Differentiation of Melipona quadrifasciata L . ( Hymenoptera , Apidae , Meliponini ) subspecies using cytochrome b PCR-RFLP patterns

Melipona quadrifasciata quadrifasciata and M. quadrifasciata anthidioides are subspecies of M. quadrifasciata, a stingless bee species common in coastal Brazil. These subspecies are discriminated by the yellow stripe pattern of the abdominal tergites. We found Vsp I restriction patterns in the cytochrome b region closely associated to each subspecies in 155 M. quadrifasciata colonies of different geographical origin. This mitochondrial DNA molecular marker facilitates diagnosis of M. quadrifasciata subspecies matrilines and can be used to establish their natural distribution and identify hybrid colonies.


Introduction
The stingless bee Melipona quadrifasciata Lepeletier 1936 (popularly known as mandaçaia) is a member of the Meliponini tribe, a group with highly eusocial organization and pantropical distribution (Michener, 2000).The exclusively Neotropical genus Melipona is the most derived group (Camargo and Pedro, 1992) and is most diverse in the Amazon basin (Silveira et al., 2002).Approximately 40 Melipona species are distributed from Mexico to Argentina (Camargo, 1979;Michener, 2000), although the biology and systematics of the described species is still poorly understood (Camargo and Pedro, 1992).
Some Melipona species are essential to pollination of Brazilian Atlantic forest (Kerr et al., 1996) and open savannah ("cerrado") flora (Silberbauer-Gottsberger and Gottsberger, 1988), and agricultural crops (Heard, 1999;Klein et al., 2003;Kremen et al., 2004) and are important for the conservation of world flora and fauna (Melendez-Ramirez et al., 2002;Corlett, 2004;VillaNueva et al., 2005).Stingless bees are also increasingly used to improve productivity in commercial crops because they require less specialized handling and maintenance procedures than bees with stings, with some stingless species having been domesticated for honey production.

Melipona quadrifasciata quadrifasciata and
Melipona quadrifasciata anthidioides (Schwarz, 1932) are subspecies commonly observed in coastal Brazil, ranging from the northeastern state of Paraíba southward to the southernmost Brazilian state of Rio Grande do Sul (Moure and Kerr, 1950).These subspecies can be easily discriminated by the yellow stripe pattern from the third to sixth abdominal tergites, the pattern consisting of 3 to 5 continuous bands in M. q. quadrifasciata and 2 to 5 discontinuous bands in M. q. anthidioides (Schwarz, 1932).
The geographic distribution of the two subspecies is very distinct, with M. q. anthidioides being distributed in warmer areas and is hence more frequent in the states of Rio de Janeiro and Minas Gerais (Melo and Campos, 1987), while M. q. quadrifasciata is found in colder regions and is hence more abundant in the states of Paraná and Santa Catarina but is also found in the states of São Paulo, Rio de Janeiro and Minas Gerais at altitudes above 1500 m (Moure, 1975).Natural hybrid zones are described in elevated areas (500 to 700 meters) in the states of São Paulo and southern Minas Gerais.Hybrid colonies originating from males and queens of the two subspecies usually consist of bees with intermediate phenotypes in the tergal banding pattern, which can lead to misidentification (Moure and Kerr, 1950;Moure, 1975;Melo and Campos, 1987).
Identification of hybrid colonies is a relevant issue because the economic value of M. quadrifasciata has encouraged the exchange of queens among commercial breeders.This practice has promoted contact between the two subspecies, leading to gene flow and development of secondary hybrid populations.Data on genetic structure of bee species are useful for conservation purposes because population genetic purity is greatly affected by beekeeping and commercial breeding practices.Genetic markers that provide information on the degree of gene flow between these two subspecies can confirm specific subspecies kept by beekeepers and provide a more realistic picture of the natural distribution of M. quadrifasciata before drastic anthropic alterations.
In this paper we report a simple method to differentiate the M. q. quadrifasciata and M. q. anthidioides matrilines using cytochrome b (cyt-b) polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) patterns.This study presents the first population analysis from a large number of colonies and focuses on the identification of a mitochondrial DNA genetic marker at the subspecies level and the detection of hybrid colonies occurring naturally or induced by beekeeping.We also present a comparison between the data herein and previous findings.

Samples
Adult workers from 155 Melipona quadrifasciata colonies were analyzed.Their geographical origin and subspecies identification (based on tergal stripe pattern) are showed in Table 1.All samples from Rio de Janeiro and Curitiba were unambiguously identified as M. q. anthidioides, while all colonies from Prudentópolis, Blumenau and Vale do Itajaí were identified as M. q. quadrifasciata.Samples from the Universidade Federal de Viçosa and from Lins as well as from two meliponaries in Ribeirão Preto were composed of colonies with quadrifasciata, anthidioides or intermediate tergal stripe patterns (samples identified as mixture).The samples from the states of Minas Gerais and São Paulo (Table 1) do not reflect the natural occurrence area of these subspecies because the bees at these sites were maintained for a long time in hives where colonies from different areas were introduced; consequently these colonies are not considered "pure" genetic pools of the M. quadrifasciata subspecies.

