OUP user menu

Craniometric Sexual Dimorphism and Age Variation in the South African Cape Dune Mole-Rat (Bathyergus suillus)

Leanne Hart, Christian T. Chimimba, Jennifer U. M. Jarvis, Justin O'Riain, Nigel C. Bennett
DOI: http://dx.doi.org/10.1644/06-MAMM-A-058R1.1 657-666 First published online: 1 June 2007

Abstract

Because absolute mammalian age is difficult to measure directly, various methods have been used for its estimation. Among these methods, the degree of molar eruption and wear are considered to be the most reliable indicators of relative age. We used the nature and extent of maxillary molar toothrow eruption and wear to assign individuals of the solitary South African endemic Cape dune mole-rat (Bathyergus suillus) collected from a single population on the grounds of Cape Town International Airport, Cape Town, South Africa, to 9 relative age classes. We then used cranial morphometric analysis, and for comparative purposes, an assessment of the nature and extent of variation in body mass and body length, to investigate the nature and extent of sexual dimorphism and age variation in this little-studied species of mole-rat. Both univariate and multivariate analyses distinguished relative age classes 2 and 3 from 6–9, but age classes 4 and 5 were intermediate between the 2 other age-class groupings, suggesting that individuals of age classes 4 and 5 may be at a point on a hypothetical growth curve where the curve begins to stabilize. Examination of our data showed the absence of sexual dimorphism in younger individuals of age classes 2–5, and its presence in older individuals of age classes 6–9. Together with a proposed study of microsatellites, our analyses may improve our understanding of the population structure of the Cape dune mole-rat.

Key words
  • age variation
  • Bathyergus suillus
  • body mass/length
  • Cape dune mole-rat
  • cranial morphology
  • molar morphology
  • morphometries
  • sexual dimorphism
  • southern Africa

Several studies have assessed the nature and extent of non-geographic variation in sexual dimorphism and variation with age in rodents (e.g., rats [Niviventer coninga (formerly coxingi)—Yu and Lin 1999]; mole-rats [Cryptomys hottentotus, C. damarensis, C. mechowi, and Heterocephalus glaberBegall and Burda 1998; Bennett et al. 1990; Davies and Jarvis 1986; Hagen 1985; Scharff et al. 1999]; mice [Peromyscus maniculatusSchulte-Hostedde et al. 2001]; and tuco-tucos [Ctenomys talarumZenuto et al. 1999]). Because absolute mammalian age is difficult to measure directly, which parameters should be used in the assessment of sexual dimorphism and variation relative to age is of fundamental importance. Although some workers have used body mass to estimate relative age, it may not be appropriate for many species, particularly for subterranean mole-rats (Bennett 1988; Bennett et al. 1990; Janse van Rensburg et al. 2004) because body mass in mole-rats is affected by the type of soil, the availability and quality of food, and in social species, by the social rank of an individual (Bennett 1988, 1989; Jacobs et al. 1991; Janse van Rensburg et al. 2004; Jarvis 1979; Morris 1972; Wallace and Bennett 1998).

The estimation of relative age using the degree of molar eruption and wear is considered to be more reliable for a wide range of mammals, particularly if a sample is from a homogenous habitat within a geographic area in order to eliminate the potential effects of geographic variation (Chaplin and White 1969; Chimimba and Dippenaar 1994; Dippenaar and Rautenbach 1986; Janse van Rensburg et al. 2004; Taylor et al. 1985). Consequently, we used the degree of molar eruption and wear to assess sexual dimorphism and variation in relative age in the solitary South African endemic Cape dune mole-rat (Bathyergus suillus). Nevertheless, body mass, which has been used to evaluate sexual dimorphism and variation relative to age in social species such as the highveld mole-rat (Cryptomys hottentotus pretoriae), as well as body length also were assessed for the solitary Cape dune mole-rat for comparative purposes.

