Cytogenetic data on six leafcutter ants of the genus Acromyrmex Mayr, 1865 (Hymenoptera, Formicidae, Myrmicinae): insights into chromosome evolution and taxonomic implications

Abstract Cytogenetic data for the genus Acromyrmex Mayr, 1865 are available, to date, for a few species from Brazil and Uruguay, which have uniform chromosome numbers (2n = 38). The recent cytogenetic data of Acromyrmex striatus (Roger, 1863), including its banding patterns, showed a distinct karyotype (2n = 22), similar to earlier studied Atta Fabricius, 1804 species. Karyological data are still scarce for the leafcutter ants and many gaps are still present for a proper understanding of this group. Therefore, this study aimed at increasing cytogenetic knowledge of the genus through the characterization of other six species: Acromyrmex balzani (Emery, 1890), Acromyrmex coronatus Fabricius, 1804, Acromyrmex disciger (Mayr, 1887), Acromyrmex echinatior (Forel, 1899), Acromyrmex niger (Smith, 1858) and Acromyrmex rugosus (Smith, 1858), all of which were collected in Minas Gerais – Brazil, except for Acromyrmex echinatior which was collected in Barro Colorado – Panama. The number and morphology of the chromosomes were studied and the following banding techniques were applied: C-banding, fluorochromes CMA3 and DAPI, as well as the detection of 45S rDNA using FISH technique. All the six species had the same chromosome number observed for already studied species, i.e. 2n = 38. Acromyrmex balzani had a different karyotype compared with other species mainly due to the first metacentric pair. The heterochromatin distribution also showed interspecific variation. Nevertheless, all the studied species had a pair of bands in the short arm of the first subtelocentric pair. The fluorochrome CMA3 visualized bands in the short arm of the first subtelocentric pair for all the six species, while Acromyrmex rugosus and Acromyrmex niger also demonstrated in the other chromosomes. The AT-rich regions with differential staining using DAPI were not observed. 45S ribosomal genes were identified by FISH in the short arm of the first subtelocentric pair in Acromyrmex coronatus, Acromyrmex disciger and Acromyrmex niger. The uniform chromosome number in the genus Acromyrmex (2n = 38) suggests that Acromyrmex striatus (2n = 22) should be transferred to a new genus. Other aspects of the chromosome evolution in ants are also discussed.


Introduction
Fungus-growing ants belong to the Atta-genus group (Ward et al. 2015) corresponding to the tribe Attini in the traditional sense. Leafcutter ants comprise a particular group of fungus-growing ants which are referred to as dominant herbivores of the Neotropics (Hölldobler and Wilson 1990). They include the genera Acromyrmex Mayr, 1865 and Atta Fabricius, 1804 and are exclusively found in the New World, primarily in the Neotropical region (Mayhé-Nunes and Jaffé 1998) and are considered the most derived group of ants arising about 8-12 million years ago Brady 2008, Mehdiabadi andSchultz 2010).
The genus Acromyrmex contains 33 described species (or more than 60 taxa if all subspecies and variations are included) (Bolton 2014, Rabeling et al. 2015. They are distributed from California (USA) to Patagonia (Argentina), excluding Chile. Most Brazilian species are widely distributed, although some of them have more restricted distribution (Gonçalves 1961, Mayhé-Nunes 1991. The genus Acromyrmex has been subdivided into two subgenera, Acromyrmex and Moellerius Forel, 1893 (Emery 1913), based on morphological traits. A phylogenetic study based on the morphological traits of this genus showed that the two subgenera formed distinct groups, of which Moellerius was considered the most derived (Mayhé-Nunes 1991). However, recent phylogenetic molecular studies of the genus Acromyrmex, including five species of the subgenus Moellerius, subdivided Acromyrmex species into distinct clusters (Cristiano et al. 2013). Only two of them, Acromyrmex balzani (Emery, 1890) and Acromyrmex landolti (Forel, 1885), were placed in the same group, suggesting that Acromyrmex and Moellerius could not be monophyletic (Cristiano et al. 2013); similar results were observed by Sumner et al. (2004). These data suggest that the two subgenera Acromyrmex and Moellerius do not represent natural groups.