Extraction, PCR-RFLP and electrophoresis and DNA sequencing
The DNA from the thorax of one adult worker per colony was extracted using a phenol-chloroform protocol (Sheppard and McPheron, 1991) and the extracted DNA used for PCR amplification of the mitochondrial cytochrome b region, using the primers 5'-TATGTACTA CCATGAGGACAAATATC-3' (forward) and 5'-ATTAC ACCTCCTAATTTATTAGGAAT -3' (reverse) described by Crozier et al. (1991).Reactions were carried out in a 25 µL final volume containing 1 µM of each primer, 250 µM of each dNTP, 2.5 mM MgCl 2 , 1x buffer (Biotools), 1 U of Taq DNA Polymerase (Biotools) and 1 µL of the extracted DNA (50-100 ng).Amplification cycles were repeated 30 times following a basic schedule of 30 s at 94 °C for denaturing, 20 s at 48 °C for annealing, and 1 min at 72 °C for extension.Digestion of the cyt-b fragment was subsequently performed in a 20 µL reaction mixture using 3 µL of the PCR product and 2 units of the enzymes in an appropriate buffer.Digestions were allowed to proceed for four hours with different restriction enzymes (Bgl II, Dra I, Mbo I, Taq I and Vsp I; all from Fermentas).Digested fragments were submitted to electrophoresis in non-denaturing polyacrylamide gels (8% (w/v) for Bgl II and 12% (w/v) for Dra I, Mbo I, Taq I and Vsp I) and silver stained.
For DNA sequencing each cyt-b haplotype found was cloned and two clones of each haplotype were forward and reverse sequenced on an ABI 3100 (Applied Biosystems) sequencer.Sequences were analyzed using Codon Code ) and Multalin (Corpet, 1988).The cyt-b nucleotide sequence variation between M. quadrifasciata subspecies were compared with GenBank entries for Melipona bicolor (AF 466146), Melipona compressipes (AF 181615) and Apis mellifera (L06178).

Results
The Melipona quadrifasciata cyt-b amplicon was 485 bp in length.Two Vsp I restriction patterns were observed (Figure 1), haplotype 1 with two fragments (119 bp and 366 bp) and haplotype 2 with three fragments (119 bp, 150 bp and 216 bp).All samples with M. q. anthidioides phenotype were associated with haplotype 1, while all samples possessing the M. q. quadrifasciata tergal stripe pattern were associated with haplotype 2 (Table 1).
Digestion of the cyt-b amplicon with Mbo I generated three patterns (Figure 1), haplotype A with three fragments (122 bp, 143 bp and 220 bp) and haplotypes B and C with four fragments (31 bp, 91 bp, 143 bp and 220 bp), the difference between B and C being the lower mobility of the 220 bp fragment in pattern C. No complete association was observed between tergal stripe phenotype and Mbo I restriction patterns, but a very close association was found between Mbo I patterns A and B with M. q. anthidioides samples and Mbo I pattern C with M. q. quadrifasciata samples (Table 1).
Digestions using Dra I generated three fragments (107 bp, 134 bp and 244 bp) while Taq I produced two fragments (138 bp and 347 bp), with no differences being detected between samples.None of the colonies analyzed showed Bgl II restriction sites.
Four composite haplotypes were obtained based on the Mbo l and Vsp I cyt-b patterns, A1, B1, B2, and C2 (Table 1).Each haplotype was submitted to sequencing reactions, with Figure 2 showing the Mbo l and Vsp I restriction sites and the differences between the haplotypes.
A cyt-b restriction map was created from the sequencing results (Figure 3), with restriction sites potentially useful for differentiating patterns A1, B1, B2 and C2 being represented on this map.For the construction of this map we considered only restriction sites with less than ten cuts per enzyme and, with the exception of the Vsp I and Mbo I sites, restriction sites shared among haplotypes are not shown.The number of enzymes restriction sites generated in the different haplotypes was six each for A1 and B1, seven for B2 and eight for C2.The two most difficult patterns to discriminate through Mbo I and Vsp I digestion (B2 and C2) could be easily differentiated using the Pac I endonuclease (Figure 3).
Relationships between M. quadrifasciata haplotypes and their similarities with M. bicolor, M. compressipes and Apis mellifera sequences are shown in Figure 4.Among M. quadrifasciata haplotypes, most substitutions were ob-served between B1/C2 (15) and A1/C2 ( 14).Fewer differences were seen between B2/C2 (4) and A1/B1 (5).It is worth to noting that the Vsp I patterns (1 vs. 2) are associated with a higher number of nucleotide substitutions throughout the cyt-b fragment than the Mbo I haplotypes (A, B and C).There were fewer substitutions between M. bicolor and M. compressipes (n = 36) than the number of substitutions between either of these species and M. quadrifasciata subspecies (ranging from 52 to 56 substitutions, respectively).This resulted in two clusters, one comprising M. quadrifasciata samples and the second encompassing M. bicolor and M. compressipes.

