Genetic divergence between Melipona quadrifasciata Lepeletier (Hymenoptera, Apidae) populations

Melipona quadrifasciata is a stingless bee widely found throughout the Brazilian territory, with two recognized subspecies, M. quadrifasciata anthidioides, that exhibits interrupted metasomal stripes, and M. quadrifasciata quadrifasciata, with continuous metasomal stripes. This study aimed to estimate the genetic variability of these subspecies. For this purpose, 127 colonies from 15 Brazilian localities were analyzed, using nine species-specific microsatellite primers. At these loci, the number of alleles ranged from three to 15 (mean: 7.2), and the observed heterozygosity (Ho) ranged from 0.03–0.21, while the expected heterozygosity (He) ranged from 0.23–0.47. The genetic distances among populations ranged from 0.03–0.45. The FST multilocus value (0.23) indicated that the populations sampled were structured, and the clustering analysis showed the formation of two subgroups and two more distant populations. The first group contained the subspecies M. quadrifasciata quadrifasciata, and the other, the subspecies M. quadrifasciata anthidioides and the two M. quadrifasciata populations with continuous metasomal stripes from northern Minas Gerais. These results confirmed that the yellow metasomal stripes alone are not a good means for correctly identifying the different subspecies of M. quadrifasciata.


Introduction
Despite their ecological importance as pollinators of native trees in Brazil (Kerr et al., 2001;Imperatriz-Fonseca et al., 2006), stingless bee populations have decreased in the Neotropics, similarly to other bee populations, due to habitat degradation and fragmentation, agricultural intensification, indiscriminate use of agrochemicals and pesticides, and introduction/spread of exotic species (Freitas et al., 2009).
In general, these studies have provided evidence that it is possible to recognize molecular differences between the two M. quadrifasciata subspecies, confirming the usefulness of the yellow metasomal stripes, which are continuous in M. quadrifasciata quadrifasciata and interrupted in M. quadrifasciata anthidioides, to distinguish them (Schwarz, 1932).
However, according to Moure and Kerr (1950), although M. quadrifasciata quadrifasciata is found in the colder regions located from Rio Grande do Sul to southern São Paulo, and M. quadrifasciata anthidioides occurs from northeastern São Paulo along eastern Brazil up to Paraíba, a hybridization zone exists in the states of São Paulo and Minas Gerais, with specimens presenting an intermediary pattern of yellow metasomal stripes.
More recently, M. quadrifasciata individuals from Januária and Urucuia (northern Minas Gerais) have been observed exhibiting continuous metasomal stripes identical to those of M. quadrifasciata quadrifasciata instead of the interrupted pattern characteristic to M. quadrifasciata anthidioides, as was expected based on their geographical localization (review in Batalha-Filho et al., 2009).The specimens from Januária, however, did not exhibit the RAPD marker present in individuals of the subspecies M. quadrifasciata quadrifasciata, being therefore genetically more similar to M. quadrifasciata anthidioides (Waldschmidt et al., 2000).
Another population bearing a continuous pattern of metasomal stripes was later on described in Sergipe and northeastern Bahia (Batalha-Filho et al., 2009).RFLP analysis and sequencing of the mtDNA COI gene again demon-strated that this population grouped together with populations from northern Minas Gerais and populations of M. quadrifasciata anthidioides (Batalha-Filho et al., 2010).
These data demonstrate that a distinction between M. quadrifasciata anthidioides and M. quadrifasciata quadrifasciata based exclusively on morphological characteristics is not reliable.
Thus, the main objective of this study was to investigate the genetic variability of some M. quadrifasciata populations from different locations in Brazil, in order to improve the knowledge about their taxonomic status.

Bee samples
Melipona quadrifasciata adult workers from 127 colonies were collected in eight Brazilian states.M. quadrifasciata quadrifasciata samples were collected in the states of Rio Grande do Sul (Porto Alegre), Santa Catarina (Içara) and Paraná (Curitiba), while M. quadrifasciata anthidioides samples were obtained in Rio de Janeiro (Miguel Pereira), Minas Gerais (Cristiano Otoni, Piranga, Caeté, Rio Vermelho and Poté), Espírito Santo (Domingos Martins) and Bahia (Caitité).We also included samples from Urucuia/Januária and São Cristovão in the analyses, representing M. quadrifasciata populations with continuous metasomal stripes from northern Minas Gerais and Sergipe, respectively.It is important to emphasize that these populations were located completely outside the distribution area of M. quadrifasciata quadrifasciata which typically exhibits continuous yellow metasomal stripes.
Samples from Ourolândia (Bahia) presenting a hybrid pattern of metasomal stripes were also analyzed.
Samples from Cristiano Otoni, Caeté, Urucuia and Januária were collected from feral colonies, directly from nature.Samples from Piranga, Rio Vermelho, Poté, São Cristovão, Ourolândia and Caitité were provided by beekeepers that maintain a small number of colonies in the original trunks or in rational boxes.These colonies were collected in their propriety wood or in the neighborhood, approximately 500-1000 m from their present-day localization.Samples from Domingos Martins, Miguel Pereira, Içara, Porto Alegre and Curitiba were obtained from colonies kept in local meliponaries.These colonies, however, were also obtained from the respective regions and did not represent translocated colonies.
The geographic distribution and the number of colonies sampled in each locality are shown in Table 1.Samples were stored in absolute ethanol at -80 °C until DNA extraction.

