Neotropical Hylid Frogs, Genus Smilisca

Usually a chorus is initiated by one duet and is quickly picked up by other individuals also calling in duets. A numerical representation of a chorus of eight frogs would approximate the following organization:

After the first one or two duets are initiated, the second individuals in the following duets usually call immediately after their respective partners have given the first notes. The other noteworthy aspect about the organization is that the entire chorus usually stops abruptly. Normally the first duet stops calling shortly before the others, but this is not invariable. Often one duet or one individual will emit several notes after the rest of the frogs have become silent. An interval of several minutes sometimes elapses before the chorus begins again. Successive choruses apparently are initiated by the same duet. Responses can be initiated artificially by imitating the call, and sometimes any loud noise will start a chorus.

Similar duets have been observed in S. phaeota. In this species the intervals are often much longer than the notes, and if two males are calling in close proximity, their calls can be mistaken for those of one individual. Smilisca phaeota does not congregate in large numbers; usually only two males call from one restricted site.

Smilisca sila has a call consisting of a primary note followed by one or more secondary notes. Males often call in duets, but not necessarily so. In a duet, the first male usually utters only primary notes until the second individual responds; then each individual produces a rapid series of secondary notes.

Smilisca puma also produces primary and secondary notes. Although individuals sometimes call alone, duets, trios, or quartets were more common. The chorus is initiated by one individual uttering primary notes until joined by the second, third, and fourth frogs. In one quartet in a marsh 7.5 kilometers west of Puerto Viejo, Costa Rica, on February 19, 1965, the same individual initiated four consecutive choruses. Each time the second member of the chorus was the same; the third and fourth frogs joined the chorus nearly simultaneously.

Individuals of S. sordida are usually irregularly situated along a stream. No duets or other combinations of individuals are apparent in the chorus structure, but once an individual calls, a frog nearby calls almost immediately; then a frog near the second individual calls, and so on. The resulting series of calls gives the impression that the sound is moving along the stream as successive individuals join the chorus and the first callers become quiet. It is not known if the same individual initiates successive choruses or if the order of calling is the same in subsequent choruses.

These limited observations on chorus structure in Smilisca show the presence of behavioral organization. The methods of establishing the organization and the significance of the call-order in breeding have yet to be discovered.

Calling males of S. baudini are often close together; some individuals have been observed almost touching one another, but no indication of territoriality or aggressive behavior has been witnessed. The more distant spacing of the stream-breeding species S. sila and S. sordida may be a function of calling-territories, but no direct evidence is available to substantiate this supposition.

Sex recognition and amplexus.—Observations on Smilisca baudini indicate that the calls of males attract females. At Tehuantepec, Oaxaca, México, a female was first observed about two meters away from a male calling at the edge of a rain pool; in a series of short hops she progressed directly towards the male, although vegetation obscured him until she was less than a meter away. When she approached to within about 20 centimeters of the male, he took notice of her, moved to her, and clasped her. At Chinajá, Alta Verapaz, Guatemala, a female swam directly across a pool about three meters wide to a calling male. Her line of movement took her within a few centimeters of a silent male, to whom she paid no attention. She stopped just in front of the calling male, which immediately clasped her. At a large muddy pond 4 kilometers west-northwest of Esparta, Puntarenas, Costa Rica, a female was observed swimming toward a small submerged tree; a male was calling from a branch about one meter above the water. The female climbed to a branch about 20 centimeters below the male, which upon seeing her there immediately jumped down and clasped her. These few observations of S. baudini show that in this species females are capable of locating calling males by means of phono-orientation; visual reception on the part of females seems to be secondary. Contrariwise, males apparently become aware of the proximity of females by seeing them; once a male sees a female he usually tries to clasp her. Possibly the males receive stimuli by means of chemo-reception, but in each observed instance the male obviously looked at the female.

Amplexus is axillary in all members of the genus. Normally amplexing males hunch their backs and press their chins to the females' backs. Clasping pairs are usually found at the edge of the water, but sometimes amplexus takes place in trees or bushes.

