Pregnancy has been studied in Carollia perspicillata bred in captivity or collected from a reproductively-synchronized wild population on Trinidad, West Indies. In both situations, periods of postimplantational embryonic diapause were sometimes observed. In captivity, this was induced by stress, or once-weekly periods of food deprivation, and resulted in gestation periods of highly variable duration (105-237 days). The normal gestation period for this species is 113-120 days. In the wild, Carollia exhibits two synchronized pregnancy periods. One includes a diapause of at least 44-50 days, but probably somewhat longer, while the other is of normal length. The diapause occurs at the primitive streak stage of development. The postimplantational timing of the diapause is associated with multiple other reproductive specializations, all of which seem to be intended to facilitate blastocyst attachment very soon after entry into the uterus, within a preferred implantation zone. Blastocysts of this species were never observed to have passed that zone or to have a prolonged, free-floating period in the main uterine cavity. This facilitates placental development within an optimally vascularized part of the uterus and the eventual production of a large, highly precocious infant. Comparative immunocytochemical and ultrastructural studies of normal versus delayed pregnancies suggest that inadequate trophoblastic differentiation within the developing placenta may play an important role in maintaining the diapause. In part this may be responsible for delayed invasion of the embryonic side of the placenta by vascularized allantoic mesenchyme. The diapause is also associated with greatly increased trophoblastic invasiveness of unknown significance.

The potential for variation and the capacity to evolve in response to ecological opportunity are important aspects of an adaptive radiation. Identifying the origin of phenotypic variation, in which natural selection might act upon, is a major goal of evolutionary developmental biology. The New World leaf–nosed bats (phyllostomids) are a textbook example of an adaptive radiation. Their cranial morphology is diverse along relative facial length, which is related to their diets. We previously used geometric morphometrics to reveal peramorphosis, a type of heterochrony, in the cranial evolution among phyllostomid bats. We then demonstrated that the mechanism of peramorphic diversity in phyllostomid rostrum length resulted from altered cellular proliferation. Here, we investigate the progenitors of the face, the cranial neural crest, and a key signaling pathway related to their proliferation and differentiation into mature tissues: the bone morphogenetic protein (BMP). With geometric morphometrics, immunofluorescence, and confocal imaging—in three phyllostomid species and one outgroup bat species—we show the molecular patterns that underlie the adaptive and innovative traits seen in phyllostomid bats. Then, with mouse genetics, we mimic the BMP molecular pattern observed in nectar feeding bats and recapitulate the elongated morphological variation in mice. Surprisingly, we also observe an expansion in the nose–tip of mice, akin to the expanding leaf–nose tissue in phyllostomid bats. These data, combined with the mouse genetics literature on BMP signaling, suggest the BMP developmental pathway plays a central role in shaping the craniofacial variation necessary for adaptation in bats. Further, we speculate that the BMP signaling pathway could underlie other bizarre facial phenotypes in mammals that are derived from frontonasal mesenchyme, such as the proboscis. Overall, this study combines a comparative framework to developmental data, with a genetic approach, to directly investigate the role of development on complex morphology.