Of additional importance in the evaluation of the nature and extent of sexual dimorphism and variation relative to age is how the derived data are analyzed in order to partition these components of nongeographic variation. Although these have in the past been assessed using a range of univariate analyses (reviewed in Chimimba and Dippenaar 1994), the partitioning of the percent contribution of the sum of squares (%SSQ) of each source of variation to the total SSQ is considered the most appropriate method (Leamy 1983). Nevertheless, there are reservations about the use of univariate analyses in the assessment of nongeographic variation because of the number of variables that must be significant before deciding on the presence of overall significance (Willig et al. 1986). Consequently, multivariate analysis of variance (Zar 1996), which uses rather than ignores correlations among variables, has been recommended as the most appropriate method for evaluating overall differences (Willig et al. 1986). Therefore, we used analysis of variance (ANOVA—Zar 1996), %SSQ, and a series of multivariate analyses to morphometrically evaluate sexual dimorphism and variation relative to age from a single population of Cape dune mole-rats.

The Cape dune mole-rat is endemic to the coastal regions of Western Cape Province, South Africa (Skinner and Chimimba 2005). A recent extermination program at Cape Town International Airport, Cape Town, South Africa, provided an ideal opportunity for a number of studies on this little-studied species. Our study forms part of this broader investigation. We assessed the nature and extent of sexual dimorphism and variation relative to age, and interpreted our findings with reference to the reproductive status of the species.

Materials and Methods

Study animals (87 males and 100 females) were obtained daily between 2003 and 2005 during a mole-rat extermination program at Cape Town International Airport, Cape Town, Western Cape Province, South Africa (33°58′S, 18°37′E). Mole-rats were trapped using a vice trap, which in essence was composed of padding along the edges of 2 armatures designed to catch the animal on either its fore or hind limbs, and were sacrificed using an overdose of halothane anaesthetic. All procedures were undertaken under the guidelines of the American Society of Mammalogists (Animal Care and Use Committee 1998) and the animal ethics committee of the University of Cape Town. Standard data recorded from the collected samples included sex, body mass (g), and body length (mm, measured from the tip of the snout to the base of the tail). Specimens were prepared as voucher specimens at the University of Cape Town and deposited in the mammal collection of the Transvaal Museum of the Northern Flagship Institute, Pretoria, South Africa.

Relative age was estimated by assigning individuals to 9 relative tooth-wear classes using molar eruption and wear on the maxillary toothrow based on the same tooth-wear classes used in the highveld mole-rat (C. h. pretoriae) defined by Janse van Rensburg et al. (2004) as follows (Fig. 1).

Fig. 1

Right maxillary molar toothrow of the Cape dune mole-rat (Bathyergus suillus) illustrating 8 tooth-wear classes (2-9) defined by Janse van Rensburg et al. (2004) and described in the text.

Tooth-wear class 1.—Only 2 cheek teeth completely erupted with a cavity where the 3rd tooth is about to emerge; dentine not deeply worn; little sign of wear on teeth; very deep grooves found on tooth surface; cusps narrow and curved.

Tooth-wear class 2.—Three cheek teeth completely erupted; only the first 2 teeth with signs of wear; no sign of wear on the 3rd cheek tooth.

Tooth-wear class 3.—Three cheek teeth completely erupted; a cavity present where the 4th cheek tooth is about to surface.

Tooth-wear class 4.—Three cheek teeth completely erupted; the 4th cheek tooth has started to surface.

Tooth-wear class 5.—All 4 cheek teeth erupted; no sign of wear on the 4th cheek tooth; dentine on the first 3 teeth slightly scooped.

Tooth-wear class 6.—All 4 cheek teeth erupted; little signs of wear on the 4th cheek tooth.

Tooth-wear class 7.—All 4 cheek teeth erupted; signs of a fair amount of wear on the 4th cheek tooth; dentine on the 1st cheek tooth deeply scooped.

Tooth-wear class 8.—All 4 cheek teeth erupted; dentine on all 4 cheek teeth deeply scooped.

Tooth-wear class 9.—All 4 cheek teeth erupted; dentine deeply scooped; all 4 cheek teeth deformed and reduced in height because of heavy wear.

Similarly, 21 linear cranial measurements defined by Janse van Rensburg et al. (2004) and an additional measurement, the width of incisors (WI—measured as the greatest width of the incisor where the incisor meets the premaxillae) were recorded by 1 observer (LH) to the nearest 0.05 mm using a pair of Mitutoyo digital calipers (Mitutoyo American Corporation, Aurora, Illinois; Fig. 2). These measurements were in turn used to assess craniometric sexual dimorphism and variation in relative age in the Cape dune mole-rat.