Egg deposition.—Oviposition has been observed only in Smilisca baudini. On the night of June 28, 1961, at Chinajá, Alta Verapaz, Guatemala, a clasping pair was observed at the edge of a shallow rain pool. After sitting for several minutes in shallow water, the female (with male on her back) swam part way across the pool and grasped an emergent stick with one hand. The female's body was nearly level with the surface of the water, and her hind legs were outstretched as deposition commenced; eggs were extruded rapidly. After a few seconds the female moved slowly to another twig a few centimeters away and deposited more eggs. This process was repeated until the female was spent. The spawn resulted in a surface film covering roughly one square meter. It is doubtful if this type of egg deposition occurs in any other species in the genus, especially those that lay their eggs in streams.

The breeding calls of the six species of Smilisca are alike in their explosive nature. Calls are emitted quickly with a short burst of air filling the vocal sac, which immediately deflates. Phonetically the calls can be described as a single "wonk" or series of such notes in S. baudini and S. cyanosticta, a low growl in S. phaeota, a relatively high pitched rattle in S. sordida, and a low squawk usually followed by one or more rattling secondary notes in S. puma and S. sila. Quantitatively, the calls of the six species differ in number of notes, duration of notes, and in pitch (Table 8, Pls. 10 and 11). Although no measurements were taken on the intensity of the calls, we observed in the field that each of the species has a loud voice. The call of S. baudini seems to carry farther than any of the others.

Call rate.—The rate at which call-groups are produced varies from one every few seconds to one in several minutes. In S. baudini, cyanosticta, phaeota, and sordida, call-groups are produced as frequently as every 12 seconds, but usually more time elapses between call groups. In S. sordida, five or more minutes sometimes elapse between call-groups. The interval is somewhat less in S. phaeota. Calls are repeated at much shorter intervals in S. puma (5-55 seconds) and S. sila (4-20 seconds).

Notes per call-group.—Except for S. puma and S. sila, the series of notes produced in any given call of a species of Smilisca is essentially the same; there is no differentiation into primary and secondary notes. Smilisca cyanosticta and S. phaeota emit only one or two relatively long notes per call-group, whereas S. baudini and S. sordida produce as many as 15 and 6 notes, respectively. Males of S. puma and S. sila often produce only the primary note; sometimes this is done several times before the secondary notes are produced. For example, one S. puma (KU 91711; tape No. 379) produced the following number of notes in consecutive call-groups: 1, 1, 1, 1, 2, 2, 3, 1, 4; secondary notes are present in only four of the nine call-groups. A typical series of consecutive call-groups in S. sila (KU 91852; Tape No. 385) has 1, 1, 1, 2, 4, 2 notes per call-group; secondary notes are present in only half of the call-groups. Smilisca puma apparently always produces at least two primary notes before emitting secondary notes; sometimes only primary notes are produced in one series of calls. The number of secondary notes following a given primary varies from one to nine; the modal number is one, and the mean is three in 27 call-groups. Smilisca sila frequently begins a series of calls with two or more primary notes, but sometimes the first primary note is followed immediately by two or more secondary notes. The number of secondary notes following a given primary varies from one to five; the modal number is one, and the average is two in 13 call-groups.

Duration.—The average duration of call-groups consisting of two or more notes is 1.18 seconds in S. baudini; 1.02 in cyanosticta, 0.91 in phaeota, 1.32 in puma, 1.48 in sila, and 1.29 in sordida. Although there is considerable variation in the lengths of the notes (only primary notes in S. puma and sila are considered here), S. cyanosticta, phaeota, and sordida have noticeably longer notes than do the other species (Table 8). The secondary notes are longer than the primary notes in S. puma (average 0.27 secs. as compared with 0.13 secs.) and in S. sila (average 0.25 secs., as compared with 0.16 secs.).

Note repetition rate.—The rate at which notes in call-groups containing two or more notes are produced varies in S. baudini from 2.5 to 7.1 (average, 3.7) calls per second; cyanosticta, 1.8-2.1 (1.9); phaeota, 2.0-2.4 (2.2); puma, 1.9-2.9 (2.2); sila, 1.3-2.4 (1.8); and sordida, 1.5-2.6 (2.1). Smilisca baudini, which has notes of short duration (0.09 to 0.13 seconds), has the fastest note-repetition rate. Although the individual notes of S. cyanosticta and S. phaeota are relatively long (average, 0.38 and 0.31 seconds, respectively), the intervals between the notes is short; consequently, their note-repetition rates do not differ greatly from those of S. puma and S. sila, which have shorter notes (average, 0.13 and 0.16 seconds, respectively) but longer intervals between notes.

