Decay of homologous chromosome pairs and discovery of males in the thelytokous fungus-growing ant Mycocepurus smithii

The prevalent mode of reproduction among ants is arrhenotokous parthenogenesis where unfertilized eggs give rise to haploid males and fertilized eggs develop into diploid females. Some ant species are capable of thelytokous parthenogenesis, a type of asexual reproduction where females develop from unfertilized diploid eggs. Thelytoky is well-documented in more than 20 ant species. Cytogenetic data are available for six species demonstrating that some thelytokous ant species are capable of producing males occasionally as well as maintaining their chromosome numbers and proper chromosome pairings. Mycocepurus smithii is a thelytokous fungus-growing ant species that inhabits large parts of Central and South America. Cytogenetic data are unavailable for M. smithii and male individuals were never documented for this species, although the presence of males is expected because genetic recombination was observed in a few sexually reproducing populations in Brazil and haploid sperm was documented from the spermathecae of M. smithii queens. This study aims at comparatively studying asexual and sexual populations of M. smithii using classical and molecular cytogenetic methods to test whether karyotype configuration is modified according to the mode of reproduction in M. smithii. Moreover, we report the discovery of M. smithii males from a sexually reproducing population in the Brazilian state Pará, diagnose the male of M. smithii, and morphologically characterize their spermatozoa. Karyotypic variation was observed within the asexual population (2n = 9, 10, or 11), whereas the chromosome number was fixed in the sexual population (2n = 14, n = 7). Identical karyotypes were maintained within individual M. smithii colonies and karyotype variation was only observed between colonies. In asexual individuals, the karyomorphs showed a decay of homologous chromosome pairs, especially in individuals with the karyomorph 2n = 11, which is potentially caused by relaxed natural selection on proper chromosome pairing. In contrast, females in the sexual population showed proper homologous chromosome pairings. In individuals of both asexual and sexual populations, we find that heterochromatin was localized in centromeric regions and on the short arms of the chromosomes, GC-rich regions were associated with heterochromatic regions, and 18S rDNA genes were located on the largest chromosome pair. This comparative cytogenetic analysis contributes to our understanding about the cytological mechanisms associated with thelytokous parthenogenesis in ants and suggests the decay of chromosome structure in the absence of meiosis and genetic recombination.

. Colonies of Mycocepurus smithii studied for the cytogenetic comparison. Information about the mode of reproduction, collection locality of the sample, number of colonies examined, number of workers, gynes and males analyzed via classical and molecular cytogenetic methods, diploid (2n) and haploid (n) chromosome numbers, and karyotypic formula are presented. Colonies of the asexual M. smithii population from Minas Gerais were kept alive in the Laboratório de Citogenética de Insetos at the Universidade Federal de Viçosa for six months (24/VI/2012 until 30/XII/2012) to obtain larvae for cytogenetics and to test whether males could be raised from asexual colonies. Males were not produced by the asexual colonies from Minas Gerais although the production of winged queens (gynes) occurred during that period. Colonies of the arrhenotokous attine ant Mycetomoellerius relictus (Borgmeier, 1934) were also kept in the laboratory during the same period and, unlike M. smithii, produced males and gynes, suggesting that the laboratory conditions were probably adequate for keeping fungus-growing ants. Colonies were maintained in plastic containers that were sealed with perforated lids to enable gas exchange and covered with moist cotton to increase humidity in artificial laboratory nests. Collections were authorized by the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) (SISBIO accession number 32459) issued to Luísa Antônia Campos Barros. Specimens were identified to species by Jacques Hubert Charles Delabie and Christian Rabeling. Vouchers were deposited in the myrmecological collections of the Centro de Pesquisas do Cacau (CPDC) at the Comissão Executiva do Plano da Lavoura Cacaueira (CEPLAC), in Bahia, Brazil, records #5706 and #5707 for Viçosa and Ponte Nova, respectively, and under the record #5714 for specimens from Belém, as well as in the Social Insect Biodiversity Repository at the School of Life Sciences at Arizona State University in Tempe, Arizona, USA.