Fig. 2

Abbreviations and reference points of 21 skull measurements as defined by Janse van Rensburg et al. (2004) and an additional measurement (WI) included in the present study: GLS—greatest length of skull, from the tip of the front incisors to the posterior part of the skull; ITC—incisor to condyle length, from the anterior surface of the incisor at alveolus to most posterior projection of the occipital condyle; BCW— widest measurement of braincase breadth; ZMB—zygomatic breadth, parietal width, ZYW—greatest zygomatic width, between outer margins of zygomatic arches, perpendicular to longitudinal axis of skull; IOB—least breadth of the interorbital constriction; WR—width of the rostrum; NA—anterior width of nasal where it joins with the premaxillae; UTR—crown length of maxillary toothrow, from the anterior edge of 1st molar to the posterior edge of the last molar; PAC—hard palate width at point of constriction immediately posterior to the last molar; NNP—distance from anterior edge of nasals to anterior edge of posterior part of zygomatic arch; GHS—greatest height of skull, perpendicular to horizontal plane through bullae; MLT—greatest length of mandible, including teeth, from posterior surface of condylar process to the tip of the incisor; MDL— greatest length of mandible (excluding teeth), from posterior surface of condylar process to anteroventral edge of incisor alveolus; MTR— mandibular toothrow length, from anterior edge of the 1st molar alveolus to posterior edge of the last molar alveolus; AFL—articular facet length to posterior edge of 4th molar; MAF—mandibular foramen-articular facet length, from ventral edge of mandibular foramen to midposterodorsal edge of articulating facet; AFA—articular facet to the middle of the angular process; MRH—mandible-ramus height, from dorsal edge of coronoid process to ventral edge of angular process; UJI—upper jaw incisor length, measured from the tip of the incisors to the base, where the teeth connect to the skull; LJI—lower jaw incisor length, measured from the tip of the incisor to the base, where the teeth connect to the skull; and WI— width of the incisor where the incisor meets the premaxillae.

Because no individuals of tooth-wear class 1 were available for our study, sexual dimorphism and age variation in craniometric data, which preliminary analyses indicated to be normally distributed, were 1st assessed simultaneously using ANOVA (Zar 1996) of samples of tooth-wear classes 2–9. All ANOVAs were undertaken after tests for normality and homogeneity of variances showed that the data satisfied the assumptions of ANOVA (Zar 1996). Where statistically significant age differences were detected by the ANOVA, nonsignificant subsets (P > 0.05) were identified by the post hoc Student-Newman-Keuls (SNK—Gabriel and Sokal 1969; Sokal and Rohlf 1981) test of ranked means. Estimates of %SSQ of 4 sources of variation (sex, age, sex-age interaction, and error [= residual]) were computed directly from the derived ANOVA tables by dividing the SSQ associated with each source of variation by the total SSQ.

Sexual dimorphism and variation relative to age also were assessed using the multivariate unweighted pair-group method using arithmetic averages (UPGMA) cluster analysis and principal component analysis (PCA) of standardized variables (Sneath and Sokal 1973). The UPGMA was based on Euclidean distances and correlation coefficients among groups, and the PCA was based on correlation coefficients among variables (Sneath and Sokal 1973). Additional analyses included standard descriptive statistics with Haldane (1955) correction for small sample sizes (n ≤ 4) being used to compute coefficients of variation. All morphometric analyses were based on the 22 cranial measurements, and were undertaken using algorithms in STATISTIC Aversion 5.0 (StatSoft Inc. 2002).

Data on body mass and body length for males and females of each age class were compared by nonparametric Mann-Whitney- U test (Zar 1996) because preliminary analyses showed that these data were not normally distributed.

Results

Univariate analyses.—Our ANOVA showed that almost all measurements differed significantly by age, sex, and the interaction between the 2 components of nongeographic variation (Table 1). However, despite 16 of 22 measurements showing significant sexual dimorphism, the largest F-values were generally associated with age rather than sex or the interaction between age and sex (Table 1), and differences among age groups were highly significant (P < 0.001) for all but 2 measurements (BCW and IOB).