Pulse rate.—Pulses vary in frequency from 78 to 240 per second in the calls analyzed (only primary notes in S. puma and S. sila), but the variation in any given species is much less than that in the entire genus (Table 8). Smilisca puma is outstanding in having a high pulse rate, which is approached only by that of S. baudini. Even in the species having the lowest pulse rates, the pulsations are not audible. The secondary notes produced by S. puma and S. sila have a slower pulse rate than the primary notes; often the pulses are audible. In S. puma the pulse rate of secondary notes is sometimes as low as 48 pulses per second, and in S. sila still lower (as low as 40 pulses per second). The upper limits of pulse rate in the secondary notes in these species merge imperceptibly with the rates of the primary note; consequently, on the basis of pulse rate alone it is not always possible to distinguish primary from secondary notes.

Frequency.—Smilisca produces noisy (as opposed to more musical) calls, and the energy is distributed throughout the frequency spectrum; the calls are poorly modulated, except in S. sordida, in which two usually discrete bands of frequency are present (Pl. 11C). For the most part the calls of Smilisca consist of little modified energy of the fundamental frequency and of its harmonics, some of which are emphasized.

The upper frequency range varies within each species and even within the calls of one individual. Smilisca phaeota has the lowest upper frequencies; no calls ranged above 4400 cycles per second (cps.), and half of the calls never exceeded 3000 cps. Smilisca cyanosticta produces calls in which the upper frequency is below 7000 cps. and usually below 6000 cps. Likewise, S. puma produces calls that are below 7000 cps., whereas S. sila has frequencies of up to 8400 cps. In both S. baudini and S. sordida, the highest frequencies attained are about 9100 cps. Variation in the highest frequencies in a series of consecutive calls by one individual frog was noted in all species. Such variation is especially prevalent in S. puma; for example one individual (KU 87771; Tape No. 376) recorded at a temperature of 24° C. at 7.5 kilometers west of Puerto Viejo, Heredia Province, Costa Rica, on July 31, 1964, produced three consecutive primary notes having upper frequencies of about 6000, 4000, and 4000 cps., respectively. Apparently in a given species the production of the higher frequencies in some notes and not in others is correlated with the amount of distention of the vocal sac and is not dependent upon the structure or tension of the vocal cords.

Although the dominant frequency in S. sordida is lower than that in S. baudini and S. cyanosticta, the call of the former is audibly higher-pitched. This is due primarily to the emphasis on certain harmonics at a high frequency (sometimes as high as 9000 cps.) in S. sordida, whereas in S. baudini and other species, if harmonics are present at those frequencies, they are not emphasized.

The fundamental frequencies are as low as 90 cps. in S. sila and S. sordida and as high as 200 cps. in S. puma (Table 8). The fundamental frequency seemingly is relatively unimportant in determining the general pitch of the call, a characteristic most dependent on the dominant frequency and emphasized harmonics in the higher-frequency spectrum. In none of the species is the fundamental the dominant frequency. In the low-pitched call of S. phaeota the dominant frequency is the third harmonic (the second harmonic above the fundamental frequency, which is the first harmonic). In all other species a much higher harmonic is dominant; for examples, in S. cyanosticta harmonics from 10 to 15 are dominant; in S. baudini, 15-19; and S. sila, 20-30.

A glance at the audiospectrographs and their accompanying sections (Pls. 10 and 11) reveals the presence of two emphasized bands of frequency in all species except S. phaeota, in which only the lower band is present. These two bands of emphasized harmonics are part of a continuous, or nearly continuous, spread of energy throughout the frequency spectrum, except in S. sordida in which the bands are usually distinct. As shown in the sections, certain harmonics in each of the bands are emphasized with nearly equal intensity. Therefore, with the exception of S. phaeota, the calls of Smilisca are characterized by two major frequencies, one of which is the dominant frequency and the other is a subdominant frequency (Table 8). The upper major frequency is dominant in all calls in S. baudini and S. cyanosticta, but either major frequency may be dominant in other species. The upper major frequency is dominant in 65 per cent of calls by S. puma, 87 per cent in S. sila, and 68 per cent in S. sordida. Individuals of these three species sometimes produce a series of calls in which the dominant frequency changes from one of the major frequencies to the other. Four consecutive notes emitted by an individual of S. sordida recorded 13 kilometers east-northeast of Golfito, Puntarenas Province, Costa Rica, had dominant frequencies of 910, 1950, and 750 cps., respectively. In each case, an alternation of major frequencies took place in respect to dominance. An individual of S. puma from 7.5 kilometers west of Puerto Viejo, Costa Rica, produced a primary note followed by one secondary note; each note had major frequencies at 600 and 1800 cps.; the dominant frequency of the primary note was at 1800 cps., whereas in the secondary note the dominant frequency was at 600 cps. The difference in emphasis on the major frequencies is so slight that shift in dominance is not audible.