Mode of reproduction
Chromosome preparation, banding, and karyotype analyses. Mitotic metaphase chromosomes were obtained following the protocol developed by Imai et al. 25 . For the asexual individuals, brain ganglia of prepupae, which are recognizable by the presence of a visible head, and post-defecating larvae of workers and gynes were analyzed. For the sexual individuals, larval brain ganglia of workers, gynes, and males, in addition to testes of pupae and larvae were used. To characterize the karyotypes of M. smithii females, the larvae and prepupae of workers and gynes were first confirmed as females as evident by the absence of testes. We then distinguished worker from queen prepupae. Queen prepupae had distinctly larger body sizes and head widths than worker prepupae. Chromosome numbers and chromosome morphologies were identified using the Giemsa staining 4%.
Chromosomes were measured and arranged in order of decreasing size, and based on the ratios of chromosome arm lengths (r = long arm/short arm) according to the classification proposed by Levan et al. 26 . The chromosomes were classified as m = metacentric (r = 1-1.7), sm = submetacentric (r = 1.7-3), st = subtelocentric (r = 3-7), and a = acrocentric (r > 7). Additionally, the karyotype assembly was conducted by pairing the probable chromosomes which suffered rearrangements. Images were edited using Adobe Photoshop version 23.2.0.
The heterochromatin regions were detected using Giemsa staining 4% according to Imai 27,28 and also according to Sumner 29 , with modifications proposed by Barros et al. 30 . Metaphasic chromosomes were stained with the sequential fluorochromes CMA 3 /DA/DAPI to detect GC and AT-rich regions based on the methodology proposed by Schweizer 31 .
The metaphase chromosomes were observed and documented using an Olympus BX 60 fluorescence microscope with a Q-Color3 Olympus® image capturing system, using the software Q capture ® with the filters WB (450-480 nm), WU (330-385 nm) and WG (510-550 nm) for analyzing CMA 3 , DAPI, and rhodamine, respectively.
Morphological characteristics of ovaries and sperm. Ten asexually reproducing queens, each one from a different laboratory colony, were dissected to study ovary morphology and development. All individuals were dissected in buffered Ringer's solution to avoid desiccation of soft morphological structures. Female reproductive parts were studied using an Olympus SZ40 stereo microscope. We examined whether queens were inseminated (empty/translucent versus filled/opaque spermatheca), and whether they were reproductively active as evident by the presence of yellow bodies (Corpora lutea) and fully developed oocytes [19][20][21] . Asexual, reproductively active females were expected to have empty spermathecae, fully developed ovaries with mature oocytes, and corpora lutea as indicators of past oviposition. Asexual females that were never reproductively active were expected to have empty spermathecae, as well as ovaries with undeveloped oocytes and without corpora lutea. In contrast, sexually reproducing females were expected to have sperm filled spermathecae, developed oocytes, and corpora lutea. However, the reproductive tract of immature sexual and asexual queens could be identical if they were dissected pre-mating and/or pre-reproduction, or if sexually reproducing individuals were dissected postmating and pre-reproduction, they would be inseminated with developing ovaries but without Corpora lutea 20 .
To study the male reproductive morphology, seminal vesicles of three males from a single colony were dissected on histological slides in phosphate buffered saline solution (pH 7.2) and, then fixed in 4% paraformaldehyde and 0.1 M phosphate buffer for 20 min. The spermatozoa were stained with 4,6-diamidino-2-phenylindole (0.6 µg/mL) for 30 min in McIlvaine buffer (pH 7.0) and then washed using the same buffer. Finally, the slides were wet mounted with 50% sucrose and covered with cover slides for nucleus measurements. The nuclei were observed and photographed using an Olympus BX 60 fluorescence microscope with a Q-Color3 Olympus ® Scientific Reports | (2022) 12:4860 | https://doi.org/10.1038/s41598-022-08537-x www.nature.com/scientificreports/ image capture system, using the software Q capture ® with the filter WU (330-385 nm). After the removal of the fluorochrome, slides were stained with Giemsa 4% for 20 min and photographed using an Olympus BX 60 light microscope to obtain the total length of the sperm and their nuclei. A total of 20 randomly selected spermatozoa were measured per male using the software package Image Pro Plus ® .
