3.2. Mitochondrial DNA

Mitochondrial DNA is a small (circa 16 000 bp), circular molecule that is transmitted intact by the queen to her offspring (workers and drones). Therefore, unlike the other markers discussed in this chapter, mtDNA is uniparentally (maternally) inherited, rather than biparentally. The non-recombining maternal inheritance renders interpretation of mtDNA data straightforward, which in combination with relatively simple and inexpensive assays has made mtDNA one of the most popular markers in honey bee genetic studies.

Variation in honey bee mtDNA has been detected using a variety of molecular methods, ranging from RFLPs (Restriction Fragment Length Polymorphisms), to PCR-RFLPs and direct sequencing. RFLP variation is revealed by digesting the entire mitochondrial genome with restriction endonucleases. These enzymes are isolated from bacteria and cut DNA at a constant position within a specific recognition site that is typically 4 or 6 bases long. A battery of 4 (HinfI) and 6-base (AccI, AvaI, BclI, BglII, EcoRI, HincII, HindII, HindIII, NdeI, PstI, PvuII, XbaI) recognition site restriction enzymes were used in early studies of European and African honey bee subspecies in the native (Smith et al., 1991; Garnery et al., 1992; Arias et al., 2006), and introduced ranges (Smith and Brown, 1988; Hall and Muralidharan, 1989; Smith et al., 1989). While these RFLP surveys roughly discriminated and supported three of the major morphology-based lineages (M, A, and C, see also Annex) proposed by Ruttner (1988) and revealed the existence of some subspecies-specific haplotypes (Smith et al., 1991), no single or composite set of restriction enzymes have proved to be diagnostic markers for subspecies identification.

The RFLP method has been replaced by PCR-RFLP mainly to overcome the disadvantage that it requires relatively large amounts of non-degraded DNA. In both methods, haplotypes are resolved by employing restriction enzymes. However, unlike RFLPs, which survey the whole mitochondrial genome, PCR-RFLP variation is revealed within a specific PCR-amplified region. Several PCR-RFLP assays (Table 4), spanning coding regions, have been developed and employed in identification of New World honey bee populations undergoing Africanization (Hall and Smith, 1991; Clarke et al., 2002; Pinto et al., 2003; Pinto et al., 2004) and in Old World honey bees of eastern European ancestry (Smith et al., 1997; Bouga et al., 2005; Ivanova et al., 2010; Stevanovic et al., 2010). These assays have proved to be useful for discriminating variation in some eastern European subspecies, previously undetected using other mtDNA methods. Specifically, amplification of COI and ND5 genes followed by digestion with NcoI/StyI and AluI, respectively, produce diagnostic patterns characteristic of A. m. macedonica (Bouga et al., 2005; Stevanovic et al., 2010). Additionally, there are differences between Greek and Bulgarian honey bees in the digestion pattern of COI with SspI, and of ND5 with HincII and FokI, which do not recognize any site in Bulgarian honey bees (Ivanova et al., 2010).

The PCR amplification of the non-coding region located between the tRNAleu and COII genes (originally named COI-COII intergenic region), followed by digestion with the DraI restriction enzyme, has been the most popular PCR-RFLP assay (Garnery et al., 1993). This assay, which is commonly known as the DraI test, has been widely used in honey bee maternal identification in both Old World and New World populations. It has been helpful in (i) phylogeographical studies (De la Rúa et al., 1998; Franck et al., 2001; De la Rúa et al., 2006), (ii) understanding the complexities underlying natural hybrid zones (Franck et al., 1998; Franck et al., 2000; Cánovas et al., 2008), (iii) detecting introgression of foreign queens (Jensen et al., 2005), (iv) tracking the temporal changes of maternal composition of honey bee populations undergoing Africanization (Clarke et al., 2002), among others.

