9 l .
u ..
‘. ﬁmﬁz.

4 {353$-

193$

[1.553. a
7 a . .- $.51
a..nﬁ.&*5h H.013... .
. !.vvu \. xv .

 

 

3E 3.1a: £53.. an} ,3 . I i...

 

 

 

2 LIBRARY
was Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

QUEEN AND WORKER INFLUENCE ON SEX ALLOCATION
PATTERNS IN THE HONEYBEE, APIS MELLIFERA

presented by

Katie Elizabeth Wharton

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. degree in Zoology,
Program in Ecology,
Evolutionary Biology, and
Behavior

 

 

41M not

Major Professor’s Signéture

6/Dl‘/o‘¢?
Date

 

 

MSU is an afﬁrmative-action, equal-opportunity employer

—4-n-0-o-I-I-u-I-I-o-o-o-l-l-I--n-c-

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

l

i

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5108 K:IProj/Aoc&Pres/ClRC/DateDue.indd

 

 

 

QUEEN AND WORKER INFLUENCE ON SEX ALLOCATION PATTERNS
IN THE HONEYBEE, APIS MELLIFERA

By

Katie Elizabeth Wharton

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOHPY

Department of Zoology
Program in Ecology, Evolutionary Biology, and Behavior

2008

ABSTRACT

QUEEN AND WORKER INFLUENCE ON SEX ALLOCATION PATTERNS
IN THE HONEYBEE, APIS MELLIFERA

By

Katie Elizabeth Wharton

Since Darwin, biologists have sought to understand the ways in which natural
selection shapes the ability of organisms to adjust their investment in the sexes. A social
insect colony, comprised of many individuals, faces the challenge of coordinating its
individuals to achieve an adaptive allocation to the sexes. In this dissertation, I focus on
the ways in which sex allocation in honeybee colonies is shaped by the actions and
interactions of the queen and workers.

Previously, attempts to understand how colony members shape colony sex
allocation patterns have generally focused on social insect species in which the queen and
workers favor different colony sex allocation optima. This queen-worker conﬂict over
sex allocation is expected to be minimal or absent in species with a highly polyandrous
queen, including the honeybee. In these species, understanding how the queen and
workers inﬂuence colony sex allocation might provide important insights into how
colonies efﬁciently and ﬂexibly respond to environmental conditions. To better
understand how queens and workers shape colony sex allocation patterns when conﬂict is
expected to be minimal or absent, I conducted experiments using the honeybee.

The queen lays nearly all of the eggs in her colony and might therefore play a key
role in her colony’s investment in males (drones). To test whether the queen’s egg-laying

decisions impact colony sex allocation patterns, I manipulated the previous egg-laying

experiences of queens and examined their subsequent tendency to lay worker or drone
eggs. My results indicate that the queen has the ability to regulate her production of
drone eggs, and in turn has the ability to inﬂuence her colony’s regulation of investment
in male reproductive ﬁmction. More generally, this work reveals that the honeybee
queen and workers share control over their colony’s sex allocation.

After the queen lays eggs, the workers rear the eggs to adulthood. However,
workers occasionally eliminate a portion of their colony’s immature males. To better
understand how brood rearing decisions by workers impact colony patterns of sex
allocation, I manipulated the abundance of older male brood in colonies and quantiﬁed
the tendency of workers to rear young larvae. Workers eliminated a greater proportion
of young male larvae when I increased the abundance of older male brood than when I
decreased it. While earlier studies suggest that male elimination reﬂects conﬂict between
the queen and workers, my results raise the possibility that male elimination might
sometimes reﬂect adaptive adjustment of male reproductive function, potentially
increasing colony efﬁciency in the interests of all colony members.

Historically, biologists have primarily investigated sex allocation theory through
the study of offspring sex ratios. I suggest that our understanding of sex allocation will
be aided by applying analyses that use the raw numbers of males and females, rather than
the ratio of males to females, as the primary variables of interest. To illustrate this point,
I examine existing datasets of offspring production and demonstrate that analyzing sex
allocation behavior can lead to ﬁmdamentally different biological conclusions than those

drawn from analyzing sex ratio behavior.

ACKNOWLEDGEMENTS

This dissertation would not have been possible without the advice and support of
many wonderful people.

First and foremost, I thank my advisor, Thomas Getty. Tom has been both a
fantastic scientiﬁc mentor and a supportive ﬁiend, and a great deal of my persistence and
motivation during grad school can be attributed to the wisdom and support that he
provided. I am deeply grateful for his advice, insights, patience, perspective, support,
humor, and enthusiasm throughout all stages of my dissertation. I feel very blessed to
have worked under his mentorship. . .even if I did have to pretend to like beet mufﬁns.

I also thank my coadvisor, Fred Dyer. Fred immediately adopted me into his lab
when I ﬁrst became interested in social insect behavior. He introduced me to the social
insect world and many of its people, encouraged my efforts to pursue research with
honeybees, and even took me on my ﬁrst beekeeping outing (during which he was stung
by a few rogue bees). Additionally, I have greatly appreciated Fred’s advice on
professional issues in academia.

I am extremely thankful for the advice and support from my other committee
members, Zach Huang and Kay Holekarnp. I ﬁrst became inspired to work with
honeybees while attending a talk given by Zach, and I thank him for facilitating my “aha”
moment of inspiration. Kay warmly greeted me to graduate school and I have always
appreciated her encouraging smiles during my committee meetings. Both Zach and Kay
have helped me to better develop and articulate my research, and my dissertation is

stronger as a result of their suggestions and input.

iv

For friendship, advice on honeybees, and fun times at the bee building, I thank
my labmates ﬁ'om the Dyer lab: Frank Bartlett, Steph Bussema, Mercedes Ramirez,
John Townsend-Mehler, and Cindy Wei. Together we enjoyed the fun and the
ﬁustrations of ﬁeldwork with the bees. For help in collecting data, I owe many thanks to
my undergraduate ﬁeld assistants: Adam Bates, Lora Brarnlett, Kim Davis, Liz Dean,
Michael Hillman, Sara Kaltz, Jenn Lieb, Jared Ruddick, Andy Taylor, Kourtney Trud gen,
Mara Trudgen, Erin Walaszczyk, and Megan Wheeler. Lora and Michael were
particularly helpful for components of my ﬁeldwork.

For advice, support, and much laughter and good times together, I thank my
labmates from the Getty lab: Natalie Dubois, Ben J anse, Jean Johnson, Eva
Lewandowski, and Lindsey Walters. Our lab meetings were very helpful and
encouraging to me, and I always enjoyed the corresponding doses of laughter and sugar
that accompanied our get-togethers. Lindsey in particular has become a very dear ﬁ'iend
and I’m extremely glad for all of her advice, support, and quite frequent conversations
throughout graduate school.

I have many people to thank for advice and logistical support. Dirk Colbry taught
me the wonders of MATLAB, Joe Riddle provided several suggestions about beekeeping,
and Dave Shuker graciously allowed me to use one of his datasets for my research on sex
allocation. The administrative staff in the Zoology department helped me with countless
miscellaneous tasks over the years, and I especially thank Lisa Craft and Sue Platte for
this assistance. I am also thankful for my funding sources, including the Department of
Zoology, the College of Natural Science, the EEBB program, and an NSF IGERT

program (DGE 0114378) in sequential decision-making.

My friends have provided great companionship and have made my time in
Lansing quite fun. Many thanks go to my friends in EEBB, my friends in the Capital
City Ringers, and my friends in the game night crew. I believe that many of us in the
latter group found that a game of doppelkopf did wonders to relieve our stress levels
during grad school. Anne Brockett was my roommate for several years and together we
shared many fun times. Robert Boes extended a hand of ﬁiendship to me throughout
grad school and is now my ﬁance. I am happy and truly blessed to have his love and
devotion in my life.

Finally, I thank Mom, Dad, and my brother Chris. Their unwavering love and

support bring a smile to my face every day.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES ............................................................................................................ x
CHAPTER 1
INTRODUCTION .............................................................................................................. 1
1. General Introduction .................................................................................................. 1
2. Colony-level patterns of drone production based on environmental factors ............. 3
2.1 Season ................................................................................................................... 4
2.2 Size of colony ....................................................................................................... 4
2.3 Food availability ................................................................................................... 5
2.4 Amount of drone brood and number of drones already present in the colony ..... 6
2.5 Queen presence/ absence ....................................................................................... 7
3. Key stages in drone production and maintenance ...................................................... 8
3.1 Workers build comb ............................................................................................ 11
3.2 Workers clean cells or ﬁll cells with nectar ........................................................ 13
3.3 Queen lays eggs .................................................................................................. 14
3.4 Workers rear/fail to rear larvae ........................................................................... 16
3.5 Workers maintain or evict drones ....................................................................... 18
4. Conclusions .............................................................................................................. 20
CHAPTER 2
THE HONEYBEE QUEEN INFLUENCES THE REGULATION OF COLONY
DRONE PRODUCTION .................................................................................................. 23
Introduction ..................................... 24
Methods ........................................................................................................................ 29
Honeybee colonies .................................................................................................... 29
General methods ....................................................................................................... 3O
Drone departure rates ................................................................................................ 33
Statistical analysis ..................................................................................................... 34
Results ........................................................................................................................... 36
Does previous comb type inﬂuence the queen’s egg-laying patterns? ..................... 36
Do the egg-laying patterns on the choice frame result from the actions of the
workers? .................................................................................................................... 38
Does previous comb type inﬂuence pupae-rearing patterns of the colony? ............. 42
Does previous comb type inﬂuence drone mortality or worker mortality? .............. 44
Did the colonies differ in their investment in drones throughout the study? ............ 44
Discussion ..................................................................................................................... 45
CHAPTER 3
MALE ELIMINATION IN THE HONEYBEE ............................................................... 51
Introduction ................................................................................................................... 52
Methods ........................................................................................................................ 55

vii

Honeybee colonies .................................................................................................... 55

General methods ....................................................................................................... 56
Survival of immature drones and workers ................................................................ 59
Statistical analysis ..................................................................................................... 60
Results ........................................................................................................................... 6O
Comparisons of the number of eggs on the focal frame and adult drones in the
colony ........................................................................................................................ 6O
Comparisons ofbrood survrval 61
Discussion ..................................................................................................................... 63
CHAPTER 4
THERE IS MORE TO SEX ALLOCATION THAN THE SEX RATIO ........................ 68
Introduction ................................................................................................................... 68
Example 1: Offspring sex allocation by honeybee queens (Apis mellifera) ................ 70
Overview of study ..................................................................................................... 70
Results ....................................................................................................................... 72
Discussion ................................................................................................................. 75
Example 2: Offspring sex allocation by female parasitoid wasps (Nasonia vitripennis)
...................................................................................................................................... 76
Overview of study ..................................................................................................... 76
Results ....................................................................................................................... 77
Discussion ................................................................................................................. 81
Conclusions ................................................................................................................... 82
REFERENCES ................................................................................................................. 88

viii

LIST OF TABLES

Table 1.1. Current understanding of how colonies adjust drone production in relation to
environmental conditions (rows) for each of the stages involved in drone production