Leafcutter ants are one of the most studied groups of fungus-growing ants (Mayhé-Nunes 1991), both in terms of biology and geographic distribution. Their status as agricultural pests has contributed to their knowledge, although taxonomic limits of different species are sometimes unclear , Bacci et al. 2009). Under these circumstances, the so-called "integrative taxonomy" can produce more consistent results by complementing data obtained by different techniques (Schlick-Steiner et al. 2010). Nowadays, ant cytogenetics is a rapidly developing research field . Cytogenetic data on fungus-growing ants with information for at least the chromosome number and morphology are available at present for 38 taxa (reviewed in Barros et al. 2011, Cristiano et al. 2013, Barros et al. 2013, 2014a, 2014b, Cardoso et al. 2014, Barros et al. 2015, corresponding to about 10% of described species (Brandão et al. 2011). In some ant genera, e.g. in Mycetarotes Emery, 1913 andCyphomyrmex Mayr, 1862, chromosome numbers are variable at the species level (reviewed in Barros et al. 2011). However, species within the genera Atta (Barros et al. 2014a, Fadini and Pompolo 1996, Murakami et al. 1998, Barros et al. 2015 and Acromyrmex (Fadini andPompolo 1996, Goñi et al. 1983) have the same chromosome numbers, 2n = 22 and 2n = 38, respectively, and similar chromosome morphology.
Cytogenetic data on the leafcutter ants are scarce. Namely, these data are available for five Atta species (Barros et al. 2014a, Fadini and Pompolo 1996, Murakami et al. 1998, Barros et al. 2015. In these species, 2n = 22 and a karyotypic formula of 2n = 18m+2sm+2st were found. Similar banding patterns were also observed in different species (Barros et al. 2014a, Murakami et al. 1998, Barros et al. 2015 which belonged to the three of four species groups defined on the basis of molecular data (Bacci et al. 2009). Cytogenetic data on Acromyrmex are also restricted but available for some taxa  (Goñi et al. 1983). All the species had the same chromosome number, 2n = 38. However, Acromyrmex (Moellerius) striatus (Roger, 1863) has recently shown 2n = 22, with a karyotypic formula of 2n = 20m+2sm (Cristiano et al. 2013), the same chromosome number found in all Atta species studied to date (Barros et al. 2014a, Fadini and Pompolo 1996, Murakami et al. 1998, Barros et al. 2015. Since A. striatus belongs to the well-supported clade which is quite distinct from other members of the genus Acromyrmex, it is suggested that this species is a sister group of all other leafcutter ants, which split before the divergence between Acromyrmex and Atta (Cristiano et al. 2013). Despite the same chromosome number, karyotypes of A. striatus and Atta species differ in morphology of two chromosome pairs as well as in their banding patterns (Cristiano et al. 2013, Barros et al. 2014a. The aim of the present study is therefore to describe chromosome sets of six species of the genus Acromyrmex to update our knowledge of karyotype evolution of leafcutter ants and Neotropical Formicidae in general.

Material and methods
Six cytogenetically studied Acromyrmex species were collected between August 2008 and March 2010 in the state of Minas Gerais -Brazil, except for Acromyrmex echinatior (Forel, 1899) which was collected in Panama (Table 1). Metaphases were obtained according to Imai et al. (1988) using larval ganglia or testes of freshly defecated larvae. To study chromosome morphology, metaphases were analyzed using conventional 4% Giemsa staining. The karyotypes were composed by arranging chromosomes according to their size and chromosome arm ratio (r) (Levan et al. 1964). Ten best metaphases per species with a similar degree of condensation were measured. Karyotypes were composed using the Corel Photopaint X3® software. The colonies and individuals analyzed are listed in Table 1. For the banding techniques, 4 to 10 individuals per species were used: C-banding for heterochromatin detection was performed according to Sumner (1972) with minor adaptations suggested by Barros et al. (2013); sequential fluorochrome staining with CMA 3 /DA/ DAPI (Schweizer 1980) was done to reveal specific GC-and AT-rich regions. To detect nucleolus organizer regions (NORs) in three species, 2 to 4 individuals of each species were studied using fluorescence in situ hybridization (FISH) with the 45S rDNA probe isolated from Arabidopsis thaliana (Moscone et al. 1996). The metaphases were observed and photographed using an Olympus® BX 60 microscope attached to a Q Color 3 Olympus® image capture system. For fluorochrome analysis, filters WB (450-480 nm) and WU (330-385 nm) were used for studying CMA 3 and DAPI staining, respectively, as well as Leica microscope DMRA2 filter Y3 (545/30 nm), attached to a D Leica IM50 Version 5 Release 190 software was used for FISH analysis.
Adult ant specimens were identified by J.H.C. Delabie and deposited in the ant collection at the Laboratório de Mirmecologia do Centro de Pesquisas do Cacau (CPDC/Brazil).
