Call Recording

Gecko vocalizations were recorded for 36 gecko species (Table 1) at a private gecko breeding facility near Tulsa, Oklahoma, USA from 10–21 June 2013. We subsequently omitted calls of the parthenogenetic Lepidodactylus lugubris, so our analyses of vocalizations are thus based upon 35 species. The ambient temperature was maintained at 28.3°C. Lighting was controlled by automatic timers. Sunrise and sunset were simulated by a 1-h period of graded light intensity in the morning and evening. The first lights turned on at about 0730 h and off at around 1930 h.

Table 1

Gekkotan species examined for this study, as listed by family: Py = Pygopodidae; C = Carphodactylidae; D = Diplodactylidae; E = Eublepharidae; S = Sphaerodactylidae; Ph = Phyllodactylidae; G = Gekkonidae. Species denoted by a pound sign (#) are those for which both advertisement and distress calls were recorded. Filled dots (•) represent adult males—these data were used to test hypotheses; open dots (*) represent females or juveniles—these data were not used in hypothesis testing (apart from the ancestral state reconstructions [ASR] for Lepidodactylus lugubris). Asterisks (*) in Columns 1 and 2 indicate taxa for which sonographic descriptions of advertisement and/or distress calls are provided for the first time. Columns numbered 1–12 indicate the species examined in the various parts of the study; the column headings referring to the five hypotheses tested, or parts thereof, as follows: 1 H1 AC DCF = Hypothesis 1, advertisement call, dominant call frequency; 2 H1 DC DCF = Hypothesis 1, distress call, dominant call frequency; 3 H2 AC FB = Hypothesis 2, advertisement call, frequency bandwidth; 4 H2 DC FB = Hypothesis 2, distress call, frequency bandwidth; 5 H3 AC = Hypothesis 3, species specificity of advertisement calls; 6 H3 AC Pac = species specificity of advertisement calls within the genus Pachydactylus; 7 H3 AC Cor = individual specificity of advertisement calls for three individuals of Correlophus ciliatus; 8 H3 AC Lim = species specificity of advertisement calls limited to only those parameters also used for distress call analysis; 9 H3 DC = species specificity of distress calls; 10 H4 AC = Hypothesis 4, correlations between laryngotracheal morphology and acoustic properties of advertisement calls; 11 H4 DC = Hypothesis 4, correlations between laryngotracheal morphology and acoustic properties of distress calls; 12 H5 ASR = Hypothesis 5, ancestral state reconstruction of laryngotracheal characters.

To collect advertisement calls, Zoom H2n handy recorders (Zoom North America) were placed in front of, atop, or inside the enclosures of identified callers and recorded through all, or part of, the night. One of us (EAR) was present at the facility from about 1900 to 0000 h each evening to identify each calling gecko.

Distress calls were recorded by day to minimize disrupting the nighttime activity of geckos, when most species produce advertisement calls. To record distress calls, we removed geckos from their enclosures by hand and stimulated them to vocalize by either tapping the snout (Bauer et al. 1992), pinching the toes (Brown 1985), and/or gently squeezing the body (Weber and Werner 1977), the latter approach being the most effective. Not all individuals emitted distress calls when prompted, but if they did calls were generally emitted within the first few seconds of capture. Individuals were captured and prompted to emit distress calls on several occasions, although never more than once per day.

Acoustic Analyses

Raven Pro v1.4 (Bioacoustic Research Program, Cornell Lab of Ornithology, Ithaca, NY) was used for the spectral and temporal analysis of vocalizations. Adjusting contrast, color, and other visual settings in Raven Pro can result in inconsistent measurement of call durations and frequencies. To compensate for this, individual chirps were measured at the 5 and 95% intervals of sound energy for duration and frequency data (Charif et al. 2010). The following parameters were recorded, wherever possible, for all advertisement calls: number of phases (i.e., number of different chirp types), number of chirps per call, call duration (s), chirp duration at 90% of sound energy (s), interchirp duration (s), dominant chirp frequency at the 5th and 95th percentiles of sound energy (kHz; hereafter referred to as chirp frequency at 5% of sound energy [kHz] and chirp frequency at 95% of sound energy [kHz], respectively), chirp frequency bandwidth (kHz), and maximum power (dB; Gramentz 2009a; Tang et al. 2001; Yu et al. 2011; Jono and Inui 2012; Phongkangsananan et al. 2014; Chen et al. 2016).

Morphological Analyses

Digital photographs of each individual for which recordings were made were taken in dorsal view to ensure correct species identification. Individuals were placed in sealable plastic bags and mass was recorded with the use of a tared spring scale. Snout–vent length (SVL), head length (HL), head width (HW), and head depth (HD) were measured with the use of a ruler or calipers.

Adult specimens of 30 gekkotan species (Appendix II) that had been formalin-fixed and preserved in 70% ethanol were obtained from museum collections and their throat regions dissected (Table 1, Column 12). Of these species, 22 were represented in our recorded vocalizations (Table 1, Columns 1–11 vs. 12). A nonvocal outgroup species, Anolis carolinensis (Reptilia: Sauria: Dactyloidae), was also dissected for comparative purposes. Twelve species for which recordings were made were unavailable for dissection (Table 1, Columns 1–11 vs. 12). Specimens were dissected from six species for which no calls were recorded in this study (Table 1, Columns 1–11 vs. 12). In these instances, calls of congeners were recorded in this study (e.g., Gekko smithii was dissected because calls of G. vittatus and G. siamensis were recorded in this study, etc.) or descriptions of their calls (e.g., Delma fraseri and H. frenatus) have been published elsewhere (Gramentz 2010; Manley and Kraus 2010).

Larynges of museum specimens were examined in situ, and subsequently portions of the trachea and tongue were excised and extraneous tissue removed following the procedures of Rittenhouse et al. (1997, 1998) and Russell et al. (2000, 2014). Pertinent anatomy was examined and photographed using a Nikon SMZ1000 dissecting microscope with attached Nikon Digital Sight DS-U2 camera (Nikon Inc., Tokyo, Japan) and a metric rule. The following morphological measurements of the entire specimen and parts thereof were recorded using NIS-Elements D 3.1 (Nikon Inc., Tokyo, Japan; Table 2): SVL, HL, HD, HW, rostrum-aditus length (RAL), larynx length (LL), larynx width (LW), anterior tracheal width (ATW), posterior tracheal width (PTW), and anterior tracheal ring width (ATRW).