(columns). D = drone; W = worker. See sections 3.1 — 3.5 of this chapter for additional
details. ............................................................................................................................... 10

ix

LIST OF FIGURES

Figure 1.1. Drone production and maintenance in a honeybee colony: a sequence of
actions taken by the workers and the queen. At each step, the bees have an opportunity to
alter the colony’s investment in drones in response to environmental conditions. Arrows
indicate typical sequences of events. Drone cells are larger than worker cells, and drones
are normally reared in these larger cells (I). The worker and drone cells are used either
for food storage (nectar and pollen; darker cells in diagram) or are cleaned in preparation
for rearing brood (II). The queen typically lays unfertilized (drone) eggs into drone cells,
and fertilized (female) eggs into worker cells (111). Workers then tend the larvae, but
don’t necessarily raise all of the larvae to adulthood (IV). Adult drones remain colony
members until they mate with a virgin queen in the population, die, or are evicted by
workers (V). See sections 3.1 — 3.5 of this chapter for details of each stage. ................... 9

Figure 2.1. Drone production and maintenance in a honeybee colony: a sequence of
actions taken by the workers and the queen. At each step, the bees have an opportunity to
inﬂuence the colony’s investment in drones. Arrows indicate typical sequences of events.
Drone cells are larger than worker cells, and drones are normally reared in these larger
cells (I). The worker and drone cells are used either for food storage (nectar and pollen;
darker cells in diagram) or are cleaned in preparation for rearing brood (II). The queen
typically lays unfertilized (drone) eggs into drone cells, and fertilized (female) eggs into
worker cells (111). Workers then tend the larvae, but don’t necessarily raise all of the
larvae to adulthood (IV). Adult drones remain colony members until they mate with a
virgin queen in the population, die, or are evicted by workers (V). In this study, we focus
primarily on the modulation of drone production that arises from actions taken by the
queen (step 111). ................................................................................................................. 27

Figure 2.2. Mean (iSE) number of drone eggs plotted against mean (iSE) number of
worker eggs on the choice frame for queens previously on worker comb (WC) and
queens previously on drone comb (DC). WC queens laid more drone eggs but the same
amount of worker eggs as DC queens. The lines represent the egg sex ratio vector on the
choice frame; the slope is the average ratio of drone eggs to worker eggs (dashed line for
WC queens, dotted line for DC queens). WC queens and DC queens did not produce
signiﬁcantly different egg sex ratios. ................................................................................ 37

Figure 2.3. Mean (iSE) area of drone comb (light bars) and worker comb (dark bars)
ﬁlled with nectar in the choice frames for WC colonies and DC colonies. The choice
frames had 825.8 cm2 drone comb and 825.8 cm2 worker comb. WC colonies and DC
colonies did not differ in the area of drone comb ﬁlled with nectar or in the area of
worker comb ﬁlled with nectar. ........................................................................................ 40

Figure 2.4. Mean (iSE) number of drone pupae plotted against mean (:tSE) number of
worker pupae on the choice frame for colonies with queens previously on worker comb
(WC) and colonies with queens previously on drone comb (DC). WC colonies raised
more drone pupae but the same amount of worker pupae as DC colonies. The lines

represent the pupal sex ratio vector on the choice frame; the slope is the average ratio of
drone pupae to worker pupae (dashed line for WC colonies, dotted line for DC colonies).
WC colonies produced a more male-biased sex ratio than did DC queens. ..................... 43

Figure 3.1. Experimental design, ﬂom a bird’s eye view of the moveable frames of comb
in a colony. Dark gray bar = focal ﬂame (composed of equal areas of drone and worker
comb). Light gray bars = treatment ﬂames (D = drone, W = worker). White bars = other
ﬂames in colony (contained mostly worker brood, pollen, honey, and empty worker
cells). The dashed line indicates the queen excluder cage and the triangle (El) indicates
the position of the queen. At the beginning of experiment (left), the queen was placed on
the focal ﬂame in the queen excluder cage and permitted to lay drone and worker eggs.
Next, the queen was placed elsewhere in the colony and each colony was randomly
assigned to one of two treatment groups (right). The treatment was applied by exchanging
ﬂames between the two groups of colonies. In the drone brood removed (D-) group, the
colony lost its drone treatment ﬂame, and a worker treatment ﬂame was inserted in its
place. In the drone brood added (D+) group, the colony received an additional drone
treatment ﬂame in the place of a worker treatment ﬂame. ............................................... 57

Figure 3.2. The survival of worker eggs and drone eggs to the larval stage on the focal
ﬂame in colonies that had drone brood removed (open circles) and drone brood added
(ﬁlled circles). The overall mean (:1: standard error) survival for each treatment group is
indicated by the correspondingly shaded squares and error bars. The survival patterns
differed signiﬁcantly between the two treatment groups (Wilks’ A1,” = 0.566, P = 0.014).
Speciﬁcally, the survival of worker eggs to the larval stage did not differ between
treatment groups (Fm, = 0.22, P = 0.642), but the survival of drone eggs to the larval
stage was signiﬁcantly lower in the colonies to which drone brood was added than in
colonies ﬂom which drone brood was removed (F 1,16 = 9.29, P = 0.008). ....................... 62

Figure 4.1. Number of published articles (between the years 1965 and March 2008)
listed in the ISI Web of Science that include either the term “sex ratio(s)” or “sex
allocation” in the article title. The data in this ﬁgure are limited to three journals that
specialize in behavioral ecology: Behavioral Ecology and Sociobiology, Animal
Behavior, and Behavioral Ecology. .................................................................................. 71

Figure 4.2. Mean (t SE) sex ratio of honeybee queens that had previously been
prevented ﬂom laying drone eggs (WC queens) and queens that had previously been
allowed to lay drone eggs (DC queens). The egg sex ratios are not signiﬁcantly different
(F132: 2.340, P = 0.140). ................................................................................................. 73

Figure 4.3. Mean (2!: SE) sex allocation of honeybee queens that had previously been
prevented ﬂom laying drone eggs (WC queens) and queens that had previously been
allowed to lay drone eggs (DC queens). The egg sex allocation patterns are signiﬁcantly
affected by the treatment (Wilks’ A122 = 0.717, P = 0.030). WC queens laid more drones
eggs but the same amount of worker eggs as DC queens (see text). ................................ 74

Figure 4.4. Data ﬂom Shuker et. a1. 2004. Female sex ratio behavior and oviposition
partner. Females were either with a female ﬂom the same strain (black bars: HV236;

xi

white bars: HV307) or were unrelated (gray bars: one HV236 female with one HV307
female). Additionally, females ﬂom the same strain were either raised on the same host
(SH) or different hosts (DH). Error bars show standard errors. ...................................... 78

Figure 4.5. Wasp sex allocation behavior in relation to oviposition partner (mean i SE
number of female and male offspring produced). Females were either with a female ﬂom
the same strain or were unrelated (one HV236 female with one HV307 female). There
was a signiﬁcant effect of relatedness on sex allocation decisions (Wilks’ A1339 = 0.923,
P<0.001). ........................................................................................................................... 80

Figure 4.6. Three ways that a female can increase (i.e., double) her production of males.
A female might a) alter her sex ratio (dashed line), b) alter her clutch size (dotted line), or
c) slightly alter both sex ratio and clutch size (dashed-dotted line). By any one of these
strategies, a female can double her production of males (i.e. switch ﬂom producing 25
males to producing 50 males). Each pie represents a female’s investment in male
offspring (gray areas) and female offspring (white areas). ............................................... 84

Figure 4.7. Wasp sex allocation behavior within the strain HV307 (mean i SE number of
female and male offspring produced). The wasp sisters were either ﬂom the same host or
ﬂom different hosts. The slopes of the dashed lines indicate the mean sex ratio

(males/total) for each group (long dashes: same host; short dashes: different host). ....... 85

xii

CHAPTER 1
INTRODUCTION

1. General Introduction

Since Darwin (1874), biologists have sought to understand the ways in which
natural selection shapes the ability of organisms to adjust their investment in the sexes.
An organism’s allocation of resources to male and female reproductive function is termed
its sex allocation (Chamov 1982). In the ﬁeld of behavioral ecology, sex allocation
theory has led to major advances in our understanding of how organisms perceive and
respond to changing environmental conditions. In this dissertation, I focus on how sex
allocation in honeybee (Apis mellzfera) colonies is shaped by the decisions of the queen
and workers.

Social insect colonies, like many individual organisms, must allocate resources
appropriately between growth and reproduction. Honeybee colonies reproduce in two
distinct ways: by swarming, a ﬁssion process in which the queen and over half of the
worker bees leave to establish a new colony, and by producing and maintaining males
(hereafter drones), whose sole apparent function is to mate with virgin queens in the
population. This latter “male” mode of reproduction allows a colony to introduce the
genes shared by its members into other colonies in the population, and is therefore an
important route by which colony members increase their inclusive ﬁtness.

Honeybee colonies actively regulate their investment in male reproductive
function. Drones typically compose 5-10% of the adult population in a honeybee colony

(Seeley and Morse 1976; Page and Metcalf 1984). The colony, however, is able to adjust

its investment in drones and workers in accordance with environmental factors (Page and
Metcalf 1984). How colonies manage to do this through the actions and interactions of
the queen and workers presents an open and interesting question in social insect biology
(Bourke and Franks 1995), and is the focus of this chapter.

The queen lays nearly all of the eggs in her colony and is therefore an important
participant in her colony’s brood production process. In honeybees and other
haplodiploid species, haploid (unfertilized) eggs develop into males and diploid
(fertilized) eggs develop into females. The queen therefore has the opportunity to
inﬂuence her colony’s sex allocation by controlling whether or not she fertilizes her eggs
(Ratnieks and Keller 1998). Despite this knowledge, biologists have long assumed that
worker honeybees control their colony’s sex allocation. In Chapter 2, I inquire whether
the honeybee queen has the ability to inﬂuence her colony’s investment in male
reproductive function.