The DraI test has shown a high power of resolution which results from a combination of length variation with restriction site polymorphisms. The COI-COII intergenic region is composed of two distinct nucleotide sequences, named P and Q, where P can also appear in several variations (P0, P1, P2). According to Garnery et al. (1993), each evolutionary lineage includes a variant of the P sequence, combined with a different copy number of the Q sequence, resulting in length polymorphisms of this mtDNA region. Honey bees of the eastern European lineage (C) have the shortest intergenic sequence, because they lack the P element and carry a single copy of the Q element (Cornuet et al., 1991; Garnery et al., 1993). Honey bees belonging to lineages M, A, Z and Y exhibit longer intergenic regions because they carry from one up to five Q elements (Rortais et al., 2011) in addition to a variant of the P element (see the BEEBOOK paper on molecular methods (Evans et al., 2013) for further details). Additional polymorphisms can be resolved following digestion of the length variants with the DraI restriction enzyme. Lineage Z was formerly known as (mitochondrial) O (see Annex).

The combination of length and restriction-site polymorphisms produced by the DraI test has resolved over 100 haplotypes (Franck et al., 2001; De la Rúa et al., 2005; Collet et al., 2006; Shaibi et al., 2009; Alburaki et al., 2011; Rortais et al., 2011; Pinto et al., 2012), which have been correctly assigned to evolutionary lineages. However, despite the higher resolving power of the DraI test compared to other assays, it is unable to identify honey bees at the subspecies level because it does not produce diagnostic haplotypes. For example, if there is no further information available, a given honey bee carrying a C1 haplotype could be identified as A. m. carnica or A. m. ligustica, because this haplotype is present in both populations, although at different frequencies (Muñoz et al., 2009). The same applies to an A1 haplotype which could be carried by an A. m. iberiensis, A. m. adansonii or other African subspecies (Franck et al., 2001; Cánovas et al., 2008).

Despite its inherent limitations, the DraI test should be adopted as the standard for honey bee maternal identification, for several reasons. First, because among all PCR-based methods, it has proven to be the most powerful and informative. Second, because COI-COII/DraI variation has been widely documented across natural and introduced honey bee ranges. Consequently, a large catalog of haplotype patterns is available in the literature (see restriction maps in Franck et al., 2001; Collet et al., 2006; Shaibi et al., 2009; Pinto et al., 2012), and sequence data for numerous described haplotypes have been deposited in GenBank. Additionally, as for the other PCR-based assays, colony identification using the DraI test can be performed in a small-sized laboratory equipped with basic equipment (Table 5), and it only takes two days for haplotype identification. A detailed protocol to implement the DraI test (from DNA extraction to digestion with DraI) is available in the BEEBOOK paper on molecular methods (Evans et al., 2013).

Sequencing is the ultimate method for assessing mtDNA variation. Sequence data generate haplotypes that are identical by descent instead of identical by state, required for phylogenetic analysis (Arias and Sheppard, 1996). Sequencing has been employed mostly for characterizing novel COI-COII/DraI haplotypes (Franck et al., 2001; Collet et al., 2006; Shaibi et al., 2009; Pinto et al., 2012). However, it should be employed more often in mtDNA surveys, because by providing further resolution may reveal diversity patterns that would otherwise have gone undetected (Muñoz et al., 2009; Martimianakis et al., 2011).

While important insights into patterns and processes of honey bee maternal diversity across natural and introduced ranges have been gained studying mitochondrial DNA variation, the sole use of this molecule for inferring honey bee evolutionary and human-mediated contemporary history (e.g. introgression) and for subspecies identification is questionable. First, because there is no single mtDNA assay that is subspecies diagnostic. Second, because it only provides the maternal component of variation and is therefore unable to detect hybridization or introgression events, which are becoming increasingly common with human-assisted gene flow through queen trading and colony transhumance. Therefore, a full and accurate identification requires employment of both mtDNA and nuclear markers (see also the BEEBOOK paper on molecular methods (Evans et al., 2013)).

Table 4. PCR-RFLP assays that have been used to identify honey bee matrilineal origins. 1Intergenic region also named tRNAleu-cox2. This intergenic region has generated over 100 distinct haplotypes. Therefore, a simple allocation into two categories, as given in this Table, is not possible. This assay is recommended for honey bee maternal identification.