View this table:
Table 1

F-values and percent sum of squares (%SSQ) of each source of variation from an ANOVA of 8 tooth-wear classes (2-9) of male and female cape dune mole-rats (Bathyergus suillus) from Cape Town International Airport, Cape Town, Western Cape Province, South Africa. Statistical significance: * = P < 0.05; ** = P < 0.01; *** _ p < o.OOl. Measurements are defined in Fig. 2.

MeasurementF-value%SSQ
Age (A)Sex (S)A × SAge (A)Sex (S)A × SError
NNP58.87***24.64***6.65***62.983.767.1226.14
ZMB34.67***15.78***3.85***53.183.465.9037.47
BCW1.692.060.426.301.091.5591.05
IOB2.14*3.101.747.451.546.0584.97
ZYW50.87***10.89**5.50***63.121.936.8330.31
WR43.34***9.59**2.80**60.241.903.9033.96
NA20.95***7.55**3.25**42.142.176.5449.14
WI33.92***6.85**3 83***58.821.556.0638.68
PAC4 49***2.861.6314.501.205.2878.90
UTR22.14***9.02**4.10***42.612.487.8947.02
LJI23.64***28.43***4.58***41.687.168.0843.08
MLT58.59***25.38***6.87***62.653.887.3526.12
MDL53.97***22.90***4.63***62.543.795.3728.31
MTR9.31***0.640.3627.220.271.0671.46
AFL6.09***1.431.2319.070.643.8576.45
AFA32.03***12.05***3.87***51.622.786.2539.37
MAP25.88***22.35***2.59*46.145.694.6243.55
MRH33 71***14.65***3.67**52.753.285.7538.22
GLS44.37***28.07***3.57**58.105.254.6731.98
ITC53.76***35.56***421***61.465.814.8127.93
UJI9 87***3.572.44*26.501.376.546.56
GHS11.42***16.88***1.7428.546.034.3661.07
X̄43.163.055.4545.99

The importance of age variation also is shown by the generally higher %SSQ values for age_(%SSQ X̄ = 43.16%; range = 6.30-63.12%) than for sex (X̄ = 3.05%; range = 0.27-7.16%) and the interaction (X̄ = 5.45%; range = 1.06-8.08%) between the 2 components of variation (Table 1). Although 14 of 22 measurements show significant interactions between age and sex at either P < 0.01 or P < 0.001, the mean %SSQ associated with the error (= residual) component (X̄ = 45.99%; range = 6.56-91.05%) is slightly higher than those for age (X̄ = 43.16%; range = 6.30-63.12%; Table 1).

The SNK tests undertaken to identify subsets of age groups that differed significantly revealed 3 contrasting trends. First, 12 of 22 measurements analyzed (ZMB, ZYW, AFA, NA, WI, GLS, ITC, LJI, MLT, AFL, PAC, and GHS) showed an orderly increase in size with increasing age from age classes 2 to 9. Second, 8 measurements (NNP, MDL, MTR, UTR, MAF, MRH, WR, and UTJ) consistently grouped individuals of the younger age classes 2 and 3 in 1 subset, and those between the older age classes 4–9 in a significantly different subset. Third, 2 measurements (BCW and IOB) showed no statistically significant differences among all 8 age classes analyzed. Among the 12 cases where there was an orderly increase in size with age, individuals of age class 4 were grouped with individuals of the younger age classes 2 and 3 for 3 measurements (AFL, PAC, and GHS).

Multivariate analyses.—The 1st principal component explained 67.2% of the total variance, and had high negative loadings (-0.728 to -0.971) on 17 of 22 measurements, and moderately negative loadings on the rest (Table 2). Only 1 measurement (IOB) had a relatively high loading (also negative) on the 2nd principal component, which only explained 5.5% of the total variance (Table 2). The 1st principal component generally represented differences with regard to size, whereas the 2nd principal component generally represented differences with regard to shape.

View this table:
Table 2

Loadings of measurements from the first 2 components of a principal component analysis of the cape dune mole-rat (Bathyergus suillus) of tooth-wear classes 2–9 from Cape Town International Airport, Cape Town, Western Cape Province, South Africa. Measurements are defined in Fig. 2.