Table 2

Morphological measurements (reported as means, in mm) for lizard species examined. n = number of specimens examined per species. ATRW = anterior tracheal ring width; ATW = width of the anterior region of the trachea; GLL = length of the glottal lips; HD = head depth; HL = head length; HW = head width, taken at the widest point of the head; LL = anteroposterior length of the larynx; LW = mediolateral width of the larynx; PTW = width of the posterior region of the trachea; RAL = rostrum–aditus length; SVL = snout–vent length.

Hypothesis Testing

All statistical analyses were carried out using JMP v11.0 software (SAS Institute, Cary, NC). We examined all data for normality of distribution using the Shapiro–Wilk W test. For our analyses we included only calls for which all parameters could be measured. The values of all chirps in a call were averaged. Only adult male calls were used; female and juvenile calls were omitted. Morphological values were averaged per species. Sampling across the Gekkota as a whole remains poor and our questions seek to reveal lineage-wide trends. Thus, hypotheses relating to sonographic properties of calls were not analyzed using phylogenetic comparative methods.

To test Hypothesis 1, that call frequency is negatively correlated to body size, we employed two Model I regressions, one for advertisement calls and the other for distress calls, using SVL as the independent variable and average dominant call frequency (DCF) per species as the dependent variable. We used data from 17 species (2 diplodactylids; 1 phyllodactylid; 14 gekkonids; Table 1, Column 1) to determine whether the DCF of the advertisement calls is correlated with body size. We used data from 23 species (2 diplodactylids; 2 eublepharids; 1 sphaerodactylid; 1 phyllodactylid; 17 gekkonids; Table 1, Column 2) to determine whether the DCF of distress calls is correlated with body size.

To test Hypothesis 2, that advertisement calls are used for intraspecific communication and distress calls used for interspecific communication, we performed a one-tailed, unpaired t-test to compare frequency bandwidth (FB) of 285 advertisement calls recorded from 17 species and FB from 260 distress calls recorded from 23 species (Table 1, Columns 3, 4). For the eight species for which both advertisement and distress calls were recorded (Ptenopus carpi, P. kochi, Afroedura namaquensis, A. loveridgei, Chondrodactylus fitzsimonsi, Pachydactylus parascutatus, P. visseri, Hemidactylus turcicus), a one-tailed, paired t-test was conducted to compare average FB per species of both call types. Calls from only one individual were used when more than one individual per species was recorded. We used a nonparametric Mann–Whitney U to test for significant differences between the groups if the data were not normally distributed.

To test Hypothesis 3, that advertisement calls are species specific, we conducted principal-component analyses (PCAs) followed by discriminant function analyses (DFAs) for both advertisement and distress calls. In the first analysis, we tested for species specificity using 285 advertisement calls from 17 species (Table 1, Column 5) and 19 individuals (three Correlophus ciliatus) using the following parameters: number of phases, number of chirps, call duration (CaD), chirp duration (ChD), DCF, call frequency at 5% of the sound energy (CF5%), call frequency at 95% of the sound energy (CF95%), and maximum power (MP). In the second analysis, we tested for species specificity within the genus Pachydactylus using 127 advertisement calls from five species (Table 1, Column 6). Call structure was complex and fairly conserved for all Pachydactylus spp. recorded, allowing for more parameters to be analyzed than for the previous analysis, which employed calls from a wide range of species with highly varied call structure. The following parameters were used in the Pachydactylus analysis: number of chirps, CaD, intercall duration (ICaD), ChD, interchirp duration (IChD), CF5%, CF95%, DCF, and MP. In the third analysis we tested for individual specificity of the advertisement calls of Correlophus ciliatus (Table 1, Column 7) with the use of 56 advertisement calls from three individuals with the following parameters: number of chirps, CaD, ChD, IChD, CF5%, CF95%, DCF, and MP. In the fourth analysis, we tested for species specificity using 285 advertisement calls from 19 individuals of 17 species (Table 1, Column 8) by limiting the number of parameters to only those used for the distress-call analysis: ChD, CF5%, CF95%, DCF, and MP. In the fifth analysis, we tested for species specificity using 260 distress calls from 23 species (the same species examined for distress-call data in Hypothesis 1, above; Table 1, Column 9) using the following call parameters: ChD, CF5%, CF95%, DCF, and MP.

To test Hypothesis 4, that variability in laryngotracheal morphology is associated with acoustic properties of phonation, we ran a series of Model I regressions using measurements of the larynx and trachea as independent variables and acoustic measurements as dependent variables. Exploration of possible correlations between laryngotracheal morphology and the acoustic properties of advertisement calls employed data from 13 species (2 diplodactylids; 1 phyllodactylid; 10 gekkonids; Table 1, Column 10). We used data from 14 species (2 diplodactylids; 1 eublepharid; 1 sphaerodactylid; 1 phyllodactylid; 9 gekkonids; Table 1, Column 11) to investigate correlations between laryngotracheal morphology and acoustic properties of distress calls.

Three males of Correlophus ciliatus were recorded, so we averaged their acoustic and morphological values for each parameter. We log-transformed nonnormally distributed data sets until they were normally distributed, or used a nonparametric Spearman's rank correlation, as appropriate. The following measurements and proportions taken from preserved specimens were used as independent variables following Russell et al. (2014): LL, LW, GLL, larynx shape (LW/LL), relative larynx length (SVL/LL), relative larynx width (HW/LW), laryngeal position (SVL/RAL), relative anterior trachea width (SVL/ATW), relative posterior trachea width (SVL/PTW), trachea shape (ATW/PTW), relative tracheal ring width (SVL/ATRW), relative length of glottal lips (GLL/LL), and relative width of glottal lips (GLL/LW). The following acoustic measurements were used as dependent variables: DCF, CF5%, CF95%, FB, and MP.

To test Hypothesis 5, that the morphology of the larynx and trachea can be used as reliable phylogenetic markers, we performed ancestral state reconstructions (ASRs) of 7 gross morphological laryngotracheal characters from the 30 species for which museum specimens were examined (Table 1, Column 12) and one outgroup species, Anolis carolinensis. The characters and character states were derived from the gross morphological observations of the laryngotracheal morphology of Afro-Malagasy geckos of Russell et al. (2000). The following morphological characters were examined: (1) shape of the cricoid (ring, plate-like, or rounded), (2) orientation of the lateral process of the cricoid (anterolateral, lateral, or posterolateral), (3) shape of the trachea (uniform diameter, tapered, grooved, or dilated), (4) continuity of tracheal rings (continuous or not continuous), (5) orientation of the ventral fibers of the m. constrictor laryngis relative to the lateral processes of the cricoid (anterolateral, lateral, posterolateral, or fan-like), (6) fusion of the cricoid with the first tracheal ring (fused or not fused), and (7) relative size of the m. constrictor laryngis (small, moderate, or large; Table 3).

