What the evidence says about breeding honey bees, from my perspective in the UK
Given that we are now into the autumn (fall, for those over the pond) in my area, there is not a great deal of beekeeping for me to do. I could write about some actual beekeeping, but have decided to turn to a topic that I keep coming back to. In the end, once you have figured out how to keep bees, success basically comes down to the weather, which is a big factor that even a walrus cannot influence, and the quality of the bees. Some bees are just better than others; they fly earlier and farther, make more honey, swarm less, rarely sting, stay healthy (with management), and come out of winter bursting with energy and ready to take on the new season. Those good ones are the ones I try to keep, but some people may prefer something else, which is their prerogative. My goal is to produce honey for my customers efficiently and profitably, not to transform the local gene pool.
Different beekeepers, similar goals
It’s useful to compare my methods and beliefs with those of another chap called Jo Widdicombe, about whom I have written previously. He is based on the south coast in Cornwall, near Plymouth, whereas I am somewhere in the middle of the country, with lakes nearby instead of the sea, although here in Cheshire, they tend to be called “meres”. We both want calm, productive bees that don’t swarm as soon as the sun comes out (something more likely to occur in Jo’s area than mine).
Jo breeds from native and near-native stock. His argument, as reported from a talk to the Scottish Native Honey Bee Society, is that “Once you have a more uniform strain, it is easier to breed true” (SNHBS). I often buy instrumentally inseminated breeder queens, graft from them, and open-mate the daughters in an apiary where I try to make sure most of the drones that virgins mate with are mine. I also graft from exceptional queens that materialise in my apiaries from time to time. Having said that, over time my queens (and associated colonies) have tended towards being quite similar in nature.
I think if I lived and worked where Jo does, I might well have adopted his approach. He is on the Rame Peninsula, quite isolated from other beekeepers, and there are plenty of near-native bees in that part of the country. Am I wrong, doing what I do, rather than trying to follow Jo? I don’t think I am, but one can never be certain. I have always found that my average “local bee” is a bit below par, being more defensive, more drippy on the comb, and more keen to swarm even when given space.
Breaking it down
The following three things tend to get tangled up when discussing this topic:
- Purity: is the bee Apis mellifera mellifera (Amm) or something else?
- Predictability: will the daughters resemble their mother?
- Performance: do they measure up to what beekeepers want (good traits)?
If I wanted to coin a catchy phrase, I could call these “the three Ps”, but that immediately triggers images of urinals in my twisted mind, so I shall restrain myself.
When I visited Jo in 2025, I found that we agreed on many things. In his SNHBS talk, he says that “by repeated selection, standards can be raised”. Colonies with a bad temper or a non-native look are “not allowed to put out drones in the area where you are trying to mate your queens.” I agree with that, apart from the non-native appearance part, and I quickly kill off any rogue queens heading evil colonies.
The real difference boils down to the choice of breeding material.
Uh oh, genetics explanation incoming
If you are learning anything about genetics from me, you’re in trouble. It’s a bit like quantum physics to me; okay up to a point, then my head explodes. But that’s not a reason to give up. Here’s what I think I know:
Traits aren’t inherited. Genes are.
Most behavioural traits, such as gentleness, depend on the interaction of multiple genes, plus the environment. That trait cannot simply be passed on to the next generation. Using the analogy of playing cards: the genes are passed on via reproduction, but for each generation they are reshuffled and dealt out, so each hand is different. That trait has to be rebuilt each time from the new hand. The environment plays a big role too (weather, forage, beekeeper).
Each gene passed on may have an additive effect on the trait, pushing it up or down the scale, and that part is inherited. However, the way two or more genes interact in combination is broken up every time a queen makes eggs. That might be why the daughter queens from an exceptional queen are often not quite as good, and the daughters from a poor queen can be better. They tend towards the population average, but there are always outliers.
“Breeding true” is what you get when a whole population draws on the same pool of gene variants, so reshuffling changes little. Purity is one route to get there, but selection within a closed population over a long period is another. One problem for me is that in my area I cannot establish a closed population, whereas Jo can on his little peninsula.
