Studies on the Onychophora VII. The early embryonic stages of Peripatopsis , and some general considerations concerning the morphology and phylogeny of the arthropoda

There appear to be four different membranes round the eggs and embryos of the species of Onychophora at various stages of their development, and not two as previously supposed. The first membrane is left behind in the ovary, two further membranes occur in the stages of Peripatopsis here described, and an outer ‘shell’ is present in oviparous species. In Peripatopsis an inner cuticular membrane is absorbed at the end of segmentation, and an outer chitinous membrane persists until birth. These membranes influence the shape and volume of the embryos, and their properties and functions are described (pp. 489, 491, 527). The uterine eggs and embryos of Peripatopsis undergo two sudden increases in size. In P. sedgwicki, P. moseleyi, P. balfouri and P. capensis the unsegmented egg swells and then remains constant in dimensions until the end of segmentation. In P. sedgwicki, P. moseleyi, P. balfouri and P. capensis the absorption of the inner membrane is followed by another dilatation which expands the blastodermic vesicle to form the large yolk sac. In P. capensis this dilatation does not take place and no large yolk sac is formed, but the membrane dilates as in the former species. In P. balfouri the dilatation of the embryo is followed by a sudden contraction which causes the early elimination of the yolk sac. The embryo then resembles the corresponding stage of P. capensis, in both cases floating within very large membranes until the embryos grow to fill them (pp. 489, 491, 501, 504, 512, 529, 537). The unsegmented egg in Peripatopsis breaks up into a ‘first blastomere’ and a number of nonnucleated ‘cytoplasmic spheres’, all of which float freely in a watery fluid (pp. 496, 529). Segmentation of the ‘blastomere’ results in a disk of blastomeres lying in a single layer against the egg membrane on one side. These cells give rise to the whole of the embryo and not to the ectoderm alone. The disk enlarges and becomes saddle-shaped (pp. 498, 530). In P. moseleyi and P. sedgwicki the disk of blastomeres spreads all round the space within the membrane until its edges meet to form a continuous blastoderm. No blastopore is formed. The disintegrating remains of the cytoplasmic spheres lie in the internal space (p. 499). In P.capensis amd P. balfouri the edges of the disk of blastomeres curl away from the membrane, and ‘large vacuolated cells’ separate from these edges and pass to the concavity. A clear blastopore is present in P. capensis and a transitory or virtual one in P. balfouri. In P. capensis the blastopore narrows to a slit and finally closes. The large vacuolated cells within form an endodermal lining to an archenteron, which may at first be partially filled by free vacuolated cells (p. 503). In P. balfouri the hemisphere of small blastomeres completely closes round the vacuolated cells; these form a solid mass, and then degenerate. Definitive endoderm arises later from a blastoporal area, as in P. moseleyi and P. sedgwicki (p. 509). The small embryo of P. balfouri then dilates while that of P. capensis does not. The gastrula of P. capensis thus corresponds with the hollow single-layered blastodermic vesicle of the other species. The large vacuolated cells form definitive endoderm in P. capensis, they are formed but degenerate in P. balfouri, and they do not occur in P. moseleyi and P. sedgwicki (pp. 530, 535, 536). Evidence in support of the view that Peripatopsis is secondarily yolkless is provided by the details here presented concerning the form of the egg, segmentation, germ-layer formation, yolk sac, size changes, etc. (p. 535). A germinal disk is formed upon a small part of the blastodermic vesicle in P. moseleyi, P. sedgwicki and P. balfouri just as it starts to dilate. The position of the disk varies specifically. It consists of a posterior thickening on which is situated a blastoporal area, and an anterior thickening from which is formed the mouth-anus. The further growth of the germinal disk and yolk sac is described for the several species. The blastoporal area gives rise to all the endoderm and mesoderm by immigration (pp. 505, 509, 510, 512, 513, 531, 536). When immigration from the blastoporal area first starts in P. capensis and P. balfouri a ‘giant cell’ 30 p in diameter sinks in. This cell is clearly glandular in P. capensis, passing droplets of secretion to its surroundings and possessing abundant mitochondria. In both species the cell disappears very soon. The disappearance in P. balfouri coincides with the sudden shrinkage of the yolk sac, which is suggestive of glandular activity of this cell. No such cell is formed in P. moseleyi or P. sedgwicki where the yolk sac remains large. This cell is possibly a specialization associated with the elimination of the yolk sac (pp. 512, 514). The formation of the mouth-anus differs in the several species, but in all it is formed after the