Ischnocnema oea (Heyer 1984)

Eleutherodactylus oeus

HOLOTYPE.—MNRio 1244, an adult male from Brazil: Espirito Santo; Santa Teresa. Collected by Augusto Ruschi in December 1942.

PARATOPOTYPES.—MZUSP 59684, USNM 235612.

DIAGNOSIS.—Eleutherodactylus oeus has indistinctly mottled posterior thigh surface patterns, E. erythromerus has light areas on the posterior faces of the thighs next to the knee joint (pattern B in Figure 11) and E. nasutus has boldly mottled thigh patterns. The outer face of the tibia of E. oeus has a dark stripe (patterns B and C in Figure 10); such dark stripes are absent in E. epipedus and gualteri. The head is also narrower in E. oeus than in epipedus and gualteri (Table 20). Eleutherodactylus oeus most closely resembles E. guentheri. At the site of sympatry, oeus differs most notably from guentheri in size (SVL measurements, male E. oeus 17.1–18.8 mm, male E. guentheri 28.0–30.5 mm at Santa Teresa).

DESCRIPTION OF HOLOTYPE.—Snout subelliptical from above, rounded in profile; canthus rostralis indistinct, lorus very slightly flared in cross section; upper tympanic annulus hidden, tympanum and annulus distinct below; vomerine teeth in short transverse series posterior and medial to choanae, vomerine tooth series separated from each other by not quite the length of a single vomerine tooth row; vocal slit present, external vocal sac barely indicated by a slight fold of skin; finger I just longer than finger II, I about equal to IV, fingers I, II, IV shorter than III; disk on finger I not much wider than digit diameter, disk width of other digits moderate, II narrower than IV narrower than III, larger disks with indented ungual flaps; fingers free; finger subarticular tubercles moderate, not pointed; outer broadly horseshoe-shaped metacarpal tubercle narrowly separated from inner ovoid metacarpal tubercle; inner base of thumb with whitish, glandular-appearing nuptial asperity; dorsal texture finely granular, upper eyelids warty-tuberculate; weak indication of dorsolateral folds behind eyes to just past shoulder region, no indication of supratympanic fold, no body glands; venter smooth, outer portion of ventral femur surface areolate; toe disks moderate, disk on toe IV largest, disks with indented ungual flaps; toe with weak lateral ridges; toe subarticular tubercles moderate; rounded outer metatarsal tubercle much smaller than ovoid inner metatarsal tubercle; tarsus lacking fold or tubercle; single pronounced heel tubercle; outer tarsus smooth; sole of foot smooth with one or two feebly developed light tubercles.

Measurements (in mm): SVL 17.1, HL 7.4, HW 6.0, EN 2.3, EE 3.6, TD 1.1, femur 10.0, tibia 11.1, foot 10.4, 3FD 0.7, 4TD 0.8.

Dorsal pattern in preservative indistinctly mottled cream, tan, and brown, irregular light cream interorbital bar and irregular light mid-dorsal blotch in scapular region, break in darker tan dorsal color to lighter cream lateral color in area of dorsolateral stripes, but no stripes indicated; tip of snout and front of eye with short dark brown stripes, upper lip with three irregular light vertical stripes, middle stripe just in front of eye broadest and most distinct, dark bordered behind; upper limbs irregularly barred; dark rectangular blotch including upper tympanum continuous with somewhat lighter dark stripe flaring into broad oblique lateral band at mid-body, dark sacral spot with short anterior projection; belly and middle portions of ventral limb surfaces light with a scattering of brown pigment, throat boldly mottled brown and light; outer tibia surface with distinct dark stripe; posterior surface of thigh indistinctly mottled, mostly brown, with light pin stripe on lower portion of thigh from below anus to mid-thigh.

ETYMOLOGY.—From the Greek oios (unique, peculiar), in allusion to the fact that many taxa have been described as new from Santa Teresa, seemingly without distributions other than at Santa Teresa. This taxon is further peculiar in that of the many frog specimens collected from the Santa Teresa area, there are only three individuals of E. oeus known, all collected in 1942.