Effect of temperature on calls.—The present data are insufficient to test statistically the correlation between temperature and variation within certain components of the calls in Smilisca, but even a crude graph shows some general correlations. The widest range of temperatures is associated with the recordings of S. baudini. Three individuals recorded at a temperature of 30° C. at Tehuantepec, Oaxaca, had pulse rates of 180 pulses per second and fundamental frequencies of 160-180 cps., as compared with an individual recorded at a temperature of 17° C., which had a pulse rate of 140 and a fundamental frequency of 135 cps. All individuals of S. baudini recorded at higher temperatures had faster pulse rates and higher fundamental frequencies. Pulse rates differ in the other species in the genus but less strikingly (probably owing to narrower ranges of temperatures at which recordings were made). In five recordings of S. sordida made at 20° C. the pulse rate is 80-90, as compared with four recordings made at 25° C. having pulse rates of 120-135. Thirteen recordings of S. sila made at 17° C. have pulse rates of 97-112 (average 105); one individual recorded at 26° C. has 120 pulses per second. Seemingly no correlation exists between temperature and other characteristics of the calls, such as duration and rate of note-repetition.

The breeding call as an isolating mechanism.—Blair (1958), Bogert (1960), Duellman (1963a), Fouquette (1960), Johnson (1959), and others have provided evidence that the breeding calls of male hylids (and other anurans) serve as isolating mechanisms in sympatric species. In summarizing this discussion of the breeding calls of Smilisca we want to point out what seem to be important differences in the calls that may prevent interspecific hybridization in sympatric species of Smilisca.

The genus is readily divided into two species-groups on morphological characters; this division is supported by the breeding calls. In the species of the baudini group the calls are unmodulated and lack secondary notes. In the sordida group the calls either have secondary notes or are modulated.

Smilisca baudini occurs sympatrically with S. cyanosticta and S. phaeota; where they occur together, both species sometimes breed in like places at the same time. We are not aware of these species breeding synchronously at exactly the same site, although S. baudini and S. cyanosticta were calling on the same nights and less than 100 meters apart in Oaxaca in June, 1964. Regardless of their respective breeding habits, sympatric species have calls that differ notably. Except for the higher fundamental and dominant frequencies, the calls of S. cyanosticta and S. phaeota closely resemble one another, but the calls of both species differ markedly from that of S. baudini. The geographic ranges of S. cyanosticta and S. phaeota are widely separated.

The calls of the allopatric species S. puma and S. sila are not greatly different. Smilisca sordida has a distinctive call and occurs sympatrically with S. puma and S. sila. In the streams in southern Costa Rica S. sordida and S. sila breed synchronously, but the high-pitched modulated call of the former is notably different from the lower, unmodulated call of S. sila.

The data indicate that the calls of related sympatric species differ more than the calls of related allopatric species. We postulate that these differences evolved to support the reproductive isolation of the sympatric species. The data are insufficient to determine geographic variation in the calls and to determine if differences in the calls are enhanced in areas of sympatry as compared with the allopatric parts of the ranges.

Other calls.—As stated previously, there is no direct evidence of territoriality in Smilisca; we have heard no calls that can be definitely identified as territorial. Single notes of S. baudini, phaeota, and sila have been heard by day, just prior to rains, or during, or immediately after rains. Such calls can be interpreted as "rain calls," which are well known in Hyla eximia and Hyla squirella. Distress calls are known in several species of Rana and in Leptodactylus pentadactylus; such calls result from the rapid expulsion of air over the vocal cords and with the mouth open. Distress calls have been heard from S. baudini. At Charapendo, Michoacán, México, a male that had one hind limb engulfed by a Leptodeira maculata emitted several long, high-pitched cries. A clasping pair of S. baudini was found in a bush at the edge of a marshy stream 2 kilometers northeast of Las Cañas, Guanacaste Province, Costa Rica. When the pair was grasped, the female emitted a distress call.

