Genetic Diversity in Honey Bee Colonies Enhances Productivity and Fitness

Abstract

Honey bee queens mate with many males, creating numerous patrilines within colonies that are genetically distinct. The effects
of genetic diversity on colony productivity and long-term fitness are unknown. We show that swarms from genetically diverse
colonies (15 patrilines per colony) founded new colonies faster than swarms from genetically uniform colonies (1 patriline
per colony). Accumulated differences in foraging rates, food storage, and population growth led to impressive boosts in the
fitness (i.e., drone production and winter survival) of genetically diverse colonies. These results further our understanding
of the origins of polyandry in honey bees and its benefits for colony performance.

Full text

Genetic Diversity in Honey
Bee Colonies Enhances
Productivity and Fitness
Heather R. Mattila* and Thomas D. Seeley
Honey bee queens mate with many males, creating numerous patrilines within colonies that are
genetically distinct. The effects of genetic diversity on colony productivity and long-term fitness are
unknown. We show that swarms from genetically diverse colonies (15 patrilines per colony)
founded new colonies faster than swarms from genetically uniform colonies (1 patriline per
colony). Accumulated differences in foraging rates, food storage, and population growth led to
impressive boosts in the fitness (i.e., drone production and winter survival) of genetically diverse
colonies. These results further our understanding of the origins of polyandry in honey bees and its
benefits for colony performance.
O
ne of the central challenges for
understanding the evolution of sociality
in bees, ants, and was ps (Order Hyme-
noptera) is the phenomenon of polyandry , or
multiple mating with different males by queens
(1). Selection for polyandry is unexpected because
it generates intracolonial genetic diversity , which
erodes high levels of relatedness among female
offspring, thereby hindering the evolution of
altruistic behavior toward kin. Nevertheless,
polyandry occurs repeatedly in social insects (2)
andtoanextremedegreeineveryspeciesof
honey bee (genus Apis)(3). Several hypotheses
have been proposed to explain how the benefits of
a genetically diverse work force could outweigh
the costs of reduced altruism resulting from low
within-colony relatedness (4). A popular hypoth-
esis suggests that genetically diverse work forces
may operate more efficiently (5) and, consequent-
ly , produce colonies with a fitness advantage over
those with uniform gene pools. However, there is
conflicting evidence that genetically diverse colo-
nies perform tasks better as a collective than
genetically uniform colonies do (6–10 ) and,
furthermor e, enhanced productivity of the work
force has never been linked explicitly with colony-
level fitness gains.
A honey bee colony propagates its genes in
two ways: by producing reproductive males
(drones) and by producing swarms, when a
reproductive female (queen) and several thou-
sand infertile females (workers) leave and
establish a new nest. Swarming is costly and
perilous; with limited resources and labor, a
swarm must construct new comb, build a food
reserve, and begin rearing workers to replace the
aging work force. In temperate climates, newly
founded colonies must operate efficiently be-
cause there is limited time to acquire the
resources to support these activities. Colony found-
ing is so difficult that only 20% of swarms sur-
vive their first year (11); most do not gather
adequate food to fuel the colony throughout the
winter and die of starvation.
W ith the challenges of successful colony
founding in mind, we conducted a long-term
study to compare the development of genetically
diverse and genetically uniform colonies after a
swarming event. Each genetically diverse colony
(n = 12) had a queen that was instrumentally
inseminated with sperm from a unique set of
fifteen drones and each genetically uniform
colony (n = 9) had a queen inseminated with a
similar volume of sperm from a single drone.
Drones were selected at random from a pool of
over 1000 individuals collected from 11 drone-
source colonies. To replicate the experience of
feral colonies, swarms were created by forcing a
queen and 1 kg of her worker offspring (~7700
bees) to cluster in a screened cage for three days,
where they were fed sucrose solution ad libitum
to simulate preswarming engorgement on honey.
Each swarm was subsequently relocated to a
combless hive that was similar to that pre-
ferred by colonies naturally (12). Colonies were
founded on 11 June 2006, during the region’s
swarming season (13). Once swarms were in
their new nest sites, we documented colony de-
velopment by measuring comb construction,
brood rearing, foraging activity, food storage,
population size, and weight gain at regular inter-
vals (14). Intracolonial genetic diversity im-
proves disease resistance (15), therefore colonies
were medicated throughout the study so that we
could examine the effects of multiple patrilines on
productivity and fitness with minimal interference
from the effects of enhanced resistance to disease.