Table 3

Character matrix used for the ancestral state reconstruction of laryngotracheal morphology in gekkotans (outgroup = Anolis carolinensis). (a) Cricoid shape: (0) ring, (1) plate-like, (2) rounded. (b) Projection angle of the lateral processes of the cricoid cartilage: (0) anterolateral, (1) lateral, (2) posterolateral. (c) Trachea shape: (0) uniform diameter, (1) tapered, (2) grooved, (3) dilated. (d) Continuity of tracheal rings: (0) continuous, (1) not continuous. (e) Fusion of the cricoid and first tracheal ring(s): (0) fused, (1) not fused, (?) unknown. (f) Orientation of ventral fibers of m. constrictor laryngis: (0) anterolaterally, (1) laterally, (2) posterolaterally, (3) fan-like, (?) unknown. (g) Relative size of the m. constrictor laryngis: (0) small, (1) moderate, (2) large.

We performed ASRs in Mesquite v3.04 (Maddison and Maddison 2015) and traced the character states using the Trace Character History function across a recently published phylogeny that was pruned to include only the taxa for which laryngotracheal data were collected. Both parsimony and ML methods were used to trace character evolution. The parsimony ASRs incorporated an unordered states assumption, which counts one step for any character state change. A Markov k-state one-parameter model was incorporated for the ML ASRs which assumes that any particular change is equally probable.

The phylogeny used is a time-calibrated phylogeny incorporating representatives of 98% (120) of gekkotan genera estimated from a Bayesian analysis of the concatenated gene data set based upon the analysis of five nuclear genes and one mitochondrial fragment (Gamble et al. 2015). Most species, including the outgroup, and all gekkotan genera examined in this study are included in this tree.

Calls

Advertisement and distress calls.—Advertisement calls were recorded for 19 males representing 17 species, 10 of which are sonographically presented for the first time (Table 1, Column 1 symbols with asterisks; 1 diplodactylid; 9 gekkonids; Table 4; Fig. 1). Most of these are monophasic (i.e., consist of one type of chirp), but Oedura marmorata, Hemidactylus turcicus, and Ptenopus garrulus emitted two chirp types, Chondrodactylus fitzsimonsi three types of chirps in three distinct calls, and Stenodactylus sthenodactylus produced four chirp types with a repertoire of three calls (Fig. 2). Most advertisement calls consisted of temporally discrete, individually countable chirps, although a few incorporated temporally fused, trilled chirps that were not segregable into discretely countable individual notes. Distress calls were recorded for 27 species and 31 individuals, of which those of 18 species are sonographically presented for the first time (Table 1, Column 2 symbols with asterisks; 1 diplodactylid; 2 eublepharids; 1 sphaerodactylid; 14 gekkonids; Table 5; Fig. 3).

Table 4

Mean and range values for advertisement calls of gekkotan species recorded in this study (see Table 1 for family designations). For species with a repertoire of calls, each call type is reported separately and denoted in parentheses. n = number of calls (if more than one individual was recorded the number of individuals is given in parentheses; see Fig. 2 for descriptions of call types). CaD = call duration; CF5% = call frequency at 5% of the sound energy; CF95% = call frequency at 95% of the sound energy; ChD = chirp duration; DCF = dominant call frequency; FB = frequency bandwidth; ICaD = intercall duration; IChD = interchirp duration; MP = maximum power.

Table 5

Mean and range values for distress calls of gekkotan species recorded in this study (see Table 1 for family designations). n = number of calls (if more than one individual was recorded the number of individuals is denoted in parentheses). CaD = call duration; CF5% = call frequency at 5% of the sound energy; CF95% = call frequency at 95% of the sound energy; DCF = dominant call frequency; FB = frequency bandwidth; J = juvenile; MP = maximum power.

Fig. 1

Sonograms of advertisement calls from gekkotan species that are depicted for the first time (asterisks in Table 1, Column 1). Species are presented in the same order as in Table 1. (a) Correlophus ciliatus, (b) Afroedura namaquensis, (c) A. loveridgei, (d) Chondrodactylus fitzsimonsi, (e) Pachydactylus parascutatus, (f) P. scutatus, (g) P. montanus, (h) P. weberi, (i) P. visseri, (j) Microgecko persicus. Scales along the x-axes are not identical for all panels; all y-axes range from 0 to 20 kHz.

Fig. 2

Advertisement calls of gekkotan species possessing polyphasic calls or call repertoires of more than one call type. (a) Oedura marmorata, (b) Hemidactylus turcicus, and (c) Ptenopus garrulus produced two types of chirp (labeled 1 and 2). (d) Chondrodactylus fitzsimonsi produced three call types, each possessing unique chirps, (di) a low-frequency “whoop” or bark, (dii) squeals, and (diii) low-frequency chirps of shorter chirp and interchirp duration than the previous call. (e) Stenodactylus sthenodactylus emitted four call types, represented by (ei) the most common emission consisting of 2–4 chirps, (eii) isolated chirps from the second call type, (eiii) entire call of the second call type, (eiv) polyphasic call. Scales along the x-axes are not identical for all panels; all y-axes range from 0 to 20 kHz.

Fig. 3

Sonograms of distress calls from gekkotan species that are depicted for the first time (asterisks in Table 1, Column 2). Species are presented in the same order as listed in Table 1. (a) Oedura monilis male, (b) Correlophus ciliatus female, (c) Coleonyx variegatus male, (d) Goniurosaurus orientalis (juvenile), (e) Eublepharis hardwickii male, (f) Teratoscincus scincus male, (g) Afroedura namaquensis male, (h) Afroedura loveridgei male, (i) Homopholis wahlbergii male, (j) Chondrodactylus angulifer female, (k) Chondrodactylus angulifer male, (l) Chondrodactylus fitzsimonsi male, (m) Pachydactylus gaiasensis female, (n) Pachydactylus gaiasensis male, (o) Pachydactylus parascutatus male, (p) Pachydactylus visseri male, (q) Gekko vittatus male, (r) Gekko siamensis female, (s) Agamura persica male, (t) Cyrtodactylus pulchellus male, (u) Hemidactylus boavistensis male, (v) Hemidactylus ruspolii male, (w) Hemidactylus triedrus male.