Bees are different
Queens reshuffle genes a lot, whereas drones can’t reshuffle at all. Queens show “10 times higher levels of recombination (19 cM/Mb) than previously analysed genomes of higher eukaryotes” (Beye et al. 2006). I had to look up eukaryotes, which are animals, plants, fungi, and protists (such as algae and amoebas), i.e. most complex life forms that we are aware of. A drone grows from an unfertilised egg and has no father. His sperm are identical copies of his single set of genes, which is itself a random half of his mother’s. So “selecting drones” really means selecting the colonies that produce them.
Extreme polyandry, drones from a wide area: In a DNA paternity study in the Peak District, “Ninety percent of the matings occurred within a distance of 7.5 km, and fifty percent within 2.5 km.” The furthest was 15 km. Even in the most isolated part of Edale, “sixty percent of all matings were to drones from Edale hives” (Jensen et al. 2005), meaning that 40% were not. Each queen also mated with many drones: the study found an average effective mating frequency of 17.2. Edale is a semi-isolated valley, very different from the open Cheshire plain where I keep bees, so I treat these distances as a rough guide rather than a measurement of my situation. Your neighbours’ bees are in the mix, whether you like it or not. Having said that, apiary vicinity mating (AVM) is a known phenomenon, so, as usual with bees, there is no rule that always applies.
Colony behaviour has two genetic sources. Breeders separate the queen effect (what the queen’s own genes do, through her laying and her pheromones) from the worker effect (the workers’ genes, half from the queen and half from whichever drones she met). In records from 14,650 Austrian Carniolan colonies, “Heritability estimates for worker effect were larger than those for queen effect” for every trait studied (Brascamp et al. 2016). So the old saying that temper comes “from the drone side” is not quite right because drones supply half the worker genes, and the other half is from the queen.
How much can selection achieve?
Here is the equation used by many breeders:
Response to selection (R) = h² × selection differential (S)
Heritability (h²) comes from the additive effects of genes passed on. As explained earlier, the other effects are subject to the reshuffle. The table below shows the h² scores for three traits. For the Austrian study I have used the combined heritability, which is the figure breeders select on. The worker-effect heritabilities on their own were higher: 0.70 for honey yield, 0.41 for gentleness and 0.36 for swarming.
| Trait | Austrian Carnica, 14,650 colonies (Brascamp 2016) | Italian Ligustica, 4,003 records (Andonov 2019) |
| Gentleness / defensive behaviour | 0.37 | 0.36 |
| Swarming | 0.06 | 0.34 |
| Honey yield | 0.27 | 0.26 |
The two studies used different statistical models, so compare within a column rather than across a row.
Here is an example using the Austrian programme’s own scale, where gentleness is scored “on a scale from 1 to 4, rounded to one decimal” and “higher values are desirable” (Brascamp 2016). The numbers that follow are my example, not from the paper. Say your apiary averages 3.2 for that trait, and you choose breeder queens whose colonies average 3.8, six-tenths of a point better. If you also choose drones from colonies that score 3.8 (different line), the resulting daughter queens will score according to this:
Improvement: 0.37 × (3.8 − 3.2) = 0.222
New score for daughter queens: 3.2 + 0.222 = 3.422, or about 3.4 on the one-decimal scale
If you only control the queen side, and leave the drones to chance, the improvement will be half of 0.222. The score of daughter queens should be 3.2 + 0.111 = 3.311, or about 3.3.
That looks feeble, but it adds up. Keep selecting for years and the average climbs, though progress usually slows as the variation shrinks.
I have found that moving colonies towards being less defensive is perhaps easier than making them less prone to swarm, but so many other factors come into play, such as swarm prevention techniques applied in good time. The two studies in the table appear to disagree sharply on swarming, but that is partly down to the type of estimate: the Austrian figure is a combined heritability, and the worker-effect heritability for swarming on its own was 0.36. Different bees and different statistical models make a direct comparison unsafe anyway.