ADULT SPECIMEN DEFINITION.—Dorsum uniform or mottled (patterns A-2 and A-6 in Figure 1); no mid-dorsal pin stripes; no broad light mid-dorsal stripes; no light dorsolateral stripes; one individual with a light snout (pattern A in Figure 4); light interocular stripes present; interrupted or continuous dark stripes on outer tibia (patterns B and C in Figure 10); posterior surface of the thigh indistinctly mottled; life colors unknown; males 17.1–18.8 mm SVL; head width moderate (Table 20); hind limb length moderate (Table 20).

ADVERTISEMENT CALL.—Unknown.

DISTRIBUTION.—Known only from the type-locality (Figure 21).

ESPIRITO SANTO. Santa Teresa (MNRio 1244, MZUSP 59684, USNM 235612).

Zoogeography

Zoogeographic understanding requires two kinds of data: distribution and relationship. I am unable, with the data as analyzed in this paper, to produce a satisfactory hypothesis of relationships among the members of the E. guentheri cluster. Attempts to cladistically analyze the various pattern characters were frustrated in that there were but a handful of characters for which polarities could be determined and the derived states had a mosaic distribution among the species. Nevertheless, some general zoogeographic features are evident from the species distributions of the cluster members.

Somewhat surprisingly, the combined distribution of all species in the cluster correlates well with the middle and southern Atlantic Forest Morphoclimatic Domain (as defined by Ab'Sáber, 1977) (Figure 27). Whereas the cluster was chosen for study because the members occurred throughout the Atlantic Forests, more southerly records (Braun and Braun, 1980) were known. I also assumed that the cluster members, specifically E. guentheri, would extend at least into the Atlantic Forest of Bahia and in various adjacent buffer zones or morphoclimatic domains. This assumption was made because other members of the genus Eleutherodactylus, which also have a direct development mode of reproduction, occur in the areas mentioned. However, the known distributions of the E. guentheri cluster members are rather restricted within the Atlantic Forest Morphoclimatic Domain.

Available data indicate that the presently known distributions are a good approximation of the actual northern distributional limits. When present, such species as E. epipedus, gualteri, guentheri, and nasutus are likely to be collected. At most places, cluster members are relatively common and although cluster members may be reproductively active at night, they are often encountered on the leaf litter during the day. Thus, negative data (specimens not collected from given localities) have more meaning for this cluster of frogs than for many other frog groups. The frog fauna from Linhares, Espirito Santo, has been reasonably sampled. This locality, on the north side of the Rio Doce, lies about 80–90 km north of Santa Teresa. No members of the E. guentheri cluster were in the MNRio collections, where collections from Linhares are deposited. The next northern locality that has been well sampled is around Ilhéus, Bahia. Bokermann (1975) reported on the Eleutherodactylus from the area. He did not collect any member of the E. guentheri cluster as defined in this paper. Bokermann (1975) described a new species, E. vinhai, from Ilhéus, indicating that it was a member of the E. guentheri group. The species has a granular belly, which might ally it with E. erythromerus. If E. vinhai does turn out to be a member of the E. guentheri cluster, it would be the most northerly member known of the E. guentheri cluster. All other Eleutherodactylus known from the northern Atlantic Forests belong to species groups that are quite distinctive from E. guentheri and its close relatives. Thus, although definition of the northern limit of the E. guentheri cluster awaits determination of whether E. vinhai is a cluster member, it seems safe to conclude that the species defined in this paper have their northern limit south of the Rio Doce.