Discussion
The complete association found between specific tergal stripe phenotypes and Vsp I cyt-b restriction patterns (Vsp I pattern 1 in anthidioides and pattern 2 in quadrifasciata) allows for simple subspecies identification.Although Mbo I patterns could not be wholly assigned to a particular subspecies, a close link between Vsp I 1 -Mbo I pattern A with M. q. anthidioides and Vsp I 2 -Mbo I pattern C with M. q. quadrifasciata was observed.The Vsp I patterns are associated with higher nucleotide substitutions along the cyt-b fragment.As the dendrogram in Figure 4 shows, this association explains the clustering of Vsp I patterns into different clades and highlights the importance of Vsp I haplotypes for M. quadrifasciata subspecies identification.
The Mbo I restriction patterns B and C showed no fragment size differences, although they could be differen- tiated by slower electrophoretic mobility in the 220 bp fragment of pattern C. The difference in electrophoretic mobility may have resulted from base substitutions, leading to double-strand conformation polymorphism (DSCP).This has been described as a potential source of variable genetic markers for termite populations (Atkinson and Adams, 1997).In DSCP, polymorphisms occur when mutations affect the curvature of double-stranded DNA molecules visualized following electrophoresis in non-denaturing acrylamide gels (reviewed in Hagerman, 1990).This is apparently common in AT-rich DNA.Thus, the difference observed in the 220 bp cyt-b fragment between haplotypes B and C is a single base substitution (C ↔ A) at position 82 (Figure 2).A similar pattern of differential fragment mobility without size difference has been observed in the same    mtDNA region of Apis mellifera subspecies (KM Ferreira, personal communication).
The cyt-b restriction patterns for Bgl II have been proposed by Weinlich et al. (2004) as diagnostic markers for M. quadrifasciata subspecies.According to these authors, a Bgl II restriction site within the cyt-b region was only detected in M. q. quadrifasciata.However, the 155 M. quadrifasciata colonies analyzed by us showed no Bgl II restriction site, a fact confirmed by sequencing analyses.The absence of a Bgl II restriction site in M. quadrifasciata cyt-b has also been reported by Moretto and Arias (2005).Waldschmidt et al. (2000) identified a random amplified polymorphic DNA (RAPD) marker in quadrifasciata absent from anthidioides colonies of Brazilian M. quadrifasciata populations sampled from the core distribution of each subspecies.In the warmer northern sampling area of Minas Gerais state, where M. q. anthidioides colonies are frequent, specimens with M. q. quadrifasciata abdominal band patterns were found without this RAPD marker.This finding was interpreted by Waldschmidt et al. (2000) as the occurrence of natural hybrid colonies.In this sense, the cyt-b haplotypes described in our current paper provides a better method to identify hybrid colonies because to a specific tergal pattern must correspond the differential cyt-b pattern expected to that subspecies.Moretto and Arias (2005) recently reported a mitochondrial variant in the cytochrome oxidase I locus as a diagnostic marker for Melipona quadrifasciata subspecies, although population analyses are needed to confirm the broader application of this genetic marker.In addition, Moretto and Arias (2005) also reported a Dra I polymorphism in the cyt-b+ND1 region enabling the differentiation of the two subspecies.This variation was not observed in our present study, probably due to the different size of the fragment analyzed (1800 vs. 485 bp).
Stingless bees are suffering population declines in many areas of the world due to habitat destruction and deforestation (Melendez-Ramirez et al., 2002;Kremen et al., 2004;Samejima et al., 2004;VillaNueva et al., 2005), with some species probably facing extinction.The high degree of environmental alteration due to human activities has led to the current crisis for pollinators of wild and agricultural crops (Kearns et al., 1998;but see Cane, 2001).Loss of genetic variability due to environmental perturbation and the implication of such loss in respect to the conservation of stingless bees needs to be understood, and for this to occur the level of gene variation and genetic structure of stingless bee populations must be quantified.
The PCR-RFLP method outlined here appears wellsuited as a molecular tool for the identification of Melipona quadrifasciata subspecies, allowing easy diagnosis of matrilines.This molecular analysis, when associated with morphological and behavioral data, will be useful in future studies that will improve our understanding of the biology, biogeography and evolution of Melipona quadrifasciata.

Figure 1 -
Figure 1 -Non-denaturing silver stained 12% polyacrylamide gel showing Melipona quadrifasciata cytochrome b PCR-RFLP patterns for Vsp I (haplotypes 1 and 2) and Mbo I (haplotypes A, B and C) endonucleases.Fragment profiles pattern (fragment sizes) were based on sequencing results.

Figure 3 -
Figure 3 -Linear restriction map for Melipona quadrifasciata cytochrome b haplotypes A1, B1, B2 and C2.The restriction sites are represented by letters (A, Alu I; E, Ear I; M, Mbo I; P, Pac I; V, Vsp I) and their relative positions are indicated.

Table 1 -
Number of colonies of Melipona quadrifasciata sampled, their geographical origin, subspecies identification based on tergal stripe patterns and the cytochrome b polymerase chain reaction restriction fragment length polymorphism (RFLP-PCR) composite haplotypes found.N = number of colonies for each site.