The workers, or non-reproductive females in the colony, perform many essential
tasks for their colony’s brood production process. For example, workers build the wax
cells that are used to rear the brood, clean the cells in preparation for deposition of an
egg, and provide food to the developing larvae. Occasionally, however, workers will
cannibalize developing larvae and therefore fail to rear all of the queen’s eggs to
adulthood (W oyke 1977). By facultatively rearing workers or drones, workers might
adaptively adjust their colony’s sex allocation in response to changing environmental
conditions. In Chapter 3, I investigate how a honeybee colony’s investment in male
reproductive function is shaped by the brood rearing decisions of the workers.

Historically, biologists have primarily investigated sex allocation theory through

the study of offspring sex ratios. In Chapter 4, I suggest that biologists might gain more .
insightful interpretations of their system of study by treating the raw number of males and
number of females (sex allocation), rather than the ratio of males to females (sex ratio),
as the variables of interest in their studies. To illustrate this point, I compare the results
of taking a sex allocation versus a sex ratio approach to datasets documenting offspring
production by honeybee queens (Wharton et al. 2007) and female parasitoid wasps
(Shuker et a1. 2004).

In the remainder of this chapter, I review the literature on how honeybee colonies
adjust their investment in male reproductive function, and I highlight promising areas of
future research within this topic. I begin by identifying ﬁve environmental conditions
and explaining how each affects colony-level patterns of drone rearing. Next, I deﬁne the
key stages involved in drone production and maintenance and focus on how drone rearing
at each stage changes in relation to the ﬁve environmental conditions. I conclude by
emphasizing the importance of studies that examine how colony sex allocation is shaped

by the actions and interactions of the queen and workers.

2. Colony-level patterns of drone production based on environmental factors

Honeybee colonies regulate their production and maintenance of drones in
accordance with several environmental factors. In this section I identify ﬁve of these

factors and discuss colony patterns of drone rearing in relation to each.

2.1 Season

The number of drones present in a honeybee colony varies over the course of a
year. A colony’s drone population increases during the spring and reaches a peak in the
late spring or early summer. Although this peak in drone production also coincides with
the peak in worker production, it is during this time of year that colonies produce the
highest proportion of drone brood (Free and Williams 1975). This peak occurs just prior
to the population’s main swarming season, when virgin queens are most abundant (Allen
1958; Allen 1965b; Page 1981; Lee and Winston 1987). As a result, colonies have the
greatest number of drones during the time of year when drones are most likely to
successfully mate. The colony’s drone population slowly decreases throughout the late
summer and into the fall, and typically dwindles to zero by winter.

The seasonal factors that inﬂuence these trends in drone production are unknown.
While colonies might adjust drone production in relation to season itself (day length and
temperature), the trends more likely arise ﬂom a combination of season and other

environmental cues (colony size and availability of food) that vary with season.

2.2 Size of colony

Honeybee colonies reﬂain ﬂom producing drones until the colony has a
sufﬁciently large population of adult workers. The population of adult workers in a
colony is naturally low following a swarming event, and swarms that establish new
colonies prevent early drone production by not constructing drone comb (the wax cells in

which drones are reared) until an average of 22 days after establishment (Lee and

Winston 1985). Once established, larger swarms build a greater proportion of drone cells
more quickly than do smaller swarms, indicating a strong effect of swarm size on drone
comb investment (Henderson 1991). In addition, small swarms that are composed of
fewer than 10,000 workers sometimes build little or no drone comb at all, indicating that
small colonies limit their investment in drones until they’ve grown larger (Lee and
Winston 1985).

As small colonies increase their production of brood, they also increase the
proportion of drone brood (Free and Williams 1975). However, once colonies reach a
large size (more than 12,000 adult workers), there doesn’t appear to be any relationship
between the number of workers reared and the percentage of drone brood in the colony
(Page and Metcalf 1984). Colonies therefore reach an upper limit in their investment in

drones.

2.3 Food availability

Drones are more costly to produce and maintain than workers (Seeley 2002;
Hrassnigg and Crailsheim 2005), so colonies should regulate drone production in
accordance with the availability of food. Seeley and Mikheyev (2003) conﬁrmed this in
a controlled experiment. They maintained a group of colonies in a food-plentiful
environment and observed that the colonies produced similar amounts of drone brood.
They then transferred the colonies to an island with very few food sources and
experimentally manipulated food availability by provisioning half of the colonies with

sugar water and leaving the other colonies to rely on only the island’s limited food

sources. On the island, the food-supplemented colonies continued to produce many
drones, but the non-supplemented colonies lowered their production of drones.
Honeybee colonies also decrease their production of workers when foraging
conditions are poor (Schmickl and Crailsheim 2001; Schmickl and Crailsheim 2002), but
it is unknown how honeybee colonies jointly regulate drone and worker production
during periods of low food availability. Since the value of drones changes in relation to
workers throughout the year, colonies might be expected display seasonal trends in how
they jointly produce drones and workers in response to food availability, but this has not

yet been tested.

2.4 Amount of drone brood and number of drones already present in the colony

Colonies adjust their production of immature drones (hereafter drone brood) in
accordance with the amount of drone brood already present. Speciﬁcally, when drone
brood is added to colonies, the colonies lower their production of additional drones (Free
and Williams 1975). This demonstrates that that the presence of drone brood in a colony
inhibits the colony’s production of additional drones. Drone production is therefore
regulated by a negative feedback process.

Studies that tested whether colonies adjust their production of drones in
accordance with the number of adult drones present have produced mixed results.
Rinderer et a1. (1985) added adult drones to colonies and found that the colonies
subsequently lowered their drone production. However, Henderson (1994) removed

drones ﬂom colonies and found that colonies did not increase drone production. These

mixed results call into question whether, and if so how, honeybee colonies can assess the
size of their population of adult drones.

Future work might better examine the trade-offs that colonies face by investing in
drones. Drone rearing does not appear to trade off with worker rearing, because colonies
that are prevented ﬂom investing in drones do not produce more workers than colonies
allowed to rear drones (Allen 1963; Allen 1965a; Page and Metcalf 1984). Drone rearing
might trade off with honey yields, an important means by which colonies survive the
winter, but there is both supporting and contradictory evidence that drone rearing affects
honey yields (Allen 1965a; J ohansson and J ohansson 1971; Seeley 2002). These results
raise questions about the life history trade-offs that honeybee colonies face when

investing in drones.

2.5 Queen presence/absence

The environmental factors discussed above have only been evaluated in colonies
where the queen is present, termed queenright colonies. In queenright colonies, the
queen produces approximately 99.9% of the colony’s adult males (Visscher 1989). If the
queen becomes inviable or dies, the colony will attempt to rear a replacement queen. If
that attempt fails and the colony becomes “hopelessly” queenless, drones become the sole
source through which the colony can pass along genes. Accordingly, the colony
increases its investment in drones relative to workers. This happens naturally because
some workers reared in queenless colonies develop ﬁrnctional ovaries and are able to lay

drone (unfertilized) but not female (fertilized) eggs.

In queenless colonies, workers begin laying eggs around 24 days after the queen’s
disappearance, and their production of drone eggs is short and synchronous (Page and
Erickson 1988). Which workers are most successful in laying eggs is difﬁcult to infer,
because workers consume or “police” many of the eggs (Miller and Ratnieks 2001 ).
Additionally, subfamilies of workers vary in their tendencies to produce and consume
eggs (Robinson et al. 1990), and younger workers appear to be more successful at laying
eggs than older workers (Delaplane and Harbo 1987). Despite the oophagy that occurs,
worker egg-laying can result in 6000 additional adult drones for a queenless colony (Page
and Erickson 1988). However, because the workers cannot produce new workers or a

new queen, the colony cannot maintain these drones for long before the colony dies.

3. Key stages in drone production and maintenance

Colonies produce and maintain drones in a multi-staged sequential process
(Figure 1.1). At each stage, either the workers or the queen have the opportunity to shape
the colony’s investment in drones. In this section I examine each of these stages in detail.
For each stage, I present a brief overview of the stage’s importance for colony drone
production, review how the bees modify the colony’s investment in drones in response to
the environmental factors listed in the previous section, and highlight aspects of the stage

that aren’t well understood. A summary of this section is provided in Table 1.1.

  

ea to a;
(g) 9.0... as. age. V

 

 

 

II III
Workers . Queen Workers Workers
Build wax cells Fill (with food) Lays eggs in cells Rear or fail to Maintain or evict
(drone or worker) or clean cells (drone g worker) rear larvae adult drones

 

Figure 1.1. Drone production and maintenance in a honeybee colony: a sequence of
actions taken by the workers and the queen. At each step, the bees have an opportunity to
alter the colony’s investment in drones in response to environmental conditions. Arrows
indicate typical sequences of events. Drone cells are larger than worker cells, and drones
are normally reared in these larger cells (I). The worker and drone cells are used either
for food storage (nectar and pollen; darker cells in diagram) or are cleaned in preparation
for rearing brood (II). The queen typically lays unfertilized (drone) eggs into drone cells,
and fertilized (female) eggs into worker cells (111). Workers then tend the larvae, but
don’t necessarily raise all of the larvae to adulthood (IV). Adult drones remain colony
members until they mate with a virgin queen in the population, die, or are evicted by
workers (V). See sections 3.1 — 3.5 of this chapter for details of each stage.

Table 1.1. Current understanding of how colonies adjust drone production in relation to
environmental conditions (rows) for each of the stages involved in drone production
(columns). D = drone; W = worker. See sections 3.1 — 3.5 of this chapter for additional

 

 

 

 

 

 

 

details.
Workers Workers clean Queen lays Workers Workers
build comb cells/fill cells eggs rear/fail to maintain/evict
with nectar rear larvae adult drones
Season D comb D cells are used D egg Survival of D D eviction is
construction is primarily for production is larvae is higher in the
highest in the brood rearing in higher in variable fall than in
spring and spring/summer spring-early across spring or
early summer and primarily summer than season; summer
for nectar in in late survival is
fall summer-fall lowest in the
fall
Size Larger Unknown Unknown Unknown Unknown
colonies build how colony
more D comb, size affects D
and build it egg
more quickly, production
than smaller
colonies
Food availability Colonies must Unknown Egg Fewer brood D eviction is
have nectar production is survive under higher in unfed
intake to build lower during poor foraging colonies than in
any comb at poor food conditions; fed colonies
all; unknown conditions; relative
how food unknown how survivals of D
affects this affects and W brood
proportion of relative are unknown
D comb built production of
D and W eggs
Amount of drone D comb alone Unknown D brood Survival of D Unknown
brood and adult inhibits D inhibits D egg larvae is
drones present comb production; lower when
in colony construction; unknown if colonies
D brood adult D contain D
enhances this inhibit D egg brood;
inhibition. production unknown
Unknown if how survival
adult D inhibit of D larvae is
D comb inﬂuenced by
construction adult D
Queen Queenless Unknown If queen is Survival of D eviction
presence/absence colonies build absent, queen-laid D higher in
less comb workers lay D brood queenright
overall, but a but not W increases in colonies than in
greater eggs queenless queenless
proportion of colonies; low colonies
the comb built survival of
is D comb worker-laid D

 

 

 

 

brood

 

 

10

 

3.1 Workers build comb

Workers build the wax cells that are used by the colony to store food and rear
bees. These cells are either small in size (worker cells) or larger in size (drone cells).
Drone cells typically constitute 15-20% of the total comb area (Seeley and Morse 1976).
Though bees occasionally reshape one type of cell into the other type under unnatural
conditions (see Free and Williams 1975), they typically don’t alter the wax comb that has
already been built. Since drones cannot be reared properly in the smaller worker cells,
the number of drone cells in a colony limits the number of drones that can be reared at
any one time. However, if colonies are provisioned with an excessive amount of drone
comb, they won’t use all of it for rearing drones, indicating that a colony’s regulation of
drones is determined by more factors than just the amount of comb in the colony (Allen
1965a). Of all of the stages involved in drone production, the construction of comb has
been the most extensively studied. For an extensive review of the factors that govern
comb construction in honeybees, see Pratt (2004).