Frequency–size relationships of the calls.—Dominant call frequency (DCF) of advertisement call (W = 0.91, P = 0.11) and SVL (W = 0.92, P = 0.13) data were normally distributed, and exhibit a negative relationship (DCF = 5.36–0.04 × SVL, R² = 0.55, P < 0.001; Fig. 4a). No correlation was detected for the DCF of distress calls (P = 0.59).

Fig. 4

Frequency characteristics of gecko calls. (a) Dominant call frequency of advertisement calls exhibits a negative relationship to SVL (P < 0.001; see Table 1, Column 1 symbols with asterisks). (b) Frequency bandwidth (kHz; reported as means ± 1 SE) of advertisement calls compared to distress calls for all species examined (bi; P < 0.001), and for the eight species for which both call types were recorded (bii; P = 0.02). See Table 1, Columns 3–4, and species names with pound signs (#).

Frequency bandwidth of the calls.—Advertisement call data for all 17 species examined (Table 1, Column 1) were not normally distributed (n = 257, Shapiro–Wilk W = 0.89, P < 0.001), so a nonparametric Mann–Whitney U test was used. The distributions of the advertisement calls and the distress calls (Table 1, Column 2) differ significantly; the ranked sum mean frequency bandwidth was 199 for advertisement calls and 354 for distress calls (DF = 543, U = 11.45, P < 0.001; Fig. 4 bi). Distress calls have a greater frequency bandwidth (x̄ ± 1 SE = 7.69 ± 0.28 kHz) than do advertisement calls (x̄ = 3.18 ± 0.08 kHz) when tested using data for the eight species for which both call types were recorded (DF = 7, t = 2.84, P = 0.02; Table 1; Fig. 4bii).

Specificity of advertisement and distress calls.—In the analysis of advertisement calls for the 17 species for which these are available (Table 1, Column 5), the first two principal components account for 54.0% of the total variance (PC1 = 33.2% and PC2 = 20.8%). The discriminant function analysis (DFA) correctly classified 92% of all advertisement calls to the calling species (Fig. 5a). For the analysis of the advertisement calls of the five species of Pachydactylus (Table 1, Column 6), the first two principal components accounted for 54.6% of the total variance (PC1 = 39.9% and PC2 = 14.7%), and the DFA correctly classified 100% of all advertisement calls to the calling species (Fig. 5b). For the three individuals of Correlophus ciliatus (Table 1, Column 7) the first two principal components accounted for 49.9% of the total variance (PC1 = 28.6% and PC2 = 21.1%), and the DFA correctly classified 95% of all advertisement calls to the calling individual (Fig. 5c).

Fig. 5

Discriminant function analysis (DFA) of gecko calls in relation to call specificity. Encompassing circles represent the 95% confidence limit containing the true mean for that group. The scale of axes for panels a, b, and d are identical, but differ from those used in panels c and e (which are identical to each other); the legend for panels a, b, and d is also the same. (a) For the 17 species analyzed for 7 parameters of the advertisement call (Table 1, Column 5), the DFA correctly assigned 261 of 285 calls (92%) to the calling species. (b) With the use of 9 advertisement call parameters for 5 species of Pachydactylus (Table 1, Column 6), the DFA correctly assigned 127 of 127 advertisement calls (100%) to the calling species. (c) Based upon 8 advertisement call parameters for 3 individuals of Correlophus ciliatus (Table 1, Column 7), the DFA correctly assigned 53 of 56 advertisement calls (95%) to the calling individual. (d) For the 17 species analyzed for call parameters of the advertisement call, the DFA correctly assigned 248 of 285 advertisement calls (87%) when using only the call parameters used in the distress call DFA (Table 1, Column 8). (e) Based upon 5 call parameters of the distress call for males of 23 species (the same as those used for the advertisement call in panel [d]), the DFA correctly assigned 133 of 260 distress calls (51%) to the calling species. Al = Afroedura loveridgei; An = Afroedura namaquensis; Ap = Agamura persica; Ca = Chondrodactylus angulifer; Cc = Correlophus ciliatus; Cf = Chondrodactylus fitzsimonsi; Cp = Cyrtodactylus pulchellus; Cv = Coleonyx variegatus; Eh = Eublepharis hardwickii; Gv = Gekko vittatus; Hb = Hemidactylus boavistensis; Hr = Hemidactylus ruspolii; Htr = Hemidactylus triedrus; Htu = Hemidactylus turcicus; Hw = Homopholis wahlbergii; Mp = Microgecko persicus; Oma = Oedura marmorata; Omo = Oedura monilis; Pc = Ptenopus carpi; Pg = Pachydactylus gaiasensis; Pga = Ptenopus garrulus; Pgu = Ptyodactylus guttatus; Ph = Ptyodactylus hasselquistii; Pm = Pachydactylus montanus; Pp = Pachydactylus parascutatus; Ps = Pachydactylus scutatus; Pk = Ptenopus kochi; Pv = Pachydactylus visseri; Pw = Pachydactylus weberi; Rl = Rhacodactylus leachianus; Ss = Stenodactylus sthenodactylus; Ts = Teratoscincus scincus.

When employing only the parameters used for the analysis of distress calls for 17 species (Table 1, Column 8), the first two principal components accounted for 67.1% of the total variance in advertisement calls (PC1 = 42.5% and PC2 = 24.6%), and the DFA correctly classified 87% of these advertisement calls to the calling species (Fig. 5d). For the distress calls of the 23 species recorded (Table 1, Column 9), when using the same five parameters examined for the advertisement calls of 17 species (Table 1, Column 8), the first two principal components accounted for 66.7% of the total variance (PC1 = 42.7% and PC2 = 24.0%), and the DFA correctly classified 51% of all distress calls to the calling species (Fig. 5e).

Laryngotracheal Morphology

We provide descriptions of the muscles and skeleton of the larynx, and the form of the trachea for an exemplar gecko, Hemidactylus turcicus, which is geographically widespread and highly vocal. These are compared to an outgroup species Anolis carolinensis (Reptilia: Sauria: Dactyloidae) which is a model organism for reptilian studies (Alföldi et al. 2011) and of comparable size to many geckos (SVL < 75 mm; Jensen 2008). Anolis has been reported to emit sounds comparable to gekkotan distress calls, but these are not regarded as being true vocalizations which pass through vocal cords and are relatively ineffective social signals (Milton and Jenssen 1979). To our knowledge, sounds that could be categorized as having a social function have not been reported for A. carolinensis.

