Allobates sp.

Kok, Philippe J. R., Hölting, Monique & Ernst, Raffael, 2013, A third microendemic to the Iwokrama Mountains of central Guyana: a new “ cryptic ” species of Allobates Zimmerman and Zimmerman, 1988 (Anura: Aromobatidae), Organisms Diversity & Evolution (New York, N. Y.) 13 (4), pp. 621-638 : 623-635

Allobates sp. “Iwokrama” Fouquet et al. 2013: tables 1–3, Fig. 3

( Figs. 2–5 , Table 1)

Holotype IRSNB 4155 (field number PK 3672 ), an adult male collected by Philippe J. R. Kok, 3 December 2012 at 1030 hours, Turu Falls , base of Iwokrama Mountains, Potaro- Siparuni District, Guyana (04°24’46”N, 058°47’02”W, ca. 160 m elevation).

Paratopotypes (n=5) Three adult males ( IRSNB 4157–59 , field numbers PK 3796 and PK 3797–98), collected by Jonathan Clegg, and Philippe J. R. Kok and Armstrong Simon, respectively, with same data as holotype ; one adult male ( MTD 47884 , field number 047), collected by E. Meyer, 25 May 2010, on Turu Falls trail, right bank of Turu creek going upstream to Turu Falls; and one adult female ( IRSNB 4156 , field number PK 3795, designated as the allotype), collected by Philippe J. R. Kok and Armstrong Simon with same data as holotype .

Referred specimen A juvenile (IRSNB 15915, field number PK 3526), collected by Philippe J. R . Kok and Lawrence Antoin, 19 March 2011, close to the top of one of the highest of the type series of Allobates amissibilis sp. nov. Abbreviations are defined in the text peaks (unnamed) of the Iwokrama Mountains (04°20’11”N, 058°46’54”W, ca. 950 m elevation) . One leg removed and preserved in 95 % ethanol for molecular analyses.

Etymology The specific name is considered a noun in apposition and is derived from the ecclesiastical Latin word amissibilis literally meaning “that may be lost”, in reference to the threats that this potentially highly restricted species may face in the near future as a result of increasing human pressure due to the aesthetic attractiveness of the locality where it occurs.

Definition and diagnosis The new species is characterised by the following unique combination of characters: (1) SVL small, adult males 16.9 ± 0.6 mm (range 16.3–17.5 mm, n =5), adult female 17.8 mm (n =1); (2) in life skin on dorsum shagreen to granular, always more granular posteriorly; (3) median lingual process absent; (4) annulus tympanicus visible anteroventrally, tympanum posterodorsally barely visible; (5) vocal sac distinct, subgular; (6) maxillary teeth present, very small; (7) distal tubercle on Finger IVabsent; (8) tip of Finger IV not reaching distal subarticular tubercle on Finger III when fingers adpressed; (9) Finger I 7–12 % longer than Finger II (n =6); (10) finger discs weakly expanded, middle section of last phalange of Finger III 80 % of Finger III disc width; (11) lateral fringes on fingers absent; (12) metacarpal ridge absent; (13) Finger III not swollen in male, overall morphology of Finger III identical in male and female; (14) supracarpal pad absent; (15) thenar tubercle conspicuous, weakly protuberant; (16) black arm gland absent; (17) nature of tarsal keel variable, usually short, tuberclelike, curved or directed transversely across tarsus and not extending from metatarsal tubercle, rarely straight or very weakly curved, extending proximolaterad from halfway between metatarsal tubercle and dermal thickening; (18) tarsal fringe absent; (19) Toe IV disc moderately expand- ed, middle section of last phalange of Toe IV 50–60 % of Toe IV disc width; (20) basal webbing only between Toes III–IV; (21) metatarsal fold absent; (22) weak pre- and postaxial fringes on Toes II–III (preaxial ones more conspicuous) and weak preaxial fringe on Toe IV (usually more conspicuous than on Toes II–III), present in male, more weakly developed (almost undetectable) in female; (23) body colouration cryptic with no distinct pattern on adult dorsum, but dorsal “hourglass” pattern in the single juvenile; (24) dorsolateral stripe absent or present, in which case as a diffuse narrow line, extending from tip of snout to vent area; (25) oblique lateral stripe present as a diffuse broad area, extending from groin to about midbody length; (26) ventrolateral stripe present, whitish to yellowish white, with no distinct dark blotches ventrolaterally; (27) paracloacal marks present, pale to orangish yellow; (28) enlarged cloacal tubercles absent; (29) distinctly enlarged tubercle on each eyelid; (30) throat colouration sexually dimorphic, pinkish grey, suffused with dark pigment in male, immaculate cream to yellow in female; (31) belly colouration not clearly sexually dimorphic, cream to yellow in both sexes; (32) dark dermal collar absent; (33) iris metallic copper with black reticulations in life, incomplete pupil ring (present only above the pupil); (34) large intestine unpigmented; (35) adult testes unpigmented; (36) mature oocytes pigmented; (37) diurnal habits, males calling by day; (38) advertisement call characterized by calls of usually 9–12 notes consisting of seven harmonics, emitted between silent intervals of usually ca. 2–6 s.

