The body size of a univoltine carabid beetle Carabus tosanus on Shikoku Island, Japan, was clearly smaller in higher-altitude populations (subspecies), which possibly represents incipient speciation. To explore the determinants of altitudinal differences in body size in this species, we studied the degree of phenotypic plasticity by conducting rearing experiments at two constant temperatures and examined genetic differences through interpopulation crosses. At 15 °C, C. tosanus had a longer developmental period and a shorter adult body than at 20 °C. Nevertheless, variation in body size due to temperature effects (phenotypic plasticity) was small compared to the interpopulation differences, which suggests substantial genetic differences between populations (subspecies) at different altitudes. In F1 offspring from crosses between a low-altitude (subspecies tosanus) and a high-altitude population (subspecies ishizuchianus), adult body length was affected by the genotypes of both parents, with an interaction effect of parental genotype and offspring sex. Further analyses revealed that adult body length was affected by sex-linked factors in addition to autosomal factors. These genetic differences in body size may have resulted from adaptations to different altitudes and may be important for the process of incipient speciation because body size differences could contribute to premating reproductive isolation.
Body size is a key trait involved in adaptation because it affects the physiological and life history traits of an organism. Geographical variation in body size is widely observed, the most common pattern being for increasing body size with latitude, which is called Bergmann's rule (Bergmann, 1847). This pattern is observed in many endotherms and in some ectotherms such as insects (Blanckenhorn et al., 2006; Stillwell et al., 2007; Stillwell, 2010). Because the body surface–volume ratio generally decreases with increasing body size, body size plays an important role in the thermoregulation of endotherms, and in starvation resistance and desiccation resistance in ectotherms. In contrast, clinal body size variation in arthropods often follows the converse of Bergmann's rule (Masaki, 1967; Mousseau, 1997). In univoltine insects, which can only overwinter at a particular developmental stage, their developmental time is restricted by habitat temperature. The decrease in body size in cooler habitats can be explained by selection for a shorter developmental time, which results in smaller body size. Therefore, the converse of Bergmann's rule is considered a result of climatic adaptation in univoltine arthropods (Masaki, 1967; Roff, 1980).
Differences in body size play a major role in reproductive isolation, particularly between incipient species (Schluter & Nagel, 1995; Wood & Foote, 1996; Nagel & Schluter, 1998; Lu & Bernatchez, 1999; Tanabe & Sota, 2008). Therefore, the evolution of body size can result in speciation. Clinal variation in body size reflects divergent local adaptations because body size is a determinant of reproductive capacity and survival, and it is constrained by developmental time. If body size diverges genetically (rather than environmentally) between adjacent local populations, it can cause reproductive isolation and result in parapatric speciation or para-allopatric speciation (Coyne & Orr, 2004). Therefore, to examine the possibility that clinal body size differentiation causes speciation, studying the genetic basis of clinal variation in body size is important. Nevertheless, few studies have determined genetic and environmental factors involved in clinal variation in body size (Stillwell, 2010). Genetic basis of large body size differences in animals with internal fertilization is difficult to study because of the difficulty in mating per se due to size differences. Thus, the role of body size differentiation in speciation has not been well explored despite its potential importance as an agent of reproductive isolation.
Ground beetles of the subgenus Ohomopterus in the genus Carabus are endemic to Japan and exhibit marked diversity in body size. Ohomopterus species are univoltine and the larvae develop from spring to summer, feeding exclusively on megascolecid earthworms (Sota, 1985). The differentiation in body size, along with marked divergence in genital morphology, serves as a major component of species diversity in this group despite the relative paucity of ecological divergence (Sota & Nagata, 2008). Generally, each species shows clinal reductions in body length (i.e. the converse of Bergmann's rule along temperature, latitudinal and altitudinal gradients; Sota et al., 2000). Because the period that is suitable for larval development is restricted at higher latitudes or altitudes, and shorter developmental periods, and consequently, smaller body sizes may be selected as a result of climatic adaptation (Sota et al., 2000). In addition, differences in body size occur among sympatric species (Sota et al., 2000) and are effective at preventing interspecific mating and reproductive interference between co-occurring species (Okuzaki et al., 2010; see also Nagata et al., 2007). Interspecific differences in body size are correlated with differences in developmental time and could be genetically based (Sota, 1985; Sota et al., 2002). However, the genetic basis and phenotypic plasticity (in terms of temperature) involved in the intraspecific variation in body size are generally unknown.
