Certain aspects of this proposed phylogeny warrant further comment. Features such as the deposition of additional bone that roofs the skull or that forms lateral projections from the frontoparietals, like those in S. baudini and phaeota, are minor alterations of dermal elements and not basic modifications of the architecture of the skull. Consequently, we hypothesize the independent development of these dermal changes in S. baudini and phaeota. Similar kinds of dermal modifications have evolved independently in many diverse groups of frogs.
Likewise, we propose the parallel development of stream-adapted tadpoles in S. sordida and sila; in both cases the tadpoles adapted to changing environmental conditions (see following section on evolutionary history). Tadpoles of S. sordida already had two rows of labial papillae before entering the streams; subsequently the tadpoles developed complete rows of papillae, ventral mouths and long tails having low fins. Possibly the tadpoles of S. sila had two rows of labial papillae prior to their adapting to stream conditions; in the process of adapting they developed ventral mouths and long tails having low fins. Similar modifications in tadpoles have occurred in many diverse groups of Middle American hylids, such as Plectrohyla, Ptychohyla, the Hyla uranochroa group, and the Hyla taeniopus group.
Our lack of concern about coloration is due to the fact that, with the exception of the blue spots on the flanks and posterior surfaces of the thighs in some species, the coloration of Smilisca, consisting of a pattern of irregular dark marks on a paler dorsum and dark transverse bars on the limbs, is not much different from that of many other Neotropical hylids. Blue is a structural color, rare among Amphibia, which is achieved by the absence of lipophores above the guanophores. Thus, the incident light rays at the blue end of the spectrum are reflected by the guanophores without interference by an overlying yellow lipophore screen. According to Noble (1931), lipophores are capable of amoeboid movement that permits shifts in their positions, between or beneath the guanophores. We do not know whether this behavior of lipophores is widespread and is effected in response to environmental changes, or whether it is a genetically controlled attribute that is restricted in appearance. If the latter is the case we must assume that the prototype of Smilisca possessed such an attribute which was lost in S. baudini, phaeota, and puma. The development of blue spots is not constant in S. sordida and S. sila; in S. cyanosticta the spots range in color from blue to pale green.
The coloration of the tadpoles is not distinctive, except for the presence of dorsal blotches on the tails of S. sila and sordida. However, the similarity in pattern cannot be interpreted as indicating close relationships because nearly identical patterns are present in Hyla legleri and some species of Prostherapis. This disruptive coloration seems to be directly associated with the pebble-bottom, stream-inhabiting tadpoles.
In the baudini group, S. phaeota and cyanosticta are allopatric, whereas S. baudini occurs sympatrically with both of those species. The call of S. baudini differs notably from the calls of S. phaeota and cyanosticta, which are more nearly alike. Although in the phylogenetic scheme proposed here S. sila is considered to be more distantly related to S. puma than is S. sordida, the calls of S. sila and puma more closely resemble one another than either resembles that of S. sordida. Smilisca sila and puma are allopatric, whereas S. sordida is broadly sympatric with both of those species. We assume that in their respective phyletic lines the differentiation of both S. baudini and sordida was the result of genetic changes in geographically isolated populations. Subsequently, each species dispersed into areas inhabited by other members of their respective groups. Selection for differences in the breeding calls helped to reinforce other differences in the populations and thereby aided in maintaining specificity.
With respect to temporal and spatial aspects of evolution in Smilisca, we have tried to correlate the phylogenetic evidence on Smilisca with the geologic data on Middle America presented by Lloyd (1963), Vinson and Brineman (1963), Guzmán and Cserna (1963), Maldonado-Koerdell (1964), and Whitmore and Stewart (1965). Likewise, we have borne in mind the evidence for, and ideas about, the evolution of the Middle American herpetofauna given by Dunn (1931b), Schmidt (1943), Stuart (1950, 1964) Duellman (1958, MS), and Savage (MS).
According to Stuart's (1950) historical arrangement of the herpetofauna, Smilisca is a member of the Autochthonous Middle American Faunal Element, and according to Savage's (MS) arrangement the genus belongs to the Middle American Element, a fauna which was derived from a generalized tropical American unit that was isolated in tropical North America by the inundation of the Isthmian Link in early Tertiary, that developed in situ in tropical North America, and that was restricted to Middle America by climatic change in the late Cenozoic.
Savage (MS) relied on the paleogeographic maps of Lloyd (1963) to hypothesize the extent and centers of differentiation of the Middle American Faunal Element. According to Lloyd's concept, Middle America in the Miocene consisted of a broad peninsula extending southeastward to about central Nicaragua, separated from the Panamanian Spur of continental South America by shallow seas. A large island, the Talamanca Range, and remnants of the Guanarivas Ridge formed an archipelago in the shallow sea. The recent discovery of remains of mammals having definite North American affinities in the Miocene of the Canal Zone (Whitmore and Stewart, 1965) provides substantial evidence that at least a peninsula was continuous southeastward from Nuclear Central America to the area of the present Canal Zone in early mid-Miocene time. South America was isolated from Central America by the Bolivar Trough until late mid-Pliocene.
Thus, in the mid-Tertiary the broad peninsula of Nuclear Central America, which consisted of low and moderately uplifted regions having a tropical mesic climate, provided the site for the evolution of Smilisca. It is not possible to determine when the genus evolved, but to explain the differentiation of the species it is unnecessary to have the ancestral Smilisca present prior to the Miocene.