Workers build drone comb mostly in the spring and early summer, indicating that
season inﬂuences drone comb construction (Free 1967). Two lines of evidence support
this assertion. First, swarms established earlier in the year produce more drone cells than
swarms (of an identical size) established later in the year (Lee and Winston 1985).
Second, colonies build the highest proportion of drone cells in May through July (Free
and Williams 1975). This occurs despite the fact that the authors in that study controlled
for both food availability and drone brood presence by supplementing their colonies with

sugar syrup throughout the year, and by removing built drone comb on a weekly basis.

11

These results therefore are consistent with the idea that season itself (day length and
temperature) inﬂuences drone comb construction.

Drone comb construction is inﬂuenced by colony size. Small colonies construct
fewer drone cells than larger colonies (Free 1967; Lee and Winston 1985; Henderson
1991). This pattern holds for established swarms as well: small swarms usually build
only a very small proportion of drone comb at ﬁrst (Taber and Owens 1970; Henderson
1991), whereas larger swarms will build drone comb more immediately (Free and
Williams 1975; Henderson 1991). Smaller colonies likely suppress their construction of
drone comb in order to allow the colony to grow large enough to support a population of
drones.

Drone comb construction is regulated by negative feedback. Workers reduce their
construction of drone comb when drone comb is already present (Free 1967; Free and
Williams 1975; Pratt 1998), and they further reduce drone comb construction if drone
brood is present as well (Pratt 1998). Based on these trends, one might expect that
workers reduce drone comb construction when the colony contains an abundance of adult
drones, but this remains an open question.

Drone comb construction is inﬂuenced by whether or not the colony has a queen.
When the colony is queenless but workers are rearing queens, the workers build a higher
proportion of drone cells (Free 1967), but less comb overall, than do queenright colonies
(Pratt 2004). This is not because the queen constructs comb or directs comb construction
(Pratt 1998), but presumably occurs because selection has acted on queenless colonies to
invest more in drones, their sole method for passing on genes. In queenright colonies,

food must be available in order for workers to construct any comb at all (Pratt 2004), but

12

how the amount of food available inﬂuences the colony’s relative investment in drone
cells to worker cells is largely unknown. One study found that colonies did not alter
drone comb construction depending on food availability, but that study used too few

colonies to make a statistical inference (Sasaki and Obara 2001).

3.2 Workers clean cells or fill cells with nectar

Worker bees perform the actions of cleaning the brood cells and storing or
removing food ﬂom the cells. When the cells have been cleaned of debris and do not
contain any pollen or nectar, they are available to receive an egg ﬂom the queen
(Winston 1987). Since the queen lays eggs continuously and workers clean and use cells
constantly, whether a particular cell is used for food storage or for rearing brood depends
upon a dynamic interplay of actions taken by both the queen and the workers.

The use of drone and worker cells for egg deposition appears to be independent of
the use of these cells for either cleaning or food storage. In one study, the decision of the
queen to deposit an egg in a drone or worker cell was not constrained by the decisions of
workers to ﬁll drone and worker cells with nectar (Wharton et al. 2007). In another
study, workers were equally likely to clean either worker or drone cells (Sasaki et al.
1996). Although the decisions of the queen and workers might be independent in some
circumstances, they are likely dependent when a colony has limited cells available for
brood or food. Future work might address whether in such cases the workers primarily
limit the cells available to receive eggs, or whether the queen primarily limits the cells

available for food storage.

13

Whether workers respond to environmental factors for cell cleaning activities or
for storing food remains largely an open question. The use of drone comb for these
activities appears to depend on season: drone comb contains mostly drone brood in the
early summer and mostly food in the early fall (Free and Williams 1975; TD. Seeley,
personal communication). This seasonal pattern might arise because workers reﬂain
ﬂom ﬁlling drone cells with food until later in the season, because queens reduce the
number of eggs they lay into drone cells later in the season, or because of some
combination of these two factors. To date, few studies have examined how the use of a
cell depends on colony size, food availability, the presence of drones, or the colony’s
queen status. This paucity of research likely stems ﬂom the difﬁculty of determining
how patterns of brood and food deposition arise ﬂom a complex interplay between the

decisions of the workers and the queen.

3 .3 Queen lays eggs

The queen honeybee contributes to colony drone production by laying drone
(unfertilized) eggs into available drone cells. She deposits worker (fertilized) eggs into
worker cells. The queen sometimes lays drone and worker eggs in a ratio that deviates
ﬂom the ratio of drone and worker cells (Henderson 1991; Wharton et al. 2007),
suggesting that the queen can shape brood production patterns by choosing whether to lay
eggs in either drone or worker comb. These eggs eventually develop into larvae, which
might be consumed by the workers (see section 3.4). Therefore, to understand how the

queen adjusts her drone egg production in response to environmental conditions, it is

14

necessary to examine the colony’s egg (rather than larvae or pupae) sex allocation. Few

studies focus on the egg sex allocation patterns within a colony, and we address potential
reasons for this lack of research in section 4. Nevertheless, a few studies have examined
queen egg-laying decisions, and I outline those results here.

A colony’s number of drone eggs depends on the season. The proportion of drone
eggs is higher in the spring and summer than in the fall (Free and Williams 1975), though
this trend might be due to higher food availability in the summer. Sasaki and Obara
(2001) examined the interaction of season and food availability on the queen’s egg laying
decisions. They found that when colonies are continuously supplied with food
throughout the year, queens lay more drone eggs in the spring and summer than in the
fall. Additionally, while the queen lays few drone eggs in the fall independent of food
condition, in the late spring she appears to produce a higher proportion of male eggs
during good food conditions than during poor food conditions. However, that study used
a within-subjects design and did not control for the order of food condition treatments.
Therefore, how the queen alters her production of drone eggs in response to food
availability should be tested in future studies. Additionally, future studies might examine
whether the queen’s production of drone eggs changes in response to colony size.

The honeybee queen’s production of drone eggs is regulated by negative
feedback. Studies concluding this had found that queens who had recently laid drone
eggs subsequently produced fewer drone eggs than queens who had recently laid only
worker eggs (Sasaki et al. 1996; Wharton et al. 2007). In the study by Wharton et al.
(2007), the authors followed the eggs to the pupal stage and found that the queen’s egg-

laying decisions inﬂuenced colony-level drone production patterns. Additionally, the

IS

authors controlled for season, colony size, and food availability, and found that the
decisions of the queens were not constrained by worker decisions to ﬁll cells with nectar.
That study therefore demonstrated that the queen altered her egg-laying patterns by
detecting and responding to either her own previous egg-laying experience or the
presence of drone brood in the colony. Which of these two factors was a greater
determinant of queen egg-laying patterns is an attractive topic for future study.

Overall, how queens alter drone production in response to environmental factors
is an area that has been largely overlooked. We discuss reasons for why this might be in

section 4 of this chapter.

3.4 Workers rear/fail to rear larvae

Workers sometimes consume a portion of their colony’s developing larvae
(Woyke 1977). Workers presumably eliminate larvae when environmental conditions are
unfavorable for rearing brood. These actions might serve to recycle colony resources
before the larvae become too expensive for the colony to support (Webster et al. 1987).
Consistent with this idea, workers selectively destroy younger larvae, which would cost
more to rear to maturity than older larvae (Woyke 1977; Schmickl and Crailsheim 2001;
Schmickl and Crailsheim 2002; Schmickl et al. 2003). Additionally, although workers
rarely consume pupae, they preferentially consume younger rather than older pupae
(Newton and Mich] 1974).

The tendency of workers to eliminate males ﬂuctuates over the season and with

the colony’s queen status. During the spring and fall, workers consume more drone

l6

brood than worker brood, but in the summer the survivals of drone and worker brood are
low and approximately equal (Fukuda and Ohtani 1977; Woyke 1977). These results
suggest that workers consume more drone larvae in the fall when drones are of little
value to the colony. In queenless colonies, the survival of queen-laid drone and worker
larvae increases (Woyke 1977), although once workers begin to lay eggs, many worker-
laid drone eggs are destroyed (Miller and Ratnieks 2001).

The tendency of workers to eliminate males is inﬂuenced by the amount of drone
brood in the colony. This was conﬁrmed in a recent study by Wharton et al. (in review).
They manipulated the abundance of older male brood in colonies and found that the
survival of younger male larvae was lower when the abundance was increased than when
it was decreased. Meanwhile, the survival of worker larvae was high across colony
conditions. These results suggest that the brood rearing decisions of workers assist the
colony in regulating its investment in male reproductive function. This study is one of
only a few to date that directly compares the survival of drone and worker larvae in an
experimental manipulation.

The tendency of workers to eliminate males likely is inﬂuenced by colony food
conditions, but how food conditions affect the relative rates of cannibalism for drone and
worker larvae awaits future study. The tendency of workers to consume worker larvae
increases during periods of poor food availability (Schmickl and Crailsheim 2001).
Interestingly, under poor colony food conditions, the workers preferentially consumed
younger (less than 3 days old) rather than older (4-5 day) larvae and also preferentially
withheld nursing ﬂom the younger larvae (Schmickl and Crailsheim 2001; Schmickl and

Crailsheim 2002; Schmickl et al. 2003). The colonies in these studies contained only

17

worker brood, so it is unknown how the workers would have treated drone larvae under
similar conditions.