Here, we focused on one Ohomopterus species, Carabus tosanus, which shows marked variation in body size. This polytypic species is confined to Shikoku Island, Japan, where it occurs throughout a wide range of elevation (alt. 50–1900 m) (Fig. 1) and exhibits marked altitudinal variation in body size (Sota et al., 2000). Populations at lower altitudes (approximately 50–1200 m) in the central to south-west part of Shikoku possess large bodies and are black in colour; they are designated as the subspecies tosanus. In contrast, populations with smaller bodies that are copper in colour occur at higher altitudes (approximately 900–1900 m); three subspecies are described from different mountain regions: ishizuchianus, kawanoi and botchan (Ishikawa, 1985; Imura & Mizusawa, 1996).
However, mitochondrial gene sequences have shown no clear differentiation among the subspecies (Sota & Nagata, 2008; Y. Tsuchiya, unpublished), suggesting that these subspecies have differentiated recently. In our field study on the life histories of the subspecies tosanus and ishizuchianus conducted in a mountain area (Ikeda et al., 2012), the subspecies were displaced at an altitude of around 1000 m without a trend for continuous intergradation in body size. Although the body size decreases with an increase in altitude and shows a clinal variation within each subspecies, there is a conceivable gap in body size between subspecies. These subspecies differed in the timing of their life cycles as well as in body size, which appeared to reflect altitudinal differences in temperature conditions (cooler at higher altitudes) and larval food sources (earthworms, which are scarcer at higher altitudes) (Ikeda et al., 2012). The difference in body size between these subspecies was so large that it might prevent hybridization. Therefore, if the body size difference is genetically based, it may facilitate speciation. Thus, it is important to show the presence of genetic difference between subspecies for the study of incipient speciation as a by-product of adaptation to different altitudes. To determine the extent to which genetic factors and phenotypic plasticity contribute to the observed difference in body size, we conducted laboratory experiments using representative populations at different altitudes.
Materials and methods
Study organisms and field sampling
Adult females of C. tosanus were collected from seven sites at altitudes of 620–1500 m, which covered the habitats of all four subspecies, using pitfall traps from April to June in 2008 and 2009 (Table 1; Fig. 1). Collected beetles were transported to our laboratory in Kyoto, Japan, within 2 days of capture and reared in an incubator at 20 °C with L16/D8 light conditions and were fed minced beef every other day. These beetles were used to produce populations of experimental offspring.
Table 1. Sampling sites, mean body length for field-collected beetles, numbers of field-collected females used in experiments and numbers of their offspring under two temperature conditions for Carabus tosanus
Population No. (code)
Locality Latitude, °N, Longitude, °E
Body length, mm
Mean ± SD (n)
Reared in 2009.
Reared in 2008.
M, male; F, female; n0, number of field-collected females; n1, number of emerging adults at 20 °C; n2, number of emerging adults at 15 °C.
Subspecies: 1—3, C. t. tosanus; 4, 6, 7, C. t. kawanoi; 5, C. t. ishizuchianus.