DNA extraction and amplification
Genomic DNA from 127 individuals (one per colony) was extracted according to the protocol of Waldschmidt et al. (1997) and amplified using nine microsatellite primers specifically designed for M. quadrifasciata in our laboratory (Table 2).
The amplification reaction solutions consisted of 10 ng DNA, 200 mM of each dNTP, 1.0 mM of each primer (F, R and M13), 1.5 mM MgCl 2 , and 0.5 unit of Platinum Taq polymerase (Invitrogen) in 10x buffer.Amplifications were performed by an initial denaturation step of 94 °C for 3 min, followed by 10 cycles of 94 °C for 15 s, with a specific annealing temperature for each primer for 20 s and 72 °C for 30 s. Subsequently, 25 cycles at 89 °C for 15 s, 53 °C for 20 s, 72 °C for 30 s, and a final extension at 72 °C for 30 min were performed.The polymerase chain reaction (PCR) products were identified using a MegaBACE automated capillary sequencer.Allele sizes were scored against the size standard ET 550R (GE Healthcare Life Sciences, Pittsburgh, PA, USA), and the peaks were analyzed using the MegaBACE Fragment Profiler program.

Statistical analysis
The Popgene version 1.32 program (Yeah et al., 1999) was used to calculate the following estimators of genetic diversity: allele frequency, expected (H e ) and observed (H o ) heterozygosity, genetic distance between populations (Nei, 1978), and F-statistics (Wright, 1978).For these statistical analyses, colonies were grouped according to locality, and each of these groups was considered to be a population.
Grouping analysis based on the UPGMA algorithm and the analysis of molecular variance (AMOVA; Excoffier et al., 1992) were carried out using the Genes program (Cruz, 2012).The cophenetic correlation coefficient that measures to which extent the clustering result corresponds to the original dissimilarity matrix was used to test the efficiency of the clustering analysis for the data (Rohlf and Fisher, 1968), and AMOVA was calculated considering two hierarchic levels and the groups formed in the clustering analysis.

