Study Site

The Khomas Hochland Plateau in central Namibia is characterized by highly dissected plains and rugged mountains. The plains are dominated by mica schists, sandstones, and granite, whereas the Auas Mountains, home to the second highest peak in Namibia (Moltkeblick, 2478 m asl), are mostly comprised of Damara origin quartzite, but also contain schist, amphibolite, marble, conglomerates, and gneiss (Miller 2008). The vegetation is dominated by Highland Shrubland Savanna (Giess 1998; Burke and Wittneben 2008). The Auas Mountain range has been identified as a national biodiversity hotspot (Irish 2002). Although locally influenced by elevation, rainfall in Namibia is more strongly driven by synoptic weather systems originating in the Atlantic and Indian Oceans; consequently, within the distribution of N. pustulatus, the rainfall gradient is predominantly west to east (172–481 mm annually; Kaseke et al. 2016). In contrast, regional temperatures are more strongly associated with elevation; within the distribution of N. pustulatus, the mean annual temperature range is 16.1–20.78C, and it is 2.5–4.68C warmer at lower elevations (,1700 m asl) than at higher elevations (.2000 m asl; Kaseke et al. 2016).

Data Collection

Surveys for N. pustulatus were conducted throughout their range from November 2014 through September 2019. Coordinates and elevation (datum = WGS84; ±3 m) were recorded for each individual. Individuals were extracted from rock crevices with every effort to maintain crevice integrity and habitat (Fig. 2A). In addition, we examined 10 specimens (SVL and head dimensions) from the existing collection at the National Museum Namibia (Appendix).

Fig. 2

(A) Namazonurus pustulatus female, juvenile, and male (left to right) freshly sampled from Düsternbrook Guest Farm; (B) a head scar on a juvenile that is missing frontal, prefrontal, loreal, and pre-ocular scales; (C) picture of scarring on extremities, missing four of five digits; (D) picture of an intact tail; (E) picture of a regenerated tail, breakage indicated by the arrow, notice the whorls are less spikey and a different color from original tail; (F) female femoral region showing femoral pores but lacking generation glands; and (G) male femoral region showing generation glands and femoral pores.

There is no information on size at sexual maturity for N. pustulatus, but our a priori break between juvenile and adult size class was informed by a late-stage embryo found through dissection in a female at SVL 65 mm (Heaton 2017), juvenile (SVL 55–75 mm) size in similar cordylid species (Mouton et al. 1998; Flemming and Mouton 2002; Nieuwoudt et al. 2003), and the frequency distribution of our captures (Fig. 3). Lizards were identified by sex (presence–absence of generation glands; Fig. 2F,G) and assigned to a juvenile (SVL, 65) or adult (SVL ≥ 65 mm) size class. Individuals smaller than 43 mm SVL could not be accurately sexed based on the presence or absence of femoral pores due to a lack of knowledge on size or age of pore development where 43 mm is the smallest SVL where pores were observed (this study). SVL, intact tail length (TL; tails could be identified as regenerated by a difference in color from the original tail, and regenerated tail whorls have duller spikes than original tail whorls; Fig 2D,E), and body total length (SVL þ TL) of field-caught specimens were measured to the nearest 1 mm by using a straight edge ruler; SVLs of museum specimens were measured using a string and straight ruler (1 mm). We measured the distance from the axilla to groin (AG; from the flex point of the forearm to the flex point of the hindlimb), head length (HL; from snout to posterior edge of the parietals), head width (HW; across the temporal region), and head height (HH; at the midpoint of the supraoculars). We also measured the length of the humerus (upper arm: outer shoulder to outer elbow), radius–ulna (lower arm: outer knee to ankle), femur (upper leg: hip to outer knee), tibia–fibula (lower leg: outer knee to ankle), and digits (Right fourth finger and right fourth toe). Measurements were recorded to the nearest 0.1 mm with a dial caliper, and all limb and digit measurements were taken on the right side. Scars were identified as broken tail, missing toes, or visible scarring on the body or head (Fig. 2B,C,E). Mass (grams) was recorded for field-captured individuals by using a spring scale (PESOLA; 10 g and 30 g at 0.1-g and 0.5-g resolution, respectively). Femoral pores and generation glands were counted in the field and/or from high-resolution photos. Sixteen voucher specimens were deposited in the National Museum of Namibia (female = 9, male = 7; Appendix), all other captured individuals were released back to their crevice.

Fig. 3

Frequency histogram of SVL distribution for field-caught Namazonurus pustulatus. Females reach a larger overall maximum SVL.