Generic allocation The new species is assigned unambiguously to the genus Allobates by the combination of the above characters # 2, 3, 10, 20, 23, 25, 32, 35 and based on molecular phylogenetic relationships (see Fig. 3 in Fouquet et al. 2013 in which the new species is listed as “ A. sp. “Iwokrama””).

Morphological comparisons with congeneric species Available data converge to suggest that several cryptically coloured Allobates have relatively restricted distributions (the authors unpubl. data, see also Caldwell et al. 2002; Lima et al. 2009; Simões et al. 2013). Allobates amissibilis has only been collected at two localities in the Iwokrama Mountains despite the fact that several intensive surveys have been conducted by the authors and by independent research groups in the entire Iwokrama Forest and peripheral areas. All surveys conducted outside the Iwokrama Mountains area were unable to secure any additional specimen of the new species. Comparisons with congeners thus focus on Allobates species from the Guiana Shield region (sensu Señaris and MacCulloch 2005) and extralimital Amazonian species, but we also compare the new taxon with known species within the same phylogenetic clade (sensu Fouquet et al. 2013). Morphological comparisons are based on examination of museum specimens (see Appendix) and/or original descriptions/redescriptions.

Eight species of Allobates are currently reported from the Guiana Shield region ( Señaris and MacCulloch 2005; Frost 2013): A. femoralis , widespread in the lowlands of the Amazon Basin and the Guianas; A. granti ( Kok, MacCulloch, Gaucher, Poelman, Bourne, Lathrop, and Lenglet, 2006) , from French Guiana and Suriname [some populations recently reported from Suriname by Ouboter and Jairam (2012) likely belong to an undescribed taxon; see also Fouquet et al. 2012a)]; A. marchesianus ( Melin, 1941) reported as widespread in the Amazon Basin, but its presence outside the type locality (Missão Taracuá, upper Rio Uaupés, north of the Amazon river, Amazonas state, Brazil) is uncertain (see Caldwell et al. 2002); A. myersi ( Pyburn, 1981) , from the Amazonian lowlands of Colombia; A. spumaponens , disjunctly distributed in central Guyana and northern Pará, Brazil ( Avila-Pires et al. 2010); A. sanmartini ( Rivero, Langone, and Prigioni, 1986) , endemic to Las Majadas, Bolívar state, Venezuela; A. sumtuosus ( Morales, 2002) , type locality (Reserva Biológica Río Trombetas) in Pará state, Brazil, a disjunct population reported from Loreto, Peru; and A. undulatus ( Myers and Donnelly, 2001) , endemic to Cerro Yutajé, Amazonas state, Venezuela.

Allobates amissibilis is immediately distinguished from A. femoralis and A. myersi by the absence of conspicuous flashmarks on dorsal surface of thighs, by the absence of dark mottling on venter (both present in A. femoralis and A. myersi ), and by smaller size (max SVL less than 18.0 mm in A. amissibilis vs more than 20.0 mm in A. femoralis and A. myersi ).

Allobates amissibilis is morphologically similar to A. granti and confusion between these species is easy, but the new species can be distinguished readily from A. granti by the following combination of characters (those of A. granti in parentheses): larger SVL in male (16.3–17.5 mm, n =5 vs 14.9–16.7, n = 8 in A. granti ), dorsal surface of hindlimbs light to moderately dark brown (dark bluish grey), belly colouration not clearly sexually dimorphic, cream to yellow in both sexes (strongly dimorphic, white in male, yellow in female), oblique lateral stripe as a diffuse broad area, extending from groin to about midbody length (oblique lateral stripe less conspicuous, not reaching midbody length, usually barely distinct), and thenar tubercle weakly protuberant (distinctly protuberant).

Allobates amissibilis is immediately distinguished from A. marchesianus by lacking a broad distinct dorsolateral stripe (when present a diffuse narrow line in A. amissibilis ), and in having cream to yellow belly in male (uniform grey in A. marchesianus ).

Allobates amissibilis is also morphologically very similar to A. spumaponens (the geographically closest cryptically coloured Allobates species), but the new species can be readily distinguished from A. spumaponens by the following combination of characters (those of A. spumaponens in parentheses): slightly larger SVL in male (16.3–17.5 mm, n =5 vs 13.2– 17.0, n = 6 in A. spumaponens ), belly colouration not clearly sexually dimorphic, cream to yellow in both sexes (strongly dimorphic, hyaline-white in male, yellow in female), throat colouration clearly sexually dimorphic, throat suffused with dark pigment in male, immaculate yellow in female (poorly dimorphic, pale with very discrete spotting on chin and margin of lower lip in male, immaculate in female), oblique lateral stripe as a diffuse broad area, extending from groin to about midbody length (oblique lateral stripe less conspicuous, not reaching midbody length), dorsolateral stripe usually absent, as a diffuse narrow line when present (dorsolateral stripe more conspicuous, broader), conspicuous dark grey blotches on flanks absent (present), and faint dark brown transverse crossbars on legs (absent).