In general, the circumstances under which workers cannibalize larvae are poorly
understood. This is largely because very few studies to date compare the survival of
drone and worker larvae in a manipulative experiment. An exception to this is the study
mentioned above by Wharton et al. (in review). I suggest that we need more
manipulative studies to determine whether and under what conditions workers selectively
destroy drone or worker larvae. To selectively destroy either drone or worker larvae,
workers would need to distinguish between the two. While it is unknown if workers can
distinguish between drone and worker larvae based on cell size alone, workers are likely
not able to distinguish between the two sexes based on chemical cues until the larvae are
3-4 days old, which is approximately seven days after the queen lays eggs (Sasaki et al.

2004). Much more work remains to be done in this area.

3.5 Workers maintain or evict drones

Besides producing drones, colonies maintain adult drones. Adult drones are
costly for the colony to maintain because a drone consumes approximately four times the
amount of food that a worker consumes (Winston 1987). Colonies will maintain a drone
until one of the following happens: the drone mates successfully with a virgin queen
(drones die upon mating), the drone dies, or workers evict the drone ﬂom the colony in a

process known as “drone eviction.”

18

During drone eviction, workers force drones out of the colony and prevent them
ﬂom reentering (Free 1957). Since the evicted drones cannot access the colony’s food
stores, they soon die of starvation. A drone’s age appears to affect his likelihood of being
evicted; Free (1957) observed that workers selectively harassed older drones while
permitting younger drones to remain. Drone eviction takes place gradually over the
autumn and is most likely to occur on warmer days when the workers are more active in
the colony (Morse et a1. 1967). Although the exact cues that trigger drone eviction are
unknown, it is thought that workers evict drones when the expected cost of maintaining
drones outweighs the expected ﬁtness beneﬁt.

Since the ﬁtness beneﬁt of keeping and maintaining drones ﬂuctuates with
environmental conditions, colonies exhibit greater levels of drone eviction under
particular environmental circumstances. Workers evict more drones in the fall than in the
spring (Morse et al. 1967); drone eviction before the fall rarely occurs. During the fall,
queenright colonies evict more drones during periods of poor food availability (Free and
Williams 1975). Additionally, when foraging conditions are poor in the fall, queenright
colonies evict more drones than do queenless colonies; when foraging conditions are
good, the colony’s queen status does not affect the amount of drone eviction (Free and
Williams 1975).

There is still much to learn about drone eviction. In particular, it is unknown how
colony size (or a reduction in colony size following a swarm event) affects drone
eviction. Observations support the idea that colony size might play a role: when ﬂames
of brood and bees were brought into a small (2-ﬂame) observation hive during mid-

summer, workers soon evicted the drones Wharton, personal observation), although this

19

doesn’t always happen (Free 1957). Additionally, future research should better elucidate
the proximate cues that trigger the workers to recognize that their colony has an excess

number of drones.

4. Conclusions

Drones are an important but costly investment for the honeybee colony, so
colonies regulate the number of drones they produce. This regulation of drone
production arises ﬂom the actions and interactions of the queen and workers, who must
respond to environmental factors that affect the expected ﬁtness value of rearing drones.
To date, research with honeybees has elucidated the responses of colony members to
several environmental inﬂuences. However, this work is by no means exhaustive, and
has only begun to shape our understanding of honeybee brood production dynamics and
colony organization. Future work in this area is crucial in order to better understand how
the individuals in social insect colonies coordinate their actions to achieve adaptive
colony decisions.

The emerging picture of brood production in honeybees is that the queen and
workers share control over their colony’s sex allocation. Historically, the queen was
viewed as an egg-laying “machine” that did not regulate colony—level patterns of sex
allocation. Recent evidence suggests that this previous view is incorrect; the queen’s
egg-laying decisions can shape her colony’s regulation of investment in male
reproductive function (see section 3.3). Additionally, workers play an important role in

colony sex allocation patterns by building and preparing wax cells to hold the brood,

20

rearing or destroying the larvae, and maintaining or evict adult drones (sections 3.1, 3.2,
3.4, 3.5). Together, the decisions of the queen and workers provide a colony with many
stages and strategies by which it might adjust sex allocation in response to environmental
conditions.

In many social insect species, the queen and workers are expected to favor
different sex allocation optima (Free 1957; Hamilton 1964; Trivers and Hare 1976;
Strassmann 2001). This queen-worker conﬂict over sex allocation arises because of the
haplodiploid system of sex determination, by which workers are typically 3 times more
related to their sisters than to their brothers whereas the queen is equally related to her
daughters and sons. In species characterized by queen-worker conﬂict, the extent to
which colony members can inﬂuence sex allocation provides insights into the distribution
of power within the colony, where power is deﬁned as the ability to do or act in situations
of conﬂict (Beekman et al. 2003; Beekman and Ratnieks 2003).

In species in which the queen mates with many males (as in the honeybee),
queen-worker conﬂict over sex allocation is theoretically very small or absent (Moritz
1985; Ratnieks et al. 2006). Species that exhibit an extremely high queen mating
ﬂequency include the honeybee (N eumann and Moritz 2000), the Florida harvester ant
Pogonomyrmex badius (Rheindt et al. 2004), and the Aﬂican army ant Dorylus anomma
(Kronauer et al. 2004). In these species, queens and workers theoretically “agree” on
their colony’s sex allocation. In species characterized by an absence of queen-worker
conﬂict, the extent to which colony members can inﬂuence sex allocation provides
important insights into how colonies efﬁciently and ﬂexibly respond to environmental

conditions.

21

Whether or not there is conﬂict between the queen and workers over colony sex
allocation, the study of queen and worker inﬂuence on colony sex allocation patterns has
great potential to advance our understanding of the dynamics and organization of social
insect colonies. Increasingly, researchers are recognizing social insects as an excellent
model system for addressing questions about how living systems organize their
individual components in order to cope with a dynamic environment (Camazine et al.
2001). Research that elucidates the effects of queen and worker decisions on colony sex
allocation will therefore broaden our understanding of how social insect colonies, and in
turn other complex living systems, coordinate their group members in order to cope with
environmental uncertainty. This research is also likely to provide key insights into the

behavioral ecology of social insects, and more broadly, the evolution of cooperation.

22

 

CHAPTER 2
THE HONEYBEE QUEEN INFLUENCES THE REGULATION
OF COLONY DRONE PRODUCTION

Wharton KE, Dyer FC, Huang ZY, Getty T (2007) The honeybee queen inﬂuences the
regulation of colony drone production Behavioral Ecology 18:1092-1099.
Oxford University Press.

23

Introduction

Social insect colonies, like many individual organisms, must allocate resources
appropriately between grth and reproduction. How colonies manage to do this is not
well understood, because little is known about how colony life history decisions are
governed by the actions and interactions of colony members (Bourke and Franks 1995).
Recent work has begun to focus on how one colony-level reproductive decision, the
investment in males, is inﬂuenced by the queen and the workers (Herbers et al. 2001;
Beekman et al. 2003; Beekman and Ratnieks 2003; Mehdiabadi et al. 2003; Pen and
Taylor 2005; Ratnieks et al. 2006). However, the attempt to understand how the queen
and workers cooperate when allocating colony resources to reproduction is complicated
by the fact that their actions may instead spring ﬂom conﬂict.

A colony produces males through the actions of the queen and the workers, who
are potentially in conﬂict over their colony’s sex allocation, or investment in males
versus females (Ratnieks et al. 2006). While there are other types of queen-worker
conﬂict (see Ratnieks et al. 2006), we consider only conﬂict over sex allocation for this
paper. The potential for this conﬂict is due to the haplodiploid system of sex
determination, whereby the queen is equally related to her sons and daughters whereas
workers are on average more closely related to their sisters than to their brothers
(Hamilton 1964). When colonies have one queen that is singly mated, workers are three
times more related to their sisters than to their brothers, leading to the prediction that the
queen and workers are in conﬂict over their colony’s sex allocation and that each party
should attempt to bias sex allocation in its favor (Trivers and Hare 1976). One way in

which workers might do so is by selectively eliminating male larvae, and workers of

24

several species of ants apparently do this (Aron et al. 1994; Aron et al. 1995; Keller et al.
1996a; Keller et al. 1996b; Sundstrbm et al. 1996). Given that workers in some species
appear to control the colony’s investment in males, whether queens can inﬂuence male
investment by modulating their egg sex ratio has recently gained interest (Aron et a1.
1995; Keller et al. 1996b; Cremer and Heinze 2002; Beekman and Ratnieks 2003;
Mehdiabadi et al. 2003; de Menten et al. 2005a; de Menten et al. 2005b; Koedam et al.
2005).

Although the queen and workers might experience conﬂict, their inclusive ﬁtness
is determined largely by colony success, so they share an interest in adaptively regulating
the colony’s investment in reproduction. So, while the ability of the queen or the workers
to inﬂuence their colony’s investment in males is consistent with the hypothesis that they
are in conﬂict, an alternative explanation is that queens and workers are assessing local
information about the value of rearing males and are cooperatively but sequentially
shaping the colony’s allocation to reproduction. For example, the queen might modulate
her egg sex ratio to ensure that the colony is well-supplied in males, and workers might
eliminate males later if conditions become unfavorable for raising sexuals. We term this
hypothesis queen-worker cooperation. Because queens and workers might agree over
their colony’s allocation to reproduction but also experience conﬂict over sex allocation
(Herbers et al. 2001), queen—worker conﬂict and queen-worker cooperation are not
mutually exclusive scenarios. Any ability of queens or workers to inﬂuence their
colony’s male investment may therefore reﬂect a strategy of biasing sex allocation in
their favor (conﬂict), a strategy of obtaining the colony’s adaptive allocation to

reproduction (cooperation), or some mixture of both. For this reason, it is difﬁcult to

25

interpret the functional signiﬁcance of queen and worker sex-biasing abilities when there
is a high potential for queen-worker conﬂict.

While cooperation between queens and workers is likely to be entangled with
conﬂict in many species, highly polyandrous species should have little potential for
conﬂict over sex allocation (Ratnieks et al. 2006). This is because when the queen mates
with many males, as in honeybees (Estoup et a1. 1994; Neurnann and Moritz 2000),
Aﬂican army ants (Kronauer et a1. 2004), and harvester ants (Rheindt et al. 2004), the
average relatedness between workers is low so workers are almost equally related to their
brothers and sisters. As a result, the queen and workers should favor a nearly equal
investment in the sexes (Moritz 1985; Ratnieks et al. 2006). Therefore, highly
polyandrous species provide an opportunity to study whether queens and workers have
evolved the ability to inﬂuence the colony’s investment in males when conﬂict over sex
allocation is not in the picture. In this paper, we investigate one such species, the
honeybee, and ask whether the queen has the ability to inﬂuence colony patterns of drone
(male) production.