Data Analysis

Head and body comparisons between adult females and adult males were examined using a Wilcoxon rank sum test with Bonferroni-adjusted P values because the data were not normally distributed. Differences in adult size and the relationship between SVL and HH against elevation with sex as a factor in adult N. pustulatus were quantified using a generalized linear model (Nelder and Wedderburn 1972). HW, HL, and TL against elevation were not tested as the variables were correlated with SVL (HW, t = 9.0097, df = 182, P ≤ 0.01; HL, t = 10.857, df = 183, P , 0.01; and TL, t = 9.7618, df = 91, P ≤ 0.01).

The relationships between SVL versus TL and head dimensions between sexes were examined using regressions. Residuals were tested for normality using the Shapiro–Wilk test. Only the residuals for HH versus SVL were nonnormal, and transformations did not result in a normal distribution; therefore, no transformations were used. Allometric relationships were examined using regressions with the interaction of sex and age with SVL by testing for differences in the slope of AG, TL, and head dimensions.

To further explore the relationship between head dimensions and SVL, segmented regression (Muggeo 2003) analyses were conducted. Segmented regression allows for the inclusion of multiple breakpoints in a linear regression when a change in slope between two variables is detected, using a maximum likelihood approach to estimate the breakpoint. The breakpoint may indicate the size (SVL in our case) at which there is a change in growth patterns, such as growth to reproduction. The existence of breakpoints was tested for using a score test-based approach (Muggeo 2016), and a nonzero difference in the slope parameter of a segmented relationship was identified using a Davies' test, resulting in a best breakpoint (Davies 1987, 2002). A breakpoint was estimated using a segmented regression that was informed by the Davies' test results (Muggeo 2003, 2008). Finally, differences between regression slopes between size/age classes was calculated by a Z-score:

where b is the respective regression slopes and SEb is the standard error of that slope.

Differences in the number of missing toes and tails between the sexes (adults only, juveniles only, and combined) of field-captured individuals were tested using Pearson's chi-square test (with Yates' continuity correction for 2 3 2 tables). The correlations between field-captured adult generation glands and femoral pores with SVL were tested using Pearson's r correlation coefficient. Statistical analyses were done in R (v3.6.1; R Core Team 2019) with the segmented (v1.0.0; Muggeo 2003, 2008) package by using a significance level of P ≤ 0.05.

Body and Head Size

Morphometric data were collected from 224 field specimens and 10 museum specimens including 113 adult females, 77 adult males, 21 juvenile females, 17 juvenile males, and 6 individuals that were too small to determine sex. The largest adult female (SVL = 87 mm; AG = 41.8 mm, and mass = 18.5 g) was larger in overall size and mass than the largest male (SVL = 82 mm, AG = 39.5 mm, and mass = 16.0 g), and on average, females were larger than males in SVL, AG, and mass (Table 1). Adults were larger at lower elevations (R² = 0.35, t = –8.27, P = 0.01) but there was no effect of elevation on HH (R² = 0.00, t = 0.74, P = 0.46).

Table 1

Morphological measurements of adult Namazonurus pustulatus and results of Wilcoxon rank sum tests comparing males and females. Range, X, and SD are in millimeters. P values were adjusted with Bonferroni correction. (*) indicates median instead of X̄. Positive % diff. values indicate that females are larger for the trait.

Adults had similar body and head growth rates but grew slower than juveniles. For juveniles, the variations in HW, HL, HH, and TL were explained by variation in SVL (R² = 0.79, 0.82, 0.55, 0.90, respectively;   Supplemental Table S1 (DeBoerEtAl_22-00008r2.2_Supplemental Table S1.docx), available online). To a lesser degree, the variation in HW, HL, and TL size in adults was explained by variation in SVL (R² = 0.30, 0.39, 0.51, respectively) and HH was not (R² = 0.05;   Supplemental Table S1 (DeBoerEtAl_22-00008r2.2_Supplemental Table S1.docx)). Allometric relationships of HW, HL, and HH indicated that heads of juveniles grew faster than adults (excluding juvenile HH versus adult male HH), but there was no difference in TL between juveniles and adults (Table 2; Fig. 3A,C,E, F). Finally, there were no differences in TL or head dimension growth rates between adult females and adult males (Table 2).

Table 2

Results of t-tests comparing regression coefficients of head width, head length, head height, and tail length versus snout–vent length among the adult sexes and age classes of Namazonurus pustulatus.

