On the morphology of Plectrohyla chryses ( Anura : Hylida ; Hylini ) , with comments on some controversial characters , phylogenetic relationships , and diagnosis of this species

We describe the tadpole and juvenile of Plectrohyla chryses and conduct an analysis of skin thickness and a general phylogenetic analysis of the Hylini to determine the phylogenetic position of P. chryses. We found that P. chryses has thin skin and that is part of the P. bistincta group. We also found that adult P. chryses have an axillary membrane and a thoracic fold, but that the rostral keel is absent. We conclude that only the character “skin thickness” is useful for the diagnosis of P. chryses, and we propose a new diagnosis for this species.


INTRODUCTION
The genus Plectrohyla contains 42 species of stream-dwelling frogs distributed in the highlands of Mexico, Guatemala, northern El Salvador, central northern Honduras, and northern Nicaragua (Duellman 2001, Frost 2015).There are two monophyletic groups recognized in Plectrohyla, the P. guatemalensis group with 18 species distributed from southern Mexico to northern Nicaragua and the P. bistincta group with 24 species, all distributed in Mexico (Faivovich et al. 2005, Frost 2015).

On the morphology of Plectrohyla chryses (Anura: Hylida; Hylini), with comments on some controversial characters, phylogenetic relationships, and diagnosis of this species
Sobre la morfología de Pectrohyla chryses (Anura: Hylidae; Hylini), con comentarios sobre algunos caracteres controversiales, sus relaciones filogenéticas, y su diagnosis Duellman et al. (2016) transferred the P. bistincta group to a newly erected genus Sarcohyla, a questionable action given that most of the species of the P. bistincta group (16 out of 24) have not been included in a phylogenetic analysis (see Faivovich et al. 2005:104 for a similar argument).
One of the lesser-known species of the Plectrohyla bistincta group is P. chryses (Adler, 1965).This species is known from 32 individuals.Its tadpole and juvenile are still undescribed.P. chryses inhabits oak-pine-fir forest in the Sierra Madre del Sur west of Chilpancingo, Guerrero at elevations between 2340 and 2600 meters (Duellman 2001, Frost 2015).Adler (1965) described Plectrohyla chryses based on four adult specimens (1 male, 3 females).Mendelson & Toal (1996), after examining the holotype, a paratype (KU 106306), and several newly discovered individuals, concluded that three characters of the diagnostic combination (thoracic fold, axillary membrane, and rostral keel) were erroneous.Presently, it is unclear what the correct condition of these characters are, or whether these characters can be used to diagnose P. chryses from other species of the P. bistincta group.Adler (1965; see also Duellman 2001 andDuellman et al. 2016) included the character "dorsal skin thickness," among others, in the diagnostic combination of P. chryses (Adler 1965).Toal & Mendelson (1995:13) questioned the utility of this character for the taxonomy and systematics in the P. bistincta group because previous conclusions about whether the species of this group have thin or thick skins were subjective (i.e., based on qualitative assessments, not measurements).
Presently, quantitative measurements have only been taken from individuals of three species of the P. bistincta group (i.e., P. bistincta (Cope, 1877), P. chryses, and P. mykter (Adler and Dennis, 1972); see Kaplan et al. 2015), so it remains unknown if the previous conclusions concerning the taxonomic distribution of the states "thin skin" and "thick skin" in the P. bistincta group are correct.
P. chryses is considered part of the P. bistincta group because it has long fingers with vestigial webbing, prominent (not projecting) prepollex, and lacks a quadratojugal (Adler 1965, Duellman 2001; but see Duellman & Campbell 1992).However, presently there is no synapomorphy supporting the inclusion of P. chryses in the P. bistincta group or in the genus Plectrohyla (Faivovich et al. 2005).Thus, the phylogenetic position of P. chryses remains in doubt.
We recently obtained tissues for DNA analysis, as well as tadpoles and juveniles of individuals of the P. bistincta group from two localities west of Chilpancingo, Guerrero.
Based on the most recent diagnosis (Mendelson and Toal 1996) they appear to be P. chryses.Herein, we conduct a phylogenetic analysis using DNA sequences in order to assess whether P. chryses is part of the P. bistincta group.We describe the tadpole and juvenile, and we reexamine the holotype and two paratypes of P. chryses.Based on our observations, we discuss what are the conditions of the characters thoracic fold, axillary membrane, and rostral keel in this species.We measure the skin thicknesses in fourteen species of the P. bistincta group and discuss whether the thickness of the dorsal skin can be correctly assessed by external examination alone and whether the character skin thickness can still be used in the taxonomy and systematics of this group.Finally, we evaluate whether a diagnosis of P. chryses that includes the above characters is still useful.

