3.3.2.2. Problems and potential pitfalls with using microsatellites

Homoplasy, the evolution of identity in state is one of the biggest problems when using DNA microsatellites. The rate of mutations in DNA microsatellites is high, due to replication slippage, and hence homoplasy is a factor of importance. Estoup et al. (1995b) demonstrated a method to quantify this process in honey bees, using interrupted microsatellite loci and bees of various subspecies. Their results show that alleles of identical length can in fact be differentiated by sequencing. Within the major (molecularly defined, see Annex) lineages A, C and M, alleles with identical length tend to also have identical sequences, but variation between subspecies is frequent. Detailed studies demonstrate that 7 different alleles all have the length of 224 in the microsatellite locus A113 (Viard et al., 1998) in eight populations. Most studies on DNA microsatellites are limited to determining the length of each allele, which will lead to an underestimation of allelic richness. This is an important caveat to consider when interpreting the results of studies between honey bee populations belonging to different lineages.

It is common in studies of bee populations to find a lower level of observed heterozygosity than expected. The term null alleles is used to describe alleles that will not yield a PCR product (Chapuis and Estoup, 2007). This is probably due to a mutation at the primer binding site. The bee will appear homozygous if a null allele occurs together with a normal allele. If the null allele occurs at a high frequency in a given population, the observed heterozygosity will decrease. However, in general there is little testing performed for the presence of null alleles. It is possible to detect the presence of null alleles directly, via the observation of parent offspring triads. This process is comparatively easy in honey bees due to the haploid state of the drone, however rarely pursued. Alternatively, a more indirect method was established in the software packages Genepop (Raymond and Rousset, 1995), and FreeNA (Chapuis and Estoup, 2007), using the presence of homozygous null alleles in individuals and the increase of homozygosity in various populations to estimate the frequency of null alleles. Not surprising, it is possible to show that the frequency of null alleles varies across a range of populations and loci (Kryger, unpublished data).

In order to determine the subspecies status of an individual honey bee, a honey bee colony, or a honey bee population, it is important to compare the results to reference material and published genotype information. Unfortunately, no standard reference material, such as a standard allelic ladder, is available for honey bees as there is for several other species (O’Reilly et al., 1996; Schnabel et al., 2000). No accepted source is available that would provide a standard set of alleles of known length to compare to your results and calibrate your own fragment sizes against. Even the use of custom made oligos of known length is not an ideal remedy, since their run times may vary depending on the base composition of the fragment. The lack of organization amongst the scientists studying honey bee populations has resulted in the use of a large variety of loci with differently labelled primers. As a result, the data from independent studies are at best difficult to compare. This is a serious limitation, and we strongly urge the community to develop commonly available standards, to increase reproducibility and comparability of data between labs and studies.