There were notable differences in the progress
of genetically diverse and genetically uniform
colonies during the early stages of colony found-
ing. Colonies with genetically diverse worker
populations built ~30% more comb than colonies
with genetically uniform populations before con-
struction leveled off after 2 weeks [(Fig. 1);
repeated measures ANOVA; F(1,19) = 25.7, P <
0.0001 (colony type); F(19,342) = 126.9, P <
0.0001 (time); F(19,342) = 31.8, P < 0.0001
(interaction)]. During the second week of colony
founding, we compared foraging rates (number
of workers returning to hive per minute for all
workers and for only those carrying pollen) be-
tween different combinations of randomly paired
colonies (one colony from each treatment, n =50
pairs per day) throughout five consecutive
mornings. Genetically diverse colonies main-
tained foraging levels that were 27 to 78% higher
than genetically uniform colonies on three of five
mornings [(Fig. 2); paired t tests with Bonferroni
adjustment; 20 June: t(49) = 4.1, P = 0.0001 and
t(49) = 5.7, P < 0.0001; 21 June: t(49) = 3.2, P =
0.002 and t(49) = 5.5, P < 0.0001; 22 June: t(49) =
5.2, P < 0.0001 and t(49) = 5.8, P < 0.0001].
Moreover , after 2 weeks in their new nest site,
genetically diverse colonies stockpiled 39% more
food than genetically uniform colonies [mean
1390 ± 120 versus 990 ± 145 cm
2
comb per
colony filled with food; t test; t(19) = 2.1, P =
0.045]. This difference was not because some
colonies lacked space (genetically diverse and
uniform colonies had mean 61 ± 1.1% and 66 ±
3.2% of comb empty, respectively); instead, it
was likely a consequence of increased foraging
activity in genetically diverse colonies. The mag-
nitude of these dif ferences in growth during the
initial 2 weeks after colony founding is impres-
sive, considering that work forces in genetical-
Department of Neurobiology and B ehavior, Cornell
University, Ithaca, NY 14853, USA.
*To whom correspondence should be addressed. E-mail:
hrm24@cornell.edu
Fig. 1. Area of comb
(means ± SEM) con-
structed by genetically
diverse (
■
) and ge-
netically uniform (
□
)
colonies after occupy-
ing new nest sites on
11 June. Dates when
groups differed signif-
icantly in comb area
are indicated by a hor-
izontal line (top).
Mean area of comb (cm
2
)
12000
9000
6000
3000
0
13 June
16 June
20 June
23 June
26 June
30 June
4 July
7 July
10 July
17 July
25 July
2 Aug.
7 Aug.
14 Aug.
21 Aug.
30 Aug.
20 JULY 2007 VOL 317 SCIENCE www.sciencemag.org362
REPORTS
on July 20, 2007 www.sciencemag.orgDownloaded from

ly diverse and genetically uniform colonies
were still similarly sized 1 month after
founding [9 July comparison of wo rker pop-
ulat ions; t test; t(19) = 1.6, P =0.12].
One month after “swarming,” there was an
isolated and brief period when abundant forage
became available (~9 to 21 July). At the start of
this honey flow , genetically diverse colonies
were already twice as heavy as genetically uni-
form colonies [(Fig. 3); repeated measures
ANOVA; F(1,19) = 39.9, P <0.0001(colony
type); F(86,1373) = 237.0, P < 0.0001 (time);
F(82,1373) = 17.4, P < 0.0001 (interaction)].
Throughout the flow, genetically diverse colo-
nies gained an average of 0.3 ± 0.05 kg/day per
colony and increased their higher initial colony
weight by 305%, whereas genetically uniform
colonies gained only 0.09 ± 0.01 kg/day per
colony and increased their lower colony weight
by only 163% before resources waned (Fig. 3;
comparison of mean daily weight gain during
flow; paired t test; t(11) = 3.9, P = 0.0007). The
influx of food sparked a resurgence in comb
construction in genetically diverse colonies, how-
ever , comb area remained unchanged in geneti-
cally uniform colonies (Fig. 1).