These descriptions serve as a baseline for the comparisons made in this study, and augment those of Moore et al. (1991), Rittenhouse et al. (1997, 1998), and Russell et al. (2000, 2014) who provided anatomical descriptions of at least some part of the laryngotracheal anatomy of geckos. The primary attributes explored in this comparison (see Fig. 6) are the basic shape of the cricoid cartilage; the form the lateral processes of the cricoid cartilage; size of the glottal lips; size, origin, and insertion of the m. constrictor laryngis and m. dilator laryngis; occurrence of fusion of the cricoid with the first tracheal ring; and the presence of a cricoid fenestra. The general shape of the arytenoid cartilages is not easily discerned through dissection and is more clearly assessed through three-dimensional reconstruction from serial sections (Russell et al. 2000), which was not undertaken in this study.

Fig. 6

The larynx and trachea of male Hemidactylus turcicus (a,b) and Anolis carolinensis (c,d) in dorsal (a,c) and ventral (b,d) views. Abbreviations: ac, arytenoid cartilage; ad, aditus laryngis; cc, cricoid cartilage; gl, glottal lips; itrs, intertracheal ring space; lp, lateral process of cricoid cartilage; mcl, m. constrictor laryngis; mdl, m. dilator laryngis; tr, tracheal ring. Scale bar = 1 mm.

Description of laryngeal structure.—The cricoid cartilage of Anolis carolinensis is round or ovoid in shape, with very small (essentially absent, as determined by examination of cleared and stained specimens) lateral processes that are not visible through the laryngeal musculature (Fig. 6c,d). Large paired, laterally curved arytenoids are clearly evident along the anterior surface of the cricoid and flare out anteriorly to circumscribe the aditus laryngis. The flat m. constrictor laryngis originates from the ventral midline of the cricoid and its fibers spread laterally in a fan-like pattern, to the lateral aspect of the cricoid. The m. dilator laryngis is extremely small and lies superficial to its ipsilateral m. constrictor laryngis, which originates at the posterolateral end of the larynx and runs along the lateral surface of the cricoid to insert on the arytenoid cartilage. Glottal lips were not readily observed. The cricoid is not fused with the first tracheal ring. There is no cricoid fenestra.

In comparison to Anolis, the cricoid cartilage of geckos bears hypertrophied lateral processes. The cricoid cartilage of H. turcicus is shaped similarly to that of most geckos for which descriptions have been furnished, being broad and ring-shaped, from which the dorsoventrally flattened lateral processes project (Fig. 6a,b). The anterior margin of each lateral process extends directly laterally from the cricoid, whereas the posterior margin extends anterolaterally, resulting in the lateral process having a shape resembling a right triangle with a rounded distal extremity. The paired arytenoid cartilages articulate anteriorly with the cricoid and circumscribe the aditus laryngis, the opening of which is situated along the anterodorsal surface of the cricoid. Each of the paired constrictor laryngis muscles is large, with its parallel fibers coursing laterally from their origin on the processus lingualis of the basihyoid (located ventral to the cricoid), around the anterior edge of the lateral process to insert on the dorsal midline of the central ring of the cricoid. Each of the paired m. dilator laryngis muscles lies superficial to its ipsilateral m. constrictor laryngis, originating at the apex of the lateral process of the cricoid, running along its anterior edge and inserting on the arytenoid cartilage. This muscle is much more conspicuous than that of A. carolinensis. The glottal lips of H. turcicus are inconspicuous. The cricoid fenestra is absent.

Noteworthy variations in cricoid structure among geckos.—Departures from the form of the cricoid cartilage exhibited by Hemidactylus turcicus are sporadically distributed among the taxa examined, with no obviously identifiable familial trends. The features of the cricoid that exhibit the greatest variation are the shape of the cricoid fenestra, the form of its central ring, and the orientation and shape of the lateral processes.

The dorsal cricoid fenestra of Chondrodactylus fitzsimonsi is square in outline and is deeply indented within the cricoid and demarcated by rounded edges. A similar, although relatively smaller and less deeply indented, dorsal cricoid fenestra is present in Gekko smithii. The dorsal cricoid fenestra of Ptyodactylus guttatus is rectangular, but it is small and triangular in P. togoensis. In Stenodactylus sthenodactylus the dorsal cricoid fenestra is reduced to a shallow midline groove.

In Lepidodactylus lugubris, the dorsal cricoid fenestra is teardrop shaped, with the apex directed posteriorly. A similar form is evident in the sphaerodactylid Teratoscincus scincus. In the eublepharid Goniurosaurus luii the dorsal cricoid fenestra is more ovoid in outline. In all three of these species, the dorsal cricoid fenestra extends posteriorly along the trachea, resulting in incomplete fusion of the anteriormost tracheal rings.

In the eublepharid Coleonyx variegatus, the central ring of the cricoid is slightly inflated posteriorly and is fused with the first tracheal ring. The anterior and posterior borders of its lateral processes deviate from being normal to the sagittal axis of the larynx, endowing them with a tapering form. This contrasts with the lateral processes of H. turcicus, in which only the posterior margin of the lateral process deviates from a line normal to the sagittal axis of the larynx. Ventrally, the cricoid ring of C. variegatus bears a conspicuous longitudinal groove that extends posteriorly onto the first few tracheal rings and houses the processus lingualis of the basihyoid.

The lateral processes of the cricoid of the carphodactylid Underwoodisaurus milii extend posterolaterally with respect to the sagittal axis of the cricoid, a condition unique among the species examined.

All three species of Ptenopus examined exhibit enlarged cricoid cartilages with enlarged lateral processes. In dorsal view the central ring of the cricoid extends posteriorly from the lateral processes to form a flat, triangular plate that sits atop its base (as noted for P. garrulus by Russell et al. 2000). In Stenodactylus sthenodactylus, the central ring of the cricoid also exhibits a highly differentiated shape, being expanded anteriorly and thus resembling a funnel, possibly as a result of an insensible fusion of the cricoid with the first few tracheal rings. The arytenoid cartilages of this species do not cover the majority of the central ring of the cricoid. The form of the cricoid of Agamura persica and Goniurosaurus luii is similar to that of S. sthenodactylus, although less markedly expanded.