The size of the drone population in a honeybee colony depends upon a sequence
of actions taken by the queen and workers (Figure 2.1). With each action, there is an
opportunity for the bees to alter the investment in drones. Workers construct the wax
combs in which brood are reared, and thus control the colony’s investment in drone and
worker cells (Pratt 2004). Drone cells are larger than worker cells and allow drones to
develop to full size; therefore, the number of drone cells provides an upper limit for the
number of drones that can be reared at any one time. Additionally, workers use some

cells for the storage of pollen and nectar, and therefore determine the availability of

26

    

 

 

V
Workers Workers Workers
Build wax cells Fill (with food) Lays eggs in cells Rear or fail to Maintain or evict
(drone or worker) or clean cells (drone g worker) rear larvae adult drones

 

Figure 2.1. Drone production and maintenance in a honeybee colony: a sequence of
actions taken by the workers and the queen. At each step, the bees have an opportunity to
inﬂuence the colony’s investment in drones. Arrows indicate typical sequences of events.
Drone cells are larger than worker cells, and drones are normally reared in these larger
cells (I). The worker and drone cells are used either for food storage (nectar and pollen;
darker cells in diagram) or are cleaned in preparation for rearing brood (II). The queen
typically lays unfertilized (drone) eggs into drone cells, and fertilized (female) eggs into
worker cells (111). Workers then tend the larvae, but don’t necessarily raise all of the
larvae to adulthood (IV). Adult drones remain colony members until they mate with a
virgin queen in the population, die, or are evicted by workers (V). In this study, we focus
primarily on the modulation of drone production that arises ﬂom actions taken by the
queen (step 111).

27

empty cells that can receive an egg. The queen may then use these empty cells for
depositing eggs. She produces a female or male egg depending on whether or not she
fertilizes the egg, and generally lays female eggs in worker cells and male eggs in drone
cells (Ratnieks and Keller 1998). The queen sometimes lays eggs in a drone to worker
ratio that deviates ﬂom the ratio of available drone to worker cells (Henderson 1991),
suggesting that she might be able to control the sex of her offspring by selecting what
type of comb to use. Next, the larvae are tended by the worker bees, who can decrease
investment in drones or workers by failing to provision larvae or by consuming them
(Free and Williams 1975; Schmickl and Crailsheim 2001; Sasaki et al. 2004). Finally, the
drones that reach adulthood may be evicted ﬂom the colony by the worker bees in the
late fall or when foraging conditions are poor (Morse et al. 1967; Free and Williams
1975). Because workers initially construct the brood cells and can later alter any queen
investment patterns by decreasing the number of immature or adult drones, it is widely
assumed that workers control colony drone production.

The value of drones varies with environmental factors. In order to increase colony
efficiency by minimizing the number of drones that will be destroyed by the workers, a
queen might be expected to adjust her egg-laying patterns to achieve her colony’s current
optimum investment in drones. Season and energy budget are important factors that
affect colony drone production (Free and Williams 1975; Seeley and Mikheyev 2003),
and there is some evidence that the queen alters her drone egg production in relation to
these (Sasaki and Obara 2001). In addition, colony drone production is inhibited by the
presence of drone brood (Free and Williams 1975), indicating that colonies regulate

drone production through a negative feedback process. Since this drone brood normally

28

originates ﬂom eggs produced by the queen, queens might use oviposition history or the
presence of drone brood as an indicator of how many drone eggs to lay. In accordance
with this, Sasaki et a1. (1996) provide evidence that suggests that honeybee queens
decrease subsequent investment in drones if they recently laid drone eggs. However,
because the authors used the same queens in multiple treatment conditions staged over
time, their data failed to exclude both season and colony energy budget as explanatory
variables for the observed queen egg-laying response. Thus, whether queens can
adaptively regulate drone egg production remains an open and relevant question.

In this study, we used a between-subj ects design to test whether honeybee queens
regulate investment in drones. We manipulated the previous egg-laying experience of
queens by either allowing or preventing them ﬂom laying drone eggs for a period of time,
and we then provided them the opportunity to lay both drone and female eggs. If the
queen’s production of drone eggs is regulated by negative feedback, we expected that
queens who were allowed to lay drone eggs for a period of time would subsequently lay

fewer drone eggs compared to queens who were initially prevented ﬂom laying drone

eggs.

Methods

Honeybee colonies

We conducted this study in the summers of 2004 and 2005 in the countryside

surrounding the Michigan State University campus in East Lansing, Michigan. Twenty

29

four colonies of Italian honeybees, Apis mellifera ligustica, were used; 12 each summer.
Because each colony had been purchased as a package in the late spring of the year it was
used in the study, no colony was used twice, and all were approximately the same size
(~13,000 bees). Each colony was housed in a standard Langstroth hive consisting of one
deep hive body with ten ﬂames of worker comb. By initially providing the colonies with
worker comb only, we ensured that drone production occurred mainly during the

experimental periods.

General methods

The general strategy was to conﬁne half of the queens to drone comb and half to
worker comb for a period of time, and then to compare the egg-laying patterns of the two
groups of queens later, when both types of comb were available.

We randomly assigned colonies to one of two treatments, drone comb (DC) or
worker comb (WC). The treatment labels refer to the type of comb to which we conﬁned
the colony’s queen before moving her onto a ﬂame with both types of comb. Therefore,
before they were moved, queens in WC colonies (WC queens) could not produce viable
drone offspring whereas queens in DC colonies (DC queens) could. These treatments
were applied on June 18 in 2004 and on June 16 in 2005. We label those dates as “day 0”
of the experiment.

On day 0, we conﬁned each queen to a ﬂame of the randomly assigned type of
treatment comb (drone or worker). The cage used to conﬁne the queens was made ﬂom

queen excluder material, which has small openings through which worker bees can ﬂeely

30

pass but the larger queen cannot. We then placed the cage in the colony and positioned it
toward an edge of the broodnest, which is where colonies normally rear both drone brood
and worker brood.

On day 4, we veriﬁed that both the WC and DC queens had laid many eggs on the
treatment ﬂame (average proportion of ﬂame area covered with eggs, WC: 0.58, range
0.21-0.91, DC: 0.48, range 0.12-0.84). We then transferred each queen onto a new ﬂame
composed of equal areas of drone and worker comb (the choice ﬂame). We placed the
choice frame into the queen excluder cage and placed the cage at the other edge of the
broodnest. Queens are able to switch between laying drones and workers multiple times
while successively laying eggs (Sasaki et al. 1996), and this occurs over just a few
minutes or hours (Wharton, unpublished data). Following the methods of Sasaki et al.
(1996), we allowed the queens to lay eggs on this choice ﬂame for 24 hours.

On day 5, after we removed each queen ﬂom the cage and placed her back into
the rest of the colony, we estimated (see below) the number of drone and worker eggs on
the choice ﬂame. We used an estimate rather than an actual count because it was faster
and thus reduced the exposure of the eggs to the hot surmner weather. After making this
estimate, we placed the cage containing the choice ﬂame back in the colony. This
prevented the queen ﬂom laying additional eggs on this ﬂame, but allowed the workers
to provision larvae that hatched ﬂom these eggs. We removed the queen excluder cage
ﬂom the colonies on day 7 and 15 of the study in 2004 and 2005, respectively.

To estimate the numbers of drone and worker eggs laid in the choice ﬂame, we
covered the ﬂame with a grid made ﬂom 1.27 x 1.27 centimeter hardware cloth, and

counted the number of grid squares containing at least one cell with an egg in the drone

31

comb portion of the choice ﬂame and the number of grid squares containing at least one
cell with an egg in the worker comb portion of the choice ﬂame. This gave us an estimate
of the area of drone and worker comb covered with eggs; this technique is commonly
used to obtain estimates of areas of comb covered with young brood (J efﬂee 1958). Since
drone and worker cells are different sizes, we multiplied the area of drone and worker
comb by the number of cells in that area (using the convention of 2.60 drone cells per
cm2 and 4.29 worker cells per cmz; Page and Metcalf, 1984) to obtain an estimate for the
number of drone and worker eggs laid by the queen, providing a more intuitive notion of
queen behavior. This transformation ﬂom area to eggs is likely to give a slight
overestimate of the actual number of eggs, but the amount of overestimation should be
similar for both treatment groups. Because this transformation is applied equally across
treatment groups and is linear, it does not affect our analyses or conclusions.

As we were collecting the data on eggs laid by the queen, we noticed that some of
the cells in the choice ﬂame had been ﬁlled with nectar, which would have deterred
queens ﬂom laying eggs in these cells. We used the method just described to estimate the
maximum area (in cm2) of drone and worker comb that was used for nectar storage.

After the eggs on the treatment and choice ﬂames reached the pupal stage (and
before these pupae emerged as adults), we removed the ﬂames ﬂom each hive and took
digital photographs of them. We used these photographs to count the number of drone
and worker pupae (indicated by capped cells; worker pupae are indicated by ﬂat cappings
and drone pupae are indicated by raised cappings) on both the treatment and choice
ﬂames for each colony. We noticed that there were some worker pupae on the drone

comb treatment ﬂames and on the drone comb half of the choice ﬂames, indicating that

32

the queen had occasionally laid worker eggs in drone cells (as reported in Page et. al.
1993). This occurrence was extremely rare on the choice ﬂames: of the colonies that had
pupae on the drone portion of the choice ﬂame, the proportion of these pupae that were
worker pupae was less than 0.029 in all but three of the colonies. In two of the three
remaining colonies, the proportions of pupae that were worker were 0.051 and 0.105, and
the remaining colony had only one pupa (a worker pupa) on the drone comb portion of
the choice ﬂame. Since the occurrence of worker pupae in drone cells was so rare on the
choice ﬂames, we are conﬁdent that our estimate of the number of drone eggs on the
choice ﬂames closely approximates the actual number of drone eggs present. In 2004, we
photographed the treatment ﬂames on day 17 and the choice ﬂames on day 20. In 2005,

we photographed both the treatment and choice ﬂames on day 15.

Drone departure rates

Although we did not provide our colonies with any drone comb prior to the start
of the experiment, colonies often build small patches of drone comb in gaps and spaces
within the hive. Our colonies had done so and consequently had reared a small number of
adult drones before the start of the experiment. Since the presence of adult drones might
inﬂuence a colony’s future production of drones (Rinderer et al. 1985) and therefore
might inﬂuence the egg-laying decisions of the queen, we measured whether the DC and
WC colonies had similar numbers of adult drones near the beginning of the study. To
estimate the prevalence of adult drones, we measured the rate of drone departures ﬂom

each colony. An observer sat near the entrance of a hive, recorded the number of drones

33

leaving during a short period of time (either 1 or 5 minutes), and visited each hive in a
ﬁxed order until each was visited for 10 minutes total. We then converted these counts to
the rate of drone departure per minute. We made these observations on sunny afternoons,
when adult drones in the hive embark on either practice or real mating ﬂights. In 2004,
we performed these counts on days 1 and 7 of the study. In 2005, we performed these
counts on days 3 and 7 of the study. Thus, for each colony, we obtained two rates of

drone departures, which were averaged to obtain the colony’s drone departure rate.

