Bioacoustics of Engystomops puyango

Variation. Variation in dorsal coloration of preserved specimens is extensive ( Fig. 6 ). Background dorsal coloration varies from light gray (QCAZ 27005, 27009) to dark gray (QCAZ 31508) or dark brown (QCAZ 35691). Irregular dark marks may be present in diverse patterns ( Fig. 6 ). A clear mid-dorsal line extends along the posterior half of the body (QCAZ 36511) but in some specimens (QCAZ 26988, 26997) it only extends along the posterior one fifth. There is variation in the abundance and arrangement of tubercles (all lighter than the background; Figs. 3 and 6 ). Tubercles coalesce into small ridges that partly enclose a mid-dorsal light blotch in QCAZ 26984, 27005, 27012. The parotoid glands are fused with the flank glands in 28 out of 32 specimens (average gland length is 16.63 mm , SD = 1.05; 56.9–66.2% of SVL). In QCAZ 27013, the glands are fused on the left side but separated on the right side. In QCAZ 26983 and 26989, the flank glands are separated from the parotoid glands on both sides. Ventral surfaces of preserved specimens have a cream to yellowish-cream background color with light gray (QCAZ 26991, 28762) to dark gray markings (QCAZ 26874, 26984). Marks are arranged in diverse patterns ( Fig. 6 ) and vary from being restricted to the anterior half of the body (darker on folded vocal sacs; QCAZ 26968) to being present over the entire venter (less abundant posteriorly, QCAZ 27008). In a few specimens, the ventral marks are arranged in well-defined large spots (QCAZ 27002, 26983). A midventral cream stripe can be present from near the tip of the snout to the gular region (QCAZ 26994) or continue to the mid venter (QCAZ 26988). In most individuals the stripe is absent (QCAZ 27008) or ill defined (QCAZ 26986). The arrangement of dark spots and tubercles on the ventral surfaces of the feet and hands of QCAZ 26975 is shown in Figure 7 . Head shape varies between rounded (QCAZ 34511) and subacuminate (QCAZ 28760) in dorsal view; in lateral view it varies between rounded (QCAZ 26972) and protruding (QCAZ 31506, 26974). Lateral head coloration varies between light gray and dark gray. The area below the eye and tympanum is cream in most specimens (e.g., QCAZ 26987–88) but can be restricted to a thin longitudinal light stripe from the jaw articulation to the parotoid gland (QCAZ 26976). Two or three vertical dark bars can be present in the loreal region and below the orbit (e.g., QCAZ 26968, 26973, 26982). The tympanic annulus is concealed dorsally and posteriorly. With few exceptions (e.g., QCAZ 26984, 34511) the tympanic annulus has tubercles; the tympanic membrane has tubercles in only five specimens of the type series (e.g., QCAZ 26989, 28758). The following morphometric data pertain to adults. In the type series, the largest male has a SVL of 30.48 mm , and the largest female 32.62 mm ; mean male SVL = 27.59 mm ( n = 45; SD = 1.17), mean female SVL = 27.98 mm ( n = 6; SD = 2.38). Amplectant pairs measured during a behavioral study at Bosque Protector Puyango (not collected) had mean male SVL = 26.76 ( n = 72; SD = 1.62) and mean female SVL = 28.68 ( n = 72; SD = 1.49). Females were significantly larger than males ( t = 6.19, df = 100, P <0.001). Snout-vent length was positively correlated between amplectant males and females (ANOVA F = 19.93, df = 71, P <0.001). Among the type specimens, Catamayo has the highest male SVL (mean = 29.31 mm , SD = 0.51) followed by Puyango (27.50, SD = 0.95) and Mangaurquillo (26.17, SD = 1.75; Table 5 ). Differences in SVL are significant between Catamayo and Mangaurquillo (Tukey’s HSD P <0.001), and Catamayo and Puyango (Tukey’s HSD P = 0.020). All other pairwise comparisons were non-significant ( P values> 0.071). Call. Males call while floating on standing water. Acoustic parameters of the advertisement call of E. puyango are shown in Table 4. The call consists of two obligatory and one facultative components with harmonic structure. The first component is characterized by a slight increase in frequency (mean = 48.0 Hz, SD = 67.24, n = 10) with decreasing amplitude modulation. Sidebands (from amplitude modulation) are evident