Eggs of S. baudini, cyanosticta, and phaeota have been found in the field, and eggs of S. sila have been observed in the laboratory. The eggs of S. puma and sordida are unknown. Insofar as known, Smilisca baudini is unique in the genus in depositing the eggs in a surface film. Each egg is encased in a vitelline membrane, but individual outer envelopes are lacking. The eggs are small; the diameter of recently-deposited eggs is about 1.3 mm. and that of the vitelline membrane is about 1.5 mm. The eggs of S. cyanosticta and phaeota are deposited in clumps, and the eggs are larger than those of S. baudini. Diameters of eggs of S. cyanosticta are about 2.3 mm., and those of the outer envelopes are about 4.0 mm. Artificially fertilized eggs of S. sila raised in the laboratory have diameters of about 2.4 mm.; the diameter of the outer envelopes is about 4.9 mm.

In order to determine the reproductive potential of the six species, ovulated eggs were removed from females and counted. The numbers of eggs recorded are: 3 S. baudini—2620, 2940, 3320; 1 S. cyanosticta—910; 3 S. phaeota—1665, 1870, 2010; 1 S. puma—518; 3 S. sila—369, 390, 473; 3 S. sordida—524, 702, 856. These limited data indicate that the large species (S. baudini, cyanosticta, and phaeota) have more eggs than do the smaller species. The stream-breeding species (S. sila and sordida) have relatively few eggs by comparison with the pond-breeders. Possibly this is a function of size of eggs rather than a correlation with the site of egg-deposition.

The tail is about half again as long as the body in S. baudini, cyanosticta, phaeota, and puma; in these species the caudal musculature is moderately heavy, and the caudal fins are deep. The caudal musculature is upturned distally in S. baudini and phaeota, and the dorsal fin extends anteriorly onto the body in these two species and in S. puma. The tail is about twice as long as the body in S. sila and sordida. In both species the caudal fins are shallow in comparison with the depth of the caudal musculature, especially in S. sordida (Fig. 14); in neither species does the dorsal fin extend anteriorly onto the body.

Mouthparts.—The mouth of S. sordida is completely bordered by two rows of papillae, whereas in the other species the median part of the upper lip is devoid of papillae. Smilisca baudini and puma have two rows of papillae; S. sila has one complete row (except medially on the upper lip) and one incomplete row, and S. cyanosticta and phaeota have only one row (Fig. 15). All species have numerous papillae in the lateral fold; the fewest lateral papillae are found in S. cyanosticta and phaeota. Although all species have two rows of teeth in the upper jaw and three rows in the lower jaw, specific differences in the nature of the rows exist between certain species. The second upper tooth-row is narrowly interrupted medially in S. sila and sordida and broadly interrupted in the other species. The first upper row is strongly arched in S. puma, moderately arched in S. baudini and sila, and weakly arched in the other species. In all species the third lower tooth-row is the shortest, only slightly so in S. sila and sordida, but only about half the length of the second lower row in S. puma.

The beaks are well developed and finely serrate in all species. The lower, broadly V-shaped, beak is slender in S. puma, rather robust in S. baudini and sila, and moderately heavy in the other species. The lateral processes of the upper beak are shortest in S. puma and longest in S. baudini and sordida. In the latter the inner margin of the upper beak and lateral process have the form of a shallow S, whereas in the other species the inner margin of the upper beak forms a continuous arch with the lateral processes (Fig. 15).

Coloration.—The tadpoles of Smilisca lack the bright colors or bold markings characteristic of some hylid tadpoles; even so, the subdued colors and arrangement of pigments provide some distinctive markings by which the species can be distinguished from one another. The species comprising the baudini group (S. baudini, cyanosticta, and phaeota) are alike in having the body brown or grayish brown dorsally and transparent with scattered brown pigment ventrally. A cream-colored, crescent-shaped mark is present on the posterior edge of the body; this mark is usually most noticeable in S. baudini and least so in S. cyanosticta. Other differences in coloration in members of the baudini group are relative and subtle. Smilisca phaeota usually is more pallid than baudini, and cyanosticta usually is darker than baudini; both species have larger dark markings on the tail than does S. phaeota. Smilisca baudini has a dark streak on the middle of the anterior one-fourth of the tail (Figs. 11-13).