Effects of temperature on adult body size and developmental time
To study the effects of temperature on adult body size and developmental time, we reared offspring from field-collected females at 15 and 20 °C under long-day conditions (L16/D8). These temperatures were selected by referring to the mean air temperatures at the field sites during the larval development (June–August) of C. tosanus. An approximate mean temperature for sites at an altitude of 1000 m was 20 °C, whereas 15 °C represented 1700-m habitats (Ikeda et al., 2012). Females that were collected in the field were reared individually in plastic cups (10 cm diameter and 10 cm height) at 20 °C with L16/D8 and were fed minced beef every other day. These females had probably copulated in the field and stored sperms in the spermatheca, and most of the eggs that were deposited in the laboratory were fertilized. The eggs were divided into two groups and were incubated at 15 °C or 20 °C under L16/D8. Hatched larvae were reared individually in plastic cases (6 cm diameter and 4 cm height) and were supplied with sufficient amounts of earthworms until they hid in the soil to pupate. The body lengths of emerging new adults were measured using an electronic digital calliper (0.01 mm unit). Developmental time was defined as the number of days from egg deposition to adult emergence.
To study the effects of temperature, sex and altitude of the source population on adult body size and developmental time, we used an analysis of covariance (ancova). Initially, the full model with all interaction terms was examined, and interaction terms without significant effects were removed from the final model. Log-transformed values for body length and developmental time were used. In addition, the relationship between adult body size and developmental time was analysed using an ancova that included sex and temperature as independent variables and developmental time as a covariate. Interaction terms without significant effects were removed from the final model. We used log-transformed mean body length and mean developmental time for each population.
Genetic basis of body size
To study the genetic basis of body size differences between high- and low-altitude subspecies, we conducted a crossing experiment using a population of the high-altitude subspecies ishizuchianus from Mt. Ibuki (No. 5 in Table 1; hereafter H) and a population of the low-altitude subspecies tosanus from Mt. Saragamine (No. 2 in Table 1; hereafter L). Adult beetles that were obtained in the previous experiments were crossed while being held at 20 °C with L16/D8 (female × male: L × L, L × H, H × L, H × H). The crosses included four pairs of H × H, six pairs of L × L and five pairs for each of L × H and H × L. The eggs that were laid were collected and incubated at 20 °C (L16/D8), and all of the hatched larvae were reared. Body weights for first, second and third instar larvae and adults were measured within 1 day after hatching, moulting or eclosion, respectively, using a semi-micro-balance (Sartorius BP 210 D) that was accurate to the nearest 0.01 mg. Body lengths of emerging adults were measured to the nearest 0.01 mm using a digital calliper. Further crosses using the F1 generation were attempted but were unsuccessful.
To determine the effects of parental genotype (H or L) and offspring sex on adult body length, a three-way anova was performed. Interaction terms without significant effects were removed from the final model. Body lengths were log-transformed. To distinguish between maternal effects and the effects of genetic factors on autosomal and sex (X) chromosomes, we tested the significance levels of four contrasts between specific cross types in offspring of either sex using anova following the method proposed by Fairbairn & Roff (2006). Note that Carabus species, including the subgenus Ohomopterus, have XY sex chromosomes (Weber, 1966; Serrano & Galián, 1998; T. Sota, unpublished). Maternal effects could be detected in female offspring by testing the contrast between H × L and L × H crosses. Autosomal effects could be detected in male offspring by testing the contrasts between H × H and H × L or between L × H and L × L. The combined effects of sex-linked and maternal effects could be detected in male offspring by testing the contrast between H × L and L × H. Sex-linked effects were attributable to the X chromosome, assuming that Y-linked effects were negligible (Fairbairn & Roff, 2006).
Effects of temperature on adult body size and developmental time
Figure 2 shows adult body lengths and developmental periods for all of the experimental populations, but only four populations (No. 2, 4, 5 and 6) were reared at both 15 and 20 °C. For these four populations, ancova showed that all three independent variables had significant effects: adult body length was larger at 20 °C than at 15 °C, larger in females than in males and larger for populations from lower altitudes (Table 2). Additionally, only the altitude × sex interaction was significant among the four possible interaction effects; sexual body size differences tended to decrease at higher altitudes. Developmental time was significantly longer at 15 °C than at 20 °C and was longer for populations from lower altitudes, but it did not differ between the sexes (Table 2). ancova showed that mean body length was positively associated with developmental time when the effects of sex and temperature were excluded (developmental time: d.f. = 1, b =1.722, F =24.75, P =0.0001; sex: d.f. = 1, F =9.56; P =0.0066; temperature: d.f. = 1, F =29.29, P <0.0001).
