Access to morphological data of herpetofauna is vital not only for systematic study, but also to help understand how taxa function in their environments (i.e., ecomorphology; Williams 1972; Karr and James 1975). Ecomorphological analyses may be complex, requiring consideration of functional and evolutionary contexts (Losos 1990; Gavrilets and Losos 2009; Dugo-Cota et al. 2019; Moen 2019), and benefit from robust morphological data with little noise. Unfortunately, robust morphological data are unavailable for many reptiles and amphibians, hampering insight into their ecologies.

Recording tortoise shell dimensions as part of their morphologies is a simple procedure; however, some taxa are difficult to sample (Branch 2007; Loehr et al. 2021). Consequently, available data may be restricted to opportunistic recordings gathered throughout species' distribution ranges over long time periods. This is the case for Karoo dwarf tortoises (Chersobius boulengeri), a South African endangered and declining endemic (Hofmeyr et al. 2018) that inhabits arid, rocky terrain in the southern interior (Boycott and Bourquin 2000). The number of Karoo dwarf tortoise voucher specimens in databases is restricted to about 113 (including specimens unfit for shell dimension recordings), collected at many different locations between 1881 and 2006 (K. Tolley, pers. comm., July 2021). Thus, publications of Karoo dwarf tortoise shell dimensions are scant and have small (n = 1–19; Duerden 1906; Archer 1968; Haagner 1990; Van Wijk and Bates 1999; Branch 2007) or unknown (Boycott 1989; Branch 1998, 2008; Boycott and Bourquin 2000) sample sizes.

We sampled a population of Karoo dwarf tortoises over a relatively short time period and recorded shell dimensions of all individuals encountered. Besides providing a robust data set, our aim was to assess sexual shape dimorphisms and to compare results with speckled dwarf tortoises (Chersobius signatus; Loehr et al. 2006). Because speckled dwarf tortoises have similar sizes as Karoo dwarf tortoises and inhabit similar arid, rocky terrain, we expected to find few differences.

Methods. — From February 2018 until March 2020, we sampled a Karoo dwarf tortoise population (56-ha area) in the Northern Cape Province, South Africa (coordinates recorded on the biodiversity database of CapeNature, Western Cape Province, South Africa). When a tortoise was located, we recorded its sex (Boycott and Bourquin 2000). Tortoises that were too small for confident sex determination were recorded as juveniles. Subsequently, we used digital calipers to measure (± 0.1 mm) its straight carapace length (SCL, straight-line distance between the nuchal and supracaudal scutes), maximum shell width (MSW), shell width at the seam between the sixth and seventh marginals (SW6–7), maximum shell height (MSH), shell height at the seam between third and fourth vertebrals (SH3–4), midline plastron length (MPL, straight-line distance between the gular scutes anteriorly to between the anal scutes posteriorly), and anal gap (AG, straight-line distance between the anal scutes to the supracaudal scute). Shell volume (SV, cm³) was calculated as π × SCL × MSW × MSH/6000 (Loehr et al. 2007). Prior to release, we photographed and notched (Boycott and Bourquin 2000) tortoises for future recognition, except juveniles with SCL ≤ 40.57 mm, which we marked with nail polish instead.

We calculated means for shell dimensions of males, females, and juveniles, and compared them using Kruskal-Wallis (KW) tests followed by Dunn's post hoc tests. Linear regressions explored relationships among shell dimensions, after log-transformation of variables and covariates (i.e., log-log) if needed to meet test assumptions. In two cases, we accepted slight deviations because regression statistics tend to be robust with respect to departures from test assumptions (Zar 1999): juvenile data were just heteroskedastic (p = 0.050), and male data were nonnormal (p = 0.014) in a regression of (log) SW6–7 vs. (log) MSW. We used analysis of covariance (ANCOVA) to compare significant regressions between males and females. Juvenile covariates showed too little overlap with males and females for ANCOVA. If regression slopes between males and females differed, Zerbe tests (Zerbe et al. 1982) determined the covariate range where slope differences were significant. Statistical tests were conducted in StatsDirect 1.9.12 (ANCOVA; Birkenhead Merseyside, UK) and SigmaPlot 12.0 (all other tests; Systat Software, San Jose, CA). We considered statistical test results significant if p < 0.05.

Results. — The population produced 52 male, 37 female, and 5 juvenile Karoo dwarf tortoises. Males, females, and juveniles had different SCL and AG values (KW test; H₂ ≥ 22.45, p ≤ 0.001). Post hoc tests revealed that SCL was larger in females than in males and juveniles, and larger in males than in juveniles, whereas AG was larger in males than in females and juveniles, and larger in females than in juveniles (Table 1). Mean MSW, SW6–7, MSH, SH3–4, MPL, and SV also differed among groups (H₂ ≥ 58.69, p ≤ 0.001), and were larger in females than in males and juveniles (Table 1).