The locality records outside of the Atlantic Forest Morphoclimatic Domain in the States of Minas Gerais, São Paulo, and the southermost two records in Rio Grande do Sul (Figure 27) occur in a mesic forest vegetation, identified as deciduous mesophytic subtropical forests of east and south Brazil by Hueck and Seibert (1972, vegetation type number 29 on the map). Eleutherodactylus guentheri cluster members do not occur throughout this vegetation type, but appear to have quite restricted distributions within it. The same general mesophytic subtropical forest vegetation occurs in the Misiones region of Argentina where the late Avelino Barrio, in particular, made efforts to collect the herpetofauna. Cei (1980), in his summary of the amphibians of Argentina, indicates that no Eleutherodactylus is known from Misiones. The localities within the mesophytic subtropical forests where E. guentheri cluster members have been collected may represent local edaphic conditions where the forest conditions approximate those found in the Atlantic Forests.

The locality outside of the Atlantic Forest Domain in the State of Paraná (Figure 27) is Volta Grande. I do not have personal experience with this locality, but it appears to be well within the Araucaria Domain (as defined by Ab'Sáber, 1977). The northern record in Rio Grande do Sul, Cambará do Sul, is definitely in the Araucaria Domain. Again, the presence of E. guentheri cluster members appears to be restricted, not widespread, within the Araucaria Domain. Pedro Canisio Braun has been working on the distributions of frogs in the State of Rio Grande do Sul for many years. The record from Cambará do Sul is the only one reported from the Araucaria Domain in Rio Grande do Sul (Braun and Braun, 1980).

A key to understanding the distributional limits of Eleutherodactylus guentheri and nasutus will lie in understanding why these two species only occur in a few restricted localities, and not throughout the mesophytic subtropical vegetation and the Araucaria Domain of Brazil.

As presently understood, the distribution range of E. guentheri almost includes the distribution ranges of all other species in this cluster. Eleutherodactylus nasutus has a broader distribution than previously known, but the southern limit of distribution is enigmatic in that there is no a priori reason, based on ecological considerations, for the species not to occur further to the south. This is also true for the extremely restricted distributions of the remaining four species of the cluster. Without a detailed, probable hypothesis of relationships among the species, it is pointless to speculate on the mode of speciation and geographic consequences involved.

The rather amazing sympatric occurrence of four members of the same species cluster at Teresópolis and Santa Teresa reflects a general pattern of Eleutherodactylus diversity in the Atlantic Forests. Although very few data points are available, they describe a pattern of reduced diversity at the northern and southern extremes with the highest diversity occurring from the Organ Mountains to the area including Santa Teresa. The available data, from north to south, are as follows: two species are known from the State of Pernambuco; Bokermann (1975) reported a total of five species from Ilhéus, Bahia; nine species occur at Teresópolis; six species at Boracéia; and only one species gets into the State of Rio Grande do Sul (Braun and Braun, 1980), just beyond the southernmost extent of the Atlantic Forest Domain. The diversity gradient correlates with temperature; rainfall patterns differ throughout the Atlantic Forests, but in a complex, not clinal, fashion. The Eleutherodactylus diversity gradient suggests that there is relatively little eco-physiological stress on Eleutherodactylus at the center of diversity and greater stress at the northern and southern extremes of the Atlantic Forests. Understanding the ecophysiology of the Atlantic Forest Eleutherodactylus may provide greater understanding of the distribution patterns and patterns of point diversity than explanations invoking competition.

A major reason in choosing the E. guentheri cluster to analyze was that it appeared from the outset that it would be possible to analyze patterns of differentiation within E. guentheri itself and to see if the variation correlated with geography. Two kinds of data are available: pattern characteristics and morphology.

The pattern characteristics were analyzed as follows. The data from Table 14 were combined with data for two states that differed, but due to sample sizes, not statistically significantly (dorsal patterns A-11, and C in Figure 1). The total number of significantly differentiated states was noted for each population and, on a geographic plot, the significant states that were shared between populations were indicated by drawing lines connecting the populations (Figure 28). These data thus provide an index of relative differentiation among populations and an index of possible relatedness through sharing of derived states (assuming significantly differentiated states are derived) among populations.