Description of tracheal structure.—The trachea extends posteriorly from the larynx and its primary morphological variations concern its overall shape and the continuity of the tracheal rings. Tracheal morphology is similar for Anolis carolinensis and Hemidactylus turcicus, and in both species is of generally uniform diameter and supported by complete tracheal rings (Fig. 6). One notable difference, however, is the relatively greater robustness and anteroposterior width of the tracheal rings of A. carolinensis compared to its intertracheal ring spaces. The tracheal rings are of uniform width and are uninterrupted around their circumference. Tracheal morphology is relatively conserved in most of the gecko species examined, with it being either of uniform width or gently tapering anteriorly.

Noteworthy variations in tracheal structure among geckos.—In a few species highly deviant tracheal morphologies are evident in comparison to that described above. Such departures—like those noted above for the cricoid cartilage—are sporadically distributed across the taxa examined, and are not clustered in discrete regions of the phylogeny. Apart from those noted below, most species have complete tracheal rings throughout the length of the trachea. The dorsal cricoid fenestra extends posteriorly in Lepidodactylus lugubris, Goniurosaurus luii, and Teratoscincus scincus, which produces an incomplete fusion of the anteriormost tracheal rings with the cricoid. Ptyodactylus guttatus exhibits a high degree of branching of the tracheal rings, a feature only occasionally encountered in other species, although not evident in the congeneric P. togoensis. The tracheal folding in Ptenopus garrulus (Russell et al. 2000) was also observed in our dissected material.

The trachea of Chondrodactylus angulifer is markedly tapered, having a vastly greater anterior, compared to posterior, width, with the tracheal rings being incomplete dorsally in the expanded anterior region (cf. Russell et al. 2000), comprising the so-called “tracheal zipper” (Busing 1990). The trachea of C. fitzsimonsi also exhibits this tapering form, but funnels back to a narrower diameter just posterior to its junction with the cricoid. Similar tracheal tapering is also evident in Oedura marmorata, although in this species the trachea is more dilated. Only a slight tapering is evident in O. monilis.

Relationships Between Laryngotracheal Morphology and Phonation

Statistically significant relationships between laryngotracheal morphology and acoustic properties of advertisement calls were found for the following (Fig. 7; see  Table S1 (175_herp-75-03-06_s01.docx) in the Supplemental Materials available online): dominant call frequency and larynx length (Fig. 7a); dominant call frequency and larynx width (Fig. 7b); glottal lip length and dominant call frequency (Fig. 7c); call frequency at 5% of sound energy and larynx length (Fig. 7d); call frequency at 5% of sound energy and larynx width (Fig. 7e); glottal lip length and call frequency at 5% of sound energy (Fig. 7f); larynx shape (LW/LL) and maximum power (Fig. 7g); relative larynx width (HW/LW) and call frequency at 95% of sound energy (Fig. 7h); relative anterior trachea width (SVL/ ATW) and dominant call frequency (Fig. 7i); relative posterior trachea width (SVL/PTW) and call frequency at 5% of sound energy (Fig. 7j); relative posterior trachea width (SVL/PTW) and call frequency at 95% of sound energy (Fig. 7k); call frequency at 5% of sound energy and tracheal shape (ATW/PTW; when P. kochi is excluded; Fig. 7l).

Fig. 7

Features of laryngotracheal morphology for which there are statistically significant relationships with acoustic properties of advertisement calls (see Table 1, Column 10). Data for those features of laryngotracheal morphology that were not normally distributed were log-transformed prior to comparison with acoustic call properties, or a nonparametric Spearman's rank correlation was used, as appropriate. Detailed statistical results are provided in  Table S1 (175_herp-75-03-06_s01.docx) of the Supplemental Materials available online.

Fewer significant relationships were found between laryngotracheal morphology and the acoustic properties of distress calls than for advertisement calls. Significant relationships were found for the following (Fig. 8; see  Table S1 (175_herp-75-03-06_s01.docx) in the Supplemental Materials available online): glottal lip length and maximum power (Fig. 8a), relative anterior tracheal ring width (SVL/ATRW) and call frequency at 5% of sound energy (Fig. 8b), relative posterior tracheal width (SVL/PTW) and frequency bandwidth (Fig. 8c), relative larynx width (HW/LW) and maximum power (Fig. 8d), relative anterior tracheal ring width (SVL/ATRW) and dominant call frequency (Fig. 8e), relative posterior tracheal width and call frequency at 95% of sound energy (Fig. 8f).

Fig. 8

Features of laryngotracheal morphology for which there are statistically significant relationships with acoustic properties of distress calls (see Table 1, Column 11). Data for those features of laryngotracheal morphology that were not normally distributed were log-transformed prior to comparison with acoustic call properties, or a nonparametric Spearman's rank correlation was used, as appropriate. Detailed statistical results are provided in  Table S1 (175_herp-75-03-06_s01.docx) in the Supplemental Materials available online.

Ancestral State Reconstructions (ASRs)

The parsimony and maximum-likelihood (ML) ASRs were generally in agreement, although the ML dendrograms generally indicated lower certainty at deep nodes than did the parsimony reconstructions (Fig. 9).

Fig. 9

Ancestral state reconstructions of Gekkota based on parsimony (left) and maximum likelihood (right). (a) cricoid shape, (b) projection angle of the lateral processes of the cricoid cartilage, (c) trachea shape, (d) continuity of tracheal rings, (e) fusion of the cricoid and first tracheal ring(s), and (f) orientation of ventral fibers of m. constrictor laryngis, (g) relative size of the m. constrictor laryngis. Pie charts at each node represent the relative likelihood of a particular state. See Table 3 for definitions and matrix of character states.

Form of the cricoid.—Both analyses indicate that the ancestor of geckos had a broad, ring-shaped cricoid, compared to the rounded cricoid present in the outgroup, and that the plate-like cricoid arose in the ancestor of Ptenopus (Fig. 9a).

Parsimony and ML methods lead to different conclusions about the ancestral state of the orientation of the lateral processes of the cricoid cartilage. The parsimony ASR indicates that the most recent common ancestor of Gekkota possessed lateral processes that project anterolaterally, and that laterally projecting lateral processes arose independently once in Diplodactylidae and once in Eublepharidae (Fig. 9b). The morphology of the lateral processes of the ancestor of the clade comprised of ([Gekkonidae + Phyllodactylidae] + Sphaerodactylidae) is equivocal, but there is strong support that the ancestor of ([Pachydactylus + Chondrodactylus] + Afroedura) possessed lateral processes that project anterolaterally, and that the ancestor of Ptenopus possessed lateral processes that project laterally. The ancestor of Carphodactylidae also had lateral processes that project posterolaterally. Contrastingly, the ML ASR indicates uncertainty about all of the attributes mentioned above for the ancestor of Gekkota, and weak support for all nodes apart from the observed states of the terminal taxa (Fig. 9b).