Allobates amissibilis is easily distinguished from A. sanmartini in having a smaller SVL (maximum SVL 17.8 mm vs 25.4 mm in A. sanmartini ), tympanum smaller than half the size of eye (larger than half the size of eye in A. sanmartini ), and Finger I longer than II (shorter in A. sanmartini ).

Allobates amissibilis is mainly distinguished from A. sumtuosus by having Finger III not swollen in male (reported as swollen in the original description of A. sumtuosus ), a broad oblique lateral stripe extending from groin to about midbody length (reported as absent in the original description of A. sumtuosus ), dorsolateral stripe usually absent, as a diffuse narrow line when present (dorsolateral stripe more conspicuous and broader in A. sumtuosus ), throat suffused with dark pigment in male (reported as immaculate in A. sumtuosus ), and faint dark brown transverse crossbars on legs (absent in A. sumtuosus ).

Allobates amissibilis is distinguished readily from A. undulatus by having a smaller SVL (maximum SVL 17.8 mm vs 25.0 mm in A. undulatus ), by lacking wavyedged dorsal markings (present in A. undulatus ), and a supracarpal pad atop wrist in male (present in A. undulatus ).

Six extralimital species of Allobates are known to occur in Amazonia, in peripheral areas to the Guiana Shield region: A. caeruleodactylus ( Lima and Caldwell, 2001), known only from three localities south of the Amazon river in Amazonas state, Brazil; A. crombiei ( Morales, 2002) , reported only from the drainage of the lower Río Xingú, south of the Amazon river, Pará state, Brazil; A. grillisimilis Simões, Sturaro, Peloso, and Lima, 2013 , known only from south of the Amazon river, in the Madeira-Tapajós interfluve, Amazonas state, Brazil; A. masniger ( Morales, 2002) , known only from the lower drainages of the Río Tapajos, south of the Amazon river, Pará state, Brazil; A. nidicola ( Caldwell and Lima, 2003), reported from south of the Amazon river in the Amazonas and Rondônia states, Brazil; and A. paleovarzensis Lima, Caldwell, Biavati, and Montanarin, 2010, reported from a few localities on both sides of the Amazon river, Amazonas state, Brazil ( Kaefer and Lima 2012). Comparisons with these species are provided below.

Allobates amissibilis is distinguished immediately from A. caeruleodactylus by larger SVL in male (16.3–17.5 mm, n =5 vs 15.0–16.3, n = 12 in A. caeruleodactylus ), by lacking sky-blue digits and blue discs on toes in male (present in A. caeruleodactylus ), and by having faint dark brown transverse crossbars on legs (absent in A. caeruleodactylus ).

Allobates amissibilis is distinguished readily from A. crombiei by smaller SVL in male (16.3–17.5 mm, n =5 vs 17.4–19.0, n = n/a in A. crombiei ), and by having a broad oblique lateral stripe extending from groin to about midbody length (absent in A. crombiei ).

Allobates amissibilis differs from A. grillisimilis mainly by larger SVL in male (16.3–17.5 mm, n=5 vs 12.8–15.9, n = 33 in A. grillisimilis ), by having the belly colouration cream to yellow in both sexes (white to translucent in both sexes in A. grillisimilis ), and the throat suffused with dark pigment in male (few melanophores located on chin only).

Allobates amissibilis is distinguished from A. masniger by the following combination of characters (those of A. masniger in parentheses): smaller SVL in male (16.3–17.5 mm, n=5 vs 17.9–19.5, n=n/a in A. masniger ), absence of conspicuous dorsolateral stripe (usually present and conspicuous), presence of a broad oblique lateral stripe extending from groin to about midbody length (usually absent), absence of dark blotches ventrolaterally (present), belly cream to yellow (grey or light grey), and lighter throat colour (usually very dark).

Allobates amissibilis is distinguished from A. nidicola by smaller SVL in male (16.3–17.5 mm, n =5 vs 18.5–21.0, n = 30 in A. nidicola ), and by the same combination of characters mentioned above to distinguish A. amissibilis from A. masniger [ A. masniger and A. nidicola are morphologically almost identical ( Tsuji-Nishikido et al. 2012)].

Allobates amissibilis mainly differs from A. paleovarzensis by the combination of much smaller SVL in male (16.3– 17.5 mm, n =5 vs 18.3–22.4, n = 31 in A. paleovarzensis ), Finger III not swollen in male (weakly swollen in A. paleovarzensis ), and lack of a broad dorsolateral stripe (usually present in A. paleovarzensis , when present a diffuse narrow line in A. amissibilis ).