Production of new workers in genetically
diverse colonies surpassed that of genetically
uniform colonies within the first month of col-
ony development [(Fig. 4); repeated measures
ANOVA; F(1,19) = 63.5, P < 0.0001 (colony
type); F(11,174) = 37.4, P < 0.0001 (time);
F(11,174) = 16.0, P < 0.0001 (interaction)].
Brood rearing by workers increased continually
in genetically diverse colonies until the end of
August, whereas genetically uniform colonies
produced consistently low numbers of workers
over the same period (Fig. 4). Consequently, ge-
netically diverse colonies had far larger worker
populations by the end of August [mean 26,700 ±
1830 versus 5300 ± 2400 individuals per colony;
t test; t(17) = 7.1, P < 0.0001]. These differences
in post-founding development likely resulted
from a combination of an enhanced capacity of
multiple-patriline colonies to construct nest ma-
terials, to rear brood, and to acquire food (given
comparable worker populations) and the momen-
tum that this lent to the pace of colony growth, a
pace that single-patriline colonies were not able
to match despite having similar opportunities
after a “swarming” event.
Colony size is closely tied to fitness; larger
colonies produce more drones, have higher win-
ter survival, and issue more swarms (16–18).
Here, intracolonial genetic diversity resulted in
considerably more populous and resource-rich
colonies, which in turn affected their fitness. Ge-
netically diverse colonies reared significantly more
drones than genetically uniform colonies before
brood rearing declined in September: mean 1910 ±
384 versus 240 ± 109 drones per colony [Fig. 4;
t test; t(19) = 3.7, P = 0.002]. The larger, genet-
ically diverse colonies also collected and stored
more food than genetically uniform colonies
and all survived a late-August cold period that
starved and killed 50% of genetically uniform
colonies (Fig. 3). The remaining genetically uni-
form colonies exhausted their food reserve and
died by mid-December, whereas 25% of genet-
ically diverse colonies survived to May (Fig. 3).
We have demonstrated that the productivity
and fitness of honey bee colonies is enhanced by
intracolonial genetic diversity . Our data confirm
and extend trends toward increased growth re-
ported in short-term studies of polyandrous col-
onies with low (≤6) numbers of patrilines (6, 7).
The benefits of a genetically diverse worker
population are especially evident during colony
founding when survival depends critically on
Fig. 2. Foraging rates
(means ± SEM) of genet-
ically diverse (
■
:all
returning workers;
●
:
only workers carrying
pollen) and genetically
uniform (
□
:allreturn-
ing workers;
○
:only
workers carrying pol-
len) colonies. Asterisks
mark days when daily
foraging rates differed
significantly between
groups.
Fig. 3. Weight (means ±
SEM) of genetically di-
verse (solid line) (n = 12)
and genetically uniform
(dashed line) (n =9)
colonies after occupying
new nest sites on 11
June. Dates when groups
differed significantly in
weight are indicated by a
horizontal line (top). Each
arrow marks the death of
acolony.
0
3
6
9
12
11 June
4 July
24 July
13 Aug.
2 Sept.
22 Sept.
12 Oct.
1 Nov.
21 Nov.
11 Dec.
31 Dec.
20 Jan.
9 Feb.
1 Mar.
21 Mar.
10 Apr.
30 Apr.
20 May
Mean weight per colony
(bees, comb, brood, food) (kg)
0
10
20
30
40
50
19 June 20 June 21 June 22 June 23 June
Mean number of foragers returning
to hive per minute
*
*
*
*
*
*
Fig. 4. Number
(means±SEM)ofwork-
ers (solid lines) and
drones (dashed lines)
produced by genetical-
ly diverse (workers:
■
;
drones:
●
) and genet-
ically uniform (workers:
□
;drones:
○
)colonies.
Acensusofindivid-
uals in capped-pupae
cells was made on each
date; pupae counted at
this time emerged as
adults during the inter-
val between that cen-
sus and the next (14).
Horizontal lines (work-
ers: single line; drones: double line) mark periods when brood rearing differed significantly between
groups.