Form of the trachea.—Parsimony methods indicate that the ancestor of Gekkota possessed a trachea of uniform diameter, and that departures from this morphology arose independently in specific terminal taxa rather than being shared at deeper nodes (Fig. 9c). The tracheal shape of the ancestor of African gekkonids (Ptenopus, Afroedura, Chondrodactylus, and Pachydactylus), however, is equivocal. Again, ML methods provide weak support for the tracheal shape of the gekkotan ancestor (Fig. 9c). The only node strongly supported is that of the ancestor of Eublepharidae, which likely had a trachea of uniform width.

Parsimony and ML methods agree that the ancestral state for tracheal rings was a series of continuous units around the circumference of the trachea and that incomplete rings arose independently in several terminal taxa (Fig. 9d).

The parsimony analysis indicates that the cricoid cartilage is fused to the anteriormost tracheal ring in the ancestor of Gekkota and that separation occurred early in ([Carphodactylidae + Pygopodidae] + Diplodactylidae) lineage and also in Phyllodactylidae and the Hemidactylus + Cyrtodactylus clade (Fig. 9e). The ML ASR indicates uncertainty for the fusion of the cricoid cartilage and the anteriormost tracheal ring of the gekkotan ancestor. The only node strongly supported is that of the ancestor of Eublepharidae, which likely possessed a cricoid cartilage fused with the anteriormost tracheal ring (Fig. 9e).

Muscle form.—The parsimony ASR indicates weak support for most of the deep nodes, but strong support for the premise that the ancestor of the clade containing Gekkonidae + Phyllodactylidae possessed posterolaterally oriented fibers of the m. constrictor laryngis on the ventral surface of the cricoid, and for the ancestor of Gekkonidae and that of the African gekkonids possessing the same morphology (Fig. 9f). Morphologies deviating from this condition arose chiefly in terminal taxa. There is strong support that the ancestor of the clade containing Pygopodidae + Carphodactylidae, and Oedura also possessed this morphology. Laterally oriented fibers arose in the ancestor of the Chondrodactylus + Pachydactylus clade, and also in the ancestor of Hemidactylus. From the ML analyses, there is weak support for all nodes apart from the observed states of the terminal taxa (Fig. 9f).

The size of the m. constrictor laryngis in the gekkotan ancestor is equivocal based on the parsimony ASR, but indicates that the ancestor of the ([Gekkonidae + Phyllodactylidae] + Sphaerodactylidae) clade possessed an intermediate-sized m. constrictor laryngis (Fig. 9g). An enlarged m. constrictor laryngis arose in the ancestor of the ([{Hemidactylus + Cyrtodactylus} + {Agamura + Stenodactylus}] + [Lepidodactylus + Gekko]) clade, whereas a diminutive m. constrictor laryngis is likely the ancestral state for Eublepharidae. The ML ASR also supports a small m. constrictor laryngis as the ancestral state for Eublepharidae, but provides weak support at all other nodes (Fig. 9g).

Outcomes of Hypothesis Testing

Hypothesis 1: Call frequency and body size.—The dominant call frequency of advertisement calls was found to be negatively proportional to body size (Fig. 4a). This size–frequency relationship has also been documented for a diversity of anuran species and presumably reflects a direct correlation between body size and the dimensions of the vocal folds (Gingras et al. 2013). No significant relationship was detected between the dominant call frequency of distress calls and body size. The frequency distribution of distress calls occupies a greater bandwidth than that of advertisement calls (Fig. 4b), and the sound energy of distress calls is more variable and diffuse (Figs. 1–3). It is likely that the dominant call frequency of distress calls is more variable within the sound energy range. As a result, the measurements for these calls are more variable and obviate a similar allometric relationship to that seen for advertisement calls.

One potential confounding factor when investigating ecological correlates across a set of closely related species is the possibility that any observed correlation is explainable as a result of the sharing of traits through phylogeny, rather than reflecting ecological adaptation (Felsenstein 1985; Maddison and Maddison 2005). Such a concern is lessened for this study because of the breadth of the phylogenetic and morphological sampling undertaken. The 17 species used in the regression analysis of advertisement calls comprise representatives of 3 families and 10 genera, and thus are not clustered in a localized area of the gekkotan tree. Additionally, although Gekkonidae was the most heavily sampled family, the taxa investigated represent a diverse range of body sizes (Table 2)—for example, Microgecko (SVL ≈ 28 mm, mass = 0.5 g), Afroedura (SVL = 50–65 mm, mass = 4.3–8.5 g), and Chondrodactylus (SVL = 74–95 mm, mass = 17–38 g). This rationale holds true for all hypotheses tested herein, including those investigating the morphological correlates of phonation.

For Hypothesis 1, we conclude that the dominant frequency of advertisement calls is negatively proportional to body size.

Hypothesis 2: Intra- and interspecific vocal communication.—Our results indicate that advertisement calls are used for intraspecific communication and that distress calls are used for interspecific communication, because the frequency bandwidth of the latter averages 7.69 kHz across all species, compared to the 3.18-kHz average frequency bandwidth for advertisement calls. Gekkotan ears (apart from those of pygopodids, many of which are sensitive to frequencies above 10 kHz; Manley and Kraus 2010) are relatively insensitive to sounds above 10 kHz (Werner et al. 2008). The high-frequency bandwidth of distress calls and the number of distress calls whose frequencies extend above 10 kHz indicate that much of their sound energy is inaudible to geckos, and that this call type is used for interspecific communication. One hypothesis is that such calls serve to deter predators that actively attack geckos, with the high-frequency bandwidth of the distress call being adaptive for startling and deterring predators with hearing sensitivities different from those of geckos (Russell et al. 2014).

For Hypothesis 2, we conclude that advertisement calls are used for intraspecific communication, whereas distress calls are used for interspecific communication. Below, we present further evidence in support of this conclusion.

Hypothesis 3: Species specificity of advertisement calls.—Strong support for advertisement calls being species specific is provided by our DFAs, indicating that such calls are used for intraspecific communication devoted to such aspects as mate attraction and/or territoriality. Our DFAs were able to identify the calling species correctly 92% of the time for advertisement calls when all measured parameters were used (Fig. 5a), and 87% of the time when the parameters were limited to only those that were also measured for distress calls (Fig. 5d). When compared to the 51% rate for correct identification of the calling species for distress calls (Fig. 5e), it is clear that advertisement calls incorporate more distinctive content than do distress calls.