Principal component axes
MeasurementIII
NPP−0.970.045
ZMB−0.932−0.025
BCW−0.389−0.451
IOB−0.182−0.810
ZYW−0.9650.080
WR−0.9150.010
NA−0.876−0.016
WI−0.8930.096
PAC−0.430−0.475
UTR−0.728−0.120
LJI−0.8780.014
MLT−0.9850.073
MDL−0.9710.065
MTR−0.542−0.069
AFL−0.5720.164
AFA−0.9270.113
MAP−0.898−0.008
MRH−0.9150.095
GLS−0.9340.078
ITC−0.9430.046
UJI−0.7300.067
GHS−0.798−0.146
% variance explainedAxis I = 67.20%Axis II = 5.50%

Because of the large number of individuals involved in the analysis, we 1st examined the biplot of PCA axis 1 versus PC A axis 2 with reference to age (Fig. 3). Tooth-wear classes 2 and 3 largely separated from tooth-wear classes 4–9 on the 1st, size-related rather than the shape-related principal component axis. There was extensive overlap among the higher age classes along both axes, but the scattergram showed a tendency for individuals of age classes 4 and 5 to fall intermediate between those of age classes 2 and 3 and those of age classes 6–9 on the 1st PCA axis.

Fig. 3

The first 2 axes from a principal component analysis of males and females showing the distribution of 8 tooth-wear classes (2-9) of the Cape dune mole-rat (Bathyergus suillus) from Cape Town International Airport, Cape Town, Western Cape Province, South Africa. Minimum convex polygons enclose individuals of each tooth-wear class. For illustrative clarity, the scatter of individuals in multivariate space and their associated sexes have been omitted.

A distance phenogram from the UPGMA cluster analysis (Fig. 4) showed 3 relatively discrete clusters. One cluster comprised largely male and female individuals of age classes 2 and 3, with some individuals of age classes 4 and 5, suggesting no sexual dimorphism within these younger age classes. A 2nd cluster comprised largely individuals of age classes 6–9, with some individuals of age classes 4 and 5, and with some minor subclusters within this cluster comprising individuals of the same sex. A 3rd cluster only included males of age classes 6–9, which are larger than females, suggesting the presence of sexual dimorphism in these older age classes.

Fig. 4

A distance phenogram from an unweighted pair-group method using arithmetic averages (UPGMA) cluster analysis of male and female Cape dune mole-rats (Bathyergus suillus; age classes 2–9) from Cape Town International Airport, Cape Town, Western Cape Province, South Africa. Group 1 comprised the following sample sizes, age classes, and sexes (M = male; F = female), respectively: 3:2F, 1:2M, 3:3F, 11:3M, 8:4F, 2:4M, 5:5F, 2:5M, 1:6F, 1:7F, and 1:8F; group 2: 7:4F, 5:4M, 11:5F, 6:5M, 14:6F, 7:6M, 17:7F, 8:7M, 20:8F, 6:8M, 6:9F, and 3:9M; and group 3: 1:2M, 1:3F, 2:6M, 6:7M, 17:8M, 1:9F, and 11:9M.

The absence of sexual dimorphism among younger individuals and its presence among older individuals was examined further by independent ANOVA and %SSQ of individuals of age classes 2–5 and age classes 6–9, respectively (Table 3). An ANOVA including only individuals of age classes 2–5 revealed that 19 of the 22 measurements differed significantly by age, but none differed by sex, with the interaction between sex and age being significant for 7 measurements (Table 3a). The importance of age variation rather than sexual dimorphism in the younger age classes 2–5 also is shown by the generally higher %SSQ for age__(%SSQ X̄ = 35.93%; range = 2.94-58.46%) than for sex (X̄ = 0.68%; range = 0.00-3.13%) or the interaction (X̄ = 5.37%; range = 1.07-12.79%) between age and sex (Table 3a). The %SSQ analysis of age classes 2–5 also showed higher mean %SSQ values for the error component (X̄ = 58.02%) than for age, sex, and or the interaction between age and sex (Table 3a). Where significant interactions were found between age and sex for both age classes 2–5 and age classes 6–9, most differences were significant at P < 0.05, which means that the possibility of type II error is high. Given that our analyses were based on 22 measurements, the general lack of the importance of this interaction may perhaps be reflected in multivariate analysis, which included all 22 variables and offered a much more robust interpretation with reference to age and sexual dimorphism in multivariate space.

