Parallel Evolution of Derived Modes of Reproduction in Amphibians Marvalee H. Wake University of California, Berkeley XI Spanish-Portuguese Herpetological Congress 6-9 October 2010

The Evolution of Viviparity I study comparative reproductive biology and development, especially the evolution of live-bearing in frogs, salamanders, and caecilians. Study of the evolution of viviparity provides the potential for examination of mechanisms by which the phenomenon arises, especially the evolution of features that are similar in different lineages. Similar traits that have evolved independently in lineages not associated with derivation from a common ancestral state are labelled homoplasies. Homoplasies are often identified (declared) by mapping characters on a tree representing the phylogenetic relationships of taxa. If features that appear “the same” occur in distantly related lineages, they are considered homoplasious. Because the research goal is often either the phylogeny itself or the identification of homoplasy, few attempts been made to assess the mechanistic basis for the evolution of homoplasious conditions.

Kinds of homoplasies Convergence--similar features develop in distantly related lineages, not derived from a shared ancestral condition, and via different generative programs. Reversal--features thought lost in lineages recur (genetic basis is retained and recalled). 3. Parallelism--similar features develop from a common (identifiable?) substrate/program, elicited independently in different lineages by the same or different “signals”. (Less well characterized…) Do these cases overlap conceptually and pragmatically? Are they kinds of homology, or is homoplasy conceptually different? Why is parallelism important in evolution?

How does live-bearing arise??? I examine both maternal and embryonic properties of live-bearing development: the condition of the mother (e. g., ovaries, oviducts, and skin), the trajectory and characteristics of development of the embryos/fetuses, and ecology and behavior. I use a diversity of techniques (developmental, morphological, analytical). 3. I examine several taxa within each of the comparator clades, rather than a single typological example, in order to assess both within-clade and across-clade similarity and variation. 4. I use the evidence drawn from my studies to generate hypotheses about the mechanisms by which the evolution of live-bearing has occurred.

Viviparity (live-bearing) has arisen many times and in many ways among amphibian taxa, including: Males that brood their young in their vocal sacs (Rhinaderma darwini) or pouches on their legs (Assa spp.); Females that brood their young in their stomachs (Rheobatrachus spp. †); *Intraoviductal retention through late larva or complete metamorphosis, often with maternal nutrition (some species now †); *Intraoviductal cannibalism in Salamandra salamandra bernardezi and S. s. fastuosa (derived recently and independently); *Back-brooding in frogs. (* = work in my lab, in part) I will quickly and very briefly show you a few examples, but then focus on back-brooding in frogs and intraoviductal viviparity in frogs, salamanders and caecilians as examples of analyses of parallelism--this will be a broad (and brief) overview.

Rheobatrachus silus † SE Australia Photo: Michael Tyler Stomach-brooding frog Rhinoderma darwini Patagonia Photo: Dante Fenolio Male vocal sac brooder Assa darlingtoni Australia Photo: Unidentified Male with metamorph emerging from inguinal pouch Eleutherodactylus jasperi † Puerto Rico Photo: M. H. Wake Pregnant female with unborn oviductal froglets (removed) Nectophrynoides asperginis East Africa Photo: Dennis Demello Ovoviviparous; range size of a football field

Example: Back-brooding through late tadpole or metamorph in the frogs Gastrotheca (Amphignathodontidae) and Pipa (Pipidae) (Research programs of E. del Pino [in particular], R. Jones, W. E. Duellman, R. Elinson, L. Trueb, M. H. Wake)

Amphignathodontidae (Hemiphractidae) Gastrotheca riobambae Photo: Luis Coloma Flectonotus pygmaeus Mauricio Rivera Gastrotheca testudinea Photo: J. Köhler Gastrotheca fissipes Photo: Celio Haddad Gastrotheca guentheri Photo: Unidentified Amphignathodontidae (Hemiphractidae)

43 of 86 species in the clade. Wiens et al., 2007. Evolution 61:1886-1899.

Gastrotheca riobambae pouch opened, dorsal pouch wall reflected to expose dorsal and ventral epidermis of the pouch; eggs removed Gastrotheca testudinea pouch opened and reflected; near-birth froglet exposed; thin bell gills cover surface (Jones et al. 1973) Photos: J. Chin/MHW

Gastrotheca riobambae po A. Gastrotheca riobambae A. Pouch opening (po) ; B. Eggs on back under skin; C. Eggs exposed (SVL 36.7 mm; ~ 70 ova, each 3.2 mm dia; two layers) Gastrotheca longipes D. Eggs under skin; E. Eggs exposed (SVL 73. 5 mm; 17 ova, each 11. 5 mm dia) Photos: J. Chin, D. Buckley, MHW E. E. D.

Pipa arrabali Photo: Adrian Garda Pipa carvalhoi Photo: Alex Haas Pipa stethlageae Photo: Peter Janzen Pipa pipa Photos: Peter Janzen Pipa parva Photo: Dante Fenolio Close-up of embedded embryos

* * * * * U U T T F T T * = Available for this study. F = Froglets; T = Tadpoles; U = Unknown Trueb and Massemin, 2000. Amph.-Rept. 22:33-54.