Inbreeding
Continual selective breeding has the effect of narrowing the gene pool of a population. When a virgin queen mates with drones from her own close family, some of her fertilised eggs may carry two identical copies of the sex-determining gene (csd), so the larva becomes a diploid drone instead of a normal worker. These larvae are quickly removed by the workers, resulting in a spotty brood pattern. This is why breeders keep good records (pedigrees) and are careful when selecting drone donor colonies at isolated mating stations or when instrumentally inseminating queens.
In 14 years of keeping bees, I have only once seen the “pepper-pot brood pattern” that can indicate inbreeding. My guess, and it is only a guess, is that the queen mated close to home with very few drones, one of them a close relative. With only a couple of mates, one unlucky match can knock out a large share of the brood, whereas with 17+ it would barely show up. I change the line of my breeder queen every season, which should keep the risk low. It just so happened that when I found that ugly brood in a nucleus colony, I was hosting a visit from my local beekeeping association! The gods play with us mercilessly, do they not?
The case for native bees
BIBBA’s committee statement says, “Locally adapted honey bees have long been known by experienced beekeepers to consistently perform better”. It also says, “random mating with different types of bees creates arbitrary crosses that often result in aggressive behaviour” (BIBBA). They also cite Büchler et al. 2014 as supporting evidence. That trial measured how long colonies survived when left untreated for varroa. Most beekeepers treat and feed their colonies, so it tells us little about how local bees would perform in a managed honey-production operation. The second point, F2 aggression, is a real phenomenon, but not one that I see very frequently.
Local bees survive longer: A European trial tested 597 colonies of 16 genotypes in 20 apiaries. “Colonies with queens from local origin survived 83 days longer compared to non-local origins” (Büchler et al. 2014).
Native bees are not extinct: In Ireland, “97.8% of sampled bees were determined to be pure A. m. mellifera”, from 412 bees across 24 counties (Hassett et al. 2018).
Pockets of Amm survive in Britain and Ireland: A 2024 whole-genome study of colonies across Britain and Ireland found “17 out of 70 colonies showed no significant introgression”. It adds that “The Island of Colonsay (Scotland) and parts of the South West of England also harbour ‘pure’ colonies” (Buswell et al. 2024).
The law already protects one reserve: On Colonsay and Oronsay it is an offence to keep “animals of the genus Apis other than Apis mellifera mellifera” (SSI 2013/279).
On those 17 out of 70, the colonies were taken from beekeepers keeping “non-imported stock”, chosen because they were likely to be native. The study indicates that pure colonies exist, but it doesn’t tell us what share of bees are pure Amm. It will certainly be less than 17 out of 70.
Jo warns about crosses. “Crosses can give enhanced vigour in the first generation, but this is not maintained” (SNHBS). That fits crosses between distinct types of bee, such as a native queen mated to imported drones. Around me most bees are already mongrels, and crossing one set of mongrels with another is not quite the same thing. What I see when my colonies requeen themselves is not a dramatic collapse, but a steady slide back towards the local average, which is just what the genetics section above would predict.
BIBBA’s aims include “Breeding and rearing of locally adapted bees as a stepping-stone towards the restoration of native and near-native bees” (BIBBA). So the destination is native or near-native, not merely local. What BIBBA doesn’t say is that every hive must be pure black. Nor does it define what “near-native” means, but from talking to Jo it would seem that colour is a clear selection tool (anything not black is out of his breeding pool).
If near-native means a dark, locally selected bee with some mixed ancestry, then that is one thing. If it means nearly pure Amm across all of lowland England, I still think that is an almost unachievable goal. Even among colonies sampled from beekeepers keeping non-imported stock, 53 of the 70 in the Buswell study showed some introgression, so hybridisation is widespread.
The case for selecting from what we’ve got
Selection works on heritable variation, and the table above shows plenty of it for temper and honey. Nothing in genetics says a population must be pure before selection can work.
However, every figure I’ve quoted comes from a pure-subspecies programme with pedigree records. I haven’t found a published study measuring progress in an open-mated, mixed population like most UK honey bees. Jo’s point about predictability is true; mixed stock gives more variable daughters, so progress is slower and noisier than with the pure subspecies.