in each harmonic but they merge at the end of the component (as a result, amplitude modulation ends; Fig. 5 ). The power spectrum along all the call shows three spectral peaks ( Fig. 5 C). Dominant frequency of the call was on the second spectral peak in calls from nine out of ten males (mean = 2022.3 Hz, SD = 30.8). Dominant frequency was 835.5 Hz in calls from the only male on which the dominant frequency was on the first spectral peak. The third spectral peak has high energy only in calls with the third component (see below). The dominant frequency of the first component was on the second spectral peak in calls of nine out of ten males (mean = 2017.6, SD = 40.5). Dominant frequency was 994 Hz in the only male on which the dominant was on the first spectral peak. Species SVL TL FL AL HL HW EN E. pustulatus ( n = 51) 27.08 ± 1.36 11.42 ± 0.59 11.50 ± 1.06 6.26 ± 0.36 8.73 ± 0.42 9.11 ± 0.80 2.75 ± 0.23 Arenillas ( n = 10) 28.33 ± 0.80 26.84–29.34 11.78 ± 0.39 11.24–12.37 11.35 ± 0.41 10.85–12.28 6.50 ± 0.24 6.22–6.99 8.99 ± 0.19 8.76–9.37 8.72 ± 0.22 8.34–9.21 2.68 ± 0.13 2.37–2.85 Huaquillas ( n = 10) 25.53 ± 1.13 23.90–28.14 10.64 ± 0.48 9.98–11.44 9.82 ± 0.50 9.02–10.73 5.98 ± 0.33 5.25–6.46 8.54 ± 0.37 8.08–9.33 7.92 ± 0.23 7.56–8.25 2.58 ± 0.07 2.49–2.72 Patricia Pilar ( n = 17) 27.43 ± 1.24 25.17–29.88 11.59 ± 0.53 10.65–12.72 12.09 ± 0.71 10.45–13.01 6.24 ± 0.41 5.69–6.84 8.73 ± 0.54 7.68–9.6 9.60 ± 0.47 8.63–10.28 2.84 ± 0.27 2.36–3.42 Puerto Rico ( n = 14) 26.86 ± 0.77 25.85–27.96 11.50 ± 0.33 10.94–12.14 12.08 ± 0.64 10.44–13.03 6.32 ± 0.26 5.76–6.71 8.70 ± 0.32 8.15–9.29 9.65 ± 0.53 8.76–10.52 2.82 ± 0.23 2.31–3.17 E. puyango ( n = 30) 27.31 ± 1.16 11.88 ± 0.58 11.10 ± 0.64 6.36 ± 0.31 8.49 ± 0.44 8.81 ± 0.61 2.65 ± 0.16 Catamayo ( n = 3) 29.31 ± 0.51 28.76–29.79 12.31 ± 0.61 11.86–13.01 11.84 ± 0.82 11.36–12.80 6.80 ± 0.32 6.58–7.17 9.11 ± 0.12 9.02–9.26 8.50 ± 0.08 8.42–8.59 2.65 ± 0.05 2.61–2.71 Mangaurquillo ( n = 4) 26.17 ± 1.75 23.78–27.67 11.47 ± 0.19 11.21–11.68 10.28 ± 0.45 9.79–10.83 6.04 ± 0.23 5.81–6.36 8.35 ± 0.52 7.74–8.97 9.36 ± 0.37 9.15–9.92 2.54 ± 0.21 2.33–2.81 Puyango ( n = 23) 27.43 ± 0.83 25.98–29.29 12.03 ± 0.47 11.33–12.77 11.29 ± 0.46 10.38–12.19 6.42 ± 0.25 5.90–6.84 8.50 ± 0.36 7.91–9.45 8.80 ± 0.57 7.54–9.8 2.68 ± 0.15 2.36–2.92 The second component follows immediately and is a whine-like note. Each harmonic is a nearly pure tone with descending frequency ( Fig. 5 ). The average decrease in frequency of the first harmonic is 448 Hz (SD = 45.6, n = 10; measured to the end of the call). The average frequency at the beginning of the second component is 937.7 Hz (SD = 41.0, n = 10) for the first harmonic and 1924 Hz (SD = 67.15, n = 10) for the second. The first spectral peak had more energy than the second in calls of 8 males out of 10 males . The number of visible harmonics in the spectrogram varies between 6 and 12. A facultative third component can be added after the whine and is characterized by a gradual increase in amplitude and a switch in dominant frequency from the first to the fifth or sixth harmonic ( Fig. 5 ). Average fundamental frequency of the third component is 509 Hz (SD = 24.9, n = 10; Table 4); average dominant frequency is 2851 Hz (SD = 135.5; n = 10). Infrequently, the third component exhibits a doubling in the number of harmonics (as shown in Fig. 5 B). The majority of males may be capable of doubling the harmonics because in captivity we recorded repeatedly calls from eight males and all of them produced at least once a call with doubling. This type of call, however, was produced infrequently: out of 12 recordings of third components from seven captive males, only two presented doubling in the number of harmonics; out of ten males recorded at Bosque Protector Puyango , only one did. Observations at Bosque Protector Puyango suggest that the third component is produced when another individual approaches the calling male. Our experiment to test the influence of the proximity of other individuals in the production of the third component shown a significant positive effect. The average proportion of calls with third component for accompanied