The morphological data used were from the discriminant function analysis, specifically the posterior classification results (Tables 17, 18). The methodology used is best explained by example. The individuals of population RJ1 that were posteriorly classified as SP1 individuals are more similar to the SP1 centroid, thus most SP1 individuals, than they are to the RJ1 centroid, thus most RJ1 individuals, in the features (measurements) analyzed. These “missed” classifications are used as an indication of morphological similarity between these two populations and the greater the incidence of “missed” classifications, the more similar the populations are assumed to be to each other. These data were figured geographically (separately for males and females) by connecting those populations in which “missed” classifications occurred (Figure 28). This index of morphological similarity, based on “missed” classifications, could be the result of two different causes: either a sharing of derived morphological states or a retention of the ancestral morphology. If geographic variation is evident among populations, either cause of morphological similarity should demonstrate a pattern of geographic variation.

If differentiation of populations correlated completely with geographic variation, the following two general patterns would be predicted. First, for both the pattern state and morphological data, the strongest connections among populations should be with the geographically closest populations, resulting in localized spider web patterns with few, if any, connections among distant populations. Second, because of the differences in kind of the pattern state and morphological data, differences of detail would not be surprising when comparing overall patterns; however, there should be concordance between the male and female morphological data set as illustrated.

Visual inspection of these data sets as analyzed and illustrated (Figure 28) indicates that the patterns of variation do not, in fact, have a series of web-like connections. Rather, there are several long-distance connections among populations resulting in patterns with a lot of noise. Some signals are discernable within the generally noisy framework, nonetheless.

The pattern state data indicate that considerable population differentiation has occurred, ranging from 1 (populations 2, 4, Figure 28) to 13 significantly differentiated states (population 7, Figure 28) per population (note that in Figure 28, the basic dichotomy of weakly differentiated [small circles, 1–5 differentiated states] or strongly differentiated [large circles, 9–13 differentiated states] populations is shown). The distribution of differientiated states is not dependent on sample sizes available for analysis. For example, sample 1 (Figure 28) from Teresópolis with a moderate sample size is well differentiated, and sample 3 (Figure 28) from Rio de Janeiro with a large sample size is weakly differentiated.

Two kinds of comparisons argue against a geographic component of differentiation. The first is that the largest number of shared significantly differentiated pattern states occurs between the populations from Serra da Bocaina and Boracéia (Figure 28, samples 5 and 7). On geographic grounds, the Boracéia population would be expected to share the greatest number of pattern states with the Paranapiacaba-Cubatão sample (Figure 28, samples 7 and 8), which in fact share no differentiated pattern states (also see p. 15). The second kind of comparison involves examining pairs of populations, which based on geography, should be most similar to each other. These are Teresópolis and Petrópolis in the Organ Mountains of Rio de Janeiro (Figure 28, samples 1 and 2) and Boracéia and Cubatão-Paranapiacaba in the same block of the Serra do Mar of the State of São Paulo (Figure 28, samples 7 and 8). For these comparisons, there is no indication from any of the data sets that these population pairs are most similar to each other.

The arguments against geographic variation accounting for the observed differentiation among populations analyzed do not preclude a component of geographic variation in fact occurring among populations of E. guentheri. The morphological data arguably suggest a component of geographic variation among the State of São Paulo samples (Figures 28, samples 5–9). The data themselves have certain liabilities that may limit their usefulness in describing geographic variation patterns. The pattern-state data were taken on samples that had been pooled over a 60-year period in some cases. Certainly there should be year-to-year variation in frequency of occurrence of pattern states that may add noise to the analysis. Also, for example, the morphological differences observed may be in response to similar adaptations of size to similar local climates, resulting in similar body forms in lowland populations at southern latitudes and higher elevation populations at northern latitudes. These morphological adaptations would have to be factored out in an analysis of geographic variation. The available data lead to the following hypothesis, which is accepted for purposes of this paper: differentiation is evident among populations of E. guentheri, but no geographic pattern is evident in the differentiation. The E. guentheri complex would be an ideal candidate for electrophoretic analysis, which would provide much more direct evidence on the degree and nature of intra- and inter-species variation and relationships.