Pipa arrabali embryo in pouch Pipa arrabali pouch skin; note lower layer Oviduct of a pregnant Pipa arrabali D. E. Ovary with corpus atreticum and ‘mature’ ovum in postovulatory Pipa snethlageae Oviduct of a postovulatory Pipa snethlageae Photos: D. Buckley/MHW

Summary of Example: In both clades, estrogen prepares the dorsal skin for ‘pregnancy’, and progesterone conditions and maintains gestation. However, the skin response differs between and even within clades. The oviducts maintain their responses to the hormones throughout gestation. Courtship and amplexus differ dramatically between these aquatic and terrestrial clades; however, in both cases, behavior has evolved so that females assume postures that allow the males to guide fertilized ova to their backs. Embryos develop modifications for respiration/gaseous exchange during gestation--bell gills in Gastrotheca, expanded, highly vascularized tails in Pipa. The mechanism of control of timing of parturition is not known. However, corpora lutea resorb well before ‘birth’ in G. riobambae, so loss of progesterone and increase in prolactin may be involved, so duration of corpora lutea may determine stage of development at birth, probably in both clades. New data suggest that the length of time the pregnancy can be maintained determines whether or not metamorphosis will be completed before birth.

Example: Intraoviductal development to metamorphosis, with maternal nutrition following yolk resorbtion, in frogs, salamanders, and caecilians (Research programs of F. Xavier, H. Greven, G. Guex, J.-M. Exbrayat, M. H. Wake)

V Od V Od V B VSt, L V B V Od V VS V Od

Nimbaphrynoides occidentalis Photo: Mark-Oliver Rödel

Oviparity Ovoviviparity Viviparity D. Buckley, unpubl.

Salamandridae; Wengen, Switzerland Salamandra atra Salamandridae; Wengen, Switzerland Pregnant female with near-birth fetus (note its gills, nearly complete limb development, etc.)

* ? Tree from Wilkinson and Nussbaum, 2006 * Indicates that the genus includes one or more viviparous species * * * *

Viviparous Caecilians Gymnopis multiplicata Costa Rica Photo: Michael Fogden Dermophis mexicanus Guatemala Photo: Sean Rovito Scolecomorphus vittatus Tanzania Photo: John Measey Chthonerpeton indistinctum Uruguay Photographer: Mirco Sole Chthonerpeton indistinctum Uruguay Photo: Mirco Sole Geotrypetes seraphinii Ghana Photo: Dante Fenolio Dermophis mexicanus Intraoviductal Fetuses Photo: M. H. Wake

Parallelisms! Pregnant Female Oviductal Epithelium Fetal Teeth (SEM) SEM of oviduct of early pregnant D. mexicanus (Lai and Wake, unpubl.) SEM of pregnant ‘uterus’ of Thamnophis ordinoides (Blackburn et al. 2002) Pregnant Female Oviductal Epithelium Dermophis mexicanus (Wake, 1980) Fetal Teeth (SEM) Parallelisms! SEM of oviduct of Salamandra atra (Guex and Greven, 1994) Salamandra atra (Greven, 1998)

Histochemistry of Dermophis mexicanus Oviduct; Ovary Non-pregnant PAS Hx D. Post-pregnant PAS Hx B. Mid-pregnant Best’s Carmine E. Ovary with maturing ovum and corpus luteum C. Mid- late-pregnant Sudan Black B Histochemistry of Dermophis mexicanus Oviduct; Ovary

Main Points of this Research: Both oviducts and skin respond morphologically and physiologically to ovarian hormones by proliferation, hyperemia, and increased vascularity. The oviductal morphology is similar across taxa, although the secretions of the cells vary in composition; i. e., morphology can be “the same”, but what it produces biochemically and physiologically (and behaviorally) can differ considerably across taxa. Little is known about hormones and effects in gestation, development, metamorphosis, and birth; several, especially prolactin, are ignored. Embryos/fetuses evolve clade-specific means of facilitating gaseous exchange, obtaining nutrients, etc. Maintenance of development in or on the body of a parent is a complex interaction of morphology, development, endocrinology, and ecology, as well as historical contingency, that is underappreciated, understudied and deserving extensive attention. It is especially important to rectify this, given that the existence of many species with derived modes of reproduction is threatened.

What, then, is homoplasy; in particular, what is parallelism?? Is it pattern, process, or the end result--the phenotypic expression? Is it a combination of these? Is it all attributable to “deep homology”? My analyses suggest that it is indeed “all of these”, and more. Any attempt to understand homoplasious features requires a hierarchical analysis of the probable mechanisms that underlie development and maintenance.

How and why selection operates in the evolution of these modes has been suggested (the “cold hypothesis”, competition for resources, reduction of larval predation, etc., etc.), but not often demonstrated, and never in the full complexity of the molecular, cellular, organismal, and ecological interactions involved. Similarly, homoplasy should not be merely an “end product” or phenotypic expression. For example, is “viviparity” in all its forms the same state, or does the term cover a diversity of processes and “end products” at several different levels of biological organization? If so, what components are homologous, and what are homoplasious?

These are not necessarily new ideas, but I try to frame them in a more holistic context. Homoplasy, including parallelism, is demonstrably a major phenomenon in evolution, now more accessible to study. Parallelism in particular illustrates both constraints--an organism can only use the material it has--and opportunities--there are innovative ways to use that material in response to selection, and they may co-occur in distantly related lineages. The study of the mechanistic basis of homoplasy opens new ways to explore the selection pressures that elicit responses that are mechanistically and phenotypically similar. The examination of homoplasious evolution requires a hierarchical research framework that examines features and mechanisms at a diversity of levels of organization.

Acknowledgements I appreciate the invitation from to speak at this most interesting meeting. I thank the many colleagues who have provided materials for my studies of the evolution of live-bearing modes of reproduction in amphibians, either via collecting or loans from the collections in their charge. I appreciate many discussions and collaborations with my students, postdocs, and colleagues about life history evolution, pattern and process of evolution, and why amphibians are the most interesting animals in the world, especially caecilians. The support of the National Science Foundation, agencies of the University of California at Berkeley, the John Simon Guggenheim Memorial Foundation, the American Philosophical Society, the Radcliffe Institute for Advanced Study, the Smithsonian Tropical Research Institute, and other groups has made the research possible. I thank you for your attention.