0
2
4
6
8
10
12
14
16
20 June
2 July
14 July
25 July
7 Aug.
19 Aug.
31 Aug.
11 Sept.
25 Sept.
7 Oct.
21 Oct.
6 Nov.
Mean number of individuals produced
(workers x1000, drones x100)
www.sciencemag.org SCIENCE VOL 317 20 JULY 2007 363
REPORTS
on July 20, 2007 www.sciencemag.orgDownloaded from

successfully accomplishing a variety of pressing
tasks. Given similar numbers of workers, envi-
ronmental conditions, and need, newly founded
colonies built comb faster , foraged more, and
stored greater amounts of food when their work
forces were comprised of many genetically dis-
tinct patrilines. Initial differences in labor pro-
ductivity amplified growth rates over time and
led to dramatic fitness gains for genetically di-
verse colonies (i.e., production of drones, colony
growth, and survival). Thus, we expect intense
selection favoring polyandry because intracolo-
nial genetic diversity improves the productivity
of the work force and increases colony fitness
during the risky process of colony founding.
Higher collective productivity of genetically
diverse colonies may be rooted in a broader or
more sensitive response from worker popula-
tions to changing conditions. The probability
that a worker will engage in a task has been
linked repeatedly to genotype [e.g. (5, 8, 19)].
Consequently, colonies with multiple patrilines
would be expected to have worker populations
that are able to respond to a broad range of task-
specific stimuli and, as a group, should be able
to provide appropriate, incremental responses to
changes in these stimuli (5). The observation
that intracolonial genetic diversity improved pro-
ductivity in colonies is consistent with predic-
tions made by models of division of labor that
rely on genotypic differences in response thresh-
olds among workers (20). Nevertheless, the ex-
tent to which genetically uniform colonies lagged
behind genetically diverse colonies in the early
stages of colony development was surprising,
considering that colonies initially lacked comb
and food reserves, and presumably, stimuli re-
flecting these needs could not have been greater.
Actual response thresholds of workers are not
well documented (20), and it is difficult to know
how they are related to the productivity of indi-
viduals and the colony as a whole. For example,
workers may vary genetically in the rate at which
they perform a task once their response threshold
is reached or they may not be “good” at tasks for
which they have high thresholds (i.e., they lack
physiological apparatuses or experience). Alter-
natively, thresholds may be so high for some
tasks that behaviors are effectively missing from
a worker’s repertoire, thus multiple patrilines
would contribute to the diversity of labor in a col-
ony, rather than division of labor among workers.
A key advantage of intracolonial genetic
diversity was revealed during infrequent periods
when food resources were plentiful (~33 days
during our study). Genetically diverse colonies
gained weight at rates that far exceeded those of
genetically uniform colonies (Fig. 3), whose slug-
gish foraging rates suggest that intracolonial
genetic diversity enhances the discovery and ex-
ploitation of food resources by work forces,
especially during periods when resources become
suddenly and abundantly available. Intracolo-
nial genetic diversity would result in more rapid
mobilization of forager work forces if, by
broadening the range of response thresholds in
colonies, it increased the probability of having
sufficient workers functioning as foragers and/or
broadened the range of conditions over which
foragers inspected, scouted, recruited to or were
recruited/reactivated to food resources. Selection
for polyandry would be strong if the genetic
diversity that it bestows on colonies enhances
the sophisticated mechanisms of honey bees for
recruiting nest mates to food. Because success-
ful colony founding by honey bees depends so
heavily on rallying foragers and the swift ac-
cumulation of resources, this could explain, in
concert with other benefits unrelated to worker
productivity (15, 21), the widespread occurrence
of extreme polyandry in all honey bee species.
References and Notes
1. R. H. Crozier, P. Pamilo, Evolution of Social Insect
Colonies (Oxford Univ. Press, Oxford, 1996).
2. J. Strassmann, Insectes Soc. 48, 1 (2001).
3. D. R. Tarpy, D. I. Nielsen, Ann. Entomol. Soc. Am. 95,
513 (2002).
4. R. H. Crozier, E. J. Fjerdingstad, Ann. Zool. Fennici 38,
267 (2001).
5. G. E. Robinson, R. E. Page Jr., in Genetics of Social
Evolution, M. D. Breed, R. E. Page Jr., Eds. (Westview
Press, Boulder, CO, 1989), pp. 61–80.
6. B. P. Oldroyd, T. E. Rinderer, J. R. Harbo, S. M. Buco,
Ann. Entomol. Soc. Am. 85, 335 (1992).
7. S. Fuchs, V. Schade, Apidologie 25, 155 (1994).
8. J. C. Jones, M. R. Myerscough, S. Graham, B. P. Oldroyd,
Science 305, 402 (2004).
