OUP user menu

B chromosomes and cranial variability in yellow-necked field mice (Apodemus flavicollis)

Vida Jojić , Jelena Blagojević , Mladen Vujošević
DOI: http://dx.doi.org/10.1644/10-MAMM-A-158.1 396-406 First published online: 15 April 2011

Abstract

B chromosomes (Bs) are supernumerary chromosomes relative to the standard karyotype. The maintenance of Bs in the yellow-necked field mice (Apodemus flavicollis) was reconsidered by examining their effects on 3 components of cranial variability: canalization, developmental stability, and morphological integration. Bs do not disturb developmental homeostasis in their carriers. Moreover, Bs play a significant role in structuring cranial variation. We suggest that direct interactions between developmental pathways in mice without Bs might be a dominant mechanism for generating covariation of cranial traits, and integration of cranial traits in B carriers could be generated primarily by parallel variation of separate developmental pathways. Integration due to parallel variation is more predisposed to modifications by natural selection than integration caused by direct interactions, which could be beneficial to B carriers under variable environmental conditions. By contributing to the genetic variability of species possessing them, Bs provide themselves with long-term presence in populations. Therefore, the Bs of A. flavicollis should be considered as symbiotic genomic elements.

Key words
  • Apodemus flavicollis
  • B chromosomes
  • canalization
  • cranium
  • developmental stability
  • fluctuating asymmetry
  • geometric morphometrics
  • modularity
  • morphological integration
  • yellow-necked field mouse

The occurrence of additional chromosomes named B chromosomes (Bs) has been known for more than a century. However, genetic consequences of Bs and the way they are maintained in populations is still a conundrum for the vast majority of species harboring them. Other than dispensability, which means that they are not a necessary part of the genome, no other feature is shared by all Bs, although non-Mendelian inheritance and absence of pairing and recombination with members of standard complements (As) are used as defining features of Bs. The appearance of Bs is the most widespread chromosomal polymorphism among eukaryotes. On average, it is estimated that 15% of species possess them. Within animals Bs have been found in all major taxonomic groups except birds (Vujošević and Blagojević 2004). In mammals they are rare, and only 1.2% of species (mostly rodents) have them. However, the genus Apodemus represents a marked contrast within mammals, because 6 of 21 species, including the yellow-necked field mouse (Apodemus flavicollis (Melchior, 1834)) feature the presence of Bs. Bs are found in almost all populations of A. flavicollis in a wide range of frequencies, from 11.0% in Slovenia (Vujošević et al. 1991) to 93.5% in northern Bohemia (Zima and Macholán 1995).

Many studies have revealed that the frequency of Bs in populations remains stable over years, and several models describing mechanisms for their maintenance in populations have been proposed; however, 2 models dominate. The parasitic model argues that the stable frequency of Bs in populations is achieved by accumulation mechanisms that are facilitated by drive and elimination due to their detrimental effects (Jones 1991; Östergren 1945). Alternatively, the heterotic model proposed by White (1973) assumes that in the absence of drive the equilibrium frequency of Bs could be maintained by the benefit that Bs bring to carriers with low numbers, whereas high numbers of Bs could be harmful. In the last 2 decades the prevailing opinion has been that Bs are genomic parasites. The equilibrium frequencies of B carriers in populations of A. flavicollis have been reported for successive years (Vujošević 1992; Zima and Macholán 1995). Investigations carried out in the last 20 years have revealed the absence of accumulation mechanisms (Blagojević et al. 2009; Vujošević et al. 1989, 2007, 2009) and the possible adaptive significance of Bs (Blagojević and Vujošević 1995, 2004; Jojić et al. 2007; Vujošević and Blagojević 2000; Zima et al. 2003), indicating that the long-term presence of Bs in populations of A. flavicollis could not be explained by the parasitic model but rather with a model closer to the heterotic one.

Emphasizing the parasitic nature of Bs, Camacho (2005) pointed out that their effects are rarely manifested overtly in the external phenotype and that Bs present in low numbers do not have a very observable impact on the host phenotype. However, hundreds of genes are expressed in the development of any complex organ, making contributions to traits that are important in aggregate but are too small to be detected individually (Weiss 2008). This could be one reason why the effects of Bs and their genetic activity in mammals have been scored in only a handful of species. Furthermore, the simplicity of their definition hides the complexity of a wide range of B systems (Jones and Houben 2003). Therefore, it is reasonable to expect that their maintenance is not the same in different species.

Although Bs are often heavily heterochromatic and genetically inert, this is not the case with Bs of A. flavicollis (Tanić et al. 2000, 2005; Vujošević and Živković 1987). Also, the effects of Bs in A. flavicollis were detected on different cranial and mandibular traits (Blagojević and Vujošević 2000, 2004; Jojić et al. 2007). Thus, based on the prediction that Bs could carry genes and transcriptional factors that affect cranial variability, we reconsidered maintenance of Bs in the yellow-necked field mouse by examining the effects of Bs on canalization, developmental stability (DS), and morphological integration. Canalization and DS are 2 elements of developmental homeostasis, which is defined as the mechanism responsible for ensuring phenotypic constancy in organisms despite the great variability of genetic and environmental features (Zakharov 1989, 1992). Canalization refers to the buffering of developmental processes against environmental and mutational perturbations (Waddington 1942, 1957; Wagner et al. 1997), and DS is the ability of developmental processes to buffer random noise that arises within the developmental processes themselves (Waddington 1957). Both canalization and DS suppress phenotypic variation. Morphological integration, as the 3rd main component of phenotypic variability (Hallgrímsson et al. 2002; Willmore et al. 2007), refers to how variability is structured by the underlying developmental, genetic, and functional connections among traits (Cheverud 1996; Olson and Miller 1958; Wagner 1996). It biases the direction of variation rather than acting to reduce it. Closely related to the concept of morphological integration is modularity. Modules are sets of traits whose parts are integrated internally by numerous and strong interactions but are independent from other modules due to relatively few or weak interactions connecting different modules (Raff 1996; Wagner 1996; Willmore et al. 2007). Because morphological traits can be integrated through developmental, genetic, and functional relationships, modularity occurs in developmental, genetic, and functional contexts. Both genetic and functional modularity contribute to evolutionary modularity (Klingenberg 2008).

Canalization could be measured by estimating the magnitude of variance among individuals. Corresponding traits in different individuals develop under different environmental and genetic conditions, and therefore the differences in the amount of among-individual variation for a trait will indicate differences in ability to canalize development against genetic and environmental stresses (Willmore et al. 2007). DS is usually measured by estimating the magnitude of the variance within individuals. Bilaterally symmetric traits develop under the same genetic and environmental conditions, and therefore the deviations from perfect symmetry are caused by developmental disruptions arising within an individual. This is the basic premise for measuring DS by fluctuating asymmetry (FA), which refers to the minor, random differences between the 2 sides in bilaterally symmetric traits (Palmer and Strobeck 1986; Van Valen 1962). Reduced among-individual variance and FA signify a higher level of canalization and DS. Because ancestral states of canalization and DS for a structure are unknown, they can be determined only by comparison with some reference state (Rutherford 2000; Willmore et al. 2007).

Beyond conventional use of the magnitude of FA as a measure of DS, FA patterns have been used recently to examine developmental relationships underlying morphological integration (Klingenberg 2002, 2003, 2004, 2008; Klingenberg and Zaklan 2000). Coordinated variation among morphological traits can have 2 developmental origins: direct developmental interactions, for instance between traits that derive from a common developmental precursor (Riska 1986) or interaction by inductive signaling (Jacobson and Sater 1988), and parallel variation in separate pathways that are subject to the same environmental or genetic variation. Because the left and right sides of bilaterally symmetric structures share the same genome and experience very similar environmental conditions, the study of FA provides almost complete control over external sources of variation (such as perturbations originating from the environment or allelic variation at the population level) that might produce parallel variation of separate pathways. Furthermore, because FA originates from random perturbations that occur spontaneously within the developmental pathways themselves, it can be transferred only among pathways by direct interaction. Thus, the patterns of covariation among traits for signed FA must result from direct interactions of developmental pathways and not from parallel variation of separate pathways, whereas the patterns of covariation among individuals can be of either developmental origin. By comparing the patterns of covariance in asymmetry with the patterns of covariance among individuals, it is possible to assess the relative importance of 2 mechanisms underlying trait integration.

Factors that can increase the level of developmental stress and disturb developmental homeostasis can be of environmental or genetic origin (Møller and Swaddle 1997). In agreement with the currently prevailing view that Bs are genomic parasites, Bs would increase the level of stress and produce detrimental effects in their hosts, as most parasites do. Consequently, B carriers would possess lower levels of both canalization and DS compared to noncarriers. In addition to using the magnitude of left–right asymmetry (FA) to test the prediction that developmental homeostasis is disturbed in B carriers compared to noncarriers, we used covariances of signed FA to infer the developmental origins of cranial trait integration within both noncarriers and B carriers, which was the secondary aim of our study. Finally, we tested the hypothesis that the cranium of noncarriers and B carriers is organized into 2 developmental modules, “basicranium” and “face.”

Materials and Methods

Data.—Yellow-necked field mice (A. flavicollis) were collected on Mount Jastrebac (Serbia) from March to October during 1989. Animals were captured and treated following procedures approved by the American Society of Mammalogists (Gannon et al. 2007). Chromosome preparations were made directly from bone marrow using a standard procedure (Hsu and Patton 1969). The presence and number of Bs were determined by analyzing 30 metaphase figures per specimen. Skulls were cleaned by exposure to dermestid beetles. In this study only adult animals with complete eruption of the 3rd upper molar (M3) were included. The sample comprised 146 specimens among which 86 animals were without Bs (referred to as B0 mice, 46 females and 40 males), and 60 were animals with Bs (referred to as B+ mice, 33 females and 27 males).

Digital images (2,272 × 1,520 pixels resolution) of the ventral surface of crania, supported by modeling clay with the palate positioned parallel to the photographic plane, were taken using a Nikon COOLPIX4500 digital camera (Refot B, Belgrade, Serbia). Thirty-three (14 paired and 5 median) 2-dimensional landmarks were digitized (Table 1; Fig. 1) using tpsDig software (Rohlf 2008) by the same observer. Most landmarks were of type I or type II according to Bookstein (1991). To assess measurement error for shape we used Procrustes analysis of variance (ANOVA—Klingenberg et al. 2002; Klingenberg and McIntyre 1998) for a set of 30 mice (∼20% of the total). Two images of each cranium were taken, and each image was digitized twice. In Procrustes ANOVA of morphological structures with object symmetry, the main effect of individuals (random factor) stands for individual variation in shape, the main effect of reflections (fixed factor) accounts for the directional asymmetry, and the individuals × reflections interaction estimates the measure of FA. The residual variance component among replicates quantifies measurement error. The mean squares for FA and individual variation exceeded the error components (imaging and digitizing) by more than 3.5-fold, and all the subsequent analyses of FA were based on a single image per cranium.

Fig. 1

Landmarks recorded on the ventral surface of the cranium of the yellow-necked field mouse (Apodemus flavicollis). See Table 1 for landmark definitions.

View this table:
Table 1

Definitions of landmarks digitized on the ventral surface of the cranium of the yellow-necked field mouse (Apodemus flavicollis).

1Anterior extremity of the suture between the nasals
2, 20Lateral margin of incisive alveolus where it intersects outline of the skull in photographic plane
3, 21Anteriormost point of incisive foramen
4Suture between the vomerine portion of the premaxillary and maxillary in the incisive foramen
5, 22Anterior extremity of the zygomatic plate
6, 23Posteriormost point of incisive foramen
7, 24Intersection between the anterior end of the premolar and maxillary
8, 25Anterior border of the posterior palatine foramen
9, 26Intersection between the posterior end of the 3rd upper molar (M3) and maxillary
10Back of palatine
11, 27Lateralmost point in the suture between the presphenoid and basisphenoid
12, 28Anterior region of the squamosal zygomatic process where it joins the zygomatic arch
13, 29Anteriormost point of foramen ovale
14Midpoint of basisphenoid–basioccipital suture
15, 30Point where the suture between the basisphenoid and basioccipital contacts the tympanic bulla
16, 31Anterior tip of the external auditory meatus
17, 32Posterior tip of the external auditory meatus
18Anteriormost point of the foramen magnum
19, 33Lateral tip of the occipital condyle

Preliminary analyses.—Cranial variability was examined by using landmark-based geometric morphometric techniques. The landmark coordinates of original and mirrored configurations of both replicates were aligned simultaneously using a generalized Procrustes analysis to extract centroid size (CS) and superimposed Procrustes coordinates. CS is the size measure for landmark configurations. In generalized Procrustes analysis all the landmark configurations were scaled to unit CS, translated so that they have a common center of gravity, and rotated to minimize the sum of squared distances between landmarks of each configuration to the corresponding landmarks of a consensus configuration (Rohlf and Slice 1990). The variation remaining in the Procrustes coordinates after reflection and superimposition represents complete shape variation (Dryden and Mardia 1998). However, for the studies of shape with bilateral symmetry the total shape variation is partitioned into components of symmetric variation among individuals and asymmetric variation within individuals. In the analyses of the shapes of morphological structures with object symmetry, such as vertebrate skulls, the Procrustes consensus for all original and reflected configurations, and the average of the original and reflected configurations of each specimen, are perfectly symmetric shapes. The variation among individuals in these averages of original and reflected configurations constitutes the symmetric component of shape variation, and the variation within individuals in the landmark deviations of the original configuration from the average of the original and reflected configuration constitutes the asymmetric component of shape variation (Klingenberg et al. 2002). Superimposed Procrustes coordinates were subjected to Procrustes ANOVA (Klingenberg et al. 2002; Klingenberg and McIntyre 1998) with additional main effects of sex and presence of Bs (0, 1, and >1 Bs).

To test for allometry we used multivariate regressions of both symmetric (individual variation) and asymmetric (FA) components of shape variation onto log CS. Allometry tests were based on regression by a permutation test with 10,000 iterations, under the null hypothesis of independence between size and shape, by randomly exchanging the value for log CS among individuals (Good 1994). Reflection, superimposition, Procrustes ANOVA, and regressions were carried out using the procedures incorporated in MorphoJ (Klingenberg 2011).

Analyses of canalization and DS.—The total sample was partitioned according to sex and presence of Bs into 4 groups: B0 males, B0 females, B+ males, and B+ females. To eliminate the influence of allometry all specimens were standardized to the mean CS for each group (Zelditch et al. 2004b) using the Standard6b program (Sheets 2001). To partition overall phenotypic variation into symmetric (individual variation), asymmetric (FA), and measurement error components, Procrustes ANOVAs (Klingenberg et al. 2002; Klingenberg and McIntyre 1998) were performed on these standardized data.

The level of canalization was estimated by among-individual size and shape variance (Willmore et al. 2006a, 2006b; Zelditch et al. 2004a). Variance in cranium size was calculated as the variance of log-transformed CS (Varlog CS). Variance in cranium shape (Varshape) was calculated for Procrustes distances using the standard metric for variance. Procrustes distances were computed from the symmetric component of shape variation. To estimate the significance of the difference in size and shape variance among groups Levene's tests were performed on absolute size differences (log CSj − log CSmean) and Procrustes distances, respectively.

The level of DS was estimated by the level of shape FA. Parametric F-tests in Procrustes ANOVAs were used to determine whether variation among individuals, FA, and directional asymmetry were significant within each of the 4 groups. To check for potential antisymmetry the signed asymmetries (asymmetric component of shape variation) were inspected for signs of bimodality using the Kolmogorov–Smirnov test of normality. The level of FA was estimated by the FA10a index of Palmer and Strobeck (1986). To estimate the significance of differences in cranial shape FA among groups Levene's test was performed on shape FA scores. Shape FA scores (in units of Procrustes distance) were calculated from the asymmetric component of shape variation. Analyses of canalization and DS were conducted in MorphoJ (Klingenberg 2011) and Sage (Marquez 2007).

Analyses of morphological integration.—Overall cranial integration and covariation between its specific parts (modules) were analyzed separately within B0 and B+ mice. To eliminate the influence of allometry residuals from multivariate regressions of both symmetric (individual variation) and asymmetric (FA) components of shape variation regressions onto log CS were used. Additionally, we corrected for the effects of sex (in B0 mice); that is, sex and 1 or >1 B (in B+ mice) by using the pooled within-group covariance matrices in all subsequent analyses (Klingenberg 2009; Klingenberg et al. 2003; Mitteroecker and Bookstein 2008).

Patterns of variation among individuals (symmetric) and FA (asymmetric) components of shape variation were compared by computing matrix correlations between the respective covariance matrices both within and between B0 and B+ mice. Analyses on matrix correlation between these covariance matrices included the diagonal entries of the matrices. The significance of these correlations was determined using the matrix permutation test with 10,000 permutation rounds against the null hypothesis of complete dissimilarity between the respective covariance matrices (Cheverud et al. 1989) by permuting landmarks. Only the paired landmarks were used for comparison of symmetric and asymmetric components. Principal component analyses were performed to visualize covariance structures of individual (symmetric) and FA (asymmetric) components of shape variation. Thin plate spline deformation grids along each principal component axis visualize shape differences between the average landmark configuration and configuration along the respective principal component axis.

The a priori hypothesis of a 2-module (basicranium–face) organization of the cranium for individual and FA components of shape variation was evaluated according to the procedures described by Klingenberg (2009). As a measure of the strength of the association between hypothesized modules, the RV coefficient of Escoufier (1973) was calculated and compared with RV coefficients obtained for alternative partitions of the configuration into subsets of landmarks. These subsets contained the same numbers of landmarks as the hypothesized modules, and alternative partitions were spatially contiguous. Spatial contiguity was obtained using an adjacency graph. If the calculated value of the RV coefficient between hypothesized modules was the lowest value or it fell in the lower tail of the distribution of RV coefficients observed for alternative partitions, the predicted hypothesis of modularity was confirmed. Analyses of morphological integration were conducted using MorphoJ software (Klingenberg 2011).

Results

Chromosome analyses.—Within B+ mice, 80% of individuals had 1 B. The remaining 20% had > 1 B (7 had 2 Bs, 3 had 3 Bs, 1 had 4 Bs, and 1 had 5 Bs).

Preliminary analyses.—No significant differences due to sex (F1,146 = 0.29, P = 0.5934) and presence of Bs (F2,146 = 0.55, P = 0.5791) were found for cranial size. However, for cranial shape significant differences caused by sex (F31,9052 = 2.66, P < 0.0001) and no significant differences due to presence of Bs (F62,9052 = 1.09, P = 0.2906) were found.

Multivariate regression of individual variation of shape on log CS was statistically highly significant (P < 0.0001), and allometry explained 21.0% of the symmetric component of shape variation. Regression of the signed asymmetries on log CS also was statistically significant (P < 0.0001), but only 1.5% of the asymmetric component of shape variation was size dependent.

Canalization and DS.—As revealed by Procrustes ANOVAs of shape (Table 2), the main effects of individuals (variation among individuals) and reflections (directional asymmetry) and the individuals × reflections interaction (FA) were all highly significant at P < 0.0001 in all groups except for directional asymmetry in B+ females, which was significant at P = 0.0085. When analyzing FA it is important to distinguish FA from directional asymmetry and antisymmetry. Directional asymmetry accounted for a fairly small percentage of the total variation in all analyzed groups (Table 2), and Kolmogorov–Smirnov tests disclosed that none out of the 66 signed asymmetry distributions showed significant departure from normality. Thus, these asymmetries could be assigned to FAs.

View this table:
Table 2

Procrustes ANOVAs of cranial shape for Apodemus flavicollis. % total = percentage of the total variation. B0 = mice without B chromosomes, B+ = mice with B chromosomes.

GroupEffectd.f.FP% total
B0 males (n = 40)Individual1,209, 2,4806.42<0.000181.03
Reflection31, 2,4802.60<0.00010.84
Individual × reflection1,209, 2,4804.71<0.000112.63
BO females (n = 46)Individual1,395, 2,8527.34<0.000182.83
Reflection31, 2,8522.79<0.00010.70
Individual × reflection1,395, 2,8524.46<0.000111.29
B+ males (n = 27)Individual806, 1,6748.38<0.000183.07
Reflection31, 1,6742.62<0.00011.00
Individual × reflection806, 1,6743.43<0.00019.92
B+ females (n = 33)Individual992, 2,0467.76<0.000182.88
Reflection31, 2,0461.730.00850.58
Individual × reflection992, 2,0463.75<0.000110.68

Analysis of the level of canalization indicated higher values for cranium size variances in B+ males compared to the other 3 groups (Fig. 2a). The lowest level of among-individual variance in cranium size was found in B+ females. B+ males have a significantly higher level of among-individual variance in cranium size (F3,142 = 2.97, P = 0.0339). Observed differences in variance in cranium shape among the groups (Fig. 2b) were not statistically significant (F3,142 = 1.87, P = 0.1366).

Fig. 2

The level of canalization and developmental stability of Apodemus flavicollis without (B0) and with (B+) B chromosomes. a) Among-individual variance of log-transformed centroid size (Varlog CS). b) Among-individual shape variance (Varshape). c) Level of shape fluctuating asymmetry (FA10a).

Although analysis of the level of DS disclosed that B0 males had a somewhat higher level of FA compared to the other 3 groups (Fig. 2c), we found no significant differences in the levels of FA (F3,142 = 1.44, P = 0.2327).

Morphological integration.—The covariance patterns of individual variation and FA were congruent only in B0 mice (r = 0.610, n = 86, P = 0.0122), but B+ mice did not exhibit any similarities between these patterns (r = 0.574, n = 60, P = 0.1594). However, covariance patterns of individual variation and FA were both significantly correlated between B0 and B+ mice (Table 3). Principal component analyses also showed agreement between patterns of FA between B0 and B+ mice (except between 3rd principal components), and patterns of individual variation between B0 and B+ mice were consistent between 2nd principal components (Fig. 3).

Fig. 3

Principal components (PCs) of individual (symmetric) and fluctuating asymmetric components of shape variation in yellow-necked field mice (Apodemus flavicollis) without (B0) and with (B+) B chromosomes. The percentages indicate the proportions of variation for which the respective principal components account.

View this table:
Table 3

Matrix correlations and results of the matrix permutation tests for Apodemus flavicollis. Ind = Individual variation, FA = fluctuating asymmetry, B0 = mice without B chromosomes (n = 86), B+ = mice with B chromosomes (n = 60).

ComparisonCorrelationP-value
Ind versus FA within B00.6100.0122
Ind versus FA within B+0.5740.1594
Ind versus Ind between BO and B+0.8500.0000
FA versus FA between BO and B+0.8330.0000

The a priori hypothesis of 2-module organization of the cranium was confirmed for individual variation in both B0 and B+ mice, where the RV coefficient between basicranium and face fell in the lower tail of distribution of RV coefficients observed for the alternative partitions (Fig. 4). In B0 mice only 296 of 40,018 alternative partitions had RV coefficients that were lower than that observed for the partition into hypothesized modules. Although subdivision of the cranium into basicranium and face was confirmed in B+ mice, the proportion of partitions with a lower RV coefficient was substantially higher (1,963 of the 40,018 alternative partitions). For FA the hypothesized modularity of the cranium was rejected in both B0 and B+ mice. Of 40,018 alternative partitions, 8,086 in B0 and 6,043 in B+ mice produced a lower RV coefficient than that observed for the hypothesized partition into basicranium and face. The RV coefficients for all possible partitions of the cranium generally vary within relatively narrow ranges (about 0.2–0.4), indicating fairly similar RV coefficients for all partitions.

Fig. 4

Evaluation of an a priori hypothesis of a 2-module (basicranium = open circles and face = filled circles) organization of the cranium of yellow-necked field mice (Apodemus flavicollis) without (BO) and with (B+) B chromosomes. Histograms of the RV coefficients for those partitions for which the subsets of landmarks are spatially contiguous. The values of RV coefficients (Escoufier 1973) observed for the partition into basicranium and face and proportions (P) of partitions with an RV coefficient lower than that observed for the subdivision into basicranium and face are indicated by arrows.

Discussion

This study examined the effects of supernumerary Bs on 3 components of cranial variability (canalization, DS, and morphological integration) in the yellow-necked field mouse. Overall, 2 important points can be gleaned from the results. Bs do not disturb developmental homeostasis in their carriers, and they play a significant role in structuring cranial variation.

Most Bs are heavily heterochromatic and therefore considered to be genetically inert. However, this is not necessarily so (Jones 1995), because the ability of Bs to cause a variety of phenotypic effects is well documented (Jones and Rees 1982). Moreover, in recent years it has been demonstrated that some Bs show transcriptional activity (Brockhouse et al. 1989; Green 1988, 1990; Jones 1995; Leach et al. 2005). Green (1990) gave a list of 27 species of insects, amphibians, and plants that carry functional genes on Bs, and this list is far from complete. Within mammals, Graphodatsky et al. (2005) identified the proto-oncogene C-KIT on Bs of 2 canid species, and the suggestion that Bs of A. flavicollis are euchromatic (Vujošević and Živković 1987) was confirmed by application of AP-PCR–based DNA profiling (Tanić et al. 2000), which revealed molecular markers specific for Bs. Furthermore, after analyzing and comparing gene expression in A. flavicollis with and without Bs, Tanić et al. (2005) found that the expression of 3 important genes was elevated in animals possessing Bs. Willmore et al. (2007) suggested that, although loss-of-function typically uncovers more detectable differences among phenotypes, more subtle differences in gene expression also will affect the phenotype in an equally complex manner.

We observed the highest level of among-individual variance in cranium size in B+ males, and the lowest level was found in B+ females. Variability could be increased if disruption of gene expression occurs for a gene whose products largely control the expression of several other genes in major developmental pathways (Willmore et al. 2007). A general view exists that variability can arise from intrinsic properties of genetic architecture, such as number and size of gene effects, dominance interactions within loci, epistatic interactions among loci, and pleiotropic patterns (Hallgrímsson et al. 2006; Kenney-Hunt et al. 2008). Estimating the additive and dominance effects of quantitative trait loci affecting skeletal characters of the cranial vault and face in house mice (Mus musculus), Leamy et al. (1999) identified 26 quantitative trait loci on 17 of the 19 autosomes that significantly affected these characters. Some quantitative trait loci are sex-limited with effects restricted to a single sex, but it is rare that a single quantitative trait locus has opposite effects in the 2 sexes (Cheverud 2006; Ehrich et al. 2003). Additionally, a special form of epistasis involves interaction of the Y chromosome, or other sex-determining chromosome, with loci on the autosomes (Cheverud 2006). Considering that Bs originate and evolve in a very similar way to sex chromosomes, it is possible that epistatic interactions between Bs and autosomes exist.

Our analyses of FA levels in cranium shape revealed that Bs do not decrease the level of DS in their carriers, indicating that the parasitic model of maintenance of Bs does not operate in A. flavicollis. According to Leamy and Klingenberg (2005) FA levels in various characters are influenced by dominance and especially by epistatic interactions among a number of genes affecting these or other characters. Additionally, it is likely that epistatic interactions affecting FA levels are character-specific (Leamy et al. 2005) and that the choice of characters in FA analyses is critical and could account for many ambiguous results in FA studies. However, evaluating the effects of Bs on nonmetric cranial traits in the same population of A. flavicollis, Blagojević and Vujošević (2004) also found that the presence of Bs does not disturb developmental homeostasis in B+ mice. The parasitic model of maintenance of Bs rests on the existence of the drive of Bs and their harmful effects on the fitness of B carriers. However, our previous results (Blagojević et al. 2009; Vujošević et al. 1989, 2007, 2009) and examination of data on controlled crosses (M. Vujošević, pers. obs.) indicate the absence of Bs accumulation mechanisms in A. flavicollis. As predicted by the parasitic model of Bs maintenance, Bs would reduce mean fitness of their carriers, and we expect that the frequency of animals with Bs would decrease with age. Contrary to this prediction, Vujošević et al. (2009) found no statistically significant differences in the frequency of Bs among 6 age categories, demonstrating that Bs do not exert negative effects on survival of their carriers. A number of studies have reported the highest frequencies of Bs in optimal habitats (Jones and Rees 1982). In localities with more favorable habitats any deleterious nature of Bs could be tolerated more easily. However, Vujošević et al. (2007) observed that the frequency of yellow-necked mice with Bs was correlated negatively with habitat quality. Additionally, the frequency of animals with Bs was correlated positively with extreme climatic conditions, with the highest proportion of 0.63 at the highest altitude (Vujošević and Blagojević 2000), and Boyeskorov et al. (1994) found the highest proportion of A. flavicollis with Bs (0.81) in a peripheral area of species distribution with unfavorable conditions. Uterine inspection showed no significant difference in the mean number of scars and embryos between females with and without Bs, implying that fecundity is not decreased by the presence of Bs (Blagojević et al. 2006). Thus, no harmful effects of Bs were detected on either viability or fertility. Moreover, possible adaptive significance of Bs and selective effects during fluctuations in population density have been suggested repeatedly for A. flavicollis (Blagojević and Vujošević 1995; Zima et al. 2003). Failure to uncover the adaptive significance of Bs in natural populations, with a few exceptions, could be the main reason why the parasitic model is still widely accepted (Blagojević et al. 2009; Vujošević et al. 2007, 2009).

Similarity or dissimilarity in the patterns of individual variation and FA can be used as evidence of whether the same developmental mechanisms are responsible for canalization and DS. The correspondence between individual and FA patterns of cranial trait variation found in B0 mice supports the hypothesis that canalization and DS are governed by the same underlying developmental processes. However, the discrepancy between these patterns found in B+ mice leads to a contradictory conclusion; that is, canalization and DS are determined by distinct developmental processes. Some earlier studies of mammalian skulls confirmed the association between phenotypic variance and FA. Good (Leamy 1993) and partial (Klingenberg et al. 2003) agreement between the patterns of covariation for FA and individual variation were revealed for the mouse mandible. Low but significant correlations between the signed asymmetry (FA) and the phenotypic matrices were observed for cranial traits in macaques and 2 strains of mice (Hallgrímsson et al. 2004), and Willmore et al. (2005) found partial correspondence between FA and phenotypic variation in rhesus macaque (Macaca mulatto) crania. Based on the analyses of primate limbs, Hallgrímsson et al. (2002) concluded that the mechanisms behind canalization and DS are closely related. However, after examining the relationship between FA and phenotypic variance of Indian human population samples, Reddy (1999) failed to support the hypothesis that FA reflects canalization. Contrary to the results obtained by Hallgrímsson et al. (2004), Debat et al. (2000) detected a difference between patterns of individual variation and FA in the dorsal crania of the house mouse. These contrasting results were ascribed by Hallgrímsson et al. (2004) to differences in methodology (use of 2- or 3-dimensional data, and use of interlandmark distances or Procrustes superimposed landmarks).

Correspondence or discrepancy between the patterns of FA and individual variation can indicate the relative importance of 2 developmental origins of morphological integration (Klingenberg 2002, 2003, 2004, 2008; Klingenberg and Zaklan 2000). Only direct interactions between the developmental pathways could generate the patterns of morphometric covariation observed within individuals (FA), but the patterns of covariance among individuals also include a contribution from parallel variation of separate pathways. Similarity in these patterns means that direct interactions of developmental pathways might be the primary mechanism that produces trait covariation among individuals as well, and dissimilarity suggests that parallel variation of separate pathways is the dominant mechanism underlying trait covariation among individuals. Therefore, the discrepancy between individual and FA patterns found in B+ mice indicates that Bs could contribute to morphological integration of cranial traits primarily by parallel variation of separate developmental pathways. Parallel variation of separate developmental pathways can originate from allelic variation in genes that participate in several different developmental processes. Hundreds of genes are known to affect craniofacial development (Weiss 2008). Tanić et al. (2005) showed that the activity of some genes in B+ mice can be associated directly or indirectly with Bs. Thus, it is possible that allelic variation in some of the genes that participate in different processes involved in cranium development in B+ mice could affect multiple developmental processes simultaneously, causing parallel variation of separate developmental pathways. In contrast, in B0 mice congruence between individual and FA patterns of variation indicates that direct interactions between developmental pathways could be a dominant mechanism for generating covariation of the cranial traits.

Matrix correlations and principal component analyses revealed similar patterns of covariation in cranial traits for both individual and FA components of shape variation between B0 and B+ mice. Investigating how Bs might influence morphological integration of the mandible in A. flavicollis, Jojić et al. (2007) also found that the pattern of correlation of mandibular traits was consistent between B0 and B+ mice. Debat et al. (2000) observed concordance of both interindividual and intraindividual skull variation among 3 samples of house mice (M. m. domesticus, M. m. musculus, and their F1 hybrids). Patterns of covariation among structures tend to be relatively stable among populations within species (González-José et al. 2004), among closely related species (Marroig and Cheverud 2001), and even between distinct species, such as mice and macaques (Hallgrímsson et al. 2004).

Our analysis of cranial modularity confirmed the hypothesis of 2-module (basicranium–face) organization of the cranium in both B0 and B+ mice for the variation among individuals, but the hypothesis of cranium modularity was not confirmed for FA. However, the confirmed hypothesis of modularity in B+ mice should be taken with caution because of the substantially higher proportion of partitions with a lower RV coefficient in comparison to the partition that corresponds to the hypothesized modules. The RV coefficients for alternative partitions varied within fairly narrow ranges. Analyses of whether the anterior and posterior parts of the mouse mandible are distinct modules revealed similar squared trace correlations (Klingenberg et al. 2003) and similar RV coefficients (Klingenberg 2009) for all alternative partitions. The limited range of RV coefficients indicates that modularity is relatively weak (Drake and Klingenberg 2010). The cranium is a highly integrated structure characterized by complex covariation both within and between regions (Cheverud 1982, 1989, 1995). Thus, as is the case with the mandible (Klingenberg 2009; Klingenberg et al. 2003), modularity of the cranium is also a question of degrees rather than an all-or-nothing phenomenon.

Each of 3 components of phenotypic variability and their interactions with each other modulate the expression of genetic variation into phenotypic variation (Willmore et al. 2006a) and can have important effects on the rate and direction of evolutionary change (Hallgrímsson et al. 2002). Dealing with the interrelationships among canalization, DS, and morphological integration, this study yielded new insights into maintenance of Bs in A. flavicollis. We found that the presence of Bs does not disturb developmental homeostasis in their carriers and concluded that the parasitic model of maintenance of Bs does not operate in this species. Moreover, Bs in A. flavicollis play a significant role in structuring cranial variation. Numerous processes that generate covariance in the mammalian skull can be regulated through many developmental pathways, each of which in turn is influenced by multiple genes (Hallgrímsson et al. 2007). Additionally, developmental processes that are involved in the spatial patterning of cranial structures can interact in different ways with genetic and environmental factors (Leamy et al. 1999). We propose that Bs could contribute to morphological integration of cranial traits primarily by parallel variation of separate developmental pathways, whereas direct interactions between developmental pathways might be the dominant mechanism for generating covariation of the respective traits in B0 mice. According to Amundson (2005), genetic and developmental systems would evolve just as much as the morphological traits they generate. Similarly, Weiss (2008) emphasized that complex traits are produced by networks of genetic pathways that provide many alternative potential routes to a given end and whose use can vary among or even within species. Integration due to parallel variation is more prone to modifications by natural selection than integration caused by direct interactions (Klingenberg 2004). This property of integration by parallel variation could be beneficial to B+ mice under variable environmental conditions with an extreme climate and frequent variation in population density. Blagojević et al. (2009) pointed to the existence of adaptive effects of Bs at the level of populations, emphasizing that Bs affect adaptability rather than produce adaptations. It was proposed that the presence of B s in A. flavicollis widens genetic variability in this species, enabling it to occupy more habitats successfully and to extend its specific response to different selective pressures (Blagojević et al. 2009; Jojić et al. 2007; Vujošević and Blagojević 2004; Vujošević et al. 2007, 2009; Wójcik et al. 2004). By contributing to the genetic variability of species possessing them, Bs provide themselves with long-term presence in populations. Therefore, the Bs of A. flavicollis should be considered as symbiotic genomic elements.

Acknowledgment

This work was supported by the Ministry of Science and Technological Development of the Republic of Serbia (grant 173003).

Footnotes

  • Associate Editor was Elizabeth R. Dumont.

Literature Cited

View Abstract