MATERIALS AND METHODS
We collected tadpoles from two localities near Carrizal de Bravo, on the Sierra Madre  (Wessner 1960).Skin thickness was measured as the average of three independent measurements using a compound microscope with a calibrated ocular micrometer.In Table 1 we ranked species according to the skin thicknesses from thin to thick.The difference in skin thickness between adjacent species in column 3 of Table 1 was expressed as the percentage increase in thickness between them.
Tadpoles were staged following Gosner (1960).Terminology of tadpole morphology follows that of Altig & McDiarmid (1999).Measurements and terminology used to describe the juvenile specimens follows that used by Duellman (2001).We define axillary membrane as the portion of skin that covers the angle formed at the junction of the humerus and trunk; the axillary membrane must be visible in dorsal and ventral views when the humerus is perpendicular to the trunk.The axillary membrane can be baggy and loose around the axilla (e.g., Tlalocohyla loquax [Figs. 1a,b]) or tight and ventrally flattened (e.g., P. charadricola and P. chryses [Fig.1c-f]).We define thoracic fold as the most anteriorly-located complete fold, if many are present, traversing the thorax (Fig. 2a).We follow the definitions of Kaplan et al. (2015) for the terms "fold" and "flexible".We define rostral keel as a vertical or nearly vertical, fleshy (i.e., formed by coalesced tubercles or corrugations of the skin) ridge-like elevation that runs along a portion, or the whole, of the midline of the rostrum (e.g., P. ixil.Fig. 2b); more than one rostral keel, and rostral keels of different sizes, may be present in an individual (Fig. 2b).We follow Duellman (2001; but see Adler & Dennis 1972) in recognizing two states of the character "dorsal skin thickness", namely "thin skin" and "thick skin".
To assess whether P. chryses is part of the genus Plectrohyla, specifically of the P. bistincta group, we conducted a phylogenetic analysis that included P. chryses and 62 closely related frog species following Faivovich et al. (2005) using sequences obtained through a BLAST search on GenBank.These sequences included nine other Plectrohyla species and various outgroups (information for all 63 individuals is located in Appendix 1).
We sequenced 421 base pairs of the cytochrome b gene of a juvenile P. chryses (UMMZ 239651) using primers MVZ15 (Moritz et al. 1992) and H15149 (Kocher et al. 1989) and standard PCR conditions (Faivovich et al. 2005:44).We aligned the sequences using MAFFT v.7.130b (Katoh & Standley 2013)  eyes large, their diameter 40% of head length, round, prominent; tympanum small, its diameter 35% of eye diameter, distinct, slightly higher than wide; tympanic annulus prominent posteriorly; supratympanic fold heavy, extending from the posterior border of eye to above the arm insertion, wider posteriorly than anteriorly, covering upper part of tympanum. Fingers