9. R. E. Page Jr., G. E. Robinson, M. K. Fondrk, M. E. Nasr,
Behav. Ecol. Sociobiol. 36, 387 (1995).
10. P. Neumann, R. F. A. Moritz, Insectes Soc. 47, 271
(2000).
11. T. D. Seeley, Oecologia 32, 109 (1978).
12. T. D. Seeley, R. A. Morse, Insect es Soc. 25, 323 (1978).
13. R. D. Fell et al., J. Apic. Res. 16, 170 (1977).
14. Materials, methods, and statistical analyses are available
on Science Online.
15. T. D. Seeley, D. R. Tarpy, Proc. R. Soc. Lond. B. Biol. Sci.
274, 67 (2007).
16. J. B. Free, P. A. Racey, Entmol. Exp. Appl. 11, 241
(1968).
17. T. D. Seeley, P. K. Visscher, Ecol. Entomol. 10, 81 (1985).
18. P. C. Lee, M. L. Winston, Ecol. Entomol. 12, 187 (1987).
19. G. Arnold, B. Quenet, C. Papin, C. Masson, W. H. Kirchner,
Ethology 108, 751 (2002).
20. S. N. Beshers, J. H. Fewell, Annu. Rev. Entomol. 46, 413
(2001).
21. D. R. Tarpy, R. E. Page Jr., Behav. Ecol. Sociobiol. 52,
143 (2002).
22. We thank K. Burke for field assistance, T. and S. Glenn
for rearing queens, and P. Barclay, M. Kirkland, and two
anonymous reviewers for comments on the manuscript.
Funded by a Postdoctoral Fellowship (H.R.M.) from
Natural Sciences and Engineering Research Council
(Canada) and a grant from the National Research
Initiative of the U.S. Department of Agriculture
Cooperative State Research, Education, and Extension
Service (T.D.S.) (no. 2003-35302-13387).
Supporting Online Material
www.sciencemag.org/cgi/content/full/317/5836/362/DC1
Materials and Methods
References
26 March 2007; accepted 8 June 2007
10.1126/science.1143046
PDZ Domain Binding Selectivity Is
Optimized Across the Mouse Proteome
Michael A. Stiffler,
1
* Jiunn R. Chen,
2
* Viara P. Grantcharova,
1
† Ying Lei,
1
Daniel Fuchs,
1
John E. Allen,
1
Lioudmila A. Zaslavskaia,
1
‡ Gavin MacBeath
1
§
PDZ domains have long been thought to cluster into discrete functional classes defined by their
peptide-binding preferences. We used protein microarrays and quantitative fluorescence polarization to
characterize the binding selectivity of 157 mouse PDZ domains with respect to 217 genome-encoded
peptides. We then trained a multidomain selectivity model to predict PDZ domain–peptide interactions
across the mouse proteome with an accuracy that exceeds many large-scale, experimental investigations
of protein-protein interactions. Contrary to the current paradigm, PDZ domains do not fall into discrete
classes; instead, they are evenly distributed throughout selectivity space, which suggests that they have
been optimized across the proteome to minimize cross-reactivity. We predict that focusing on families
of interaction domains, which facilitates the integration of experimentation and modeling, will play an
increasingly important role in future investigations of protein function.
E
ukaryotic proteins are modular by nature,
comprising both interaction and catalytic
domains (1, 2). One of the most frequently
encountered interaction domains, the PDZ do-
main, mediates protein-protein interactions by
binding to the C termini of its target proteins (3–6).
Previous studies of peptide-binding selectivity
have placed PDZ domains into discrete function-
al categories: Class I domains recognize the
consensus sequence Ser/Thr-X-y-COOH, where
X is any amino acid and y is hydrophobic; class
II domains prefer y-X-y-COOH; and class III
domains prefer Asp/Glu-X-y-COOH (5, 7).
More recent information has suggested that these
designations are too restrictive and so additional
classes have been proposed (8, 9). The idea that
domains fall into discrete categories, however,
raises questions about functional overlap:
Domains within the same class are more likely
to cross-react with each other’s ligands. To
resolve this issue, we characterized and modeled
PDZ domain selectivity on a genome-wide scale.
We began by cloning, expressing, and purify-
ing most of the known PDZ domains encoded in
20 JULY 2007 VOL 317 SCIENCE www.sciencemag.org
364
REPORTS
on July 20, 2007 www.sciencemag.orgDownloaded from