What I do
- Buy instrumentally inseminated breeder queens from strongly selected lines (recently from Sweet Apiaries, Jolanta (Denrosa) and occasionally Northumberland Honey Co.).
- Add any of my queens whose colonies stand out.
- Graft from their larvae.
- Open-mate the virgins in a place where I try to flood the area with drones from my colonies.
- Requeen defensive colonies so they stop putting drones into the pool.
If I can’t control the fathers, I can at least know the mother, and she’s a good one. This does mean that I buy at least one expensive breeder queen each season, but I’m fine with that. As previously hinted, if I leave them to requeen themselves for a few generations (by swarming or supersedure) they become poorer bees. The task of raising the quality of my local bees through selection, in an area with multiple beekeepers and all sorts of mixed-up bees, is daunting. However, my bees are “good bees” and by removing any queens heading defensive colonies I am hopefully contributing something beneficial to the locality.
Where it falls short
My drone flooding attempts are probably ineffective. Austrian researchers put drone colonies on an island in Lake Neusiedl, about 1.8 km from the mainland (to which it is joined by a dam), and mated queens there. On average only 63% of the offspring carried the cordovan colour marker of the island drones, which the authors used as a stand-in for paternity. The authors concluded the island “is not suitable for establishing a mating station for honey bee breeding programmes” (Sprenger et al. 2025). My apiaries are not on an island, and with half of all matings within 2.5 km, my neighbours’ drones get a big say. The reverse is also true: my drones go into the local mix, which is one reason I requeen anything defensive.
Temper has a larger worker effect than queen effect, and half the worker genes come from the drones. My daughter queens vary (not hugely) because their mates vary. I get a mix of colours from the same breeder queen, and maybe about 10% are a little snappier than I’d like. We rarely use smoke, but if the bees are grumpy, we fire up the smoker, and that almost always calms everything down. A real “colony from hell” is very rare for me, but it has happened a couple of times.
Continual purchasing: A potential drawback of what I do is the cost of buying breeder queens every season. It also makes me dependent on a small number of breeders. If their supply or prices changed, my plan would be to use my best stock, and maybe do some queen swaps with other beekeepers.
Full control: the New World Carniolan
The best example I know of selection from mixed stock is Susan Cobey’s New World Carniolan (NWC) programme in the USA. It began in 1981 from queens of mixed US and Canadian backgrounds, run as a closed population with no outside queens added.
Every mating is controlled. Each year 30 to 60 breeders are chosen, and their drones pooled for instrumental insemination. Between 200 and 250 inseminated colonies are then tested and the top third kept. They are scored on brood, temper, build-up, honey, wintering, varroa and hygiene.
Ruttner called them “the most Carniolan like bees” he had seen in North America, though they don’t meet the standard for pure Carnica. Sue Cobey warns that selection narrows the gene pool and risks inbreeding depression. After about 20 generations, the programme brought in Carnica semen from Austria, Germany, and Slovenia to spice up the genetic mix.
So, under certain controlled conditions, uniformity can be built from mongrels. It requires control of both queens and drones, and a broad, genetically diverse founding base of breeder queens. My beekeeping domain is not like that. If I were twenty years younger I think I’d get into instrumental insemination, but it seems a bit late now that I’m 63. There are people using it in the UK to good effect. Jolanta and her team from Denrosa Apiaries went over to Washington State a few years ago to learn from Susan Cobey, so it will be interesting to hear from her about what difference it has made.
What you can do
- If you live in an area where native or near-native bees form the bulk of the population, join the local queen breeding group and get involved. Your drones matter to everyone within at least 7.5 km, so join in rather than go against the flow. Breeder queens can probably be obtained from somebody in your area who is already selecting and breeding queens.
- In an area with very mixed-up bees like mine, rear your own queens rather than buying in mated ones of unknown origin. Graft from a UK-bred breeder queen, from your best colonies, or both. Keep records so that you can track progress over the years.
- Requeen bad-tempered colonies so that their drones are out of the breeding population.