Smilisca puma is distinctive in having a grayish brown body and dark gray reticulations on the tail. Smilisca sila and sordida are distinctive in having pairs (sometimes interconnected) of dark marks on the dorsal surfaces of the caudal musculature, and in dorsal view the tail appears to be marked with dark and pale creamy tan transverse bars. These dark marks, as well as the small flecks on the tail, are brown in S. sila and red in sordida. Smilisca sila has dark brown flecks on the dorsal surface of the body and small white flecks laterally; these markings are absent in S. sordida (Fig. 14).

Descriptions of the coloration of living tadpoles are given in the accounts of the species.

Hatchlings of three species (S. baudini, cyanosticta, and phaeota) are available. These larvae have non-functional eyes and large oral suckers. By the time the larvae have developed to stage 21, external gills are present, the caudal musculature and caudal fin have been differentiated, and the head is distinguishable from the body. In stage 21 oral suckers and a large amount of yolk are still present.

The developmental data on the four species show no significant variations; consequently, we will describe the development of only one species, Smilisca phaeota (Table 11, Figs. 13 and 16).

Stage 21.—Bulging cream-colored yolk mass, transparent cornea, and moderately long, unbranched filamentous gills, and oral suckers present; mouth having irregular papillae on lower lip; teeth and beaks absent; caudal myomeres distinct; pigmentation uniform over body and caudal musculature; caudal fin transparent with scattered small flecks.

Stage 25.—Operculum complete; gills absent; sinistral spiracle apparently functional; cloacal tail-piece, nasal capsules, and external nares present; gut partly formed; mouth bordered by single row of papillae, except medially; small papillae present in lateral fold of lips; two upper and three lower tooth-rows present, but not fully developed; beaks apparently fully developed; depth of dorsal and ventral fins less than depth of caudal musculature: tip of tail upturned; pigment on body most dense on dorsum and sides; faint, nearly pigmentless crescent-shaped mark on posterior edge of body; concentrations of pigment forming small spots on tail.

Stage 28.—Mouthparts complete; limb bud about half as long as thick; other structural features and coloration closely resemble those in stage 25.

Stage 30.—Limb bud approximately twice as long as thick; body as deep as wide; dorsal fin deepest just posterior to body; ventral fin deeper than caudal musculature; tail sharply upturned distally; anal tube dextral; brown pigment sparse on flanks.

Stages 34, 36, 37, and 38.—Stage 34, foot paddle-shaped with four toe buds; stage 36, five toe buds; stages 37 and 38, lengthening of toes. In all four stages, spiracle persistent, and pigmentation resembling that of early stages.

Stage 39.—Metatarsal tubercle present; greatest total length (36.9 mm.) attained.

Stage 40.—Subarticular tubercles prominent; skin over forelimbs transparent; cloacal tail-piece and spiracle absent; outer tooth-rows degenerating; caudal fins shallower than in preceding stages; distal part of tail nearly straight; size of dark markings on tail decreased; pigment present on hind limb.

Stage 43.—Forelimbs erupted; larval mouthparts absent; corner of mouth between nostril and eye; transverse bands present on hind limbs; tail greatly reduced (about 8 mm. in length).

Stage 44.—Sacral hump barely noticeable; tail reduced to a stub; corner of mouth at level of pupil of eye; dorsal surfaces pale olive-green; venter white.

Changes proceed in a definite pattern during the growth and development of tadpoles. Larval teeth are absent in hatchlings; the inner tooth-rows develop first, and the third lower row last. At metamorphosis the third lower row is the first to be lost. The tail increases gradually in length relative to the body. In stage 25 the tail is 52.1 per cent of the total length, and in stage 36, 64.6 per cent. In later stages the tail becomes relatively shorter through resorption. Duellman and Klaas (1964:320) noted a great size-variation in Triprion tadpoles in stage 25. No such variation is apparent in any stage of any of the species of Smilisca studied.

The growth and development of the other species of Smilisca do not differ significantly from that of S. phaeota. The tadpoles of S. sila and sordida from streams have relatively longer tails at hatching. For example, in tadpoles of S. sordida the average length of tail is 64.0 per cent of the body-length in stage 25, and in stage 37, 67.0 per cent.