Two described species have recently been assigned to a phylogenetic clade that is shared with Allobates amissibilis (see Fouquet et al. 2013, under the name A. sp “Iwokrama”): A. conspicuus ( Morales, 2002) , from eastern Peru and western Brazilian Amazonia; and A. insperatus ( Morales, 2002) , from the eastern slopes of the Andes in Ecuador.

Allobates amissibilis is distinguished immediately from A. conspicuus by the combination of a larger SVL in male (16.3–17.5 mm, n =5 vs 15.4–16.5, n =n/a in A. conspicuus ), lack of a broad dorsolateral stripe (present in A. conspicuus , when present a diffuse narrow line in A. amissibilis ), presence of a broad oblique lateral stripe extending from groin to about midbody length (reported as absent in the original description of A. conspicuus ), and throat colour sexually dimorphic (not sexually dimorphic in A. conspicuus ).

Allobates amissibilis is distinguished readily from A. insperatus by the lack of a broad dorsolateral stripe (present in A. insperatus , when present a diffuse narrow line in A. amissibilis ), and a cream to yellow belly in male and female (white in both sexes in A. insperatus ).

Description of the Holotype An adult male 16.3 mm SVL ( Figs. 2a,b ; 3a–c , additional measurements in Table 1) in good condition, except for a few superficial scars on posterior body and a large incision in the left thigh for tissue sample; body slender; head 1.1x wider than long; head length 30.0 % of SVL; head width 33.1 % of SVL; snout broadly rounded in dorsal view, acutely rounded in lateral view, extending past lower jaw, snout 57.1 % of head length. Nares located laterally, opening posterolaterally; canthus rostralis rounded, loreal region slightly concave, flaring slightly at upper lip; internarial distance 38.9 % of head width; eyenaris distance 30.6 % of head length; 68.2 % of eye length. Tympanum subcircular, directed posterolaterally, 40.1 % of eye length, separated from eye by a distance equal to 25.6 % of tympanum diameter; supratympanic fold absent, supratympanic area slightly concave; tympanic annulus visible anteroventrally, posterodorsal aspect of tympanum barely visible. Tongue attached anteriorly, broadly rounded posteriorly, longer than wide, median lingual process absent. Choanae very small, subcircular, lateral. Vocal slits long, lateral. Very small teeth present on maxillary and premaxillary, dentigerous process of vomers absent. Cloacal tubercles absent; vent at level of upper thighs, a small anal flap above it. Skin granular on dorsum, granules weakest on head, largest posteriorly; one distinctly enlarged tubercle on each eyelid; belly smooth.

Forelimb slender, skin smooth; metacarpal ridge absent; ulnar fold absent; hand length 23.3 % of SVL; Finger I longer than II when fingers adpressed; fingers unwebbed, lateral fringes absent; Finger III not swollen; tip of Finger IV not reaching distal subarticular tubercle on Finger III when fingers adpressed; terminal discs weakly expanded, slightly wider than long, equal to or slightly larger than the width of the distal phalanges; width of disc on Finger III 0.40 mm; discs with distinct dorsal scutes. Relative lengths of adpressed fingers III> I> II> IV; palmar tubercle large, ovoid, 0.56 mm in its largest diameter, 14.7 % of hand length, anterior periphery pigmented; thenar tubercle distinct, but not protuberant, ovoid, anterior periphery pigmented, about half the size of the palmar tubercle and narrowly separated from it. One subarticular tubercle on Fingers I, II, and IV; two subarticular tubercles on Finger III; subarticular tubercle on Finger I largest, subarticular tubercle on Finger II and basal subarticular tubercle on Finger III subequal; subarticular tubercle on Finger IV and distal subarticular tubercle on Finger III smaller, distal tubercle on Finger III smallest ( Fig. 4 ).

Hindlimb robust, skin granular; tibia length 46.6 % of SVL; heels overlapping when hindlimbs are flexed at right angles to sagittal plane of body; foot length 41.1 % of SVL; relative length of adpressed toes IV> III> V> II> I; Toe I very short, its tip reaching the base of the subarticular tubercle on Toe II when toes are adpressed on left foot, failing to reach the base of the subarticular tubercle on Toe II when toes are adpressed on right foot; discs on Toes II, III, IV, and V larger than width of distal phalanges; disc on Toe I equal to width of distal phalange. Width of disc on Toe IV 0.70 mm; rudimentary webbing present only between Toes III-IV, webbing slightly pigmented; weak pre- and postaxial fringes on Toes II-III (preaxial ones more conspicuous) and weak preaxial fringe on Toe IV (more conspicuous than on Toes II-III). Inner metatarsal tubercle oval, 0.53 mm in length, distal portion pigmented, outer metatarsal tubercle round, projecting, 0.33 mm in diameter, entirely pigmented; medial metatarsal tubercle absent on both feet, a weak metatarsal “bump” on right foot. Two subarticular tubercles present on Toes III and V, here on Toe IV, and one on Toes I and II. Subarticular tubercles on Toes I and II largest; basal subarticular tubercle on Toe IV very small on both feet. Metatarsal fold absent. Tarsal keel well defined, short, tuberclelike, directed transversely across tarsus, located 1.10 mm from proximal edge of inner metatarsal tubercle, not extending from it ( Fig. 4 ).