Segmented Regressions

Variation in HH was not explained by variation in SVL for adults (R² = 0.05; Fig. 4E;   Supplemental Table S1 (DeBoerEtAl_22-00008r2.2_Supplemental Table S1.docx)), and there was no difference in TL growth rates between any age class or sex (Table 2; Fig. 4F). As a result, a segmented regression analysis was used only on HW and HL (adults and juveniles combined without identification by sex) versus SVL to determine if a breakpoint could be identified. Score tests indicated a single significant breakpoint for HW (P , 0.01) and HL (P , 0.01) and the Davies test for best breakpoint, which informed the segmented regression, was 70.0 mm and 68.5 mm, respectively. Segmented regressions identified a significant breakpoint in slopes at 16.4 mm HW (SVL = 70.0 mm, SE = 1.2 mm, y₁ = 0.3x – 1.8, y₂ = 0.1x þ 9.4, R² = 0.88; Fig. 4B) and at 18.1 mm HL (SVL = 68.3 mm, SE = 1.3 mm, y₁ = 0.2x þ 3.5, y₂ = 0.1x þ 11.6, R² = 0.89; Fig. 4D). Z-scores indicated that the slopes before and after the break point were significantly different (HW, Z = 7.17, P , 0.01; HL, Z = 7.19, P , 0.01), for which the slope before the breakpoint is steeper than the slope after the breakpoint.

Fig. 4

Regressions of (A) head width, (C) head length, (E) head height, and (F) tail length against SVL for adult female, adult male, juveniles, and adult sexes combined of Namazonurus pustulatus. Segmented regressions of (B) head width and (D) head length against SVL for all individuals.

Scar Frequency

Scar frequencies were similar between sex and age. There were no significant differences between sexes or size classes for missing toes (27 of 108 adult females, 22 of 74 adult males, chi-square = 0.29, df = 1, P = 0.59; 4 of 20 juvenile females, 1 of 14 juvenile males, chi-square = 0.30, df = 1, P = 0.58; adults and juveniles, chi-square = 0.03, df = 1, P = 0.87), broken tails (54 of 110 adult females, 39 of 75 adult males, chi-square = 0.06, df = 1, P = 0.81; 4 of 21 juvenile females, 5 of 17 juvenile males, chi-square = 0.13, df = 1, P = 0.72; adults and juveniles, chi-square = 0.15, df = 1, P = 0.70), or either missing toes or broken tail (72 of 127 adult females, 50 of 87 adult males, 6 of 20 juvenile females, and 5 of 14 juvenile males; adults, juveniles, and combined, chi-square, 0.01, df = 1, P = 1.00).

Epidermal Glands

In adult males, the number of generation glands and femoral pores were correlated with SVL (generation glands, n = 72, Pearson's r = 0.33, P , 0.01; femoral pores, n = 72, Pearson's r = 0.33, P , 0.01). In adult females, the number of femoral pores were correlated with SVL (n = 103, Pearson's r = 0.29, P , 0.01).

Body Size

We provide quantitative evidence that Namazonurus pustulatus is a sexually dimorphic species. Adult females were the larger sex, having a higher maximum and average SVL, AG, and mass than adult males. The magnitude of the size difference increased with increasing body size, suggesting that fecundity selection acts on female N. pustulatus, as implicated in other reptiles, such as snakes and turtles (Shine 2000; Fairbairn et al. 2007; Ceballos et al. 2013; Jimenez-Arcos et al. 2016) and cordylids (Van Wyk and Mouton 1998; Costandius and Mouton 2006).

These findings are consistent with those of seven other cordylid species in which females were larger and their size was ascribed to fecundity selection (Fell 2005; Costandius and Mouton 2006; Broeckhoven and Mouton 2014) and was also related to temperature (Mouton and Van Wyk 1993; Mouton et al. 1999; Mouton et al. 2000; Costandius and Mouton 2006). Mouton and Van Wyk (1993) suggested that when resources are limited, such as at cooler temperatures at higher elevations, females allocate more energy to the growth of reproductively significant characters. Namazonurus pustulatus is considered a high-elevation species and is found at the peak of Moltkeblick (2,478 m asl; –22.65068S, 17.18038E; datum = WGS84) in the Auas Mountains, and cooler temperatures may result in more limited resources. However, female-biased size dimorphism is present in species that are not limited to high elevations and cooler temperatures (Costandius and Mouton 2006).