Results
The nine microsatellite loci analyzed presented 65 alleles, ranging from three (Mquad24) to 15 (Mquad7) alleles (mean: 7.2 alleles/locus).All loci showed deviations from the Hardy-Weinberg equilibrium (p < 0.05) at one or more sites, but no single locus showed deviations in all cases.Table S1 shows the allele frequencies observed for these loci, considering all the analyzed populations.
The most frequent alleles of the different loci were present in both populations of M. quadrifasciata quadrifasciata and M. quadrifasciata anthidioides, as well as in the populations with continuous metasomal stripes from northern Minas Gerais and Sergipe and in the hybrid morphotype from Bahia.In general, however, the populations exhibited different allelic constitutions, with some fixed and other private alleles (Table S2).
In the different analyzed populations, the observed heterozygosity (H o ) ranged from 0.03-0.21,while the expected heterozygosity (H e ) ranged from 0.23-0.47.
Clustering analysis demonstrated that populations of both subspecies, although clustering together in a single major group, can be divided into two subgroups, one containing the populations of the subspecies M. quadrifasciata quadrifasciata and the other the populations of the subspecies M. quadrifasciata anthidioides and the two M. quadrifasciata populations with continuous metasomal stripes from northern Minas Gerais (Figure 1).Only populations from São Cristovão and Ourolândia, which showed higher genetic distance values than the other populations (Table 3), were isolated in this analysis.The cophenetic correlation of this clustering was high (0.87), further showing a good adjustment between the genetic distances and the graphically represented distances.
The F ST multilocus value obtained when populations of both subspecies were considered together (F ST : 0.23) also showed that they were structured, and the AMOVA test demonstrated that the interpopulation variation (52.5%) was higher than the intrapopulation variation (47.5%) of the subgroups obtained in the clustering analysis.Differences in the allele frequencies among populations as well as their private alleles probably contributed to these results.
This high number of alleles, however, was evidenced only when all populations were considered together.When each population was considered separately, only few alleles were found in each of the loci analyzed, always with one of them at a higher frequency.Consequently, a low frequency of heterozygotes was detected, which may explain the differences between the expected and observed heterozygosity values found.Additionally, despite having sampled 13 to 18 colonies in some localities, this sample size can be considered relatively small, which might also explain this difference, as well as the observed deviations from the Hardy-Weinberg equilibrium.
Notwithstanding, the lower level of genetic variability detected in the M. quadrifasciata populations (that gen-   et al., 2003), for example, may be due to biological differences between the species, such as the facultative polygyny of M. bicolor (Cepeda, 2006;Velthuis et al., 2006), the multiple mating of honeybee queens (Estoup et al., 1994) or the limited flight dispersion distance of stingless bees (Araújo et al., 2004).Although M. quadrifasciata presents a widespread geographic distribution, occupying distinct biomes such as the Atlantic rainforest, gallery forests (seasonal forests) and arborous caatinga, environment degradation may also be contributing to their reduced genetic variability.This hypothesis should be considered, as it was already observed in M. rufiventris and M. mondury populations (Tavares et al., 2007) that environmental degradation may isolate populations in small forest fragments, leading to inbreeding and local extinction.An indication of this local extinction is provided by Batalha-Filho et al. (2009) who, in contrast to Kerr (1951), found no M. quadrifasciata specimens in Pernambuco and Paraiba, despite intensive searching.
It is worthy of note that many of the private alleles detected in the present study were found in samples from Urucuia/Januária (northern Minas Gerais).Similarly, allele A of locus Mquad24 was found only in M. quadrifasciata population from São Cristovão (Sergipe) and in the hybrid population from Ourolândia (Bahia).As mentioned above, although the samples from Urucuia, Januária and São Cristovão contain M. quadrifasciata populations with continuous metasomal stripes, they were located completely outside the distribution area of M. quadrifasciata quadrifasciata.
Therefore, the presence of these private alleles is evidence of restricted genetic flow between these and the other populations of M. quadrifasciata anthidioides and could explain the genetic differentiation detected for the populations of São Cristovão and Ourolândia.Genetic differentiation among some M. quadrifasciata populations has also been detected using ISSR markers (Nascimento et al., 2010), PCR-RFLP assays of mtDNA (Batalha-Filho et al., 2009), and mtDNA COI gene sequencing (Batalha-Filho et al., 2010).According to these authors, geographic barriers separating close neighbors of this species can explain this differentiation.
The clustering analyses showed that M. quadrifasciata populations from Curitiba, Içara and Porto Alegre, which correspond to the subspecies M. quadrifasciata quadrifasciata, form a separate subgroup from the M. quadrifasciata populations of Urucuia and Januária, which exhibit a continuous metasomal stripe pattern similar to this subspecies, but are genetically more similar to M. quadrifasciata anthidioides populations.
The two other morphologically atypical M. quadrifasciata populations (São Cristovão and Ourolândia), however, were found to be isolated in the clustering analyses performed in this study.These same populations were clustered with M. quadrifasciata anthidioides in the phylogeographic study of Batalha-Filho et al. (2010), although the presence of exclusive haplotypes was detected in populations with continuous metasomal stripes from Sergipe.This result may represent differences between the two molecular markers used (mitochondrial vs. nuclear).In the present case, as microsatellites represent nuclear markers with a high mutation rate, the presence of two or three exclusive alleles was sufficient to differentiate populations.Further studies are necessary to elucidate the taxonomic status of these populations.
Overall, the results obtained in the present study corroborate those of previous studies demonstrating genetic differences between the two M. quadrifasciata subspecies (Waldschmidt et al., 2000;Moreto and Arias, 2005;Souza et al., 2008;Batalha-Filho et al., 2009, 2010).Additionally, as already observed by Batalha-Filho et al. (2009, 2010), our results also indicated that the M. quadrifasciata populations from northeastern Bahia and Sergipe are genetically different from the M. quadrifasciata quadrifasciata samples, even though they show the characteristic continuous metasomal stripes pattern.

Table 2 -
Characteristics of the nine microsatellite loci of Melipona quadrifasciata used in the present study.
F and R: forward and reverse primers, respectively; Ta: annealing temperature; k: number of alleles.

Table S1 -
Alllelic frequencies of the nine microsatellite loci, estimated from fifteen populations of Melipona quadrifasciata

Table S2 -
Distribution of the alleles present in the nine loci analyzed in the different populations of Melipona quadrifasciata.Codes correspond to those from Table 1.Private alleles are in bold