Comparison With the Cycloramphus Model

The speciation and zoogeographic model proposed for the stream associated species of Cycloramphus (Heyer and Maxson, 1983) consists of the following elements. Each species has a restricted, localized distribution; collectively, the species distributions correlate with areas of sharp topographic relief within the middle and southern extent of the Atlantic Forest Domain where small mountain brooks occur; the allopatric model of speciation best accounts for the patterns of relationships and distributions as described by Heyer and Maxson (1983:367):

The outstanding feature of the allopatric model of speciation for Cycloramphus is the small sizes of geographic areas of isolation and speciation. The scale is local areas of high relief…. Each local area had a unique history in terms of geological formation (when and how they were formed), hydrology, extent of Atlantic Forest cover during cooler and more arid times, and colonization, adaptations, and extinctions of Cycloramphus populations. No single zoogeographical pattern is apparent. Local patterns of distribution and relationships predominate. This very local effect results from some aspect in the life history of riparian Cycloramphus that confines occurrence to a very narrow and precisely defined microhabitat. The larval phase of the life cycle is the assumed limiting aspect, because the larvae are adapted to the wet surface covered rock splash zone of small brooks. Occurrence is, thus, limited to mountain brooks within areas of high relief and occurrence is discontinuous between areas of high relief.

The specialized larval ecology in stream-associated Cycloramphus was a dominant factor in the development of the Cycloramphus model. Thus, the interplay of ecology and history was considered crucial to understanding the zoogeography of stream associated Cycloramphus. Using the model, several predictions were made, two of which involved possible tests with members of the genus Eleutherodactylus (Heyer and Maxson, 1983:370):

This narrow specialization [splash zone tadpole], combined with a long evolutionary history in an unstable area (geologically and climatically), suggests that Cycloramphus exhibits an extreme example of speciation in very local areas. We predict that this same pattern would only be repeated in other groups which have life history features that limit their distributions to very patchily distributed habitats. Some other stream associated insect groups might be expected to show a pattern very similar to that seen in Cycloramphus. The areas of isolation and differentiation should be at a somewhat larger scale for fishes and frogs that have larvae that live in the waters of the streams. Thus, for groups that have occurred in the Atlantic Forest Domain throughout the Cenozoic, such as Hylodes, the total distributional range should be greater, individual species ranges should be larger, and there should be fewer species per comparable geographic region than for Cycloramphus.…For stream frogs (with aquatic larvae) that have relatively short histories in the Atlantic Forest Domain, individual species ranges should be large and intraspecific variation should occur due to Pleistocene isolation of population units…. At another level, frogs with life histories not tied to patchily distributed habitats within the Atlantic Forest Domain, such as Eleutherodactylus, should have broader distributions both locally and geographically than either Cycloramphus or Hylodes. Further, there should be fewer species within the Atlantic Forest Domain per major lineage than for either Cycloramphus or Hylodes.

The Cycloramphus model predictions are not entirely borne out by the Eleutherodactylus data. A comment is appropriate at this point comparing the nature of the Cycloramphus and Eleutherodactylus guentheri cluster data bases. Both data bases have the same reliability of taxonomic assessment and precision of distributional understanding. The data bases differ in two ways. The Cycloramphus data set included genetic estimates of relationships; the relationships among E. guentheri cluster members remain unknown. The combination of relationship and distribution data for Cycloramphus allowed a rather inclusive and detailed zoogeographic model to be constructed. This model, as seen above, allowed predictions regarding other Atlantic Forest associated groups. Whereas there are not enough data for the E. guentheri cluster to propose a model comparable to that developed from the Cycloramphus data, there are elements of the Eleutherodactylus data that can be used to test the Cycloramphus based model. The second way the data sets differ is that data were not adequate to analyze intraspecific variation for any Cycloramphus species in the manner done for E. guentheri.