Chromosome measurements revealed morphological differences between similar karyotypes (Table 1). Several chromosome pairs were easily recognizable among different species, i.e. the first metacentric pair, the two largest subtelocentric pairs and the largest (or unique in some species) acrocentric pair (Fig. 1). In the karyotype of A. balzani, the first metacentric pair was larger than in any other studied species (Fig.  1). The size of the first metacentric and the first subtelocentric chromosome pairs were similar in all species (Fig. 1b-f), again except for A. balzani (Fig. 1a). No geographical intraspecific differences between the karyotypes were found (Table 1).    The C-banding results of A. balzani, A. coronatus, A. disciger (Mayr, 1887) and A. rugosus (Smith, 1858) indicated bands in some chromosomes: in the short arms of the submetacentric and subtelocentric and also in the centromeric regions of the metacentric chromosomes (Fig. 2). The largest subtelocentric pair (denominated as ST1) showed bands in all the species. A. disciger (Fig. 2c), A. rugosus (Fig. 2d) and A. coronatus (Fig. 2b) had bands in the telomeric region of the short arm. In A. balzani (Fig. 2a) and A. echinatior (Fig. 2f) karyotypes, the heterochromatic bands were observed in the short arm of ST1 pair. A. niger (Smith, 1858) (Fig. 2e) showed bands in the telomeric region of the short arm of the ST1 pair and also in the long arm. Moreover, A. rugosus and A. niger had additional bands in the pericentromeric region and in the long arm region of the second subtelocentric pair, respectively.
The short arms of the ST1 pair revealed differences in the banding patterns among the species with fluorochrome CMA 3 . A. disciger (Fig. 3c) and A. coronatus (Fig. 3b) in the telomeric regions. A. balzani (Fig; 3a) in the short arms; and A. echinatior in the interstitial region (Fig. 3f). However, A. niger (Fig. 3e) and A. rugosus (Fig. 3d), besides the telomeric regions of the ST1 pair, also showed bands in additional chromosomes. A. niger (Fig. 3e) had additional bands in the long arm of the ST1 pair and in the long arm of the second larger subtelocentric pair. A. rugosus (Fig. 3d) showed small bands in the telomeric regions of ST1 and in at least three other chromosomes.  DAPI-positive bright bands which could correspond to AT-rich regions were not revealed ( Fig. 3a-f). Instead, DAPI-negative regions which co-localized with CMA 3positive bands were visualized. FISH with the 45S rDNA probe visualized telomeric bands in ST1 pairs of A. coronatus, A. disciger and A. niger (Fig. 4).

Discussion
Karyotypes of Acromyrmex species observed in this study can be distinguished only on the basis of chromosomal measurements. Differential heterochromatin growth is therefore responsible for small but robust differences in chromosomal morphology, and these differences could not be observed using classification proposed by Imai (1991). This classification is based on heterochromatin distribution, and it does not reflect differences in the chromosome size among similar karyotypes. However, some chromosomes had the chromosome arm ratio (r) within the limits of the classification (submetacentric or subtelocentric). In the ants A. coronatus and A. niger the chromosomes were classified according to the greater ratio (r), i.e., as subtelocentrics.
The largest metacentric pair of A. balzani strongly differs in size from that of other species probably due to complex chromosomal rearrangements that need to be further investigated. This can be explained by the fact that A. balzani forms a separate clade together with A. landolti according to the molecular phylogeny presented by Cristiano et al. (2013).
The six studied species showed heterochromatic segments on the short arms of ST1 pair. It was observed that these GC-rich heterochromatic regions correspond to NOR which is, in turn, confirmed by FISH with the 45S rDNA probe in the chromosomes of A. coronatus, A. disciger and A. niger. In the latter species, this technique revealed a single NOR, although additional multiple CMA 3 -positive bands also were observed. This means that these additional bands are not related to the ribosomal genes. NORs are generally GC-rich and CMA 3 -positive in different organisms (Reed and Phillips 1995). However, CMA 3 -positive regions are not always rDNA clusters (Sumner 1990), as was observed in A. niger. Multiple CMA 3 -positive bands and a single NOR revealed by FISH were observed in the fungus-growing ant Mycocepurus goeldii (Forel, 1893) (Barros et al. 2010(Barros et al. , 2012; however, ribosomal gene mapping studies of Formicidae of the Neotropical region using FISH are scarce.
The nonspecific banding pattern of DAPI staining revealed in the present work is similar to those observed for other fungus-growing ants such as M. goeldii (Barros et al. 2010), A. striatus (Cristiano et al. 2013), Trachymyrmex fuscus Emery, 1934(Barros et al. 2014 and Atta species (Barros et al. 2014(Barros et al. , 2015. Up to now, 12 Acromyrmex species (plus the only subspecies) are cytogenetically studied. All of them show 2n = 38, including both subgenera Acromyrmex and Moellerius (Fadini andPompolo 1996, Goñi et al. 1983). However, A. striatus with 2n = 22 differs from other already known species (Cristiano et al. 2013). The latter chromosome number is also characteristic of all Atta species. Both Acromyrmex and Atta are considered the most derived genera of fungus-growing ants Brady 2008, Mehdiabadi andSchultz 2009). Since differences in chromosomal morphology and banding patterns can be observed within Acromyrmex, it differs in this respect from Atta.
