small number of species of woody second-growth plants that has managed to survive along the well-traveled dirt roads on low- land coastal El Salvador (site 72). These roads lie between cotton fields that are frequently sprayed with parathion and other very toxic insecticides. Not only have all the insects been eliminated from the roadside strips of vegetation by the combination of dust and insecticides, but the colonies of obligate acacia-ants have been gone so long that a large A. hindsii can show no sign of ever being located by searching queens. The thorns only stay on the 120 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 106.-Acacia kindsii: A, unoccupied 1-year-old sucker shoot (Mexico, site 6); B, unoccupied 2-year-old seedling (Mexico, site 9); c, occupied 2-year-old sucker shoot (Honduras, site 59); D, lateral branch from crown of occupied adult (Mexico, site 6). NUMBER 13 121 acacia for a maximum of about 4 years, so this means that the ants have been missing at least that long. Since these fields are 10 to 30 years older than that, I suspect that the ants and most other insects have been missing equally long. The dust along other Central American roads is generally not as bad, since those with as much traffic as El Salva- doran roads are usually paved. Despite the absence of herbivorous insects, the A. hindsii do very poorly because they are frequently used as vine standards by wild plants. At site 85 in Nicaragua, the south- ernmost local population of A. hindssii displays an extreme case of insecticide protection. This 30-acre patch of nearly pure A. hindsii (with scattered A. FIGURE 107.-~, B, Acacia hindsii (Mexico, site 6) : A, pair of 2-m-tall subadult occupied sucker shoots; B, adult 4-m-tall occupied tree, with the life form characteristic of open pastures; c, brushy pasture with a 3-m-tall occupied A. hindsii on the left and A. pennatvla at right of center (Guatemala, site 37). 122 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE lOS.-Acacia hindsii: A, looking up through the crown of an occupied adult (Mexico, site 17) ; B, fully leafed crowm of an occupied subadult (Guatemala, site 35). cornigera) was almost certainly established from seed brought by man and is an island in the center of many hundreds of acres of cotton. The acacias are of all ages antl many are reproducing. Three species of obligate acacia-ants are in the foothills 25 miles east, but after searching for 2 hours at site 85 I could find no evidence that any searching queen had located the patch during the past 4 years. If any did, they were killed by insecticide before they could attempt to cut an entrance hole in a thorn. There were almost no other insects in this patch of vegetation, antl almost every plant had no herbivore damage; cotton in this area is almost uneconomical to grow because of the severity of insect damage. Seedlings of A. hindsii in natural disturbance sites, and in brushy pasture, tend to attain a height of 10 to 40 cm over 1 to 3 years of growth unoc- cupied by obligate acacia-ants. Such unoccupied seedlings are very stunted in appearance (Figures 106~, B) with damaged leaves and many short branches that grew a few cm before having the shoot tip eaten off by an insect. In contrast to the heavily shaded unoccupied seedlings of A. allenii and A. melanoceras, whose slow growth is as much from a lack of sunlight as defoliation, the slow growth of unoccupied A. hiudsii seedlings can be attributed almost entirely to defoliation by insects. It appears that somewhat less than 1 percent of the seedlings of A. hindsii ever become occupied; most die within the first 4 years. Cursory examination of habitats where they are common indicated that the fully insolated unoccupied seedlings live longest, despite the dryness of the site. This indicated that once established, energy relationships during the wet season are more important than desiccation during the dry season. Once occupied by a single-queen colony, the young seedling often produces a small intact bush (Figure 1OGc) with multiple branches but quickly one of the vertical branches becomes the central trunk (Figures 107~, c). The central trunk grows rapidly upward; there may be 1 to 2 m of vertical elongation during a 5-month iainy season. During the second and third years of occupied growth, the crown begins