Taxonomic characterization of the Mycocepurus smithii males. Mycocepurus individuals were examined and measured using a Leica M205 C stereomicroscope fitted with an ocular micrometer. Measurements were taken at 50× and 63× magnification and recorded to the nearest 0.01 mm at the maximum magnification allowed for each measurement without exceeding the bounds of the micrometer. Composite images were generated using a Leica DFC450 digital camera mounted to a Leica M205 C stereomicroscope and assembled using Leica Application Suite (version 4.5) and Helicon Focus (version 6.6.1) software packages. Morphological terminology, measurements and indices used for the taxonomic description follow recent taxonomic studies of fungus-growing ants 34,35 .

Results
Asexual population. Dissections revealed that the spermathecae of M. smithii queens from the Minas Gerais population were translucent, indicating that they were empty. The ovaries were fully developed and yellow bodies were present, indicating recent oviposition activity of the dissected queens ( Fig. 1). In combination, the egg laying activity, the presence of corpora lutea, and the absence of sperm from the queens' spermathecae demonstrate that the studied colonies from Minas Gerais reproduce thelytokously. In addition, laboratory colonies never produced males, which indirectly supports the absence of sexual reproduction in the Minas Gerais population.
Interestingly, the karyotypes of M. smithii females showed a variable number of chromosomes. Individuals from Ponte Nova, Minas Gerais had chromosome numbers of 2n = 9, 10, or 11, whereas the karyotypes of individuals from Viçosa, Minas Gerais had chromosome numbers of 2n = 10 or 11 (Table 1, Fig. 2). The assembly of karyotypes with the probable chromosome pairings is available in Fig. 3. The variation in chromosome number was only observed between different colonies, whereas the number and morphology of the chromosomes was maintained within the same colony. The chromosome morphology was also maintained among colonies with identical chromosome numbers.
The 18S rDNA gene clusters were located in the interstitial regions of the long arms of the largest submetacentric chromosome pair (pair 1) for the karyomorph 2n = 9 (Fig. 6a) and on the long arms of the submetacentric chromosomes (1 and 1b) for the karyomorph 2n = 11 (Fig. 6b). The region of the 18S rDNA genes colocalized with the CMA 3 + interstitial markings on the chromosomes 1 and 1b. The presence of a clear secondary constriction that corresponded to the ribosomal genes and GC-rich regions was observed for both karyomorphs.    Fig. 7). Secondary constrictions were observed on the larger submetacentric chromosome pair using Giemsa and fluorochrome DAPI staining. Heterochromatic blocks were observed on short arms of some chromosomes (2nd and 3rd submetacentric and the subtelocentric pair) and strong interstitial markings were also observed on the largest submetacentric chromosome pair (Fig. 8a,b). GC-rich regions were visualized on the short chromosome arms and intrachromosomal regions that colocalized with heterochromatin blocks. In addition, we observed an interstitial band on the long arm of the largest submetacentric chromosome pair (Fig. 8c), which colocalized with 18S rDNA gene clusters (Fig. 9). AT-rich regions with differential staining with DAPI were not observed, therefore AT-rich regions were observed in uniform patterns (Fig. 8d). Only negative regions, complementary to GC-rich regions, were detected.  (Fig. 10). Body surface opaque, medium to light brown, densely and finely reticulate-punctate; appendages light to yellowish-brown. Pilosity of body surface appressed. Dorsal and lateral surfaces of mesosoma and dorsal surface of head with weakly pronounced rugae. Antennae 13-segmented with short scapes (SI 68-74). Funiculus long with first funicular segment only slightly longer than wide, funicular segments 2-11 approximately 2.5 × longer than wide. In full face view, head shape trapezoidal, wider than long (CI 107-110), occipital corners with minute denticles. Eyes large, convex (EL 0.3). Ocelli slightly elevated above posterior margin of head, all ocelli of equal size. Clypeus broadly triangular, conically elevated between antennal insertions, with long median seta. Mandibles elongate, triangular, with three teeth, including a tiny basal and a large, broad apical tooth. Maxillary   (Fig. 11).

Discussion
Karyotypes in asexually reproducing M. smithii. Our comparative cytogenetic study of asexually and sexually reproducing M. smithii colonies reveals that in an asexual population from Minas Gerais karyotypes were constant within M. smithii colonies but variable between colonies. We identified intraspecific variation between sympatric M. smithii colonies with three distinct karyomorphs where the number of diploid chromosomes equaled 9, 10, and 11. In addition, the karyomorphs differed not only in their numbers but also in the morphology of the chromosomes. Although the chromosome numbers were constant within the same colonies, the described variation was observed between colonies that were located in the same population.
Upon examining the first sample, we first believed that the karyomorph 2n = 11 was a haploid karyotype because it presented an uneven chromosome number and almost all the chromosomes were unpaired. Closer study revealed that this karyotype belonged to female individuals, which are expected to be diploid in arrhenotokous Hymenoptera. However, studying the prepupae and larvae of queens confirmed that the observed individuals were in fact females due to the absence of testes.

Karyotypes in sexually reproducing M. smithii.
In contrast to the asexual individuals, the karyotypes of M. smithii females from sexually reproducing colonies were constant. The worker and gyne larvae from the sexually reproducing M. smithii population in Belém, Pará had identical diploid chromosome numbers (2n = 14), whereas male karyotypes were characterized by half the chromosome number (n = 7), which is expected for haploid males produced via arrhenotokous parthenogenesis. Female karyotypes also exhibited the proper homologous chromosome pairings, which is consistent with female karyotypes reported for most sexually reproducing ant species 40,41 . In addition, variation in chromosome number was not observed between individuals of different sexually reproducing M. smithii colonies, which contrasts with our observation in the asexual population.
Chromosome decay and heterozygosity in unpaired chromosomes. The karyotype differences observed between asexual and sexual M. smithii populations, as well as within the asexual M. smithii population, were characterized by numerical and morphological differences probably resulting from centric fission and other chromosome rearrangements. In the absence of meiosis, centric fission can increase the chromosome number, whereas further minor and independently occurring rearrangements on both the rearranged chromosome and the larger ancestral chromosome removes traces of homology between them, leading to the decay of the diploid chromosome structure. The observed karyotypes in asexual M. smithii populations were consistent with this expectation because the karyomorph 2n = 9 was characterized by three pairs of homologous chromosomes, whereas the karyomorph 2n = 10 presented only two homologous chromosome pairings, and the karyotype 2n = 11 showed only a single set of properly paired homologous chromosomes. The loss of homologous chromosome pairings is indicative of a decay in the diploid structure of the individual karyotypes, and the most extreme form of decay was observed in karyotype 2n = 11. In contrast, a decay of homologous chromosome pairs could not be observed in the sexually reproducing population of M. smithii, and in that population the chromosomes were perfectly paired with a haploid chromosome number of n = 7 and a diploid set of 2n = 14.
Our results indicate decaying karyotypes in an asexual M. smithii population, which is consistent with the hypothesis that natural selection is relaxed in the absence of meiosis 5,42 . Under those circumstances, chromosomes accumulate structural rearrangements and can become heterozygous in the absence of forced pairings during meiosis 5,7 , which has also been discussed as a potential mechanism for sexual chromosome diferentiation 43,44 . Ultimately, the accumulation of heterozygosity on individual chromosomes should lead to the loss of homologous chromosome pairs. Our results are consistent with this prediction because in asexual M. smithii population, we Mycocepurus smithii constitutes of a mosaic of asexual and sexual populations and the loss of sexual reproduction likely evolved repeatedly and independently from the ancestral sexual population 21 . Because asexual populations likely evolved convergently, we hypothesize that additional chromosomal variations will be recognized in other asexual populations. In contrast, we would expect that the karyotypes in sexual M. smithii populations are uniform. Among the different asexual colonies from Minas Gerais bearing karyomorphs of 2n = 10 and 2n = 11, variation in chromosomal morphology was not observed, suggesting a single origin of these karyomorphs in our study population (Table 1).
Cytogenetic data is available for at most three Mycocepurus species including M. goeldii (2n = 8) 45 , Mycocepurus sp. (2n = 8) 46 , and the asexual (2n = 9, 10, 11) and sexual (2n = 14, n = 7) populations of M. smithii studied here. Considering that cytogenetic data is available for two sexually reproducing species with a karyotype of 2n = 8, we initially expected that the asexual karyotype would be 2n = 8. The asexual karyomorph of 2n = 9 represents the smallest amount of decay in diploid chromosome structure. The observed diploid chromosome number of 2n = 14 in the sexual M. smithii population is the highest number of chromosomes observed in Mycocepurus ants so far.
Chromosome rearrangements were previously reported for ants such as in sexually reproducing bulldog ants in the Myrmecia pilosula species complex 25,47 . Interestingly, variation in chromosome number was reported from individuals belonging to the same colony even when only few individuals were sampled. Imai and colleagues 25,47 suggested that the observed variation in chromosome number was a consequence of variable chromosome  www.nature.com/scientificreports/ numbers in the parental generation, which is different from the variation observed in the asexual M. smithii where variation is likely caused by the decay of chromosome structure due to the lack of meiosis and where variation occurred between but not within colonies. Outside the Hymenoptera, cyclic parthenogenesis is observed in aphids where sexually and asexually reproducing generations alternate with each other. In contrast, many members of the aphid Tribe Tramini reproduce exclusively asexually 48 , and a decay of karyotype diploidy with a high degree of chromosome diversification can also be observed in these species.
Cytogenetic data from other thelytokous ant species. Among ants, thelytokous parthenogenesis has been documented for at least 20 species in four subfamilies including the Dorylinae, Formicinae, Myrmicinae, and Ponerinae 10,12-15 . Cytogenetic information is available for six thelytokous ant species including Ooceraea biroi (Forel, 1907), Paratrechina longicornis (Latreille, 1802), Platythyrea punctata (Smith, 1858), Pristomyrmex punctatus (Smith, 1860), Vollenhovia emeryi Wheeler, 1906, and Wasmannia auropunctata (Roger, 1863) 40,[49][50][51][52][53][54][55] . In contrast to the results obtained here for M. smithii, all of the abovementioned thelytokous ant species maintain their chromosome numbers and their homologous chromosome pairings 40 . In addition, all of these species are characterized by regular or occasional male production via arrhenotoky, except for P. longicornis where males are clones of their fathers 56 , suggesting that meiosis is still functional in those species 10 . In contrast, the production of males was exclusively observed in sexually reproducing M. smithii populations, and males do not seem to be produced in asexually reproducing M. smithii colonies suggesting that meiosis may be dysfunctional in these asexual populations.