When comparing the calls of five species of Pachydactylus (Fig. 5b), our DFA correctly identified 100% of advertisement calls to the calling species. Recently, it has been shown that geographically isolated and morphologically distinct populations of what was, at the time, regarded as Gekko gecko (the red and black tokays) have distinct components to their calls (Yu et al. 2011). Subsequently, Black Tokay Geckos were elevated to specific status (G. reevesii; Rösler et al. 2011). Our findings that advertisement calls are species specific corroborate those of Yu et al. (2011).

Yu et al. (2011) also reported individually specific components in the advertisement calls of the gekkonids G. gecko and G. reevesii. We found that advertisement calls of the diplodactylid Correlophus ciliatus also conveyed information as to the individual caller, as exemplified by the advertisement calls of three individuals, of which our DFA correctly identified 95% of calls to the calling individual (Fig. 5c). Because of the small number of individuals used in our investigation, the question of individual specificity of calls warrants further investigation.

For Hypothesis 3, our results are consistent with advertisement calls being species-specific forms of communication.

Hypothesis 4: Morphological effects on phonation.—Our morphological examination of laryngotracheal morphology revealed several correlations between morphology and call properties. Increasing the loudness (i.e., maximum power) of calls is likely the result of modifications of laryngotracheal morphology that increase the velocity and pressure of air flow (Russell et al. 2014). Only larynx shape (LW/LL) was positively correlated with the loudness of advertisements calls (Fig. 7g), whereas the loudness of distress calls was positively correlated with the log of the length of the glottal lips (Fig. 8a), and was negatively correlated with relative laryngeal width (HW/LW; Fig. 8d). These data indicate that geckos with larger (specifically, wider) larynges typically produce louder calls. Wider larynges would collect and pass more air during vocalization and forcibly expel it under high pressure through the aditus laryngis, thereby producing louder calls. All three species of Ptenopus that we examined exhibit an extremely large, triangular plate-shaped cricoid cartilage, which appears to be an autapomorphic feature of the genus (Fig. 9a), and they produce the loudest advertisement calls (up to 99.1 dB) of all species recorded. The calls of these species can be heard by humans at distances of several hundred meters (Haacke 1969). Although increased laryngeal size is correlated with increased call loudness, a negative correlation was detected for the frequency (DCF; CF5%, and CF95%) of advertisement calls (Fig. 7a,b,d,e).

Tracheal width relative to body size is positively associated with several frequency measurements (Fig. 7i–k). Russell et al. (2014) noted the highly aberrant tracheal dilatation of Uroplatus and related it to the explosive and energetic distress call of Uroplatus henkeli. Several species examined in this study possess similar tracheal morphologies. The trachea of Chondrodactylus angulifer is greatly expanded anteriorly and this species readily produces explosive distress calls resembling growls. The trachea of Oedura marmorata is dilated like that of Uroplatus, although not quite as distinctively, and its advertisement call resembles the distress call of Uroplatus by exhibiting an elongated, explosive series of chirps.

Apart from Ptyodactylus hasselquistii, no diurnally active species regularly vocalize, and no advertisement calls have been reported for eublepharid geckos (Marcellini 1977; Frankenberg 1978a; Metallinou et al. 2015). No notable features distinguish the larynx and trachea of the diurnally vocal species of Ptyodactylus from those of most crepuscular or nocturnally vocal gekkonids. Additionally, when compared to the moderate to extensive musculature of the gekkotan larynx, the nocturnal, but nonvocal, eublepharid geckos exhibit relatively weakly developed laryngeal musculature. Furthermore, the lateral processes of the cricoid cartilage project only moderately in the eublepharids examined. Russell et al. (2000) suggested that the evolution of laryngotracheal specialization (as seen in Ptenopus and Uroplatus), and probably the expression of vocal abilities, has occurred independently on several occasions within Afro-Malagasy gekkonids alone. Such a contention supports claims that vocal organs in reptiles are evolutionarily highly plastic and represent a random assemblage of morphologies with no central evolutionary tendency (Gans and Maderson 1973; Vogel 1976). This hypothesis concurs with the results of our ancestral state reconstructions.

For Hypothesis 4, our findings are consistent with the idea that variability in laryngotracheal morphology is associated with acoustic properties of phonation.

Hypothesis 5: Ancestral state reconstruction.—Ancestral state reconstructions using parsimony and maximum-likelihood (ML) methods did not always accord with one another. Apart from a few exceptions, ML generally provided weak support for most deep nodes. A lack of phylogenetic signal indicates highly plastic adaptive characters (Losos 2008), which would be expected if laryngotracheal morphology that affects phonation and vocalization is under intense sexual selection. In this case, mate choice might drive the evolution of laryngotracheal morphology, possibly resulting in many evolutionary reversals. Therefore, we would expect a high degree of uncertainty of character states at ancestral nodes, as is evident in the phylogenetic analyses presented here.

In applying Hypothesis 5 to a broad grouping of geckos, we refute the idea that the morphology of the larynx and trachea can be used as reliable phylogenetic markers. Highly derived laryngotracheal structure seems to be a reliable characteristic, however, for some taxa at the generic, or slightly more inclusive, level (Fig. 9).

Conclusions

Overall, we found that advertisement calls are species and individual specific, and that dominant call frequency is negatively correlated with body size. The frequency bandwidth of distress calls extends over a much broader range than the bandwidth of advertisement calls, presumably because they are directed in a nonspecific manner to assist in startling and deterring a variety of predators that potentially exhibit a broad spectrum of auditory ranges. These conclusions support previous hypotheses that have been proposed, but not previously tested, against a large data set representative of Gekkota as a whole.

The gross morphology of the larynx and trachea was found to be fairly consistent across species, although some deviations, such as overall laryngeal size and tracheal dilation, might influence phonation characteristics. Ancestral state reconstructions using parsimony and ML methods traced across a known gekkotan phylogeny revealed that the gekkotan ancestor possessed a ring-shaped cricoid cartilage with distinct lateral processes and tracheal rings that extend continuously around the circumference of the trachea. Reconstruction methods, however, were not always in agreement. Incongruent ancestral state reconstructions under the two analytical approaches might reflect the fact that laryngotracheal morphology does not exhibit strong phylogenetic signal. This interpretation is consistent with claims that reptilian vocal organs are readily influenced by selective forces and represent an assemblage of features that are molded more by function than by phylogeny.