to fill in with small branches (Figures 107c, 108~) if not under strong lateral competition from the crowns of other plants. If there is much adjacent woody vegetation, the vertical elongation KUMBER 13 123 of the central trunk may continue until the tree is 6 to 12 m tall. Lateral branches are long and bear large leaves and swollen thorns (Figure 106~). Old A. hindsii almost invariably have dense crowns (Figures 107~, 108~). The development described for A. hindsii is very similar to that for A. cornigera. In areas where niultiple-queen species of obligate acacia-ants are primary occupants of A. hindsii, the development is somewhat different. Here, most new FIGURE 109.-A clump (clone?) of adult A. hindsii occupied by a multiple- queen species of obligate acacia-ant (Mexico, site 16). plants are sucker shoots from lateral roots of older acacias and are occupied by part of the colony in the parent shoot. This produces a dense thicket (Figure 109) of A. hindsii that can cover as much as several acres. In a new habitat, where the A. hirtdsii is becoming established through seeding, the early stages of shoot development are not much different from the case with single-queen obligate acacia-ants, but within a few years, the thicket be- 124 40 - 35 - - v 2 30- SMITHSONIAN CONTRIBUTIONS TO BOTANY W u z W V V) W E 0 25' I+ 8 0 0 n 0 n .... 0 0 0. 0 0 0 0 PI, , , , W cl 150 200 250 300 NUMBER OF FLORETS PER INFLORESCENCE FIGURE 110.-Distribution of 4314 florets among 20 inflores- cences from one adult A. hindsii (Mexico, site 5). gins to form. The multiple-queen situation prevails in areas with particularly severe dry seasons coupled with wet seasons very favorable to acacia growth, as in western Mexico. Acacia hindsii does not bear flowers until the fifth to seventh year after becoming occupied, and v) W 40 u v) W 0 I4 a 30 k 20 a 10 W B 2 1234 6 many trees occupied by multiple-queen species do not produce their largest seed crops until they are 10 to 30 years old. In contrast, A. cornigera and A. collinsii bear flowers in the fourth to sixth year of occupied growth and have their heaviest seed crops when 6 to 15 years old. The inflorescences of A. hindsii, like those of other dry-land acacias, are visited by many species of small wild bees. The inflorescences vary greatly in length, but not so greatly in the number of florets per inflorescence, (Figure 110) , but a maximum of 8 pods per inflo- rescence (Figures 11 l~, 112~) is observed in most pod crops. When more than 2 pods develop on a single inflorescence, they become so tangled when they dehisce (Figure 14~) that not all the seeds are ex- ternally visible to the birds that disperse them. The number of seeds per pod (Figure 111~) is repre- sentative of acacias in general and swollen-thorn acacias in particular. As with other dry-land acacias, these seeds are heavily attacked by bruchid beetles. The swollen thorns of A. hindsii show very high variation in volume, as do those of A. collinsii. The volume of type A thorns (Figure 113) tends to be about the same throughout the range. In areas where there are only multiple-queen ant colonies, all thorns are type A and rather monotonous in form. Where single-queen colonies are present, how- ever, and A. hindsii is represented primarily by well-defined individual trees, there is great varia- tion in thorn volume. The volume for some of the largest type B thorns encountered are graphed in 6 I4 0 62 B E A NUMBER OF PODS PER INFLORESCENCE B NUMBER OF SEEDS PER POD FIGURE 111.--Acacia hindsii (Guatemala, site 29): A, distribution of 187 mature seed pods among 100 inflorescences on on,e tree; B, distribution of 268 seeds among 22 of the mature pods. NUMBER 13 125 FIGURE 112.-Infructesccnccs with immature pods: A, A. globulifera (Guatemala, sitc 13) ; R, A. hindsii (Honduras, site 59). Figure 114, and in contrast to the large volumes for the dune ecotype of A. sphncrocephala, the large space is all useful to the ants because the thorn is flattened such that the thorn walls are only 2 to 4 mm apart. The thorns of A. hiiidsii are highly variable in shape, as are those of A. cornigera and A. collinsii. Almost all seedlings have thorns that are cylindrical or only slightly flattened in cross section (e.g., Figure 106~). Thorns on young sucker shoots are the same (Figure 115~, middle). This difference between young and old thorns is probably asso- ciated with maximizing thorn volume and wall strength with a fixed small amount of energy to be expended on the thorn. Type A thorns in the crowns of mature A. hindsii show strong geographic variation (Figures 82~, 83c, 115, 116, 117, 118~, C) but have low variation at any particular locality. Their geographic variation in morphology is as great as that displayed between different species of more geographically restricted swollen-thorn acacias. - 130 E B 3: I-. 