View this table:
Table 3

F-values and percent sum of squares (%SSQ) of each source of variation from an ANOVA of male and female cape dune mole-rats (Bathyergus suillus) of tooth-wear classes 2–5 and tooth-wear classes 6–9 from Cape Town International Airport, Cape Town, Western Cape Province, South Africa. Statistical significance: * = P < 0.05; ** = P < 0.01; *** = P < 0.001. Measurements are defined in Fig. 2.

F-value%SSQ
MeasurementAge (A)Sex (S)A x SAge (A)Sex (S)A x SError
Tooth-wear classes 2—5
NNP34.11***0.045.38**58.310.029.1932.48
ZMB29.38***0.181.8258.460.123.6237.80
BCW1.000.330.334.900.541.6392.93
IOB0.670.982.93*2.941.4212.7982.85
ZYW20.71***0.681.2750.240.563.0946.10
WR17.13***0.121.0246.060.112.7551.08
NA8.61***0.020.2130.950.020.7568.29
WI11.90***0.601.9435.770.595.8357.81
PAC43.41***0.121.4417.490.165.7976.55
UTR32.23***0.335.23**56.980.199.2433.59
LJI11.84***3.212.2234.653.136.5955.63
MLT32.46***0.051.4861.310.002.7935.89
MDL30.97***0.170.407.870.271.9089.96
MTR29.65***0.141.5659.000.093.1037.81
AFL2.130.010.239.980.011.0788.93
AFA18.67***1.071.2347.550.913.1448.40
MAP19.27***1.293.67*45.471.028.6744.84
MRH33.50**0.172.1713.710.228.8277.25
GLS22.69***2.003.66*49.311.457.9641.29
ITC20.85***2.133.12*47.611.627.3943.38
un10.60***0.333.04*32.380.349.2758.01
GHS4.90**1.610.7019.512.132.8075.56
X35.930.685.3758.02
Tooth-wear classes 6—9
NNP91.66***3.03*11.3337.853.7547.07
ZMB4.67**60.58**1.087.3031.581.6859.44
BCW0.273.370.620.682.811.5494.97
IOB4.09**3.120.379.432.400.8687.32
ZYW934***54.10***3 99**18.130.357.7473.79
WR9.42***42.59***1.5214.9822.591.9860.45
NA5.10**29.99***3.61*9.0017.636.3667.01
WI5.76**35.28***2.419.9420.304.1765.59
PAC2.186.79*2.224.895.064.9785.08
UTR1.4942.14***1.952.7025.313.5268.47
LJI8.30***53.82***4.03**12.1526.285.9155.66
MLT11.35***86.75***3.61*13.8635.324.4146.41
MDL9 94***69.41***2.80*13.4531.323.7951.44
MTR1.802.610.154.402.130.3893.09
AFL1.306.78*463.015.263.4088.32
AFA6.72***53.25***0.3610.6928.250.5760.49
MAP7.34***49.25***0.7711.7426.261.2360.77
MRH5.68**48.30***1.279.3026.372.0862.25
GLS7.28***63.87***1.5010.7031.272.2055.82
ITC11.27***93.57***2.69*13.5637.513.2345.70
un3.25*13 77***1.846.819.633.8579.71
GHS5.96***34 09***2.0210.3919.823.5266.27
X9.4720.243.2367.05

However, an ANOVA including only individuals of age classes 6–9 (Table 3b) showed that 17 of the 22 measurements differed significantly by age, 19 were sexually dimorphic with males being larger, and 7 had statistically significant interactions between age and sex. Unlike the earlier ANOVAs, this analysis yielded higher F-values for contrasts based on sex than on age (Table 3b). The greater importance of sexual dimorphism than variation associated with age in the older age classes 6–9 also is shown by the generally higher %SSQ for sex (%SSQ X̄ = 20.24%; range = 0.35-37.85%) than for age (X̄ = 9.47%; range = 0.68-18.13%) or the interaction (X̄ = 3.23%; ragne = 0.38-7.74%) between age and sex (Table 3b). Similar to the %SSQ analysis of age classes 2–5, the analysis of age classes 6–9 showed the highest mean %SSQ values for the error component (X̄ = 67.05%; Table 3b).