Statistical analysis

Every queen laid eggs on both the treatment and choice ﬂames, so we included all
of the colonies (n = 24) in our analysis. Since we collected data in both 2004 and 2005,
we ﬁrst checked for differences in our data between years. For each treatment group, we
found no differences in our dependent variables across years, so we pooled our data ﬂom
2004 and 2005 for the statistical analyses.

To test whether DC and WC queens differed in egg-laying decisions on the choice
ﬂame, we used a MANOVA to test for the effect of treatment on queen decisions on the
choice ﬂame (number of worker eggs, number of drone eggs). A queen may alter her
production of drone eggs through two methods: by skewing the sex ratio of her eggs
while holding total eggs constant, or by increasing her total production of eggs while
holding her egg sex ratio constant. Whereas a test for treatment differences in egg sex

ratio can only capture the former method, a MAN OVA, which tests for differences in

34

centroid locations among treatment groups, can capture either method. We used post-hoe
analyses to provide further insights into the decisions made by the queens.

To assess whether the egg patterns we observed on the choice ﬂame may have
arisen because of worker decisions to ﬁll areas of comb with nectar, we used Student’s t-
test to assess the difference between DC and WC colonies with respect to the areas of
both worker comb and drone comb that were used for nectar storage.

To test whether DC and WC colonies differed in pupae-rearing patterns on the
choice ﬂame, we used a MANOVA to test for the effect of treatment on colony decisions
on the choice ﬂame (number of worker pupae, number of drone pupae). We used post-
hoc tests to provide ﬁrrther insights into the colony decisions.

To assess whether worker honeybees might have altered patterns of queen
investment differentially across treatments, we used Student’s t-test to determine the
difference between DC and WC colonies with respect to both drone mortality and worker
mortality on the choice ﬂame. We deﬁned drone mortality and worker mortality as the
proportion of drone and worker eggs, respectively, that did not reach the pupal stage. We
obtained each hive’s measure of drone mortality by taking the difference between the
numbers of drone eggs and drone pupae and dividing that by the number of drone eggs;
worker mortality was obtained in a similar way with worker eggs and pupae. For this
analysis, we used data ﬂom 2005 only, because in that year we kept the choice ﬂame in
the queen excluder cage for long enough to ensure that any pupae we counted on our
digital pictures were ﬂom eggs that the queen laid while she was conﬁned to that ﬂame.

Finally, we used Student’s t-test to determine the statistical signiﬁcance of the

difference between the DC and WC colonies with respect to both drone departure rate

35

and the total investment in pupal males by the colonies (the sum total of drone pupae on
the treatment and choice ﬂames).
All analyses were performed using MATLAB 7.0. Signiﬁcance was set at the

0.05 level, and results are reported as mean :I: 1 standard deviation.

Results

Does previous comb type inﬂuence the queen’s egg-laying patterns?

Once moved to the choice ﬂames, WC queens and DC queens differed in their
egg-laying decisions (MANOVA, Wilks’ A132 = 0.717, P = 0.030; Figure 2.2).
Speciﬁcally, WC queens laid a greater number of drone eggs than did DC queens
(number of drone eggs, WC queens: 925.0 :1: 665.4, DC queens: 337.9 i 390.3; F132 =
6.950, P = 0.015). However, WC and DC queens did not differ in the number of worker
eggs they produced (number of worker eggs, WC queens: 558.7 :t 534.0, DC queens:
762.3 i 466.1; F132 = 0.995, p = 0.329).

To see if the difference in queen behavior was due to queens skewing their egg
sex ratio, we compared the egg sex ratio (male eggs divided by total eggs) across
treatment groups. WC queens tended to lay a more drone-biased sex ratio (0.59 i: 0.40)
than did DC queens (0.35 i 0.36), although this trend was not signiﬁcant (Flgz = 2.340, P
= 0.140). To see whether the difference in queen behavior was due to queens altering
their total production of eggs, we compared the number of eggs produced across

treatment groups. WC queens laid a signiﬁcantly larger number of total eggs than did DC

36

1 200 i i 1

 

 

 

 

 

 

 

 

 

 

D WCqueens //
I DCqueens /
1000- / .
+/
/
.3 .
g 800- /
‘4 /
5 500' // ‘
S / '
to /
C!
LS, 400' // T
/
/
/
200’ / -
/ "
/_..«
U! r r r r r
0 200 400 500 800 1000 1200

Eggs on worker comb

Figure 2.2. Mean (:tSE) number of drone eggs plotted against mean (:I:SE) number of
worker eggs on the choice ﬂame for queens previously on worker comb (WC) and
queens previously on drone comb (DC). WC queens laid more drone eggs but the same
amount of worker eggs as DC queens. The lines represent the egg sex ratio vector on the
choice ﬂame; the slope is the average ratio of drone eggs to worker eggs (dashed line for
WC queens, dotted line for DC queens). WC queens and DC queens did not produce
signiﬁcantly different egg sex ratios.

37

queens (total eggs, WC queens: 1483.8 i 337.4, DC queens: 1100.2 i 439.7; F132 =
5.747, P = 0.025). This egg production, which occurred over a period of 24 hours, is
consistent with reported values of queen egg production, which usually range between

1000 and 2000 eggs per 24 hours (Nolan 1925; Seeley 1985).

Do the egg—laying patterns on the choice frame result from the actions of the workers?

The results we reported in the above section could arise ﬂom the workers rather
than the queen if the egg-laying patterns were simply a byproduct of one of the following
(refer to Figure 2.1): worker decisions to selectively clean drone or worker cells, worker
decisions to selectively ﬁll drone or worker cells with nectar, or worker decisions to
selectively destroy the queen’s drone eggs or worker eggs. We address each of those
possibilities in this section.

First, the queen’s egg-laying pattern might arise ﬂom worker decisions to
selectively clean drone or worker cells. Workers clean cells by removing debris such as
old cocoons, and this cleaning process prepares a cell to receive an egg ﬂom the queen
(Winston 1987). One way to control for this possibility would be to conﬁne the queen to
comb in the absence of workers, but workers tend the queen in many ways, including
feeding her. When workers are absent queens lay very few eggs at all (Wharton,
unpublished data), but that is a highly unnatural situation and need not have anything to
do with the effects of the workers on the cells. In our study, the ﬂames we used had been
cleaned by other colonies and stored indoors before they were used for our experiment,

so were most likely already prepared to receive eggs ﬂom the queens. Even if workers do

38

play a role in preparing cells for egg-laying, previous work has demonstrated that
workers do not differentially prepare worker or drone cells (Sasaki et al. 1996). In
summary, no evidence suggested that workers in our study selectively prepared drone or
worker cells for the queen.

Second, the queen’s egg-laying pattern might arise ﬂom worker decisions to
selectively ﬁll drone or worker cells with nectar. For example, if the workers in WC
colonies ﬁlled fewer drone cells with nectar than did the workers in the DC colonies, the
signiﬁcant treatment difference in drone egg production might arise because WC queens
had more cells in which to lay drone eggs than did DC queens. However, the area of
drone comb ﬁlled with nectar did not differ between the two treatment groups (area of
drone comb ﬁlled with nectar, WC colonies: 17 a: 20 cm2, DC colonies: 25 d: 34 cm2; t =
0.659, df = 22, P = 0.516; Figure 2.3). Similarly, the area of worker comb ﬁlled with
nectar did not differ between the two treatment groups (area of worker comb ﬁlled with
nectar, wc colonies: 325 i 187 cm2, DC colonies: 219 i 126 cm2; t = 1.625, df= 22, P =
0.118; Figure 2.3). Furthermore, since each choice ﬂame had 825.8 cm2 of worker comb
and 825.8 cm2 of drone comb, queens in both treatment groups still had plenty of
available cells in which they could lay eggs. This trend was particularly notable in the
drone comb section of the ﬂame, where on average less than three percent of the total
area of drone comb was used for nectar storage, leaving almost all of the drone cells
available to the queen. Thus, queens were able to make egg-laying decisions
independently of worker decisions to ﬁll cells with nectar.

Third, the egg-laying patterns we observed might have resulted ﬂom worker

decisions to selectively destroy some of the drone eggs. For example, our observation

39

E

 

- Worker comb
[: Drone comb -

sass

u—L
01
D

100

Area of comb with nectar (square centimeters)
8

 

 

 

WC colonies DC colonies

Figure 2.3. Mean (iSE) area of drone comb (light bars) and worker comb (dark bars)
ﬁlled with nectar in the choice ﬂames for WC colonies and DC colonies. The choice
ﬂames had 825.8 cm2 drone comb and 825.8 cm2 worker comb. WC colonies and DC
colonies did not differ in the area of drone comb ﬁlled with nectar or in the area of
worker comb ﬁlled with nectar.

40

that there were more drone eggs in WC colonies than in DC colonies might be explained
by a tendency of the workers in DC colonies to selectively destroy drone eggs. We
sampled the eggs when they were 0-24 hours old, so we cannot exclude the possibility
that workers removed drone eggs before we took our estimates. However, in order for the
workers to selectively destroy drone eggs, they would need to perceive the sex of each
egg. They might distinguish the sex of eggs based on chemical cues, but there is no
difference in removal rates for drone and worker eggs when both are transferred into
drone cells (Oldroyd and Ratnieks 2000). Furthermore, there is evidence that workers
cannot use brood pheromones to distinguish between drone and worker larvae until seven
days after the queen has laid eggs (Sasaki et al. 2004). Alternatively, workers might
distinguish the sex of eggs based on cell size alone, but Sasaki et al. (2001) observed
food-starved observation hives and did not witness any cannibalization of eggs in drone
cells. Additionally, Woyke (1977) demonstrated that there was no difference in removal
rates for eggs in drone and worker cells during the spring or summer (when our study was
conducted), and although he found selective removal of immature drones in the spring, a
sex-speciﬁc difference in removal rate didn’t occur until the larval stage. For all of these
reasons, we ﬁnd it unlikely that workers in our study were selectively removing drone
eggs. In general, whether social insect workers can distinguish the sex of eggs remains an
open question in social insect biology (Passera and Aron 1996), and is an attractive area
for future research. Other social insect studies have measured egg sex ratios ﬂom eggs
that were 0-96 hours old and had been exposed to workers (Aron et al. 1994; Aron et al.
1995; Sasaki et al. 1996; Sasaki and Obara 2001), so the methods in our study are

consistent with other studies that examine queen egg-laying decisions.