Colour of the Holotype in life Dorsum chestnut brown with granules slightly darker than background; dorsal surface of head similar in colour to dorsum, but with a slight metallic tint. Wide lateral black band from tip of snout to vent, containing ca. two-thirds of tympanum, widening posteriorly from axilla. Very thin, barely distinguishable light line above the band, apparently extending from tip of snout to vent. Broad, diffuse pale oblique lateral stripe embedded in the black band, extending from groin to ca. midbody length. Flanks whitish; ventrolateral stripe present, whitish, barely visible. Upper lip whitish, suffused with tiny melanophores. Throat pinkish grey, suffused with dark pigment. Belly cream.

Upper surfaces of thighs and shanks dark brown, with a few darker blotches and three faint dark brown transverse crossbars (one on thigh, one on shank, and one on tarsus, which appear as a single stripe when legs are bent). Pale paracloacal marks present, pale orangish yellow. Pale yellow spot on groin. Forelimbs light orangish brown, with a short dark brown line tapering from arm insertion to elbow. Toes dark brown with some light grey blotches. Fingers light brown with some pale blotches. Palm brown, sole dark brown. Iris metallic copper with black reticulations, pupil ring present only above the pupil.

Colour of the Holotype in preservative Dorsum dark brown with granules slightly darker than background; dorsal surface of head slightly darker than body. Wide lateral black band from tip of snout to vent. Broad, diffuse pale oblique lateral stripe embedded in the black band, extending from groin and reaching arm insertion. Flanks white; ventrolateral stripe indistinct. Upper lip white, finely suffused with tiny melanophores (best seen under magnification). Throat white, finely suffused with dark pigment (best seen under magnification). Belly white.

Upper surfaces of thighs and shanks similar in colour to dorsum, with a few darker blotches, three faint dark brown transverse crossbars (one on thigh, one on shank, and one on tarsus). Pale paracloacal marks present. Forelimbs very light brown, with a short inconspicuous brown line tapering from arm insertion to elbow. Toes dark brown with some white blotches. Fingers very light brown with some white blotches. Palm light brown, sole dark brown.

Male secondary sexual characters Males are smaller than the only known female (mean SVL in males 16.9 mm, female 17.8 mm) and have a pinkish grey throat suffused with melanophores in life (throat immaculate yellow in living female) ( Fig. 5 ). These melanophores are less conspicuous, but still well visible in preservative and allow distinction between sexes in preserved specimens. Lateral fringes on Toes II–IV (preaxial only on the latter) are weak but visible in males, almost undetectable in the single female.

Variation among paratypes Except the sexual dimorphism in size, the nature of lateral fringes on toes, the throat colour (see above), and insignificant differences in morphometrics (measurements of the type series are provided in Table 1), all paratypes conform to the description of the holotype, except as follows: in life, dorsal colour varies from light brown to chestnut brown; lateral dark band varies from dark brown to black; colour of upper surfaces of thighs and shanks varies from brown to dark brown; colour of forelimbs varies from very light brown to orangish brown; ventral colour varies from cream to bright yellow; in MTD 47884 (male) the oblique lateral stripe fails to reach midbody length ( Fig. 2 ). In preservative, dorsal colour varies from light grey to dark brown ( Fig. 3 shows this variation, which is not related to sex). The enlarged tubercle on eyelid was well visible in all specimens in life (see Fig. 2e for instance), but in some cases becomes less conspicuous in preserved specimens (that character is not sexually dimorphic as it can be observed in both males and females, however, whether the distinctness of the eyelid tubercle is subject to physiological plasticity cannot be fully assessed given the comparatively small sample size). Aweak medial metatarsal tubercle is present in two specimens. The single juvenile collected (referred specimen) further differs in having an “hourglass” pattern on dorsum.

Advertisement call

Temporal structure 6 min, 20 s of call recorded from the holotype include 52 calls emitted between silent intervals and containing 1–19 notes each ( Fig. 6 ). The most common number of notes within a call is 10 (27 % of calls) followed by 11 (21 %), 9 (15 %), and 12 (13 %). Additional calls of 1, 5, 6, 8, 13, 15, or 19 notes each represent 2 to less than 6 % of the total number of calls, and ca. 23 % of the total number of calls if considered altogether.

Eleven representative calls (covering the observed variation in number of notes/call) have been analyzed, totalling a sample of 109 notes (see Table 2 for measurements of acoustic parameters). Note duration ranges from 0.024 – 0.044 s (mean 0.035 s ±0.005); call duration ranges from 0.029 to 7.976 s (mean 3.411 s ±2.0); inter-note interval within a call ranges from 0.277 to 0.623 s (mean 0.344 s ± 0.06). Call rate is 8.3 calls/min in the call sequence recorded. Silent interval between calls ranges from 1.627 to 19.709 s (mean 3.715 s ±2.72), and is usually ca. 2–7 s.