Independent PCAs of age classes 2–5 and of age classes 6–9 showed clearer indications of the lack of sexual dimorphism in the younger tooth-wear classes 2–5 (Fig. 5a) and its presence in the older tooth-wear classes 6–9 (Fig. 5b) than the PCA based on the total sample (Fig. 3).

Fig. 5

The first 2 axes from a principal component analysis of males and females of a) tooth-wear classes 2–5 and b) tooth-wear classes (6-9) of the Cape dune mole-rat (Bathyergus suillus) from Cape Town International Airport, Cape Town, Western Cape Province, South Africa. Dashed and continuous minimum convex polygons enclose male (M) and female (F) individuals of each tooth-wear class, respectively.

Body mass, body length, variation relative to age, and sexual dimorphism.-—Given that the craniometric analyses suggest the absence of sexual dimorphism in individuals of age classes 2–5 and its presence in individuals of age classes 6–9, these results were independently evaluated with reference to body mass, body length, variation relative to age, and the sexes. We found an increase in body mass and body length with increasing variation relative to age in both sexes. The younger tooth-wear classes 2–5 did not show sexual dimorphism in bodyJmass, and tooth-wear classes 2–6 did not show sexual dimorphism in body length, whereas males and females of olderftooth-wear classes 6–9 differed in body mass and those in agetdasses 7–9 differed in bodv length (Fig. 6).

Fig. 6

Relative age and a) body mass (g) and b) body length (mm; all ± SE) in male and female individuals of the Cape dune mole-rat (Bathyergus suillus) from Cape Town International Airport, Cape Town, Western Cape Province, South Africa, representing 8 relative age classes (2-9). Relative age classes are adopted from Janse van Rensburg et al. (2004), illustrated in Fig. 1, and defined in the text. Males are represented by stripped bars and females by solid bars. Statistical significance: * = P < 0.05; **=P< 0.01; *** = P < 0.001; NS = not statistically significant.

Discussion

Both univariate (ANOVA and %SSQ) and multivariate (UPGMA cluster analysis and PCA) analyses showed statistically significant variation with age, but no statistically significant sexual dimorphism, among individuals of the Cape dune mole-rat of age classes 2–5. In these age classes, sexual dimorphism contributed only 0.66% to, the total variance observed, whereas age contributed 36.0% of the variation. An analysis of individuals within age classes 6–9 showed statistically significant differences both among age classes and by sex. In these age classes, age contributed only 9.47% of the total variance observed whereas sexual dimorphism contributed 20.24% of the variation.

Our results show the presence of sexual dimorphism in adult but not in juvenile and subadult individuals of the Cape dune mole-rat. Our PCA of age classes 2–5 showed some indication of sexual dimorphism in individuals of age class 2, but this age class was only represented by a few individuals (n = 4) and the difference may likely reflect an artifact of small sample size. Generally, our results showed a separation of age classes 2 and 3 and 6–9, with age classes 4 and 5 being largely intermediate between the 2 distinct age-class groupings. This suggests that individuals of age classes 4 and 5 may be at a point where the growth curves for each sex begin to diverge.

The results of our study are similar to those of a study by Yu and Lin (1999), who found body mass in the spiny Taiwan niviventer (N. coninga [formerly coxingi]) to be sexually dimorphic in older individuals rather than in juveniles and subadults. A study on growth in the social Ansell's mole-rat (Cryptomys anselli) showed a statistically significant difference in the maximum growth rates between males and females (Begall and Burda 1998). Both males and females in this species showed a similar rate of growth from birth up to the 18th-20th week, after which males developed considerably faster than females.