41

Because it is unlikely that workers in different treatment groups differentially
prepared cells, ﬁlled cells with nectar, or destroyed eggs, the egg-laying patterns we
observed on the choice ﬂames were unlikely due to the actions of the workers, but

instead were due to the actions of the queen.

Does previous comb type influence pupae-rearing patterns of the colony?

On the choice ﬂames, WC colonies and DC colonies differed in their pupae-
rearing patterns (MANOVA, Wilks’ A122 = 0.639, P = 0.009; Figure 2.4). Speciﬁcally,
WC colonies raised a greater number of drones than did DC colonies (number of drone
pupae, WC: 549.6 i 351.8, DC: 134.0 :E 210.8; F132 = 12.322, P = 0.002) and also
produced a more male-biased sex ratio (pupal sex ratio, WC: 0.58 :t 0.36, DC: 0.24 a:
0.26; F132 = 5.701, p = 0.028). WC and DC colonies did not differ in the number of
worker pupae they produced (number of worker pupae, WC: 458.0 :1: 453.6, DC: 475.6 i
544.7; F132 = 0.007, p = 0.932) or in the total number of pupae reared on the choice
ﬂame (number of total pupae, WC: 1007.6 i 564.3, DC: 609.6 :1: 628.0; F122 = 2.667, p =
0. l 1 7).

Thus, just as WC queens had produced more drone eggs on the choice ﬂames
than did DC queens, WC colonies produced more drone pupae on the choice ﬂames than
did DC colonies. This result indicates that the negative feedback pattern of drone egg

production persisted at the colony level.

42

 

1200 . . i 1 - .

 

 

 

 

 

 

 

 

Cl WC colonies //
I DC colonies / .
1ooo- / -
/
/
/
. / -
800 /
° /
E. /
:- 500- / -
§ /
r: /
400- /
/ .."'..-..
20m // +
/ + ‘
UK”... 1 J r 1 r
0 200 400 500 803 1000 1 200
Workerpupae

Figure 2.4. Mean (iSE) number of drone pupae plotted against mean (:l:SE) number of
worker pupae on the choice ﬂame for colonies with queens previously on worker comb
(WC) and colonies with queens previously on drone comb (DC). WC colonies raised
more drone pupae but the same amount of worker pupae as DC colonies. The lines
represent the pupal sex ratio vector on the choice ﬂame; the slope is the average ratio of
drone pupae to worker pupae (dashed line for WC colonies, dotted line for DC colonies).
WC colonies produced a more male-biased sex ratio than did DC queens.

43

Does previous comb type inﬂuence drone mortality or worker mortality?

These results are for data taken in 2005. We did observe both drone and worker
mortality on the choice ﬂame, meaning that our estimates of drone and worker eggs were
higher than the numbers of drone and worker pupae we counted. Speciﬁcally, the WC
colonies failed to rear 54.2 i 13.9 percent of the drone eggs and 53.2 d: 12.0 percent of
the worker eggs on the choice ﬂames, and the DC colonies failed to rear 57.0 i 28.7
percent of the drone eggs and 60.0 i 24.1 percent of the worker eggs on the choice
ﬂames. Some of this estimated mortality is most likely a byproduct of our overestimation
of the number of eggs (see methods), which is expected to be equal for the two treatment
groups. The remainder of the estimated mortality might have arisen ﬂom the failure of
the worker honeybees to rear some of the drone and worker larvae. If workers were
failing to rear brood in a treatment-dependent manner, we should see a difference in
mortality between the treatment groups. However, treatment had no effect on the amount
of drone mortality (t = 0.190, df = 7, p = 0.855) or the amount of worker mortality (t =
0.619, df = 10, p = 0.550). This suggests that any decisions that the workers made about
drone or worker removal were independent of the queen’s recent egg-laying decisions.
Therefore, the net differences between treatments in the pupae-rearing patterns of the

colonies can be at least partially attributed to actions of the queens.

Did the colonies diﬂer in their investment in drones throughout the study?

We measured a drone departure rate at the beginning of the experiment of 0. 1 6 i
0.37 drones per minute in the WC colonies and 0.27 :l: 0.25 drones per minute in the DC
colonies. These rates are not signiﬁcantly different (t = 0.816, df = 22, P = 0.423),
indicating that the two treatment groups had roughly the same number of adult drones
present at the beginning of the experiment. Therefore, any treatment differences in queen
egg-laying decisions or colony brood-rearing patterns did not result ﬂom a difference in
abundance of adult drones in the colonies.

By the end of the experiment (day 20 in 2004 and day 15 in 2005; see methods),
we could determine the total number of drone pupae reared throughout the study by
summing the drone pupae on both the treatment and the choice ﬂames. WC and DC
colonies did not differ in this total (total number of drone pupae produced during the
experiment, WC colonies: 564.8 d: 343.4, DC colonies: 593.3 1 548.5; t = 0.153, df =
18.5, P = 0.880). Therefore, even though WC colonies were given only one half of a
ﬂame of drone comb whereas DC colonies were given three times that amount, the
treatment groups produced a similar ﬁnal number of pupal drones. This suggests that the

colonies regulated their production of drones.

Discussion

It is often assumed that worker honeybees regulate their colony’s investment in
drones, since they build the wax cells for rearing drones and also tend to the larvae. The
results of this study suggest that honeybee queens also contribute to the regulation of

their colony’s drone production, through modulation of their egg-laying decisions.

45

Queens that were prevented ﬂom laying drone eggs for a brief period of time (WC
queens) later produced more drone eggs than queens who had not been prevented ﬂom
laying drone eggs (DC queens). As the eggs on the choice ﬂame matured into pupae,
there is no evidence that the workers selectively killed the larvae in a treatrnent-
dependent manner, so the heightened production of drone eggs by WC queens (when
given a choice of egg type to lay) allowed the WC colonies to “catch up” on the number
of drones that were being reared. As a result, both treatment groups reared approximately
the same total number of drone pupae throughout the experiment.

Our ﬁnding that the colonies invested equally in pupal males over the duration of
the experiment is consistent with studies indicating that colonies across a population
produce similar amounts of drone brood (Page and Metcalf 1984; Henderson 1991). Our
colonies were approximately equal in size and therefore are expected to invest equally in
drones. It is remarkable that our colonies were able to accomplish this even though they
differed greatly in the amount of drone comb available (DC colonies were provided with
three times the amount of drone comb as WC colonies). Therefore, our study provides
further evidence that honeybee colonies are adept at regulating drone production.

As for how this regulation of drone production occurs, our study suggests a
greater role for the queen than is often assumed. Previous studies have shown that the
presence of drone brood suppresses the further production of drone brood in honeybee
colonies (Free and Williams 1975); brood production might also be suppressed by a large
quantity of adult drones in the colony (Rinderer et. al., 1985; but see Henderson, 1991).
Together, these studies suggest that colonies regulate drone production via a negative

feedback mechanism. However, these previous studies assessed colony drone production

46

by counting numbers of drone larvae or pupae, which may reﬂect a combination of queen
and worker decisions. Therefore, these studies do not directly address the role that the
queen plays in regulating colony drone production. Our study ﬁlls this gap by focusing
on whether the queen regulates her drone egg production. We found that the queen
compensates for her own production of drone eggs based on the amount of drone eggs
she recently laid. Furthermore, the workers in our study did not block (via using cells for
nectar storage) or alter (via failing to rear larvae) the queen’s decisions in any sort of
treatment-dependent manner. Taken together, these results suggest that the regulation of
drone brood production at the colony level may emerge at least in part by a negative
feedback process of drone egg production by the queen.

Queens might increase their drone egg production by laying a greater total
number of eggs, by creating a more drone-biased egg sex ratio, or some combination of
these two. In our study, queens did differ in the total number of eggs they laid, which was
due to a difference in the number of drone but not worker eggs that were laid. The
difference in egg sex ratio was not statistically signiﬁcant, but we did ﬁnd a trend in the
predicted direction, in that queens who were prevented ﬂom laying drone eggs
subsequently produced a more drone-biased egg sex ratio. Because we had a relatively
low sample size and queen behavior was quite variable, we suspect that this sex ratio
trend may reﬂect a biologically real behavior. In short, honeybee queens can modulate
their investment in drones by altering the total number of drone eggs they lay and
possibly also by changing their egg sex ratio.

Whether social insect queens can adaptively regulate the egg sex ratio is an

important line of future research. This ability has been found in queens of species that

47

have a high potential for queen-worker conﬂict over sex allocation (Aron et al. 1995;
Keller et al. 1996b; de Menten et al. 2005a; de Menten et al. 2005b). In species that
experience this queen-worker conﬂict, the queen may manipulate her egg sex ratio in
order to exercise power, where power is deﬁned as the ability to control reproduction
when conﬂict exists (Beekman et al. 2003; Beekman and Ratnieks 2003). Alternatively,
she might manipulate her egg sex ratio in order to inﬂuence the colony’s allocation to
reproduction in agreement with the workers. Future studies that examine whether queens
can inﬂuence colony male production should attempt to distinguish between these two
ﬁmctional hypotheses.

Our study opens up some fascinating questions that deserve further work. For
example, what proximate cues does the queen use to alter her egg-laying decisions in
relation to previous egg-laying decisions? As one possibility, a queen might use some
type of memory to keep track of previous ovipositions, and could stop producing more
drone eggs after she remembers laying many drone eggs. Alternatively, queens might use
brood pheromone cues to indicate the amount of drones in her colony, and could stop
producing more drone eggs when her chemical sensory system detects a large amount of
drone brood. In our study, we did not remove any eggs before allowing queens to make
egg-laying choices, so our experimental design does not allow us to distinguish between
the hypotheses that queens are using memory or brood pheromones. Identifying the
proximate mechanisms by which social insect queens keep track of the colony’s
abundance of males is an attractive study for future research, because it has implications
for the ability of queens to react to male removal by the workers. In species that

experience conﬂict, for example, if workers selectively remove males, we predict that the

48

queen might not rely on oviposition memory because it will mislead her estimate of the
number of developing male larvae.

This work sets the stage for examining how the actions and interactions of the
queen and workers inﬂuence a colony’s investment in reproduction. Investing in males
can be likened to a supply chain management issue for the colony, where a major goal
should be to produce an adaptive amount of a product (males) through the most efﬁcient
use of resources. Brood production should be an efﬁcient process if both the workers and
the queen are able to actively and separately respond to environmental conditions that
convey information about the value of rearing males. In honeybees, some of these
conditions include season, colony size, availability of food, whether or not a queen is
present, and the number of drones present (Free and Williams 1975; