We analyzed the karyotype, mitochondrial and nuclear genetic markers, and the morphology of the Wood White butterfly L. sinapis
. This is a common species widely distributed from Portugal and Spain in the west to Siberia in the east [34
]. From this territory different chromosome numbers have been reported in literature ranging from n = 28 to n = 41 [35
]. However, these results are impossible to interpret in practice because of the discovery in 1993 of a cryptic sympatric species (L. reali
) in Europe and Asia [36
]. As all karyotype data for L. sinapis
were published before this date, it is unclear whether reported chromosome numbers reflect inter- or intraspecific variability.
Our study covers populations from different parts of the L. sinapis
distribution (Figures , ), as well as the closely related species L. reali
and L. morsei
as comparison. We discovered that diploid chromosome number ranges in L. sinapis
from 2n = 106 in Spain to 2n = 56 in eastern Kazakhstan in a longitudinal cline (Figure ; for more details, see Additional file 1
). These findings are based on the examination of 209 male specimens, with metaphase plates observed in 35 individuals, out of which 23 had unambiguous chromosome number counts (Spain - 4, France - 2, Italy - 2, Romania - 8, Kazakhstan - 7). We also found that chromosome numbers are not stable within some populations from Italy, Romania and Kazakhstan. Specimens with different chromosome numbers were found within each of these populations, and the great majority of the individuals were chromosomal heterozygotes displaying from one to six multivalents in metaphase I of meiosis (Additional file 1
, Figure S1). In the heterozygotes, we observed no abnormalities in the anaphase I stage of meiosis, and the first division of meiosis resulted in normal haploid metaphase II cells where, as expected, two types of metaphase plates with different chromosome numbers were observed. Therefore we conclude that chromosomal rearrangements are not fixed in several of the populations studied, and there seems to be no strong selection against chromosomal heterozygotes. Interestingly, chromosome number range overlaps between some studied populations separated by hundreds of kilometers, e.g. in Kazakhstan between the population from Landman (2n = 56-61) and the population from Saur (2n = 56-64).
Figure 1 Chromosomal cline in Leptidea sinapis across the Palaearctic region. a. Sampling sites and karyotype results. Metaphase plates were observed in 35 individuals, out of which 23 had unambiguous chromosome number counts: Spain - 4, France - 2, Italy - 2, (more ...)
Figure 2 Male genitalia morphology of Leptidea sinapis reveals no significant intraspecific differences. One-way ANOVA for a. phallus length/vinculum width and b. saccus length/vinculum width. The sibling species L. reali is included as positive control. Only (more ...)
In certain species, variation in chromosome number may be caused by the presence of so-called B-chromosomes (=additional chromosomes, =supernumerary chromosomes) [37
]. B-chromosomes consist mainly of repetitive DNA and can be usually found in low numbers (one to five) in a percentage of the individuals of a given population. Although they are dispensable, they can sometimes accumulate through processes of mitotic or meiotic drive [38
]. B-chromosomes can be distinguished from normal A-chromosomes because they are usually smaller and can be seen as additional chromosomes present in only some of the individuals in a population. The best diagnostic feature is their identity at meiosis, where they may be found as univalents, or in various pairing configurations (bivalents or multivalents), but never pairing with A-chromosomes. Thus, meiotic analysis is critical to distinguish between B-chromosomes and normal A-chromosomes [37
]. Although we cannot totally exclude that B-chromosomes can be found in L. sinapis
, especially taking into account that they are known in other genera of the family Pieridae [39
], there is good evidence that B-chromosomes are not a valid explanation for the chromosome number cline found in L. sinapis
. This is due to the fact that in the Spanish population, where the number of chromosomes is maximal (and correspondingly where the highest number of B-chromosomes would be expected), they seem to be completely absent: the chromosome number is stable within as well as between individuals, and no univalents have been observed during meiosis. Moreover, no univalents have been observed during meiosis in any of the other populations studied. Additionally, the following clear pattern was observed: the higher the chromosome numbers in a population, the smaller the size of chromosomes, and vice versa (Figure ; Additional file 1
, Figure S1). This regularity indicates that chromosomal fusions/fissions (but not B-chromosomes) were the main mechanism of karyotype evolution.
can be distinguished from its closest relative L. reali
by the length of the phallus, saccus and vinculum (in male genitalia) or of the ductus bursae (in female genitalia) [36
] as well as by molecular markers [41
]. Therefore, to exclude the possibility of cryptic species involved in the formation of the extraordinarily high chromosomal variability and to demonstrate the conspecificity of the populations studied, we performed morphological and molecular analysis of each studied individual.
The measured variables of the male genitalia showed no significant difference or apparent trend between chromosomal races according to one-way ANOVA (Figure ) and to discriminant analysis (DA) (Figure ). 100% of the L. reali
were correctly classified to species with the DA, but within L. sinapis
, between 0 (France and Italy) and 62.5% (Kazakhstan) of specimens were correctly assigned to region (Additional file 1
, Table S1).
The mitochondrial Cytochrome Oxidase I
) and nuclear carbamoyl-phosphate synthetase 2/aspartate transcarbamylase/dihydroorotase
) and internal transcribed spacer 2
) markers analyzed did not reveal deep intraspecific levels of divergence (maximum uncorrected p distance of 0.61% for COI
, 0.7% for CAD
and 0.16% for ITS2
) suggesting the absence of cryptic species (Figures and Figure ). The COI
haplotype network (Figure ) shows that the maximum connection steps are only four, and that the most common haplotype is found in all the studied regions. The observed genetic variability is rather low for an almost pan-Palaearctic species (e.g. [42
]), even more so since L. sinapis
is considered a non-migratory poor flyer. The fact that the same low variability is shown by several independent markers rejects a recent mitochondrial genetic sweep and strongly suggests a very recent geographic expansion. Coalescence-based dating with each marker and with all the markers combined estimates that the time to the most recent common ancestor of all the populations is only 8,500 to 31,000 years. Thus, we conclude that there is no evidence for multiple species involved in the formation of the discovered cline, and that its origin is very recent.
Figure 3 Maximum Likelihood tree of Leptidea sinapis based on the combined analysis of mitochondrial COI and nuclear CAD and ITS2 according to the HKY model (log likelihood score = -3159.19036) and 100 bootstrap replicates. The scale bar represents 0.003 substitutions/position. (more ...)
It is known that in some systems, variation in chromosome number may be a result of ongoing hybridization between different, chromosomally diverged species [29
]. Therefore, the chromosome number variability discovered may be a consequence of hybridization between L. sinapis
and its sibling species L. reali
. This explanation may seem possible given that the presence of putative F1
hybrids between L. sinapis
and L. reali
was suggested [44
]. However, these results [44
] were based on some apparent mismatches between DNA-based identifications (which were congruent for RAPD markers and COI
) and morphometry of the male genitalia. The classification of the sequenced specimens based on their genitalia was made by employing a bivariate plot, which took into account only the lengths of the phallus and saccus. A recent comprehensive morphometrical study on L. sinapis
and L. reali
from Central Italy [40
] highlighted the limitation of the "phallus and saccus" approach, which can lead to ambiguous classifications. The same study showed that this limitation can be corrected when using additional genitalic characters (especially the vinculum width) and performing multivariate analyses. Therefore, the report of possible hybrids between L. reali
and L. sinapis
requires confirmation since it may actually represent an artifact caused by the interpretation of insufficient morphological traits. Moreover, in case of interspecific hybridization we can expect that some individuals would be heterozygous for species-specific nuclear molecular markers and specimens with intermediate morphology of genitalia should be found. None of the specimens studied in our work has shown these characteristics (see above). Due to genitalic morphological constraints between the two species, introgression is likely to be unidirectional with female L. sinapis
potentially inseminated by male L. reali
]. Finally, several studies dealing with the mating behaviour of L. sinapis
and L. reali
reported that females of both species exclusively mated with conspecific males, suggesting the presence of strong precopulatory barriers [36
]. Therefore, we can conclude that interspecific hybridization is an unlikely explanation for the origin of the discovered chromosomal cline.
The clinal distribution of chromosome numbers in L. sinapis
is statistically significant (p < 0.0001) and it is very unlikely to have arisen by chance (Figure ). Interestingly, the cline is longitudinally oriented (Figure ), indicating either the direction of selective pressure involved in its formation, or the direction of population dispersal, or both of these processes. According to our dating, the moment of this dispersal would correspond to the upper Pleistocene and the Holocene, a period characterized by a strong glaciation in northern Europe and the Alps [47
]. Thus, our estimates indicate that the dispersal of L. sinapis
could have occurred before or after the last glacial maximum (24,000 to 17,000 years ago).
Figure 4 Variation of L. sinapis chromosome number across geographical longitude. Chromosome number is inversely correlated with longitude according to a linear function (r = 0.826; p < 0.0001). Results based on 23 specimens with unambiguous chromosome (more ...)
Several other cases of broad intraspecific chromosomal polymorphism have been described in animals [6
] and plants [8
]. However, all these cases differ from the cline found in L. sinapis
by the essentially smaller range of karyotype variability and by the possible existence of two or more cryptic species involved in the formation of the polymorphic chromosomal system. In order to demonstrate the intraspecific nature of karyotype variability, the following three criteria should be met simultaneously: 1) segregating chromosomal polymorphism within a population should be demonstrated, 2) molecular markers should not suggest the presence of potential cryptic species, and 3) species-diagnostic morphological differences should be lacking. To our knowledge, only studies on the common shrew and the house mouse have met all these criteria, but chromosomal races within these mammals have essentially smaller differences in chromosome number and apparently evolved through a step-by-step accumulation of single chromosomal rearrangements [9
] rather than through wide intraspecific and intrapopulation chromosome number polymorphism.