0 z 1 120 5110 g 100 2 LL: I I I I 1 1 I 0.8 1.0 1.2 1.4 1.6 1.8 2.0 TOTAL THORN VOLUME (cc) FIGURE llS.-Distribution of thorn volume among type A thorns from rcprescntative latcral branchcs in the crown of an adult occupied A. hintlsii (Mexico, site 17). 126 170 - h 8 E X b W z w cl - 150- E 130- 0 X + cl 0 b ;s 110- SMITHSONIAN CONTRIBUTIONS TO BOTANY I I I I 150 h Y n < 110- k 0 ** 8. B FIGURE 114.-A. hindsii (Mexico): A, distribution of 84 cc of thorn volume among 17 of the largest type B thorns on an occupied adult (site 9); B, distribution of 62 cc of thorn volume among 11 of largest type B thorns on an occupied adult (site 11). FICIRE ll5.-Geographic variation in typc A thorns from occupied adult A. hindsii: A, Mexico, site 17; R, Guatemala, site 33; C, Honduras, site 50; D, Guatemala, site 25A. NUMBER 13 I27 FIGURE 116.-Geographic varia- tion in type A thorns from oc- cupied adult A. hindsii: A, Mexico, site 11, B, Mexico, site 13; c, Guatemala, site 33. FIGURE 117.-Geographic var- iation in type A thorns from occupied adult A. hindsii: A, Mexico, site 8; B, Guatemala, site 35; c, Mexico, site 13; D, Guatemala, site 35. 128 SMITHSONIAN CONTRIBUTIONS TO BOTANY FIGURE 118.-Acacia hindsii (Mexico): A, type B thorns on short lateral branches from main trunk of an occupied adult (site 9) : B, white type A thorns on main trunk and lateral branches of occupied subadult (site 17); c, black type A thorns on main trunk and lateral branches of occupied subadult (site 17). Much of the variation is probably adaptive in the context of the developmental physiology of the lateral branch, rather than the ant colony. It is possible that the details of the thorn shape are nonadaptive, but unlikely, since so much of it is clinal in nature. The production of type B thorns appears highly adaptive in the context of multiple- queen colonies; the single-colony queen lives in the type B thorns. There is, however, a great deal of geographic variation in type B thorn shape (Figures 118~, 119) that probably had the same basis as that displayed by type A thorns. NUMBER 13 FIGURE 119.-Geographic variation in type B thorns from occupied adult A. hindsii: A, B, F, Mex- ico, site 11; c, Guatemala, site 31; D, Mexico, site 13. E, Type B thorns of A. collinsii that are convergent in shape on those of A. hindsii. 129 130 Literature Cited SMITHSONIAN CONTRIBUTIONS TO BOTASY Allen, P. H. 1956. The Rain Forests of Golfo Duke. 417 pages. Uni- versity of Florida Press, Gainesville. 1842. Notes on Mimoseae, with a Synopsis of Species. The London Journal of Botany, 1:494-528. 1922. Ants in Their Diverse Relations to the Plant World. Bulletin of the American Museum of Natural His- tory, 45:333-583. 1969. Variation in Behavior among Obligate Acacia-Ants from the Same Colony (Pseudomyrmex nigrocincta) . Journal of the Kansas Entomological Society, 42:58- 67. 1928. Mimosaceae. North American Flora, 23:84-120. 1854. Primitiae Florae porlobellensis sive enumeratio plan- tarum vascularum quas juxta oppidum Portobello in isthmo panamensi Americae centralis mense Aprili anno 1826 legit Joh. Em. Billberg. Svenska Vetens- kapsakademlen, Handlingar, 42:105-148. Bentham, G. Bequaert, J. Beulig, M. L., and D. H. Janzen Britton, N. L., and J. S. Rose Buerling, P. J. Chodat, R., and W. Vischer 1920. La vegetation du Paraguay. Resultats scientifiques d?une mission botanique suisse au Paraguay. IX. Ur- ticiflores. Bulletin de la Societe Botanique Geneve 11 :226-258. 1967. Olmec Civilization, Veracruz, Mexico: Dating of the San Lorcnzo Phase. Science, 155:1399-1401. 1825. Prodromus Systematis A?aturalis Regni Vegetabilis. 2: 644 pages. Paris. Fiebrig, K. 1909. Cecropia peltata und ihr Verhaltnis zu Arteca alfari, zu Atta sexdens und anderen Insekten, mit einer Notiz uber .4nieisendornen bei Acacia cavenia. Bi- ologisches Zentralblatt, 29:l-16, 33-55, 56-77. 