Recombination by crossing-over is a genetic trait and the recombination frequency commonly varies both within and between populations . As a result, the recombination rate can evolve and respond to natural selection. Laboratory experiments have demonstrated that the recombination rate can be selected readily, and directional selection has had a tendency to elevate recombination rates in domesticated animals and plants . The findings that the recombination rate in the honeybees and ants is higher than in other insects [7–11] have led to suggestions that it has been selected for in eusocial insects. Our present results from V. vulgaris are in agreement with this suggestion even though the role of selection still remains to be demonstrated.
Beye et al.  estimated that the average recombination rate in the honeybee (Apis mellifera) genome is 19 cM/Mb with little variation among its 16 chromosomes (but with considerable variation of the local recombination rate along the genome). A somewhat lower estimate (16.0 cM/Mb) was calculated by Wilfert et al. . Comparative genetic maps indicate that the recombination rate in another honeybee species, A. florae, is similar to that in A. mellifera , These estimates are four to five times higher than the estimate from the primitively eusocial bumblebee Bombus terrestris, 4.4 cM/Mb . We earlier constructed linkage maps and estimated the recombination rates as 14.0 cM/Mb in the harvester ant Pogonomyrmex rugosus  and 6.2 cM/Mb in the leaf-cutting ant Acromyrmex echinatior . The present estimate from the wasp V. vulgaris (9.7 cM/Mb) thus falls in between the two estimates from ants.
These estimates from the advanced eusocial bees, ants and wasps are higher than in other insects . The estimates in non-social hymenopterans are within the range 2.5 - 5.4 cM/Mb (4 species of parasitoid wasps) and in other insects 0.1 - 6.1 cM/Mb (15 species) . The recombination rates in the four advanced eusocial species (honey bee, leaf-cutter ant, seed harvester ant, yellow jacket wasp) are significantly larger than in the other hymenopterans (including the bumblebee, probability of no overlap in the values is P = 0.016) or in other insects in general (P = 0.0005). Bees, ants and wasps belong to aculeate Hymenoptera and are thus not phylogenetically independent. However, the lineages have diverged a long time ago and eusociality has evolved separately in them. According to Brady et al. , ants, bees and wasps had a common ancestor about 160 Mya, and the lineages leading to ants and eusocial wasps diverged about 140 Mya. The two ants in which the recombination rate has been estimated had their common ancestor about 80 Mya , and Apis mellifera and A. florea at least 8-10 Mya  and probably about 20 Mya . Even though we cannot exclude the possibility that a high recombination rate is an ancestral state, there has been ample time for selection to modify the rates if they had any adaptive significance.
The use of AFLP-marker data deserves discussion because the methodology has gained criticism e.g. due to frequent occurrence of non-homologous fragments with the same amplicon length [23, 24]. Study on Nasonia jewel wasp  revealed 41.5% shorter map size when SNP-markers were used instead of RAPD/AFLP markers . Similarly in Bombyx mori silk moth recombination map estimates vary from 3432 cM (simple sequence repeat; ) to 1413.4 cM (SNP; ) depending on the method and the number of the markers. On the other hand the original estimate of the honeybee map size was based on RAPD markers (3500 cM ) and the subsequent estimates based on microsatellites (4000 cM ) or genome sequencing (4553 cM ) have not decreased it. There is thus no universal trend that the RAPD/AFLP markers would overestimate recombination, and the data from most insects used in our comparisons were obtained with these methods, making the results comparable.
Sherman  suggested that a high recombination rate could be adaptive in eusocial insects either because recombination equalizes the fractions of genomes shared by colony members or because it generates a larger number of different multilocus genotypes. Sherman particularly considered the advantage of genotypic diversity in caste and task specialization, and Schmid-Hempel  suggested that the same can also apply to defense against parasites. As noted by Schmid-Hempel, both recombination and multiple mating by females increase the genotypic diversity among the offspring and can be beneficial to eusocial insect colonies under selection by parasites. The difference between the two factors is that unlike recombination, mating with many males also increases allelic diversity. Multiple mating by queens is known in many eusocial insects but the average mating frequency is generally rather low . It is noteworthy that the species in which high recombination rates have been estimated, have all monogynous societies, i.e. societies with a single queen. They also have large colonies with clear queen-worker dimorphism and elaborate division of tasks among workers, and the queens are typically highly polyandrous. The estimated number of effective matings is up to 17.6 in the honeybee A. mellifera, 1.9 in Vespula vulgaris, 4.7 in Pogonomyrmex rugosus and 5.3 in Acromyrmex echinatior . These estimates are clearly higher than the mean estimates for eusocial insects in general. One could thus suggest that these species benefit from genotypic diversity within colonies and that this has selected both for polyandry and for a high recombination rate
Sherman  initially hypothesized an association between recombination rate and sociality because of effects on genomic multilocus relatedness. The point is that the expected relatedness among full sisters (r = 0.75) in a single locus is in fact a mean between complete identity (r = 1) when the sisters received an identical allele from the mother and 'half identity' (r = 0.5) when the sisters received different maternal alleles and share only the paternal allele. Lack of recombination could result in genetic cliques within which sisters are unusually highly related over many loci. If there is any kin discrimination within the societies and nepotistic behaviour based on this discrimination, such genetic cliques could lead to nepotistic conflicts and harm the function of the colony. Nepotistic behavior based on kin discrimination has been doubted but some evidence for it has been presented recently [32, 33]. Whether the effect of recombination on the distribution of pair-wise relatednesses among colony members could affect kin recognition and discrimination remains to be studied.