Heterochromatin distribution pattern. Heterochromatin plays an important architectural role in chromosome structure and it is enriched in tandem repetitive sequences. The heterochromatin distribution on chromosomes is related to specific functions. For example, centromeric heterochromatin is important for the accurate segregation of chromosomes 57 . In addition, other regions with accumulations of highly repetitive DNA can be observed throughout the genome extending beyond the centromere. Heterochromatin can influence the frequency of structural rearrangements in neighboring regions 58 , and consequently, it is essential for chromosome stability 47 . According to the Minimum Interaction Theory (MIT) proposed by Imai and colleagues 47 , www.nature.com/scientificreports/ chromosome fissions are the principal rearrangements in the karyotype evolution of ants because they reduce chromosome sizes and thus reduce deleterious chromosome interactions in the interphase nucleus. The posterior heterochromatin growth after fission plays an important role in allowing telomeric maintenance and chromosome stability, yielding chromosomes with heterochromatic arms. The repetitive DNA sequences that constitute the heterochromatin in fungus-growing ant chromosomes, including M. goeldii, show richness of GC-base pairs, which is an uncommon trait in other ant species. It was suggested that this GC-rich heterochromatin originated in the common ancestor of the fungus-growing ants, but this hypothesis requires further study 41,45,59 . Notwithstanding, the results obtained in this study demonstrate that both asexual and sexual populations of M. smithii also possess a GC-rich heterochromatin composition similar to M. goeldii, which adds to the number of attine species with this characteristic trait and strengthens the hypothesis of a common origin of GC-rich heterochromatin in fungus-growing ants.
The heterochromatin distribution pattern on chromosomes aides to recognize possible chromosomal pairs in asexual M. smithii individuals. In addition, it strongly suggests the occurrence of centric fission in M. smithii according to Imai's Minimum Interaction Theory (MIT) 47 because we observed heterochromatin on short arms of rearranged chromosomes in sexual (2n = 14) and in asexual populations (2n = 9, 10, 11). In addition, the karyomorph 2n = 9, i.e., the karyotype with less chromosomal decay, presented heterochromatin on centromeric or pericentromeric regions of all the chromosomes and this pattern is similar to the one observed in M. goeldii (2n = 8) 45 . Thus, the information about heterochromatin distribution on the chromosomes provides important insights into the pathways of karyotype evolution in ants 27,47 . 18S rDNA gene distribution patterns. Ribosomal genes are important chromosomal markers used in different organisms, including ants, because these genes follow specific patterns of chromosomal organization. A single chromosome pair bearing all the rDNA genes is the most common and plesiomorphic pattern observed in diploid karyotypes of ants 60 . In M. smithii, 18S rDNA genes were located on a single chromosome pair in asexual individuals with the karyomorphs 2n = 9 (chromosome pair 1) and 2n = 11 (chromosomes 1 and 1b), as well as in the sexual individuals with the karyomorph 2n = 14 (chromosome pair 3). In M. smithii males the rDNA was located on a single chromosome (chromosome 3). These data corroborate our findings that asexual individuals are indeed diploid. In asexual individuals with the karyomorph 2n = 11, the homologous chromosomes showed the same morphology (submetacentric), but we also observed a size difference between them suggesting the occurrence of centric fission in the rDNA-bearing chromosome pair.
Both asexual and sexual populations had a secondary constriction in at least one of the homologous chromosomes that were colocalized with the 18S rDNA clusters and GC-rich heterochromatic bands. According to Teixeira et al. 60 , the presence of single rDNA sites on the chromosomes is influenced by their position on the chromosome because intrachromosomal regions are less prone to rearrangements compared to terminal regions. The analysis of rDNA clusters in further asexual populations should be promising to better understand the selective pressures on rDNA clusters in the genome.