1970. Insect Associations with the Swollen Thorn Acacias. Transactions of the Royal Entomological Society of London, 122211-255. Coe, M. D., R. A. Diehl, and M. Stuiver de Candolle, A. P. Hocking, B. Janzen, D. H. 1966a. Coevolution of Mutualism between Ants and Acacias in Central America. Evolution, 203249-275. 1966b. La hormiga del ?Cornizuelo? en Veracruz. Znstituto Nacional de Investigaciones Agricolas, Mexico, Hoja Ilustrada, 4 pages. 1967a. Fire, Vegetation Structure, and the Ant x Acacia Interaction in Central America. Ecology, 48:26-35. 1967b. Interaction of the Bull?s Horn Acacia (Acacia corni- gera L.) with an Ant Inhabitant (Pseudomyrmex ferruginea F. Smith) in Eastern Mexico. University of Kansas Science Bulletin, 47:315-558. 1967~. Synchronization of Sexual Reproduction of Trees within the Dry Season in Central America. Evolu- tion, 21:620-637. 1968. Host Plants as Islands in Evolutionary and Con- temporary Time. American Naturalist, 102:592-595. 1969a. Allelopathy by Myrmecophytes: the Ant Azteca as an Allelopathic Agent of Cecropia. Ecology, 50: 147- 153. 1969b.Birds and the Ant x Acacia Interaction in Central America, with Notes on Birds and Other Myrmeco- phytes. Condor, 713240-256, 1969~. Seed-Eaters Versus Seed Size, Sumber, Toxicity, and Dispersal. Evolution, 23: 1-27. 1971. Cassia grandis L. Beans and Their Escape from Pred- ators in Time and Space. Ecology, 52:964-979. 1972. Protection of Barteria (Passifloraceae) by Pachysima Ants (Pseudomyrmecinae) in a Nigerian Rainforest. 1973. Evolution of Polygynous Obligate Acacia-Ants in Western Mexico. Journal of Animal Ecology, 42:727- Ecology, 53~885-892. -. 750. Ms. The Obligate Acacia-Ants of Central America. Linnaeus, C. 1753. Species Plantarum. 1: 560 pages. Stockholm. Lundell, C. L. 1937. The Vegetation of Peten. Publication 478 of Carnegie Institution of Washington. 244 pages. McVaugh, R., and J. Rzedowski 1966. La Vegetacion de Nueva Galicia. Contribution5 the University of Michigan Herbarium, 9: 1-123. Monod, T., and C. Schmitt he of 1968. Contribution a l?etude des pseudo.galles formicaires chez quelques Acacias africains. Bulletin de l?lnstitut Francais d?Afrique Noire, 30:953-lb12. Rehr, S. S., P. Feeny, and D. H. Janzen 1973. Chemical Defenses in Central American Non-Ant- Acacias. Journal of Animal Ecology, 42:405-416. 1968. Nuclear and Cytoplasmic Tubules in Corticle Cells of Leaf Beltian Bodies. Journal of Cell Biology, 38: 1969. Developmental Aspects of the Shoot Apex, Leaf, and Beltian Bodies of Acacia cornigera. American Journal of Botany, 56:195-200. 1971. Glycogen Plastids in Mullerian Body Cells of Ce- cropia peltata-a Higher Green Plant. Science, 173: 344-347. Rickson, F. R. 47 1-474. Robinson, B. L. 1913. V. Diagnoses and Transfers among the Spermato- phytes. Proceedings of the American Academy of Science, 49:502-517. Rudd, V. E. 1964. Nomenclatural Problems in the Acacia cornigera Complex. Madrofio, 17: 198-201. Safford, \V. E. 1010. Bull-Horn Acacias in Botanical Literature, with a Description of Two Sew Species. Science (A?.S.), 31: 1914. Acacia cornigera and Its Allies. Journal of the Wash- ington Academy of Science, 43356-368. 676-677. NUMBER 13 131 1915. Xew or Imperfectly Known Species of Bull-Horn Acacias. Journal of the Washington Academy of Sci- ence, 5:355-360. 1922. Ant Acacias and Acacia Ants of Mexico and Central America. Annual Report of the Smithsonian Znsti- tution . . . for 1921, pages 381-394. Sauer, J. 1967. Geographic Reconnaissance of Seashore Vegetation Along the Mexican Gulf Coast. Coastal Studies Znsti- tute, Technical Report, 56: 64 pages. Louisiana State University, Baton Rouge. 1913. Acaciae myrmecophilae novae. Fedde?s Repertorium, 12:370. Schenck, H. Schlechtendal, D. von, and A. von Chamisso Standley, P. C. 1830. Plantarum Mexicanum. Linnaea, 5:554-625. 1928. Flora of the Panama Canal Zone. Contributions from the United States National Herbarium, 27: 1-416. 1940. Studies of American Plants-X. Publications of the Field Museum of Natural History, Botanical Series, 22~65-129. Wheeler, W. M. 1942. Studies of Neotropical Ant-Plants and Their Ants. Bulletin of the Museum of Comparative Zoology, Harvard, 90: 1-262. 1806. Species Plantarum. Volume 4. Berlin. U?illdenow, C. L. $? U.S. GOVERNMENT PRINTING OFFlOE 1974-516-226/29