The Cycloramphus model predicted that Eleutherodactylus species would have: (1) broader local distributions, (2) broader geographic distributions, and (3) fewer species per major lineage than for Cycloramphus. These predictions are discussed in turn.

1) Eleutherodactylus do have broader local distributions than stream associated Cycloramphus. Eleutherodactylus guentheri occurs throughout the forest floor, including stream-side locations, while C. semipalmatus is only found next to streams at Boracéia, for example.

2) Only E. guentheri and nasutus have broader geographic ranges than stream-associated Cycloramphus. The distributions of E. epipedus, erythromerus, gualteri, and oeus are similar to those of stream associated Cycloramphus. These latter distributions violate the predictions for Eleutherodactylus drawn from the Cycloramphus model.

3) The major lineages of Atlantic Forest Eleutherodactylus are not understood at present and the following may require revision if the E. guentheri cluster turns out to be a component of a larger lineage. There are 16 species of stream associated Cycloramphus, which appear to form a major lineage (Heyer, 1983; Heyer and Maxson, 1983). This figure of 16 is considerably greater than the six species comprising the E. guentheri cluster. For the present, this Cycloramphus model prediction is assumed to be validated by the Eleutherodactylus data.

The pattern of intraspecific differentiation and variation within E. guentheri is consistent with the very local centers of differentiation proposed for the Cycloramphus model. These E. guentheri data are therefore consistent with the hypothesis that during the Pleistocene, the Atlantic Forest vegetation was extremely fragmented and differentiation of E. guentheri occurred very locally, rather than within more extensive forest refugia. Present knowledge of Pleistocene forest refugia locations along coastal Brasil is not adequate to test this hypothesis.

The comparison of the Cycloramphus and Eleutherodactylus data indicates that there is some concordance between the Cycloramphus model predictions and confirmation with Eleutherodactylus data, but that there are enough differences to require re-assessment of the Cycloramphus model. The Cycloramphus model gave somewhat equal weight to the factors of ecology and history in understanding the zoogeography of the Atlantic Forest fauna. Rather than the Eleutherodactylus data showing a markedly different pattern than stream-associated Cycloramphus, as predicted from the model, several Eleutherodactylus species have geographic distribution patterns just like those of stream-associated Cycloramphus. Also, the pattern of differentiation within E. guentheri is consistent with a very local level of differentiation, like that seen at the species level in Cycloramphus. The ecologies of stream-associated Cycloramphus and Eleutherodactylus are very different. Both Cycloramphus and Eleutherodactylus have shared histories within the Atlantic Forest Domain, however. Thus, the restricted species distribution patterns common to Cycloramphus and Eleutherodactylus are likely due to common historical factors. Thus, the Cycloramphus model needs revision to the extent that in certain cases, historical factors leading to restricted distributions seemingly override ecological factors.

The combined Cycloramphus and Eleutherodactylus zoogeographical data indicate that the interplay of history and ecology is paramount to our understanding of the Atlantic Forest biotic distributions. Ecology is critical to the understanding of local distributions. Ecology is a component, but may not be as important as history, in understanding species ranges. At this point, I believe the challenge to understanding fully the zoogeography of the Atlantic Forest biota lies with knowing, in detail, the historical factors associated with population fragmentations and dispersals.

The Eleutherodactylus data suggest one additional insight into Atlantic Forest faunal zoogeography. Eleutherodactylus nasutus is an open formation associated species, with its close relatives being closed forest associates. The ecology of E. nasutus suggests that it could occur throughout open formations whether associated with the Atlantic Forest Domain or the buffer mesophytic subtropical forest. That E. nasutus shows as much fidelity to the Atlantic Forest Domain as it does indicates that there is a faunal element that is ecologically adapted to open formations, but zoogeographically restricted to the Atlantic Forest Domain.

Ischnocnema oeus is a species of frog in the family Brachycephalidae. It is endemic to Brazil. Its natural habitat is subtropical or tropical moist lowland forest. It is threatened by habitat loss.