Female-biased dimorphism in body size has also been attributed to a terrestrial lifestyle, whereas male-biased dimorphisms in body and head size have been attributed to a rupicolous lifestyle. Mouton et al. (2000) suggested that females are the larger sex in the terrestrial species C. macropholis because, in general, selection favors small body diameter due to habitat constraints, but longer bodies can hold more or larger embryos. Chamaesaura anguina, a terrestrial snake-like cordylid, also has larger females, and there is a positive correlation between female SVL and clutch size (du Toit et al. 2003). Male-biased dimorphisms in the rupicolous species O. cataphractus are attributed to male competition in response to space limitation whereby a larger body and a larger head size have fitness advantages (Mouton et al. 1999). However, sexual dimorphism cordylid studies are contradictory because some rupicolous species have female-biased size dimorphism (Costandius and Mouton 2006; Broeckhoven and Mouton 2014). As a rupicolous species, N. pustulatus females being larger is also counter to the terrestrial female body-size link. With the apparent conflicting relationships between fecundity selection with either temperature or lifestyle, it is likely that multiple biological and ecological factors influence sexual dimorphism in N. pustulatus.

There have been no studies on resource availability for N. pustulatus, but it likely varies along environmental gradients (i.e., temperature and precipitation) and may contribute to overall body size. In our study population, regional temperatures are 2.5–4.68C warmer at lower elevations than at higher elevations, but annual precipitation increases from the west to east across Namibia (172–481 mm annually; Kaseke et al. 2016). We observed a strong correlation between SVL, TL, HL, and HW versus elevation; adult females and males were larger at low elevations (1200–1700 m asl) than those at high elevations (2000–2500 m asl) but showed increased population density at higher elevations (Heaton et al. 2017). Reduced size at high elevations could be due to fewer resources, which is observed in lizards and snakes (Case 1979; Wikelski et al 1997; Amarello et al. 2010). More information on resource availability is required to see if resources are a factor in overall size for N. pustulatus.

The female-biased size dimorphism could be due to a delay in maturation whereby females attain a large body size before sexual maturity (Tinkle et al. 1970). Delayed maturation is common in viviparous lizards and is often accompanied with long gestation periods, large offspring, and fewer births per year (Tinkle et al. 1970). Namazonurus pustulatus is a viviparous lizard, and gestation data on follicle and embryo development, as well as parturition, indicate a long (6–8 mo) gestation period and large neonate SVL relative to adults (50–53%, female versus male respectively; Heaton 2017). These patterns are also reported in the Ouroborus–Karusasaurus–Namazonurus–Hemicordylus–Cordylus clade but contrast to those in the Ninurta–Chamaesaura–Pseudocordylus clade and Smaug clade, which have larger clutches of smaller offspring (Mouton et al. 2012). Namazonurus pustulatus likely gives birth to a small clutch of offspring (n = 1–2) once a year (Heaton 2017), but females may not reproduce every year, as observed in other cordylids (Van Wyk 1991). Delayed maturation may play a large role in the body size of N. pustulatus females, but no sexual maturity information is available on the species apart from estimates based on size (Branch 1998; this study, below).

Head Dimensions

Male N. pustulatus has larger HW and HL than females. Larger male head size is attributed to sexual selection in which increased male head size could have fitness advantages in male competition or mating (Cooper and Vitt 1989; Gullberg et al. 1997; Gvozdik and Van Damme 2003). These findings are consistent with those of other cordylids in which male head dimensions are larger and are ascribed to male competition and sexual selection (Costandius and Mouton 2006; Mouton 2011; Broeckhoven and Mouton 2014). In the congener N. peersi, adult males have slightly larger HW than adult females (male n = 64, female n = 59), but no differences are identified in other head dimensions (Fell 2005; Mouton 2011). Male N. pustulatus with larger HW and HL may have an increased advantage for mating. Specifically, HW and HL are positively associated with bite-force in cordylids (Broeckhoven and Mouton 2014), which is commonly associated with reptile mating behaviors (Herrel et al. 2001; Husak et al. 2006; Lappin et al. 2006; McLean and Stuart-Fox 2015).

The lack of a difference in HH between male and female N. pustulatus could indicate other selective pressures influencing head shape, such as their rock-crevice niche. For instance, head morphology in the lizard family Gymnophthalmidae (microteiids), comprising several fossorial and niche-restricted species, is a consequence of microhabitat rather than sexual dimorphism or diet (Barros et al. 2011). In cordylids, HH is constrained by habitat for rupicolous species (Losos et al. 2002).