endoderm is established. In P. capensis the mouth-anus probably forms by a reopening of the closed blastopore. In P. balfouri, P. moseleyi and P. sedgwicki the mouth-anus arises de novo by a fusion of the ectoderm and endoderm to form the lips of this organ. Its size varies greatly in the different species. It may open widely and then divide to form mouth and anus, or it may not open until after this division. The mouth-anus does not give endoderm by invagination from the lips (pp. 513, 515, 518, 530, 531, 532, 545). The formation of the mesoderm and of the mesodermal somites is described more fully than before, and with particular reference to the head end of the body. Details are given of the formation, growth and migrations of the somites, and of the origin of their coelomic cavities. The formation of somites is very uniform, and the process is simpler than that occurring in not only most Arthropoda but in the majority of Annelida as well. There is no trace of primary and secondary metamerism. The antennal somite is differentiated in a post-oral position and then migrates to a pre-oral one. The antennal somite is considered to be serially homologous with those following it. The origin of a coelomic cavity from one or several initial spaces is correlated with the size of the embryos (pp. 509, 519, 521, 546). Germ cells do not arise in the endoderm, as previously supposed. They are differentiated at different stages in the several species, and arise from undifferentiated mesoderm or directly from the blastoporal area before any mesoderm has appeared. The subsequent history of the genital rudiment is described. A small proportion of its cells become incorporated into the walls of a variable number of mesodermal somites and the remainder degenerate (pp. 520, 525, 533). Individual and bilateral variations are frequent, and many abnormal embryos have been found (pp. 497, 526). No evidence has been obtained of the existence of a pre-antennal segment. It is suggested that no significance can be attached to the single case here recorded of asymmetrical mesoderm in front of the antennal somite, or to the embryo described by Evans for Eoperipatus, as the Onychophora have been shown to be subject to such extensive variations (pp. 525, 555). The composition of the head in the Onychophora is considered. Support is given for the view that the antennal segment is the first segment of the body, and that it is the only one to become pre-oral. Evidence is presented against the suggestion by Snodgrass that the antennal segment is part of an unsegmented acron (pp. 555, 556). The data here presented indicate the existence of an underlying uniformity in ontogenetic processes displayed by the different species of Onychophora. Bouvier’s suggested triphyletic evolution within the group is not supported. The range of variations found within the single genus Peripatopsis is remarkable. The methods of endoderm formation appear to show some of the responses of these species to an increasing amount of yolk in the egg which have persisted after a secondary reduction of the yolk. P. capensis appears to be the most advanced in the complete elimination of a large yolk sac, although it shows the more primitive method of endoderm formation; P. sedgwicki shows the least reduction of the yolk sac but has eliminated its primitive endoderm formation; P. moseleyi resembles P. sedgwicki except that the yolk sac is smaller; and P. balfouri has further reduced the yolk sac and shows an endoderm condition intermediate between the above extremes (pp. 536, 539). The theory of head evolution in the Arthropoda presented by Snodgrass in criticized. The evidence for the view that the unsegmented acron includes the anterior region and the antennular and antennal segments in Crustacea, and the antennal segment in Myriapoda and Insecta, is considered to be unsound. It is suggested that this theory raises profound difficulties when applied to the Arthropoda. On the evidence now available, the Onychophora appear to show the lowest grade of arthropodan head evolution, three cephalic segments being present, only one of which has become pre-oral (pp. 555, 556). The foundations of the theory of head segmentation in Annelida and Arthropoda put forward by Henry, Ferris and Hanke are criticized (p. 561). The theories concerning primary and secondary metamerism are reviewed and a somewhat different interpretation of the facts is suggested. It is pointed out that primary and secondary metamerism must tend to appear in association with the possession of early-swimming larvae in which a premium is put on the early functioning of some parts of the body. The phenomenon is not considered to be primitive, as has been claimed. It is here suggested that the absence of a differentiation into primary and secondary segments in the Onychophora and certain Annelida is a primitive feature (p. 546). The origin of many-segmented animals (p. 546) and the origin of metamerism are considered (p. 552). The suggestion by Snodgrass that metamer