Table 1. Means, standard deviations (SD), ranges, and sample sizes (n) of shell dimensions (all in mm) and shell volume (cm³) for males, females, and juveniles in a population of Karoo dwarf tortoises (Chersobius boulengeri) in 2018–2020. Slopes and elevations columns contain significant analysis of covariance results between males and females for maximum shell width, maximum shell height, midline plastron length, and shell volume with straight carapace length as covariate, of shell width at the seam between sixth and seventh marginals with maximum shell width as covariate, and of shell height at the seam between third and fourth vertebrals with maximum shell height as covariate. Numbers between brackets in the slopes and elevations columns are covariate regions where slopes differed and mean variable differences between males (M) and females (F), respectively. Anal gap was not significantly related to straight carapace length in females.

View Fullscreen

Shell dimension recordings indicated various sexual shape dimorphisms (Table 1). Recordings of MSW, MSH, MPL (log-log), and SV (log-log) of males, females, and juveniles correlated with SCL (linear regressions; F_1,≥₃ ≥ 32.36, p ≤ 0.006) and, relative to SCL, females had larger MSH and SV than males (ANCOVA; slopes F_1,85 ≤ 0.74, p ≥ 0.39; elevations F_1,86 ≥ 62.11, p ≤ 0.001). In females, MPL and MSW increased faster with SCL than in males (ANCOVA; slopes F_1,85 ≥ 5.51, p ≤ 0.021). AG (log-log) correlated with SCL in males (linear regression; F_1,47 = 95.50, p ≤ 0.001), but not in females and juveniles (F_1,≤3 ≤ 7.78, p ≥ 0.068; Fig. 1). Sexual shape dimorphisms also included shell shape in the vertical, but not the horizontal plane; SH3–4 correlated with MSH in males, females, and juveniles (linear regression; F_1,≥3 ≥ 151.51, p ≤ 0.001) and, relative to MSH, females had larger SH3–4 than males (ANCOVA; slopes F_1,81 = 3.55, p = 0.063; elevations F_1,82 = 39.85, p ≤ 0.001; Fig. 2). In contrast, SW6–7 (log-log) correlated with MSW in males, females, and juveniles (linear regressions; F_1,≥3 ≥ 1828.42, p ≤ 0.001), but relationships were similar between males and females (ANCOVA; slopes F_1,81 = 0.88, p = 0.35; elevations F_1,82 = 0.23, p = 0.63).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Discussion. — Measurements of SCL in our population of Karoo dwarf tortoises were within maximum lengths previously reported for the species based on range-wide opportunistic recordings (90 to > 100 mm in males and 102–110 mm in females; Duerden 1906; Boycott and Bourquin 2000; Hofmeyr et al. 2005), although Branch (2008) reported a much larger maximum length of 160 mm. Other shell dimensions available in the literature are shell heights of 40 and 50 mm for males and females, respectively (Boycott and Bourquin 2000), which were similar to our records. Compared with congeneric speckled dwarf tortoises (Hofmeyr et al. 2017), Karoo dwarf tortoises were larger (average SCL, MSW, MSH, and MPL 8%–16% larger, and average SV 33%–40% larger; Loehr et al. 2006). However, this may reflect the scarcity of small individuals in our population, as indicated by 2%–7% smaller maximum SCL in male and female Karoo dwarf tortoises compared with records for speckled dwarf tortoises. Because male and female maximum SV differed +10% and –20%, respectively, for Karoo dwarf tortoises and speckled dwarf tortoises, the 2 species may display different patterns of sexual dimorphism corresponding to differences in population density, sex ratio, or mating strategy (Djordjević et al. 2011).

Chelonians are the only animals that have bodies enclosed by endochondral bone (Orenstein 2012), resulting in restrictions to limb and tail movement, and limiting expansion of the body to accommodate follicles and eggs. To help overcome restrictions, males often have relatively large shell apertures and females have relatively large shell volumes (Lagarde et al. 2001; Kaddour et al. 2008; Djordjević et al. 2011; Barros et al. 2012). Karoo dwarf tortoises were no exception; males had relatively small plastrons and large anal gaps that might facilitate tail movement during copulation, and females had relatively broad, high, voluminous, and posteriorly domed shells able to accommodate reproductive volume. Because plastral concavities, flatter shapes, and large apertures of male shells increase vulnerability to physical stress, the smaller male size in Karoo dwarf tortoises may help offset shell weaknesses (Vega and Stayton 2011).

Tribute to Retha Hofmeyr. — We dedicate this article to the late Prof. Dr. Retha (M.D.) Hofmeyr. V.J.T.L. first met Retha at the 1999 symposium of the Herpetological Association of Africa in Stellenbosch, after which she provided important feedback on a new study of speckled dwarf tortoises that eventually ran for 14 yrs. This study developed into a doctoral study supervised by Retha and Dr. Brian Henen, which was instrumental for all further tortoise work conducted by V.J.T.L. T.K. first met Retha at the University of the Western Cape, where she supervised his project on angulate tortoises (Chersina angulata) and then his doctoral study of Kalahari tent tortoises (Psammobates oculifer). Retha was an immensely strong person with tremendous integrity and she was a cornerstone in the study and conservation of South African chelonians. Her warmth, honesty, and passion for conservation will remain with us, and we feel that we honor her most when we continue to help conserve South Africa's chelonians.