males was 0.682 (SD = 0.1943, n = 6) while the proportion for lonely males was 0.085 (SD = 0.140, n = 6). Differences between proportions were significant (Wilcoxon’s z = -2.207; P = 0.027). Two components with eigenvalues> 1.0 were extracted from the PCA of calls from 49 males of E. puyango and E. pustulatus . The two PCs accounted for 76.6% of the total variation. Principal Component I loads mostly on frequency variables while PC II on call duration ( Table 2 ). The acoustic space of E. puyango is distinctive from that of E. pustulatus as shown by significant differences between both species in scores for PC I ( t = 5.34, df = 32, P <0.001) and PC II ( t = 6.34, df = 13, P <0.001; Fig. 4 ). In the DFA classification procedure, calls from all 10 males of E. puyango were correctly classified as E. puyango and calls from all 29 males of E. pustulatus were classified as E. pustulatus . The absence of incorrect assignments confirms the high acoustic distinctiveness between the calls of both species. Morphometric comparisons. Three components with eigenvalues> 1.0 were extracted from the PCA of 30 specimens of E. puyango , 51 of E. pustulatus , and 11 of E . sp. B. The three PCs accounted for 76.1% of the total variation. The highest loadings were head width and femur length for PC I, arm length for PC II, and head length for PC III ( Table 6 ). The morphometric space of E. puyango overlaps with E. pustulatus populations from Arenillas and Huaquillas ( Fig. 8 ) but only slightly with E. pustulatus from Puerto Rico and Patricia Pilar. In contrast, E. sp. B overlaps widely in morphometric space with E. puyango as shown by the lack of significant differences along both PCs (PC I: t = 0.44, df = 23, P = 0.661; PC II: t = 0.55, df = 15, P = 0.587). Comparisons between E. puyango and E. pustulatus of the variables with the highest loadings show significant differences in residual head width ( t = 2.20, df = 65, P = 0.03) and residual femur length ( t = 2.45, df = 78, P = 0.016) but not in residual arm length ( t = 1.38, df = 65, P = 0.17). In the DFA classification procedure, 16 out of 30 specimens of E. puyango were classified correctly. The misclassified specimens were assigned to E. pustulatus (four specimens) and E. sp. B ( 10 specimens ). Six out of 11 E. sp. B were correctly classified; all misclassified specimens were assigned E. puyango . Overall, the multivariate analyses indicate some morphometric differentiation between E. puyango and E. pustulatus but not between E. puyango and E. sp. B. Variable Size-free morphology

Distribution and ecology. Engystomops puyango has been recorded in southwestern Ecuador (Provincia Loja and Provincia El Oro) between 320–1291 m above sea level ( Fig. 9 ). Maximum straight-line distance between localities is 120 km . The occurrence of E. puyango in Peru is highly probable because Bosque Protector Puyango and Zapotillo are in the border with Peru . The geographic range of E. puyango is characterized by scant and highly seasonal precipitation. The rainy season lasts between February and April ( Lynch & Duellman, 1997 ). Localities are in Evergreen Lower Montane Forest of the Western Andes (Catamayo) and Foothill Semideciduous Costa Forest ( Puyango , Mangaurquillo, and Zapotillo; vegetation types are as defined by Sierra, et al. , 1999 and Cerón, et al. , 1999). The Evergreen Lower Montane Forest is a transitional form between humid forest and the dry forests that predominates in the Andes of southern Ecuador . The Foothill Semideciduous Costa Forest has scattered trees, less than 20 m tall, and a dense understory dominated by herbaceous plants. All individuals were found in open to sparsely forested areas where the original natural vegetation has been partly or completely removed by humans. At Bosque Protector Puyango , choruses were found in February 2004 and February 2005 . Males were calling from water in ponds and ditches. Amplexus and egg deposition take place at the same sites where choruses call. Engystomops puyango constructs floating foam nests during amplexus. While the female discharges the egg masses, the male beats them with his legs to produce the foam.