Other studies on subterranean rodents based on body mass, and external or cranial measurements, or both, with reference to age structure and reproductive status, have found sexual dimorphism in the social Damaraland mole-rat (Cryptomys damarensisBennett et al. 1990), the solitary Namaqua dune mole-rat (Bathyergus janettaDavies and Jarvis 1986), and in the Los Talas tuco-tuco (Ctenomys talarumZenuto et al. 1999). Other rodents in which sexual dimorphism in body size occurs include the deer mouse (Peromyscus maniculatus), the bushy-tailed wood rat (Neotoma cinerea), and the southern red-backed vole (Myodes [formerly Clethrionomys] gapperiSchulte-Hostedde et al. 2001). In contrast, no sexual dimorphism has been shown to occur in the social common mole-rat (Cryptomys hottentotus hottentotusBennett et al. 1990), the highveld mole-rat (C. h. pretoriaeJanse van Rensburg et al. 2004), the naked mole-rat (Heterocephalus glaberHagen 1985), the solitary Cape mole-rat (Georychus capensisTaylor et al. 1985), and the silvery mole-rat (Heliophobius argenteocinereusScharff et al. 1999).

The 1st principal component and the distance phenogram in our study suggest that the differences between males and females and between age classes are largely due to overall body size rather than body shape. Because male-male interactions may increase during the reproductively active phases of the life cycle when male Cape dune mole-rats compete for reproductive opportunities (Bennett and Faulkes 2000), it may be advantageous for a male to invest more resources into its growth and the maintenance of its body condition. The generally larger body size observed in adult males of the Cape dune mole-rat of age classes 6–9 may be reproductively advantageous with respect to mating opportunities. Dissection of male Cape dune mole-rats showed extensive fatty tissue deposits around the neck region (L. Hart, in litt.). These fatty tissue deposits may cushion males from incisor-inflicted damage that occurs during male-male fighting during competition for mating opportunities. Under laboratory conditions, male-male interactions are witnessed quite frequently and under field conditions, interlocking skulls have been found on various occasions (L. Hart, in litt.).

The use of molar eruption and wear to age animals has been considered an unreliable indicator of age in some mammals (Morris 1972) such as bats (Myotis lucifugusHall et al. 1957), elk (Cervus elephusKeiss 1969), and white-tailed deer (Odocoileus virginianusGilbert and Stolt 1970). However, in the absence of data on absolute age, the use of molar eruption and wear in the assessment of relative age within homogeneous populations as in our investigation may be appropriate.

Similarly, body size also has been shown to be a poor indicator of age in other mammals (Cameron-Smith 1965; Chaplin and White 1969). In the Cape dune mole-rat, there was a trend for an increase in body mass and body length with increasing relative age in both sexes. Because our study was on a solitary species, there was no additional constraint on body mass and body length due to the social rank of an individual, as has been found in social mole-rats (Bennett et al. 1990). Of particular relevance in our study is that, similar to the craniometric data, sexual dimorphism with regard to both body mass and body length also was shown to be absent in younger age classes but present in the older age classes. Unlike the social mole-rat species cited above, these results suggest that body mass and body length also could be good indicators of relative age and sexual dimorphism in this solitary species of mole-rat.

Although our analyses showed age differences among the 8 age classes examined, the absence of sexual dimorphism in younger individuals of age classes 2–5, and its presence in older individuals of age classes 6–9, the large error (= residual) component in the %SSQ analyses indicated other sources of variation may be influencing the nature and extent of nongeographie variation in the Cape dune mole-rat, and may require further investigation. In addition, our analyses were based on a single population of the Cape dune mole-rat, and there is a need to assess additional populations before generalizations in this species should be made. Studies of additional populations, together with analyses of microsatellites, may provide additional tools to improve our understanding of population structure in the solitary Cape dune mole-rat.

Acknowledgments

We thank the Airports Company of South Africa for permission to use mole-rats captured at Cape Town International Airport. We are grateful for field and technical assistance given by J. Flanagan, F. Netshidzati, V. Netshidzati, A. O'Flaherty, and P. Streydom. We thank L. Janse van Rensburg for assistance with statistical analyses, Y. van der Merwe for assistance with illustrations, and anonymous reviewers for constructive comments on the manuscript. Funding was provided by the South African National Research Foundation (GUN no. 2069070) and the University of Pretoria to NCB, and Airports Company of South Africa and the University of Cape Town to JUMJ and JO, and is gratefully acknowledged.

Footnotes

  • Associate Editor was Nancy G. Solomon.

Literature Cited

View Abstract