DISCUSSION
We considered all the collected specimens to be a single species because the tadpoles of the Carrizal de Bravo and Yextla populations are nearly identical.We assign all the collected specimens to Plectrohyla chryses by their shared phenotypic features, including that juvenile specimens (UMMZ 239648-51) have the following characters: thoracic fold; eyes prominent; vent flap short, wide, thin, not grooved medially; vent opening at upper level of thighs; skin of dorsum smooth; tympanum and tympanic ring evident; webbing of fingers vestigial; webbing of toes I2 --2 1/2 II1 3/4 -3 1/3 III2-3IV3-1 1/2 V (UMMZ 239649); ulnar tubercles distinct, forming row; tarsal fold distinct; Contributions to the morphology of Plectrohyla chryses flanks and dorsum with same coloration and pigmentation pattern in preservative (Straughan & Wright 1969).Juvenile specimens (UMMZ 239648-51) differ from the diagnostic combination of P. chryses by having a smaller tympanum, dark small blotches on dorsum of body, head and limbs, and a well-defined mottling on flanks and dorsum, and by lacking an axillary membrane.
The phylogenetic analysis (Fig. 3) shows that P. chryses is nested within the P. bistincta group, i.e. it is inside the thickskinned species.Although bootstrap values are low throughout the tree, likely due to the small number of base pairs examined, P. chryses falls within the P. bistincta group in both analyses, and Plectrohyla itself is monophyletic in both analyses.Thus, although the monophyly of the P. bistincta group, and the inclusion of P. chryses within this group, is supported by molecular synapomorphies, no concordant morphological synapomorphy is presently known.Duellman & Campbell (1992) proposed that the frogs of the P. guatemalensis group and the thick-skinned frogs of the P. bistincta group, i.e. all species except P. calvicollina, P. charadricola, P. chryses, P. labedactyla, P. sabrina, and P. thorectes, form a monophyletic group.This hypothesis was supported by four synapomorphies: medial ramus of pterygoid in contact with otic capsule; dorsal skin thick: absence of median papillary hiatus; presence of accessory labial papillae (=submarginal papillae) between marginal papillae and tooth rows.Our phylogenetic analysis suggests that Duellman & Campbell's (1992) clade is paraphyletic with respect to P. chryses and also with respect to P. thorectes because this thin-skinned species (Table 1) appear to be nested within the thick-skinned species of the P. bistincta group (Figs 3 and  4).Our morphological observations also suggest that P. chryses should not have been excluded from Duellman & Campbell's (1992) clade because it has three of the four synapomorphies that support it: medial ramus of pterygoid in contact with otic capsule, a tadpole lacking median papillary hiatus, and a tadpole having submarginal papillae between marginal papillae and tooth rows (Figs. 4 and 6).Finally, our results suggest that the validity of the fourth synapomorphy supporting Duellman & Campbell's clade, i.e. "thick skin" is questionable because some species in this clade, e.g.P. ixil and P. matudai, have thinner skins than those of some species excluded from this clade such as P. charadricola.
Adler (1965) used the character "dorsal skin thickness" in the diagnostic combination of P. chryses.Toal & Mendelson (1995:13) questioned the usefulness of this character for the taxonomy of any species of the P. bistincta group because previous conclusions as to whether a species has thin or thick skins were reached subjectively using qualitative external examinations, not quantitative measurements.Our results (Table 1, columns 4 and 6) show that previous assessments of the skin thickness using external examination are largely correct and are thus credible.Only in P. ameibothalame was the qualitative external assessment erroneous.Therefore, we consider the character "skin thickness," and its states "thin skin" and "thick skin," and the proposed taxonomic distribution of these states currently valid for use in the taxonomy and systematics of the P. bistincta group.In other words, the state "thin skin" is only present in P. calvicollina, P. charadricola, P. chryses, P. labedactyla, P. sabrina, and P. thorectes.
Toal & Mendelson (1995) also questioned the utility of the character "dorsal skin thickness" for taxonomy and systematics in the P. bistincta group by suggesting that the recognition of the states "thin skin" and "thick skin" is arbitrary because the skin thickness varies continuously in this group.Table 1 suggests that the variation of skin thickness in the P. bistincta group is discrete because the difference in thickness between the skins of some species is much larger than that between the skins of other species such as P. charadricola and P. sabrina whom are different by 68.6% versus a difference between P. sabrina and P. thorectes of 7.7%.However, it is still premature to conclude that the variation in skin thickness is discrete in the P. bistincta group or that only the states "thin skin" and "thick skin" exist because not enough species and individuals per species have yet been examined.The skin thickness of P. arborescandens, P. cyclada, P. hazelae, and P. thorectes was never assessed using external examinations or quantitative measurements.Our results (Table 1, co lumn 6) suggest that P. thorectes has thin skin and P. arborescandens, P. cyclada, and P. hazelae have thick skins.
Mendelson & Toal (1996) concurred with the original diagnosis of P. chryses (Adler 1965), except that they found that adults have a thoracic fold and that an axillary membrane and rostral keel are absent.We found that the adult individuals of P. chryses  have a thoracic fold (Fig. 2a) and an axillary membrane (Figs.1e, f), but that a rostral keel is absent (both paratypes present few small, dispersed, not-coalesced tubercles on their rostrum).We also found that all juveniles have a thoracic fold and that the axillary membrane and rostral keel are absent (Figs.5a, b).These findings confirm that a thoracic fold and an axillary membrane are present in P. chryses and that a rostral keel is absent and suggest that the character "presence of axillary membrane" in this species should be redefined as "presence of axillary membrane in adults".
It appears that the characters "presence of thoracic fold", "presence of axillary membrane in adult", and "absence of rostral keel" cannot currently be used for the diagnosis of P. chryses because, when narrowly defined (see Materials and Methods), as the correct conditions of these characters are still unknown in most species of the P. bistincta group.Also, it is still unclear whether a thoracic fold is present in juveniles of species of the P. bistincta group that are considered to lack this character (a distinct possibility, see Kaplan et al. 2015: 218).Only a comprehensive assessment of the correct conditions of the narrowly defined characters "thoracic fold," "axillary membrane," and "rostral keel" in most species of the P. bistincta group, and a better understanding of their intraspecific, sexual, and ontogenetic variation, will allow us to determine if these characters have any value for the diagnoses of P. chryses and the other species of the P. bistincta group.
With the removal of the characters of the thoracic fold, axillary membrane, rostral keel, and tarsal fold (see  (contra Mendelson & Toal [1996] adults and juveniles preset mottling [Fig.2c); flanks and dorsum with same, or very similar, coloration and pigmentation pattern; pale blotches traverse upper part of anal flap; foot webbing formula I2-2 1/2 II2-3 1/3 III2-3IV3-1 1/2 V, eyes prominent.In light of the challenge to the characters thoracic fold,   et al. 2002).Adler & Dennis (1972:15) doubted that these specimens are P. mykter because they lack a rostral keel, webbing on fingers, and mottling on venter and flank.However, this conclusion is questionable because a rostral keel can be either present or absent in P. mykter, the webbing on the fingers is vestigial and hard to compare (Kaplan et al. 2015), and specimens can often lose their pigmentation over time (Simmons 1993).The juvenile specimens UIMNH 38023-25 appear to be P. chryses because they have a mottled pattern on the dorsum (Adler & Dennis 1972, Kaplan et al. 2015, this study), a character that is absent in P. bistincta and less clearly defined in P. mykter.Examination of UIMNH 38023-25 is needed to fully identify these specimens.
Finally, the adult female UMMZ 125376 appears to be P. mykter because it has a dark venter in preservative and small body size (likely SVL< 40 mm) (Adler & Dennis 1972).Adler & Dennis (1972) did not identify UMMZ 125376 as P. mykter because it lacks a prepollical spine, as determined via X-Ray.However, the prepollical spine could still be present in this specimen but not visible through X-Ray because of poor ossification, which we found in several individuals of P. bistincta (e.g., UMMZ 85453, USNM 114520, CNHM 111076).