Diagnosis. A member of Engystomops , clade Duovox. The assignment to Engystomops is based on the molecular phylogeny published by Ron et al . (2006) which shows high support for the inclusion of E. puyango (referred as “ Engystomops sp. D”) within Engystomops . The following morphological synapomorphies ( Cannatella, et al. , 1998 ) also support this assignment: (1) presence of flank glands; (2) presence of parotoid glands; and (3) warty skin. Engystomops puyango ( Fig. 2 ) is characterized by: (1) mean SVL 27.59 mm in males (range 23.78–30.48; n = 45), 28.68 mm in females (range 25.41–32.68; n = 72); (2) skin on dorsum bearing a mixture of scattered pustules and minute tubercles; (3) snout varying between rounded and protruding in dorsal view and rounded to subacuminate in lateral view; (4) vomerine teeth and odontophores absent; (5) maxillary and premaxillary teeth present; (6) parotoid and flank glands present, usually fused, mean length = 16.63 mm (SD = 1.05; n = 28; 56.9–66.2% of SVL); (7) tarsal tubercle absent; (8) nuptial pads present; (9) Finger I shorter than II; (10) tympanic annulus evident, concealed dorsally and posteriorly; (11) tympanic membrane almost always not tuberculate. Engystomops puyango Engystomops pustulatus Chromosome Relative Arm Centromeric Type Relative Centromeric Type number Length ± SD Radio ± SD Index ± SD Length ± SD Index ± SD 1 14.08 ( ± 0.08) 1.36 ( ± 0.02) 0.42 ( ± 0.01) m 15.45 ( ± 0.10) 0.43 ( ± 0.03) m 2 12.80 ( ± 0.12) 1.81 ( ± 0.05) 0.36 ( ± 0.06) sm 12.54 ( ± 0.09) 0.37 ( ± 0.02) sm 3 11.66 ( ± 0.01) 1.93 ( ± 0.05) 0.34 ( ± 0.04) sm 10.9 ( ± 0.04) 0.34 ( ± 0.04) sm 4 10.53 ( ± 0.17) 2.70 ( ± 0.06) 0.27 ( ± 0.07) sm 10.77 ( ± 0.11) 0.45 ( ± 0.05) m 5 9.25 ( ± 0.22) 1.32 ( ± 0.04) 0.43 ( ± 0.06) m 10.44 ( ± 0.28) 0.34 ( ± 0.03) sm 6 9.25 ( ± 0.09) 1.60 ( ± 0.03) 0.38 ( ± 0.03) m 10.03 ( ± 0.17) 0.31 ( ± 0.03) sm 7 9.10 ( ± 0.07) 1.29 ( ± 0.04) 0.44 ( ± 0.05) m 9.02 ( ± 0.12) 0.42 ( ± 0.07) m 8 8.82 ( ± 0.026) 2.10 ( ± 0.03) 0.32 ( ± 0.02) sm 8.62 ( ± 0.015) 0.37 ( ± 0.04) sm 9a, 9b 8.61 ( ± 0.32) 1.63 ( ± 0.07) 1.77 ( ± 0.06) 0.38 ( ± 0.04) 0.36 ( ± 0.03) m sm 6.52 ( ± 0.23) 0.44 ( ± 0.08) m 10 5.90 ( ± 0.19) 1.37 ( ± 0.02) 0.42 ( ± 0.01) m 5.69 ( ± 0.22) 0.43 ( ± 0.07) m Engystomops puyango is most similar to E. pustulatus . They differ in dorsal skin texture (less tuberculate in E. puyango ; Fig. 3 ), chromosomes shape and size ( Table 3 ), and advertisement call (shorter and with higher frequency in E. puyango ; Fig. 4 ). Calls are significantly different in total duration ( t = 9.68, df = 26, P <0.001), duration of the first component ( t = 4.01, df = 24, P <0.001), frequency of the 2nd spectral peak of the call ( t = 910.47, df = 43, P <0.001), and frequency of the 2nd spectral peak of the first component ( t = 10.65, df = 45, P <0.001; Table 4). Genetically, large distances, typical of species pairs in Engystomops , separate both species. Pairwise uncorrected p -distances between E. pustulatus and E. puyango (range 0.065–0.068) are higher than distances between sister species E. montubio and E. randi (0.027–0.032) or even between nonsister species like E. guayaco and E. montubio (0.059), and E. montubio and E. coloradorum (0.057–0.060). However, genetic distances should not be used as the only evidence for species delimitation ( Vences & Wake , 2007 ). In our dataset, nevertheless, the genetic data is corroborated by the pattern of differentiation in skin texture, advertisement calls, and chromosome morphology and is consistent with the recognition of E. puyango as a separate species from E. pustulatus .