The B4 project
The aim of this project was to ascertain the genetic make-up of honey bees kept and managed by beekeepers from the B4 project in Cornwall. We investigated whether honey bees kept by B4 beekeepers could be genetically distinguished from locally managed honey bees. In addition we investigated how the genetic make-up of both honeys bees kept by B4 beekeepers and locally managed honey bees compared to the wider Apis mellifera mellifera population.
Twenty paired samples of queen honey bee wing clippings were taken and analysed from colonies either managed within the B4 breeding programme (B4) or not managed within a breeding programme (non-B4) in Cornwall (Table 1). Three additional samples were received without pairs; one from a B4 colony (sample 20) and one from non-B4 colonies (sample 1A), giving a total of 21 B4 colonies and 21 non-B4 colonies. All samples were provided by Andrew Brown.
Within the paired samples (B4 and non-B4), wing clipping were taken from queens that were present at the same time to avoid cross breeding between the two groups and the distance between B4 and non-B4 colonies ranged from within the same apiary to 1.8km apart. Wing clippings were collected either dry or in 90% ethanol and all samples were stored at -70°C prior to analysis
Extraction of DNA from the wing clippings followed published Chelex methods (Chaline et al., 2004: Gregory & Rinderer, 2004) which had been shown by the authors to successfully extract DNA from this sample type. The method has been previously used at Fera to extract DNA from similar sample types with low DNA content, such as hair follicles. Briefly, the wing clipping was frozen in liquid nitrogen and broken up using a micropestle. One hundred microlitres of 5% Chelex solution (in molecular biology grade water) and 4 µl of 20mg/ml proteinase K was added, followed by incubation at 56 °C overnight on a thermomixer with agitation at 750 rpm. Following incubation, samples were vortexed, heated to 99 °C for 10 minutes, then centrifuged at 8000 g for 3 minutes before removing 80µl supernatant for the SSR amplification.
Samples (3 µl) were amplified for 12 microsatellite marker primer pairs in individual reactions in a total volume of 15 µl, comprising 0.4 µM each forward and reverse primer, 0.06 mM dNTPs , 1.5 mM magnesium chloride, 0.4 U Red Hot Taq DNA polymerase , 1 × Buffer IV (supplied with the Red Hot Taq DNA polymerase), and 4 µg bovine serum albumin. Reactions were run with the following thermal cycling protocol: 94 °C for 5 min, followed by 35 cycles of 94 °C for 30 s, annealing temperature (50 °C or 55 °C) for 30 s, 72 °C for 30 s, with a final stage of 72 °C for 10 min. Use of fluorescent labels on the primers allowed products to be multiplexed before electrophoresis through a 36 cm capillary filled with POP-7 polymer (Applied Biosystems, U.K.) mounted in an Applied Biosytems 3130xl genetic analyser. SSR profiles were analysed using GeneMapper v3.7 (Applied Biosystems, U.K.).
The microsatellite data were investigated for population differentiation between the B4 samples and the non-B4 samples. As the microsatellite data were poor, only FSt and GSt were calculated, as more sophisticated methods (e.g. STRUCTURE, Pritchard et al 2000) seemed inappropriate. The data were also plotted on a Factorial Correspondence Analysis (FCA) plot to visualise possible relationships between the groups. The above analyses were done in Genetix 4.05.2. (Belkhir et al 2004).
In addition, the subspecies identity of the individual bees were investigated using an existing reference database of bees genotyped at Fera, including reference groups of Apis mellifera melifera (from France, UK – Northumberland, UK – Colonsay), A. m. carnica (from Hawaii, Slovenia and Greece), A. m. ligustica (from Hawaii, New Zealand, Australia), a bee subspecies from Greece (possibly cecropia) and A. m. iberiensis from Spain. This was done in BAPS6 using the “Admixture based on pre-defined populations” tool, using 500 iterations, 100 individuals per reference population and 20 individuals per reference individual.
Table 1. The bees analysed by Fera for the B4 project.
There was poor amplification of the microsatellites from the samples, likely because the amount of DNA extracted from the wing clips was insufficient. Of the 44 samples, 11 samples failed completely, five from the B4 bees (numbers 7, 13, 14, 16, 21) and six from the non-B4 bees (10A, 16A, 17A, 21A, 3A, 7A), leaving 33 samples with data of any kind. Of this remainder, a further five individuals (3, 6, 10, 19, and 20) had fewer than five successful microsatellite scores. Within the remainder, there was a bias towards better amplification (higher number successful microsatellites) in the non-B4 bees (8.2 successful microsatellites out of 12) compared to the B4 bees (5.7 successful microsatellites out of 12).
The pairwise measures of population differentiation for the dataset including all samples that had at least one microsatellite score was FSt =-0.00115 (with the confidence intervals -0.04529 to 0.03551), and pairwise GSt = 0.0447. Excluding all samples with fewer than 5 successful microsatellites, FSt = -0.00902 (-0.05497 – 0.02742) and GSt = 0.0440. The FCA visualisation of the relationship between the individuals (coloured by batch) is shown in Figure 1 for the data with 5 or above successful microsatellite markers.
Figure 1. Factorial Correspondence Analysis visualising the relationship between the microsatellite genotypes of individuals (for those with 5 or more successful microsatellites). In yellow are the B4 bees, in blue are the non-B4s. The axes are the first two primary axes of variation. There is no obvious clustering of individuals by group (B4 versus non-B4).