Table 2. Results of ancova for the effects of altitude of source population and temperature on adult body size and developmental time in Carabus tosanus
Adult body length
Altitude × Sex
Genetic basis of body size
The experimental crosses conducted using the L population (low elevation; Mt. Saragamine) and the H population (high elevation; Mt. Ibuki) produced 11 female and 12 male offspring from the L × L pairs, 16 female and 13 male offspring from the H female × L male pairs, 20 female and 11 male offspring from the L female × H male pairs and eight female and seven male offspring from the H × H pairs. Figure 3 shows differences in body weights at four developmental stages, developmental times from oviposition to four developmental stages, growth rate between hatching and adult emergence, and adult body length among the four cross types in each sex. The differences in adult body length among crosses (Fig. 3d) paralleled those in body weight at adult emergence (Fig. 3a). The developmental time from oviposition to adult emergence did not differ much among cross types (Fig. 3b), whereas growth rate (body weight increment per day; Fig. 3c) was smaller for offspring from the H × H pairs.
anova showed that both paternal and maternal genotypes (populations) had significant effects on offspring adult body length (Table 3). The paternal genotype × offspring sex interaction had a significant effect, but the maternal genotype × offspring sex interaction did not. To explore this point, autosomal, sex-linked and maternal effects on adult body length were distinguished from contrasts between specific cross types in offspring of either sex (Table 4). Autosomal effects and combined sex-linked and maternal effects were detected, and two of these were significant after controlling the false-positive rate. The lack of a maternal effect on this trait in female offspring suggested that the combined sex-linked and maternal effects in male offspring were attributable to the sex-linked (X-linked) effect.
Table 3. Three-way anova for the effects of parental genotype (population) and offspring sex on adult body length
Offspring sex × Father population
Table 4. Contrasts tested by anova to distinguish between the effects of autosomal (A), sex-linked (X) and maternal (M) effects on adult body length
Cross/offspring sex (effect)
Significant at α=0.05 after controlling for the false-positive rate by B–H method (Benjamini & Hochberg, 1995).
Adult body size in C. tosanus showed plasticity with temperature and became larger when individuals were reared at 20 °C compared to 15 °C. This result apparently contradicts a previous study with a related species Carabus (Ohomopterus) yaconinus, in which body size did not change between 15 and 20 °C under L16/D8 and slightly decreased at 25 °C with L16/D8 (Sota, 1986). A smaller adult body size with a shorter developmental period at a higher temperature is known as the ‘temperature–size rule’ (Atkinson, 1994), but many exceptions exist (e.g. Angilletta & Dunham, 2003). The plastic response of body size to temperature may show divergence, even among related insect species and among populations within a species (Kingsolver et al., 2007).
Plastic changes in body size due to differences in temperature were small compared to differences in body size between high- and low-altitude populations (subspecies). The plastic decrease in body size with increasing altitude and decreasing habitat temperature may partly explain the altitudinal variation in body size within subspecies in the field (Ikeda et al., 2012; Y. Tsuchiya, unpublished). However, the phenotypic plasticity appears to contribute little to the observed difference in body size between high- and low-altitude subspecies, most of which was attributable to genetic differences as evidenced by the crossing experiment between subspecies ishizuchianus and tosanus. Adult body length was affected by the genotypes of both parents, but a significant offspring sex × paternal genotype interaction was noted (Table 3), suggesting that body length was affected by some genetic factors on sex chromosomes given the presence of an XY sex chromosome system in this species. The analyses of contrasts between crosses suggested an X-linked effect, in addition to autosomal effects, on adult body length (Tables 4), provided that the Y chromosome had a negligible effect, as is often assumed. Thus, the difference in body length between high-and low-altitude populations (subspecies) could be governed by genetic factors on the X chromosome as well as by autosomes. The known cases of loci linked to the X chromosome affecting body size in animals have been reported (Lin et al., 2008; Kitano et al., 2009). However, because our evidence for the X-linked effect was not robust due to small sample sizes and the lack of F2 or backcrossed populations, further experiments are needed.
