selves break down to form the secretion, differ from the salivary glands which secrete a liquid without undergoing disintegration.

5. Pith and Other Vegetable Tissues.-Dahl (1901) describes certain ants of the Bismarck archipelago and their larvæ as feeding on

FIG. 172. Cecropia adenopus (Schimper.) A, Tip of branch with leaves cut off; t, trichilia at base of leaf petioles; r, prostoma or depression in internode; x', stoma or opening to hollow internode made by Azteca muelleri at the prostoma. B, Longitudinal section of stem showing the hollow internodes and at (y) the septa perforated by the ants.

the pith in the twigs of Clerodendron, and von Ihering (1907) finds that Azteca muelleri eats the tissue that grows over the perforation through which it enters the hollow twigs of the Cecropia. There is, of course, no myrmecophilous adaptation on the part of the plants in these cases.

Turning now to the ants which are supposed to take advantage of the inducements offered by the plants, we find in both hemispheres many species that are very fond of wandering over the vegetation and visiting the nectaries, food-bodies, etc. In the tropics whole genera have become largely or exclusively arboreal, but this does not mean

that the food of the insects is obtained exclusively or even in great part directly from the plants, for, as will be shown in a future chapter, many ants visit plants mainly for the purpose of feeding on the excrement of the aphids, coccids, etc. In the Old World the exclusively, or at any rate, very largely arboreal genera are Ecophylla, Cataulacus, Sima and Polyrhachis, in the New World Azteca, Pseudomyrma, Cryptocerus, Myrmelachista and Allomerus, and in both hemispheres Dolichoderus, Camponotus, Cremastogaster and Iridomyrmex. The only forms, however, which are so exclusively arboreal as to show unquestionable structural adaptations to this habit, belong to the genera Azteca, Pseudomyrma, Sima and to Colobopsis, a subgenus of Camponotus. Concerning the habits of Azteca and Pseudomyrma Forel (1904) says: "The species of Azteca show very disparate conditions in the castes of the same species. Sometimes the head of the female is elongated, sometimes greatly broadened; and in like manner vary the proportions of the large and small workers. Emery was the first to call attention to this fact in his excellent monograph of the genus Azteca [1894k]. I believe that these differences are correlated with differences in habit. Just as the species of Eciton are the robbers of the soil in the primeval forest and the Atta species are the destroyers of the foliage of the neotropical woods, so the species of Azteca and Pseudomyrma are the true monarchs of the trees. To my knowledge, none of the species of Azteca and only one Pseudomyrma (P. elegans) nest in the ground. But what a varied arboreal existence is led by these little monkeys among the ants as they climb and scurry about everywhere on the trees! Some of them build carton nests on the trunks and branches, others nest in great cavities in the trunks; others (A. hypophylla) nest under the leaves of certain vines with these organs closely applied to the trunks, and close up any openings at the edges of the leaves with carton. Others, again, make use of the cavities of dead branches, while still others nest in the natural medullary cavities of living Cecropia trees or any hollow swellings or spaces in all kinds of plants. Finally Mr. Ule has discovered and described antgardens in which grow certain epiphytes that are sown by species of Azteca. Now I believe that the long, narrow head of the female and of the large workers of many members of this genus, as well as that

FIG. 173

Base of leaf petiole of Cecropia adenopus. (Schimper.) t, Trichilium, or hairy cushion in which the Müllerian bodies (m) are formed.

of many species of Pseudomyrma point to a life in very narrow, tubular branches and twigs. The small Azteca worker is small enough to enter and leave such openings without the great elongation of the head, which in the much larger queen is necessary for the accommodation of the powerful mandibular muscles. The nearly brainless and jawless male

FIG. 174. Carton nest of Azteca muelleri in the main trunk of Cecropia adenopus; Santa Catharina, Brazil. From a specimen in the American Museum of Natural History. (Original.)

does not require this adaptation. A broad, depressed head points to a life in much flattened cavities (A. hypophylla), etc. There are, to be sure, other differences in the form of the head (trigona and aurita, both cartonbuilders) that cannot be accounted for in this way." At least one species of Azteca (A. viridis) is green, a very unusual color in ants and evidently an adaptation to life. among living leaves. In Pseudomyrma species the whole body is greatly elongated. These are, in fact, the most slender of ants and anyone who has seen colonies of them filling narrow twigs and stems like so many sardines packed in a box, will be sure to regard the lengthening of the body as an adaptation to life in small tubular cavities. The species of the Old World genus Sima resemble the species of Pseudomyrma very closely in structure and habits. The soldiers of Colobopsis, as I have shown in a former

chapter (p. 211) have singularly truncated heads, adapted to fitting into and closing the perfectly circular entrances to the galleries of the nests which are always in wood or in the hollow culms of sedges. Perhaps the soldiers of many species of Cryptocerus and Cataulacus use their wonderful heads for the same or similar purposes.

The foregoing facts, it must be confessed, do not furnish a very solid foundation for the myrmecophily hypothesis that has been built upon them. At most they disclose considerable adaptability on the part of the ants, and a rather dubious or clumsy counter adaptation on the part of the plants. But the authors who would convince us that there is a definite symbiosis between such very different organisms, advance as their chevaux de bataille a few cases in which certain exquisitely arboreal ants live in a definite association with certain plants that present unusual structural characters. Such are the association of the Brazilian Asteca muelleri with Cecropia adenopus, that of the Malayan Iridomyrmex myrmecodia with Myrmecodia and Hydnophytum and that of Pseudomyrma with Acacia in Central and with Triplaris in South America. We must therefore consider these cases in somewhat greater detail.

The relations of Azteca muelleri to Cecropia adenopus have been studied by Fritz Müller (1876, 1880), Schimper (1888), and von Ihering (1891, 1907). The tree

known as the "imbauba" or "imbauva" belongs to the Urticacæ and is very slender and candelabrashaped, growing to a height of 1215 m. It is most abundant along the Brazilian littoral. The trunk and branches are hollow except at the nodes, where there are thin transverse septa (Fig. 172). The sap is colorless, not milky nor rubber-containing, as stated by some authors. The crown of foliage is meagre and consists of large, palmately lobed leaves. At some time. of its life each node bears a leaf, the long petiole of which has at its base a hairy cushion, known as the trichilium (Figs. 172t, 173t), in which the yellow Müllerian bodies (m) are imbedded. The cavities of older and larger trees are almost without exception tenanted by Azteca muelleri, which perforates the septa and thus causes all the internodal cavities to communicate with one another, both in the trunk and branches. The ants do not, however, live in the smallest, still actively growing twigs. The just-fecundated queen enters the branches while the tree

(Dahl.)