For the BAPS admixture analysis, the results are summarised in Figure 2 and Table 2. There is a difference between the proportion of A.m. mellifera ancestry (France, UK- Colonsay, UK – Northumberland) between the B4 bees and the non-B4 bees, with the B4 bee samples having a greater proportion of A.m. mellifera ancestry. For samples with 7 or more successful microsatellite genotypes, the average A. m. mellifera ancestry is 0.67 and the average non-B4 is 0.47; however, the numbers are not significantly different when tested with a one-tailed T-Test (P>0.076), although the number samples included in the test is fairly low (seven individuals for B4, 11 for non-B4).
Figure 2. BAPS6 estimate of the ancestry of the bees with more than 7 microsatellite markers. Each bar is an individual (sample reference code beneath each bar), and the proportion of the amounts of the different colours in the bar is the proportion attributable to that sub-species. Blue is A.m. mellifera ancestry, pink A.m. carnica, orange A.m. ligustica and yellow A.m. iberiensis and the Greek bees. The greater the amount of blue in a bar, the greater the A. m. mellifera ancestry that individual has.
Sample Mellifera Northumberland Mellifera Colonsay Mellifera France Overall mellifera ancestry Carnica Ligustica Other
B4 -2 0.04 0.54 0.35 0.93 0.02 0.05 0
B4 -8 0.26 0.35 0 0.61 0.08 0.04 0.27
B4 -11 0 0.35 0.32 0.67 0.06 0.27 0
B4 -12 0.11 0.43 0 0.54 0.08 0.17 0.21
B4 -15 0 0 1 1 0 0 0
B4 -17 0 0.16 0.06 0.22 0.08 0.52 0.18
B4 -22 0.27 0.37 0.02 0.66 0.04 0.29 0.01
Average B4 (scores above 7/12) 0.05 0.19 0.10 0.66 0.05 0.19 0.10
Non- B4 12A 0.3 0.14 0 0.44 0.25 0.21 0.1
Non- B4 13A 0.31 0.12 0 0.43 0 0.33 0.24
Non- B4 14A 0.29 0.18 0.14 0.61 0.04 0.35 0
Non- B4 15A 0 0 0 0 0 0 1
Non- B4 19A 0.21 0.32 0.01 0.54 0.09 0.25 0.12
Non- B4 1A 0.17 0.03 0.03 0.23 0.32 0.18 0.27
Non- B4 22A 0.04 0.69 0 0.73 0.23 0.04 0
Non- B4 4A 0 0.61 0.22 0.83 0 0.17 0
Non- B4 5A 0.11 0.53 0.08 0.72 0.04 0.11 0.13
Non- B4 8A 0.46 0.03 0.05 0.54 0.12 0.3 0.04
Non- B4 9A 0.02 0.09 0 0.44 0.13 0.72 0.04
Average non-B4 (scores above 7/12) 0.17 0.25 0.05 0.47 0.11 0.24 0.18
Table 2. the proportion of Apis mellifera mellifera ancestry per individual bee, as estimated using BAPS6 (see methods section). Samples with 7 or more successful microsatellites are shown.
The data suffered from the poor extraction of DNA from the wing clippings and the subsequent poor amplification of the microsatellite markers. There was no significant population differentiation detectable between the two groups (B4 and non-B4 bees) based on the FSt and GSt statistics (the FSt confidence intervals overlapped 0), or in the visualisation of the data in the Factorial Correspondence Analysis. However, these figures are population level numbers which rely on large numbers of samples, and the lack of data makes these results difficult to interpret, as medium to small population differences would not be apparent with the amount of data available.
A second form of analysis, the ancestry estimates from BAPS6, are done at the individual level and therefore can be reliably derived for individuals with an adequate number of microsatellite scores (here taken as 7 or more markers). Using this data, it is possible to look at the sub-species affiliation of individual bees, and to see a trend towards greater A.m. mellifera ancestry in the B4 bees than the non-B4 bees. However, this trend was not statistically significant. It should be noted that this analysis is limited by the content of the reference database the bees were compared to; there could, for example, be a genetic component of a different bee subspecies in some of the individuals sampled, but this will remain undetected if that subspecies is not present in the reference database. In addition, if a bee is allocated an ancestry that is, for example, French A.m. mellifera, it does not necessarily mean that that bee has French ancestry but that its ancestry most closely resembles French A.m. mellifera than that from Colonsay or Northumberland.
Belkhir, K., et al. (2004)”GENETIX 4.0. 5.2., Software under Windows™ for the genetics of the populations.” University of Montpellier, Montpellier, France.
Chaline, N., et al. (2004). Non-lethal sampling of honey bee, Apis mellifera, DNA using wing tips. Apidologie 35: 311-318.
Corander, J., Marttinen, P., Sirén, J., & Tang, J. (2013). BAPS6: Bayesian analysis of population structure. Manual version 6.0.
Gregory, P. G. and Rinderer, T. E. (2004). Non-destructive sources of DNA used to genotype honey bee (Apis mellifera) queens. Entomologia Experimentalis et Applicana 111: 173-177.
Pritchard, J. K., Stephens M., and Donnelly, P. (2000)”Inference of population structure using multilocus genotype data.” Genetics 155: 945-959.