Amplitude modulations are variably evident in the oscillograms and range from none to two ( Fig. 7 ).

Spectral structure Seven harmonics are visible, with the dominant frequency being located in the second harmonic ( Figs. 6–8 ); the highest (seventh) harmonic is sub-ultrasonic (spectral energy in the ultrasonic range, i.e. slightly above single call containing 12 notes. d Spectrogram of the same call. Temperature during recording was 25 °C

20,000 Hz, on at least one occasion, see Table 2). The dominant frequency ranges from 5,064 to 5,845 Hz (mean 5,522.26 Hz±97.40); the frequency of the highest (seventh) harmonic ranges from 18,447 to 20,481 Hz (mean

Note # Number of Note Inter-note Call Dominant Highest harmonic harmonics duration (s) interval (s) duration (s) frequency (Hz) frequency (Hz)

1-note call 1 7 0.029

5-note call 1 7 0.033 0.485 2 7 0.039 0.317 3 7 0.032 0.345 4 7 0.032 0.338 5 7 0.033

6-note call 1 7 0.040 0.323 2 7 0.033 0.293 3 7 0.032 0.294 4 7 0.043 0.292 5 7 0.034 0.300 6 7 0.043

8-note call 1 7 0.038 0.336 2 7 0.036 0.300 3 7 0.032 0.293 4 7 0.032 0.296 5 7 0.033 0.300 6 7 0.041 0.346 7 7 0.032 0.372 8 7 0.032

9-note call 1 7 0.033 0.323 2 7 0.032 0.367 3 7 0.031 0.305 4 7 0.033 0.289 5 7 0.032 0.435 6 7 0.032 0.331 7 7 0.035 0.320 8 7 0.036 0.365 9 7 0.032

10-note call 1 7 0.032 0.291 2 7 0.029 0.496 3 7 0.029 0.328 4 7 0.030 0.305 5 7 0.037 0.309 6 7 0.031 0.322 7 7 0.031 0.315 8 7 0.037 0.328 9 7 0.037 0.323 10 7 0.035

11-note call 1 7 0.029 0.297 2 7 0.042 0.351 3 7 0.042 0.290 4 7 0.033 0.287 5 7 0.044 0.309 6 7 0.044 0.281 7 7 0.044 0.285 8 7 0.034 0.353 9 7 0.037 0.358

0.029 5,465 19,341 5,418 19,555 5,465 19,056 5,418 18,913 5,441 18,961

1.653 5,489 19,531 5,463 19,543 5,512 19,543 5,645 19,692 5,645 19,708 5,612 19,527

1.728 5,629 19,576 5,513 19,393 5,513 19,709 5,608 19,476 5,513 19,593 5,489 19,659 5,536 19,526 5,489 19,559

2.519 5,489 19,493 5,489 19,593 5,536 19,460 5,631 19,393 5,465 19,692 5,607 19,593 5,489 19,476 5,631 19,476 5,513 19,725

3.032 5,489 19,576 5,631 19,460 5,536 19,493 5,513 19,642 5,631 19,576 5,631 19,659 5,655 19,792 5,631 19,676 5,631 19,659 5,631 19,659

3.346 5,655 19,510 5,631 19,346 5,536 19,484 5,845 20,481 5,608 19,698 5,465 19,079 5,489 19,555 5,512 18,984 5,465 19,650 5,465 19,460 Note # Number of Note Inter-note Call Dominant Highest harmonic harmonics duration (s) interval (s) duration (s) frequency (Hz) frequency (Hz)

10 7 0.033 0.342 11 7 0.034

12-note call 1 7 0.032 0.327 2 7 0.026 0.394 3 7 0.032 0.311 4 7 0.036 0.324 5 7 0.033 0.339 6 7 0.031 0.300 7 7 0.031 0.378 8 7 0.032 0.332 9 7 0.031 0.312 10 7 0.031 0.391 11 7 0.031 0.402 12 7 0.031

13-note call 1 7 0.037 0.315 2 7 0.039 0.292 3 7 0.032 0.324 4 7 0.036 0.299 5 7 0.032 0.324 6 7 0.033 0.346 7 7 0.035 0.292 8 7 0.041 0.314 9 7 0.031 0.317 10 7 0.042 0.310 11 7 0.036 0.444 12 7 0.033 0.390 13 7 0.033

15-note call 1 7 0.034 0.307 2 7 0.038 0.304 3 7 0.042 0.470 4 7 0.041 0.375 5 7 0.042 0.294 6 7 0.042 0.329 7 7 0.041 0.286 8 7 0.041 0.303 9 7 0.041 0.313 10 7 0.043 0.290 11 7 0.044 0.277 12 7 0.041 0.291 13 7 0.043 0.312 14 7 0.041 0.295 15 7 0.042

