by William C. Roberts and Otto Mackensen
U.S.D.A., Agr. Res. Adm., Bureau of Entomology and Plant Quarantine*
(*In cooperation with the Wisconsin Agricultural Experiment Station and Louisiana State University.)

III. Sex Determination and Bee Breeding

EVERY beekeeper knows that the economically productive colony has a large population of worker bees. In order to provide this force the queen must lay a large number of fertile eggs, and they must hatch and be nurtured to develop into adults.

Many years ago Dzierzon discovered that male bees develop from unfertilized eggs. Drones have a mother but no father. This is known as parthenogenesis. It is not peculiar to bees, for many other animals
have this method of reproduction. The female bees, workers and queens, develop from fertilized eggs. They have both father and mother. From each parent they receive 16 chromosomes and so have 16 pairs of chromosomes. The drone, who receives all his inheritance from his mother, has only 16 single chromosomes. Having paired chromosomes, the workers and queens are called diploid individuals. Since the chromosomes of drones are not paired, they are called haplold individuals.

Recent work indicates that the fundamental determiner of sex in bees is not the number of chromosomes or whether or not the egg is fertilized. Sex is determined by the action of certain genes at one locus on one pair of chromosomes. In the wasp Habrobracon, a relative of the honey bee, the females are diploid and the males are normally haploid, but under certain genetic conditions diploid males have been produced. No diploid males have been discovered in honey bees, but there is experimental evidence that sex determination in honey bees is similar to that in Habrobracon.

The position of a gene on a chromosome is called a locus. On one of the chromosomes there is a locus that is called the X, or sex-determining, locus. At this locus in honey bees there is a series of multiple allelic genes - instead of only two alleles, such as capital W and small
w, which are alleles to each other, there are many alleles. We will call this the X locus and the alleles Xa, Xb, Xc, etc. To simplify matters we can drop the X and use only the letters a, b, c, etc., remembering always
that these are the X alleles.

According to the theory of sex determination in bees, females develop from fertilized eggs that are heterozygous at the X locus. These eggs have two unlike alleles, such as ac, ad, bc, bd, etc. Any fertilized egg that happens to be homozygous at this locus would be a drone if it developed. However, it does not develop but dies instead. Thus all fertilized eggs that are homozygous - that is, having two sex alleles alike, such as aa, bb, cc, etc. - are lethal and do not hatch. Drones develop from unfertilized eggs and are haploid, having only one of these alleles - a or b or c or d, etc.

If a queen with sex alleles ab is mated to a drone having sex allele c, the fertilized eggs from this queen will be ac and be. Since these eggs are heterozygous, their hatchability is near 100 per cent. An ab queen mated to an a or b drone has fertilized eggs that are ab and aa. Since the aa eggs do not hatch, the viability of fertilized eggs from this queen will be only 50 per cent.

It can thus be seen that egg hatchability and the brood quality of a queen are determined by the sex alleles of the queen and the drone or drones that mate with this queen. To illustrate this point let us set up a breeding example. A beekeeper chooses a breeder queen from which he will raise a number of daughter queens. This queen has high-quality brood because she mated with a drone or drones whose sex alleles were different from hers. Since the queen is diploid and heterozygous for the sex-determining locus, we will assume that she is ab. Having good-quality brood, we know that she mated with a drone or drones that were not a or b. Let us assume that she mated with two drones, one c and one d. Fertilized eggs from this queen will be ac, ad, bc, and bd. Notice that all are heterozygous for the X locus, an indication of high hatchability and good-quality brood. Let us further assume that these queen daughters ac, ad, bc, and bd are allowed to mate with drones from another unrelated queen, which we will designate as yz, and that each queen mates with two drones - that is, drones y and z, y and y, or z and z. Their fertilized eggs will be either ay, cy, by, and dy or az, cz, bz, and dz. Consequently the brood quality will be good because the queens and drones have different sex alleles.

Let us assume that the next year the beekeeper raises queen daughters from one of these queens and drones from one of the other queens and allows them to mate, together. These are called matings of first cousins. The sex alleles of the daughters of ac x yz will be az, cz, ay, and cy in
equal proportions, if the queen is ac and mated to one y and one z drone and received equal amounts of sperm from each.

Assume that the queen selected to produce the drones for mating with these virgins had sex alleles ad and that she was mated to z and y drones. Since drones develop from unfertilized eggs, all drones from an ad queen will be either a or d. If a large number of drones are produced, it is likely that half of them will be a and the other half d. If each virgin - az, cz, ay, or cy - mates with two drones, matings will be to drones a and d, or a and a, or d and d.