Figure 3 .
Figure 3. Phylogenetic relationship of P. chryses within the Hylini.Numbers on the branches represent bootstrap values.

Figure 4 .
Figure 4. Phylogenetic relationship of P. chryses within Plectrohyla.Numbers on the branches represent bootstrap values.

Table 1 .
Skin thicknesses, differences in skin thicknesses, originally assigned character state, and newly assigned character state in frogs of the Plectrohyla bistincta group.
chryses was likely nested within Plectrohyla we analyzed smaller subsets of the taxa to see if we could determine the close relatives of P. chryses.*Adler(1965);**Caldwell(1974);***Adler and Dennis (1972); **** Canseco-Marquez et al. (2002); ***** Duellman (2001).RESULTSPhylogenetic analysis.-Includingall63 of the most-closely related DNA sequences in a phylogenetic analysis, P. chryses is nested within the Plectrohyla clade (Figs.3 and 4).Bootstrap scores are low, but they are generally low throughout the tree.When only Plectrohyla species and close relatives are analyzed, P. chryses falls close to P. mykter, P. calthula, and P. bistincta, i.e., within the P. bistincta group.Skin thickness.-Theskinthicknesses of the examined species, and the differences of skin thicknesses between species, appear in Table1.P. chryses has the thinnest skin of all the examined species.
(Adler & Dennis 1972, Kaplan et al. 2015)al fold (seeKaplan et al., 2015for discussion), a re-evaluation of the diagnoses of most species of the P. bistincta group is in order.Heimes 2011 showed it is neither P. arborescandens nor P. cyclada).We think KU 87605 is Charadrahyla nephila because it is nearly identical to the tadpole of this species(compare Duellman  2001; Figs.184B and 185B to Mendelson &  Campbell 1999; Fig. 3) and it inhabits the Sierra de Juarez in Oaxaca.Adler & Dennis (1972)suggested that the identities of several individuals from the Sierra Madre del Sur west of Chilpancingo, Guerrero (i.e., IPN CB 149-156; UIMNH 38023-25; UMMZ 125376), all of which had been previously referred to P. bistincta(Duellman 1964, Adler 1965), were still unknown.These authors were unable to assess if these specimens constitute one species, nor were they able to assign them to any of the known species of the area such as P. chryses or P. mykter.Nor could they determine if the specimens represent a new species because, with one exception (i.e., UMMZ 125376), the specimens are juveniles or poorly preserved adults(Adler & Dennis 1972), making them not comparable to species with undescribed juvenile forms.We did not examine the specimens IPN CB 149-156, UIMNH 38023-25, or UMMZ 125376 in this study.However, it would appear that the adult specimens IPN CB 149-156 are P. mykter because males have nuptial excrescences on the prepollex and on all the fingers(Adler & Dennis 1972, Kaplan et al. 2015), a very rare condition in Plectrohyla known only in P. ameibothalame (Canseco