19-note call 1 7 0.024 0.413 2 7 0.028 0.410 3 7 0.031 0.406 4 7 0.031 0.388 5 7 0.031 0.403 6 7 0.033 0.356 5,489 19,508

3.570 5,465 19,389 5,346 18,910 5,465 19,161 5,441 19,493 5,489 19,875 5,489 19,593 5,536 19,609 5,489 19,427 5,560 19,443 5,655 19,476 5,631 19,476 5,655 19,627

4.181 5,489 19,559 5,612 19,443 5,629 19,460 5,529 19,443 5,612 19,510 5,596 19,526 5,612 19,759 5,612 19,759 5,496 19,709 5,596 19,626 5,596 19,510 5,612 19,642 5,596 19,676

4.428 5,496 19,360 5,496 19,227 5,479 19,377 5,479 19,377 5,496 19,410 5,446 19,061 5,463 19,410 5,463 19,410 5,430 19,360 5,463 19,161 5,463 19,344 5,430 19,443 5,463 19,360 5,446 19,061 5,463 19,078

5.064 5,430 19,360 5,064 18,447 5,280 18,447 5,380 18,929 5,380 18,845 5,430 19,028 5,430 19,012

19,444.79 Hz±275.74). The dominant frequency is slightly modulated upward ( Fig. 8 ).

Comparisons with calls of cryptically coloured congeneric species from the Guiana Shield and peripheral areas The advertisement calls of 9 of the 12 cryptically coloured Allobates species currently reported from the Guiana Shield and peripheral areas are known. However, comparison between species is often difficult because call descriptions are sometimes very brief or incomplete. Most analyses lack the necessary standardisation for reliable diagnoses. We relied on original descriptions for the comparisons provided below (except for A. granti and A. spumaponens for which we have call recordings).

The advertisement call of Allobates amissibilis is distinguished mainly from the call of A. granti (described by Kok et al. 2006, see also Kok and Ernst 2007) in having one pulse/note (two distinct pulses/note in A. granti ), and a shorter note duration ranging from 0.024 from 0.044 s (0.068 – 0.070 s in A. granti ) (see also Fig. 8 ); from the call of A. spumaponens (described by Kok and Ernst 2007) in being emitted in groups of notes between silent intervals (call continuous in A. spumaponens ), in having a shorter note duration ranging from 0.024 to 0.044 s (0.060 – 0.070 s in A. spumaponens ), and in having seven harmonics visible (two in A. spumaponens ) (see also Fig. 8 ); from the call of A. marchesianus (described by Caldwell et al. 2002) by containing 1–19 notes per call (21–24 notes in A. marchesianus ), in having longer inter-note interval ranging from 0.277 to 0.623 s (0.119 – 0.212 s in A. marchesianus ), and in having seven harmonics visible (three in A. marchesianus ); from the call of A. undulatus (described by Myers and Donnelly 2001) by containing 1–19 notes per call (2 notes in A. undulatus ), and in having a higher dominant frequency ranging from 5,064 to 5,845 Hz (ca. 3,300 –3,900 Hz in A. undulatus ); from the call of A. caeruleodactylus (described by Lima and Caldwell 2001) in being emitted in groups of notes between silent intervals (call usually continuous in A. caeruleodactylus ), and in having a dominant frequency below ca. 6,000 Hz (above ca. 6,000 Hz in A. caeruleodactylus ); from the call of A. grillisimilis (described by Simões et al. 2013) in being much slower, with a call duration of ca. 3 s on average (always less than ca. 0.3 s in A. grillisimilis ), and with much longer inter-note interval (within a same call) ranging from 0.277 to 0.623 s (0.010 – 0.043 s in A. grillisimilis ); from the call of A. masniger (described by Tsuji-Nishikido et al. 2012) and the call of A. nidicola (described by Caldwell and Lima 2003, see also Tsuji-Nishikido et al. 2012) in being emitted in groups of notes between silent intervals (call continuous in A. masniger and A. nidicola ), and in having a dominant frequency always above ca. 5,000 Hz (always below ca. 5,000 Hz in A. masniger and A. nidicola ); and from the call of A. paleovarzensis (described by Lima et al. 2010) in being slower, with a call rate of 8.3 calls/min (12–20 calls/min in A. paleovarzensis ), a call duration of ca. 3 s on average (always less than, or maximum ca. 3 s in A. paleovarzensis ), and with much longer inter-note interval (within a same call) ranging from 0.277 to 0.623 s (0.065 – 0.266 s in A. paleovarzensis ).