Body size in C. tosanus populations was positively correlated with developmental time when temperature remained constant. When habitat temperature is lower at higher altitude, smaller genetic body size requires shorter developmental time. These facts support the hypothesis that the clinal variation in body size in univoltine insects is a result of selection for shorter larval developmental periods in cooler habitats (Masaki, 1967). In the field, the high-altitude population H fully exploited the season available for reproduction and larval development, whereas the low-altitude population L used only part of the available season (Ikeda et al., 2012). This suggests that the larval developmental period is limited and that optimal body size decreases at higher altitudes. Our experimental data indicate that a 1-day reduction in the larval period will result in approximately a 1-mm decrease in adult body length at 20 °C. Thus, reductions in the larval period under cooler conditions could be associated with substantial decreases in body size. However, our crossing experiment between high- and low-altitude populations suggested that genetic difference in growth rate also existed. It is possible that growth rate per se is subject to selection and the limitation of developmental period is not be the sole selective factor for smaller body size. Smaller bodies require smaller amounts of food and can be advantageous at higher altitudes where food resources for Ohomopterus larvae (earthworms) are limited (Ikeda et al., 2012). Further studies are needed for the selective factors on body sizes in C. tosanus.
The genetic differentiation of body size in C. tosanus provides insight into ecological speciation as a result of differentiation in body size at different altitudes. It is likely that high- and low-altitude subspecies diverged relatively recently from each other (Sota & Nagata, 2008) and gradual dispersal resulted in their current distributions, but they have diverged body sizes, which can cause reproductive isolation. Our study indicated that hybridization between high-altitude and low-altitude populations (i.e. the subspecies ishizuchianus and tosanus) produced beetles of intermediate size. However, intermediate populations (or hybrid swarms) between these subspecies have not been observed in the field. This may be attributable to topographic barriers due to the steepness of slopes at altitudes around 1000 m (Ikeda et al., 2012). In addition, the difference in body size may hinder hybridization (Okuzaki et al., 2010) and may reduce the chances for gene flow between these subspecies. In other words, the segregation between subspecies at different altitudinal ranges may have been maintained by the topological gaps, and occasional contacts by a few migrants may not have cause effective gene flow due to the body size differences. In our study, an attempt at backcrossing using F1 males failed to produce viable larva, probably because the F1 males were sterile (Y. Tsuchiya, unpublished). Thus, post-zygotic reproductive isolation may have already evolved to some extent between these subspecies because of restricted gene flow caused by habitat segregation between altitudes and/or prezygotic isolation due to differences in body size. Further crossing experiments among subspecies populations are needed to examine the possibility of speciation associated with adaptation to different altitudinal habitats in C. tosanus. Lastly, given some post-zygotic isolation between the high- and low-altitude subspecies, there might be a period of allopatry between them, which facilitated their genetic divergence. Although it is hard to discriminate between truly parapatric speciation (with continuous gene flow) or para-allopatric speciation (with allopatric phases) (Coyne & Orr, 2004), the phylogeographic aspects of different subspecies need to be studied for understanding the evolutionary history of body size divergence in C. tosanus.
We thank N. Nagata for helping with the collecting of beetle, W. Blanckenhorn and an anonymous referee for their helpful comments. This research was supported by Grants-in-Aid from the Japan Society for the Promotion of Sciences (nos. 20370011, 23370011 to TS) and the Global COE Program A06 ‘Formation of a Strategic Base for Biodiversity and Evolutionary Research; from Genomics to Ecosystems’ from the Ministry of Education, Culture, Sports, Science and Technology, Japan.