Overall, selective pressures on head size and SVL are complex, and the fit of our regression models of head size against SVL is comparatively weak compared to other studies of sexual dimorphism in cordylids (Cordes et al. 1995; Van Wyk and Mouton 1998; Mouton et al. 1999, 2000, 2005; Costandius and Mouton 2006; Broeckhoven and Mouton 2014). These results are consistent with other rupicolous reptiles that have conflicting selective pressures on head dimensions where males have larger heads for male competition, but the HH is restricted by microhabitat or niche constraints (Losos et al. 2002; Lappin et al. 2006). We also identified no differences in head growth rates in N. pustulatus. This finding is similar to that in Hemicordylus capensis in which adult males have larger heads, but head growth rates between sexes were the same (Van Wyk and Mouton 1998). Conflicting selective pressures cannot optimize all functions simultaneously leading to trade-offs in morphology (Herrel et al. 2007, 2009), and in N. pustulatus, adult females and adult males have equivalent absolute HH, indicating that there might be an optimal head size that balances habitat (rock crevices) and sexual selection pressures in cordylids (Cooper and Vitt 1989).

Sexual Maturity

Our study did not directly seek to determine age at sexual maturity, but the results show a clear change in growth rates that is indicative of sexual maturity in reptiles (Halliday and Verrell 1988; Bjorndal et al. 2013). Our regression analyses indicate significant differences between the slopes of HW and HL versus SVL between adults and juveniles. Furthermore, segmented regressions indicate a change in the rate of head growth relative to SVL at 70.0 mm for HW and 68.3 mm for HL. These analyses suggest a point of differential resource allocation from body growth to reproduction (Cooper and Vitt 1989) in the range of 68.3 – 70.0 mm. The estimated breakpoint is just below that in a report by Branch (1998) on an adult SVL size of 70 – 75 mm. The shift in resource allocation is further supported by the difference in allometric head growth relative to SVL in juveniles versus adults and is also observed in other cordylids (Cordes et al. 1995; Mouton et al. 1999, 2005; Costandius and Mouton 2006). However, some individuals are capable of reproducing before 70 mm, which was demonstrated by a late-stage embryo in a smaller female (SMR 2677, SVL = 65 mm; Heaton 2017), but there could be reproductive size variation among populations. This information means that although we can identify when there is a change in growth rates, we may not be able to statistically determine the breakpoint of differential growth until after sexual maturity has been reached.

Scar Frequencies

The total scar frequency of N. pustulatus is not significantly different between the sexes when comparing adults, juveniles, or the two combined. The total scar frequency of 62% (110 of 179) in N. pustulatus is much higher than that in any other cordylid species recorded to date (Cordes et al. 1995; Mouton et al. 1999, 2000; Costandius and Mouton 2006). In general, high frequencies of tail loss are associated with higher predator pressures (Arnold 1988; Fox et al. 1994; Pafilis et al. 2008; Bateman and Fleming 2009), whereas toe loss is associated with intraspecific competition (Vitt et al. 1974; Schoener and Schoener 1980; Vitt and Cooper 1985). Tail loss is greater than toe loss in N. pustulatus, suggesting that predation pressures are greater than intraspecific competition.

Epidermal Glands

Generation glands are sexually dimorphic in N. pustulatus and are present in males and absent in females. Within Namazonurus, femoral pores (n = 8–30) and generation glands (n = 10–32) were present in males of all species, but generation glands were absent in females (Mouton et al. 2010). For male N. pustulatus, the number of femoral pores (n = 7–11, median = 8) and generation glands (n = 9–23, median = 16) were lower than most Namazonurus spp., but closest to their nearest relative (Stanley et al. 2011), Namazonurus campbelli (femoral pores = 8, generation glands = 10; Fitzsimons 1938). The number of generation glands and femoral pores in N. pustulatus increased with increasing SVL, which is congruent with other cordylids (Mouton et al. 2010; Mouton 2011). However, generation gland development before sexual maturity is uncommon in cordylids (Mouton et al. 2010), and only a few species have generation glands at birth (Mouton et al. 1998). We found eight generation glands on an individual as small as 43 mm SVL (SMR 10731), indicating its sex as male, but not on any smaller individuals (n = 4, SVL = 37–39). We did not dissect any embryos, neonates, or small individuals that could not be sexed by the presence of generation glands (SVL, 43 mm), which may reveal if generation glands are present at birth. Whether for communication or another undetermined function, the presence of generation glands on nonreproductive males suggests that N. pustulatus has an increased need for the early development of generation glands in comparison to cordylid species that develop generation glands later in life.