Distribution and ecology

Allobates amissibilis is currently known only from two localities: Turu Falls at the foothills of the Iwokrama Mountains, and close to the top of one of the highest unnamed peaks of the Iwokrama Mountains ( Fig. 9 ), between 160 and 950 m elevation, in the Iwokrama Forest Reserve, Guyana. The species is not abundant at any of the known sites, and only a few males were heard calling, very sporadically, in March, May and December. Intensive surveys led in the Iwokrama Forest during the year 1997 (Donnelly et al.

showing the seven harmonics. Arrow Spectral energy in the ultrasonic range (i.e. slightly above 20,000 Hz) for the last (seventh) harmonic. Temperature during recording was 25 °C

2005a, b) were unable to collect any specimen of this species, which further suggests low population density and geographic restriction.

Allobates amissibilis was found in open mixed lowland forest and in premontane forest with low and open canopy, in the gorge and on one peak of the Iwokrama Mountains. Interestingly, at both sites Allobates amissibilis was found in close vicinity to freshly dropped tapir ( Tapirus terrestris ) faeces. Whether frogs benefit from the presence of tapirs by feeding on insects that are attracted by their faeces or by using water filled tapir foot prints/tracks as alternative larval/reproduction habitats is unknown.

Males call from the ground, often sitting slightly above the substrate, such as on a rock or an exposed root. Advertisement calls are apparently triggered by rain, but a few males were heard calling during dry periods. No male carrying tadpoles has been found, the tadpole is unknown, and virtually nothing is known about the reproductive biology of the species.

Phylogenetic relationships

Molecular phylogenetic relationships of the new species are illustrated in Fouquet et al. (2013: Fig. 3 , under the name Allobates sp. “Iwokrama” using sequence data from the paratype MTD 47884). We therefore refrain from presenting a separate phylogenetic analysis until additional data on species not included in Fouquet et al. (2013) are available. These authors highlighted a sister relationship between the new species (under the name Allobates sp. “Iwokrama”) and

(b) Allobates spumaponens from Mabura Hill Forest Reserve, Guyana (holotype, SMNS 12511) and (c) Allobates granti from Haute Wanapi, French Guiana (paratype, MNHN 2005.0268, call recorded by Philippe Gaucher). Temperatures during recording were 25 °C [a, b (a), and b (b)], and ca. 24 °C (Bc)

a clade containing A. conspicuus , A. insperatus , and an undescribed species from Cuyabeno (NW Ecuador). The clade containing these four species is sister to the trilineatus clade of Grant et al. (2006).

Figures

Fig. 3 a–f Allobates amissibilis sp. nov. a Dorsal, b Ventral, and c Lateral views of the male holotype (IRSNB 4155). d Dorsal, e Ventral, and f Lateral views of the female allotype (IRSNB 4156). Photographs by P.J.R.K. Bar 5 mm

Fig. 2 a–g Allobates amissibilis sp. nov., showing intraspecific variation in colour and pattern in life. a Dorsolateral view of IRSNB 4155, male holotype 16.3 mm snout-vent length (SVL) (a' Ventral view of the same specimen). b Dorsolateral view of IRSNB 4156, female allotype 17.8 mm SVL. c Dorsolateral view of IRSNB 4157, male paratype 17.5 mm SVL. d Dorsolateral view of IRSNB 4158, male paratype 16.4 mm SVL. e Dorsolateral view of IRSNB 4159, male paratype 16.9 mm SVL. f Dorsolateral view of MTD 47884, male paratype 17.4 mm SVL. g Dorsolateral view of IRSNB 15915, juvenile 10.5 mm SVL. Photographs by P.J.R.K.

Fig. 4 Allobates amissibilis sp. nov. Left hand (left), and left foot (right) of the male holotype (IRSNB 4155). Photographs by P.J.R.K. Bar 5 mm

Fig. 5 Allobates amissibilis sp. nov. Ventral views of the female allotype (IRSNB 4156, left), and a male paratype (IRSNB 4158, right), showing sexually dimorphic throat colouration. Photographs by P.J.R.K. Grid squares 5 mm

Fig. 6 a–d Vocalisation of Allobates amissibilis sp. nov., holotype. a Oscillogram showing five calls emitted between silent intervals (sequence of 24 seconds). b Spectrogram of the same sequence. c Oscillogram of a

Fig. 7 a–d Vocalisation of Allobates amissibilis sp. nov., holotype. a,b Expanded oscillograms of two notes from different calls, showing variation in amplitude modulation. c,d Spectral slices of the same notes

Fig. 8 a Vocalisation of Allobates amissibilis sp. nov., holotype. Slightly expanded oscillogram and spectrogram of one note, showing frequency modulation (slightly modulated upward). b Comparison between 1 s-call sequences (showing dominant frequency only) of three morphologically similar Guiana Shield Allobates species (a) Allobates amissibilis sp. nov. from Turu Falls, Guyana (holotype, IRSNB 4155)

Fig. 9 a–c Distribution and habitat of Allobates amissibilis sp. nov. a Map of the Iwokrama Forest Reserve in Guyana. Black rectangle Iwokrama Mountains (enlarged in b). The line crossing the Reserve is the road between Linden and Lethem. b The two sites where the new species was collected in the Iwokrama Mountains. Triangle Type