Patterns of heterochromatin distribution on short arms of some submetacentric and subtelocentric chromosomes of Acromyrmex species suggest that centric fissions which contributed to the origin of the derived karyotype with 2n = 38, probably occurred in the karyotype of the most recent common ancestor of this group. Moreover, recent molecular phylogenetic reconstruction by Cristiano et al. (2013) also suggests that A. striatus is a sister group to the remaining leafcutter ants. The above-mentioned fissions were followed by heterochromatin growth which played an important role in maintaining telomeric stability according to the minimum interaction theory proposed by Imai et al. (1994). Differential heterochromatin growth in Acromyrmex is responsible for interspecific variation in the size of heterochromatic blocks. However, different species of this genus retain the same chromosome number (2n = 38), except for A. striatus with 2n = 22 (Cristiano et al. 2013).
Besides data on the chromosome numbers, multiple GC-rich segments were observed in the fungus-growing ants M. goeldii (Barros et al. 2010), Sericomyrmex sp. (Barros et al. unpublished data), T. fuscus (Barros et al. 2014b) andA. striatus (Cristiano et al. 2013). Nevertheless, single CMA 3 -positive bands were found in the short arms of ST1 pairs in all Acromyrmex spp. and in the fourth chromosome pair of Atta species. Multiple GC-rich segments observed in A. niger do not represent NORs, and therefore are probably derived.
Cytogenetic data permitted the differentiation among four of the six Acromyrmex species studied. A. balzani, included in the Moellerius subgenus, showed the largest metacentric chromosome pair with lower size compared with the other species. A. echinatior, besides the higher quantity of acrocentric chromosomes, also had interstitial bands in the ST1 pair for the fluorochrome CMA 3 , differing from the other species, which suggest the possibility of inversion. A. niger showed multiple CMA 3 -positive bands: in the telomeric regions of the short arms of the ST1 pair, in the pericentromeric regions of the long arm of the ST1 pair and in the second largest subtelocentric pair. A. rugosus had a greater proportion of metacentric chromosomes compared with the other Acromyrmex and also showed small bands in the telomeric regions of at least three other chromosomes. A. coronatus and A. disciger could only be cytogenetically differentiated from the other species by slight differences in the morphology that are probably due to the differential growth of heterochromatin on the short arms of the chromosomes.
Five of the six Acromyrmex species studied in this paper were collected in a particular area in the South East of South America. However, A. echinatior was collected in Central America, which is more than 5,000 km from the main study area. Moreover, another three species analyzed by Goñi et al. (1983) were collected in Uruguay, expanding the knowledge of South American Acromyrmex. The chromosome number is uniform among different species of Acromyrmex (sensu stricto), although chromosome morphology and banding patterns of these ants allow the identification of some species via their karyotypes. However, karyotype structure of A. striatus suggests that it belongs to a different lineage and therefore, according to sequences obtained from nuclear genes, this species does not belong to the "true" Acromyrmex lineage but to the sister group to the remaining leafcutter ants (Cristiano et al. 2013). A new genus therefore could be erected due to the karyotypic features of Acromyrmex which are further supported by combining the already published cytogenetic and molecular data (Cristiano et al. 2013) together with the additional karyological information. In this case, cytogenetics shows its importance as an additional tool in integrative taxonomy.
Conserved chromosome numbers were found in certain ant genera, as in Pogonomyrmex Mayr, 1868 in which 13 of 15 studied species had the same chromosome number. The two other species were transferred to another subgenus Ephebomyrmex Wheeler, 1902 (Taber et al. 1988). Camponotus (Myrmothrix) spp. also presented uniform chromosome number (Mariano et al. 2003), as did the other members of the genera Pheidole Westwood, 1839, Lasius Fabricius, 1804and Iridomyrmex Mayr, 1862(reviewed in Lorite and Palomeque 2010. Other animal groups, such as most birds and different genera of insects of the order Lepidoptera also demonstrated conservatism in respect to the chromosome number (White 1973), as did some bee genera, such as Melipona Illiger, 1806 (reviewed in Rocha et al. 2007) and Partamona Schwarz, 1939(reviewed in Martins et al. 2009).
Our data confirmed uniformity of the chromosome number (2n = 38) in the studied Acromyrmex species. However, chromosomal rearrangements such as heterochromatin growth are likely to be responsible for karyotypic differentiation in this ant group. Location of rDNA clusters of other leafcutter ants (especially A. striatus) also needs to be determined using molecular cytogenetic techniques (FISH). Moreover, cytogenetic studies of other members of fungus-growing ants, e.g. of the genus Trachymyrmex Forel, 1893 which represents the sister group to leafcutter ants, will be important for better understanding of chromosomal evolution of this group and Neotropical Formicidae in general.