The queen az that mated with one a and one d drone will produce the
following types of fertilized eggs with equal frequency: aa, ad, za, and zd. Since ad, za, and zd are heterozygous, they hatch, but the homozygous aa fertilized egg dies. Thus one fourth of the fertilized eggs produced by this queen fail to develop. Consequently she has low-quality brood and does not develop a colony with a large population of worker bees. This queen would have produced high-quality brood and a large colony if she had mated with two d drones instead of one a and one d, but only 50 per cent of her eggs would have hatched if she had mated with two a drones.

It is quite evident that the productivity of a colony or of a queen depends to a great extent on the sex alleles represented in the queen and in the drones that mated with the queen. Three sister queens can be identical genetically, and yet if each queen is mated to two drones of a group of brother drones, their egg hatchability can be 50, 75, or 100 per cent, depending upon the sex alleles in these drones. Other things being equal, if only 50 per cent of the fertilized eggs of a queen hatch, the colony is very unproductive. On the other hand, if the eggs are 100 per cent hatchable, very populous colonies will result. If these queens
should mate only once, there would be only 50 per cent and 100 per cent hatchability of fertilized eggs from the several queens, with no 75 per cent hatchability matings.

What then can be expected of a breed-improvement program attempted by a queen breeder? Let us assume that from his own or selected stocks he chooses three mated (tested) queens as the source of all future breeding stock. He will try to improve his stock by breeding the best to the best. He will control matings by isolation so that they will be between progenies of these queens. It is assumed that each of these queens has different sex alleles. If each queen is mated to two drones and all these drones have sex alleles different from any of those in the three selected queens, there will be four different sex alleles in the progeny of each breeder, and a total of 12 alleles for all breeders. These sex alleles are given the following distribution:

Queen No.	Queen	Drones Mated to Queens	Daughter Queens
1	ab	c and d	ac, bc, ad, bd,
2	ef	g and h	eg, eh, fg, fh,
3	jk	l and m	jl, jm, kl, km

Each of these queens with her offspring is a family. From each family the queen breeder may produce drones and queens the first year and allow the matings to occur at random or he may produce only drones from one or two families and queens from the other one or two families. Furthermore, he may segregate the drones and thus divide his population into three mating groups. Now it can be shown that the relationship of matings for the third year is essentially the same by any of these methods. In the first two years the breeder has an opportunity to select between three groups of sister queens and the three groups of brother drones produced by these queens.

If the breeder has a separate yard for drones of each family and allows queens from one of the other families to mate at this isolated location, he can avoid poor viability of brood due to the mating of individuals having sex alleles in common for the first two years only. This can be illustrated as follows: First-year daughters of queen 1 mate with drones of queen 2; second-year queens (1 x 2) mate with drones from family 3. The third-year queens (1 x 2) x 3 must mate to drones produced by (2 x 3), (1 x 3), or (1 x 2) queens. A queen (1 x 2) x 3 will have one sex allele from the 3 family and the other from either the 1 or 2 family.

Let us assume that a daughter queen has one allele from family 1 and one allele from family 3. She is designated as having sex alleles ak. She could have any one of several combinations, but she must have one allele of the four possessed by family 3 - that is, j, k, l, or m. The other of her sex alleles must come from family 1 or 2 and could be any one of the eight alleles originally present in these lines.

The drones with which this queen mates are sons of (2 x 3) queens. If a large number of (2 x 3) queens produce these drones, then the e, f, g, h, j, k, l, and m drones will occur with equal frequency. If this ak queen mates with only one drone and that drone happens to be k, then the eggs of this queen will be only 50 per cent hatchable, since the fertilized eggs will be ak and kk in equal proportions, and all kk eggs will fail to hatch. However, should she mate with a single drone having the sex allele h, her eggs would then be 100 per cent hatchable.

If a large number of queens ((1x2) x3) mate with (2x3) drones, half of them with only one drone and the other half with two drones, then 11.3 per cent of the queens will have 50 per cent viable brood, 14.8 per cent of them will have 75 per cent viable brood, and 73.8 per cent will have 100 per cent viable brood.

If the queen producer has maintained three separate groups, each group will be similar to the above, for ((2 x 3) x 1) daughters will be mated to drones produced by (3 x 1) queens and ((3 x 1) x 2) daughters will mate at the yard that produces (1 x 2) drones. The ratio in each yard will average the same for brood viability.