A prior population genetic study of M. smithii suggested that individuals in asexual populations were genetically identical across multiple generations and that males and sexual recombination were absent from asexual populations 21 . The strict clonality of asexual M. smithii populations suggested either apomixis or automixis with central fusion, i.e., a form of asexual reproduction with low recombination rates, as the cytological mechanism underlying thelytokous parthenogenesis in M. smithii 21 . Our cytogenetic results are consistent with this general interpretation and lend further support to the apomixis hypothesis because the karyotypes of asexual M. smithii individuals show that heterozygous chromosome configurations likely resulted from chromosome rearrangements in the absence of meiosis. In contrast, the presence of aneuploid individuals would strengthen the hypothesis that asexual M. smithii reproduce via automixis with central fusion, however, such individuals were not detected. Improbable and unpaired chromosomal rearrangements are not expected to be maintained in the absence of meiosis in eukaryotic genomes and its loss would allow for the decay of diploid structures, as observed in asexual M. smithii individuals. The comparative study of spermatozoa is useful for understanding the reproductive biology of ants in more detail and to obtain additional morphological characteristics that could be used for comparative studies of closely related species. The sperm length in ants varies from 53 µm in Pseudomyrmex termitarius (Smith, 1855) 61 to 230 µm in Apterostigma fungus-growing ants 62 . Information related to spermatozoa in ants is still scarce 62,63 . Nevertheless, comparing the structure of these cells can be insightful when characterizing a species and solving taxonomic problems of cryptic species existing in sympatry, such as Neopoera inversa (Smith, 1858) and Neoponera villosa (Fabricius, 1804) that showed species specific differences in the length of the sperm nucleus 64 .

Males of
The spermatozoa of attine ants are highly variable in length with 67.06 µm in Atta sexdens (Linnaeus, 1758), being the smallest known sperm cell in the fungus-growing ants, to 230.49 µm in an unidentified Apterostigma species, being the longest known sperm cell in fungus-growing ants 62 . With a total length of 69 µm, the spermatozoa of M. smithii are among the smallest observed in the fungus-growing ants, similar in length to spermatozoa Scientific Reports | (2022) 12:4860 | https://doi.org/10.1038/s41598-022-08537-x www.nature.com/scientificreports/ of Atta sexdens. In general, in the lower attine ants spermatozoa length decreases with the increase of colony size for the queens that copulate with a single male, which suggests that the production and storage of sperm affect the evolution of sperm length 62 . The genus Apterostigma is a member of the paleoattine fungus-growing ants which also includes Mycocepurus and Myrmicocrypta 65,66 , and the size difference of sperm length between different species of Apterostigma ants is remarkable, ranging from 138.06 to 230.49 µm 62 . The nucleus length of M. smithii spermatozoa measures 12.2 (± 06.4) µm. The nucleus length in other ant species ranges from 9 to 50 µm in Dolichoderus and Nesomyrmex, respectively 63 . The available data for fungusgrowing ant sperm refers exclusively to the total length, preventing further comparisons.

Conclusions
Cytogenetic studies are important for revealing the cytological mechanisms and consequences underlying the evolution of asexual reproduction in ants 67 . To the best of our knowledge, we here report the first cytogenetic analysis for ants that demonstrates chromosome number variation within a species that reproduces via thelytokous parthenogenesis. These chromosomal data obtained for the asexual population of M. smithii corroborate the decay of diploid chromosome structure in the absence of meiosis due to relaxed natural selection on chromosome architecture. Comparative cytogenetic data obtained for both sexual and asexual populations of M. smithii indicate that centric fission plays a role in the origin of chromosomal variation observed in M. smithii, which is consistent with Imai's Minimum Interaction Theory 47 . The maintenance of unpaired chromosomal homologs in the karyotypes of asexual M. smithii individuals further supports the hypothesis that apomixis is the cytological mechanism underlying thelytoky in M. smithii, but automixis with central fusion cannot be ruled out. Our study supports the hypothesis that asexual reproduction becomes obligate in M. smithii populations once the ability to reproduce sexually is lost and that asexually reproducing individuals are unlikely to revert to sexual reproduction.
Furthermore, we confirm the existence of sexually reproducing M. smithii populations in the Amazonas region of Brazil. We describe the first karyotypes of males and females from a sexually reproducing population, contributing to the knowledge about chromosomal architecture in the species. Furthermore, we provide the first taxonomic diagnosis of the M. smithii males and describe the structure of their spermatozoa.
In summary, our comparative cytogenetic study demonstrates the rapid decay of chromosome architecture in the absence of meiosis and genetic recombination, contributing to our understanding about the evolution and the cytological mechanisms associated with thelytokous parthenogenesis in fungus-growing ants.