Genes

Restriction Enzyme

Cleaved fragment

Uncleaved fragment

Authors

Cytochrome b

BglII

European

African

Crozier et al., 1991

Ls rRNA

EcoRI

A. m. ligustica,
A. m. carnica,
A. m. caucasica

A. m. mellifera,
A. m. iberiensis
of lineage M

Hall and Smith, 1991

COI

HincII

A. m. mellifera,
A. m. iberiensis of lineage M

A. m. ligustica,
A. m. carnica,
A. m. caucasica

Hall and Smith, 1991

 

XbaI

A. m. ligustica,
A. m. carnica,
A. m. caucasica

A. m. mellifera,
A. m. iberiensis
of lineage M

Hall and Smith, 1991

 

HinfI

A. m. lamarckii

Non-A. m. lamarckii

Nielsen et al., 2000

 

NcoI

A. m. macedonica

A. m. adami
A. m. cecropia
A. m. cypria

Bouga et al., 2005
Stevanovic et al., 2010

 

StyI

A. m. macedonica

A. m. adami

A. m. cecropia

A. m. cypria

Bouga et al., 2005

Stevanovic et al., 2010

 

SspI

Greek

Bulgarian

Ivanova et al., 2010

ND5

AluI

A. m. macedonica

A. m. adami

A. m. cecropia

A. m. cypria

Bouga et al., 2005

 

HincII

Greek

Bulgarian

Ivanova et al., 2010

 

FokI

Greek

Bulgarian

Ivanova et al., 2010

1COI-COII

DraI

 

 

Garnery et al., 1993

 
Table 5.
Summary of characteristics of the different methods used for identification of honey bee subspecies.

Characteristic

Morphometrics

Allozymes

MtDNA

Microsatellites

SNPs

Number of individuals per colony

10-15

10

1

1 or more (depending on the goal)

1 or more (depending on the goal)

Characters/loci usually screened

Up to 41;

wing venation

MDH, ME, EST, PGM, HK, ALP

COI-COII/DraI, COI/ NcoI/StyI/ SspI, ND5/AluI/ HincII/FokI, 16s rDNA/EcoRI

Hundreds available. However, for most studies less than 20 screened (e.g.  A7, A24, A28, A88, A113, B124, Ap43, A14, A107, A35, Ap55, Ap66)

 

Hundreds (1536 for Golden Gate Assay of Illumina) to thousands (with the Infinium Assay of Illumina)

Inheritance

Biparental

Biparental

Maternal

Biparental

Biparental

Dominance

 

Co-dominant

N/A

Co-dominant

Co-dominant

Polymorphism

 

Low

Very high in COI-COII intergenic region, otherwise medium to low

Very high

Can be high

Number of alleles

 

Multi-allelic

Multi-allelic

Multi-allelic

Biallelic

Abundance in the genome

 

Low

 

Medium

Very high

Cross-lab/study comparisons

Cross-checking recommended

Easy

Easy

Requires special preparation and cross calibration

Easy

Time to complete lab protocol

Depends on character suite typically 1 sample per day for full suite

1 day

Depends on assay, up to 2 days

1 locus or one multiplex up to 2 days

3 days

Main software packages

SPSS, Systat, Statistica, Morpheus, NTSYS, MORPHOJ

GenAlex,  Genepop, and others

GenAlex,  Genepop, Network, Structure, and others

Genepop, Arlequin, Structure, GenAlex, GeneClass, Adegenet, and other R packages

Plink, Structure, Admixture

Main Equipment

Microscope, camera, measuring software, computer

 

Centrifuge, Electrophoresis Unit, incubator

Thermal Cycler, Electrophoresis Unit, Centrifuge, Water Bath

Thermal Cycler, Electrophoresis Unit,

Centrifuge, Automated Sequencer

Thermal Cycler,

Analyst Plate Reader, Hybridization Oven,
Bead Array Reader

Cost of equipment

Low

Low

Medium

High

Very high

Cost of genotyping

Low

Low

Medium

High

Very High