If these stocks are maintained separately or united into one breeding population, the viabilities will be similar in all yards after the second year. Only 12 alleles were present originally in the stocks, and as long as these are present in equal ratios, the brood viabilities of mated queens will approximate the figures given above (also see Table I). These percentages are based on the assumption that half of all queens mate once and the other half mate twice. If the proportion differs, the percentage of 50 per cent viable matings will vary accordingly, but the percentage that will give 100 per cent viable brood will not change.

It is thus evident that one fourth of all mated queens will produce brood of low viability (50 or 75 per cent egg hatchability) after inbreeding begins, if attempts are made to improve a line by breeding within a closed population started from three selected breeder queens. During the first two seasons there is no inbreeding, for the three lines are being crossed only with each other.

The queen breeder has an opportunity to select for desired type or color, but cannot increase the percentage of matings with high brood viability because he does not know what alleles are carried by each queen.

It can be seen from Table I that the percentage of matings that give high-viability brood increases wlth the number of sex alleles in the breeding population. However, even with 40 sex alleles in a breeding population, approximately 5 per cent of all single-mated queens and 10 per cent of all double-mated queens will produce 50 or 75 per cent of viable brood. The only sure way to have high-viability brood from all queens is to mate queens of one sex-allele combination with drones originating from queens having other sex alleles.

Figure 1. Brood of high viability - 96 per cent. Queen and drones carried unlike alleles, for example, ab queen mated to c and d drones.

Figure 2. Brood of low viability - 48 per cent. Queen and drones carried same sex alleles, for example, cd queen mated to c and d drones.

Insofar as other characteristics of economic importance are concerned, little progress can be made after the first few generations by selection within a closed population. Homozygosis, or fixation of genetic characteristics, will increase slightly each generation in the first few generations after inbreeding is begun. The strain of bees developed at the end of 10 to 15 years of breeding by this method will still be variable. However, selections will have reduced some of the variability for
visible traits such as color, and the stock will appear more nearly uniform than at the beginning of the breeding program. Progress may also have been made in selecting for temper and other highly heritable characteristics. However, it is very doubtful whether significant progress can be expected in characteristics such as vigor or honey production for, like most economically important characteristics, they have low heritability. Characteristics that are highly heritable, such as color, are varied only slightly by environmental factors. Low-heritability characteristics, such as egg production, are varied considerably by the environment.

Table 1. EXPECTED BROOD VIABILITY WITH RANDOM MATINGS IN POPULATIONS HAVING EQUAL FREQUENCIES OF VARIOUS NUMBERS OF SEX ALLELES
Number of Sex Alleles	Single- Drone Matings		Two-Drone Matings			Matings Half Single and Half with Two Drones
	50%	100%	50%	75%	100%	50%	75%	100%
2	100	0	100	0	0	100	0	0
4	50	50	25	50	25	37.5	25	37.5
6	33.3	66.7	11.1	44.4	44.4	22.2	22.2	55.6
8	25	75	6.3	37.5	56.3	15.6	18.5	65.6
12	16.7	83.3	2.8	27.8	69.4	9.7	13.9	76.4
16	12.5	87.5	1.6	21.9	76.5	7	10.9	82.1
20	10	90	1	18	81	5.5	9	85.5
24	8.3	91.7	.7	15.3	84	4.5	7.6	87.9
28	7.1	92.9	.5	13.3	86.2	3.8	6.6	89.6
32	6.3	93.7	.4	11.7	87.9	3.3	5.9	90.5
36	5.6	94.4	.3	10.5	89.2	2.9	5.2	91.9
40	5	95	.2	9.5	90.3	2.6	4.8	92.6
n	2 - n	n-2 - n	(2)2 - n	2(2) - n (n-2) - n	(n-2)2 - n

It should be pointed out that the small real gain by this method of breeding is made at the expense of brood viability, owing to matings of relatives having sex alleles in com-mon. If the breeder starts with three selected queens and mates their offspring together, one fourth of all matings will result in brood of poor viability. This is a high price to pay for commercial experiments in bee breeding.

The bee breeder is thus unable to establish a superior breed of bees by using the methods that have been successful with cattle and hogs. Breeding of the best to the best and continuing with their descendants results in poor-quality brood in a large number of matings. The animal breeder fixed his breeds by inbreeding within closed populations. The bee breeder using this method fixes some characteristics, but cannot fix the breed for the important characteristic of high egg hatchability. This characteristic is necessary for quality brood and thus populous colonies.

The next article will tell how the bee breeder can insure high-quality brood in all colonies by adopting a breeding plan based on controlled hybrids.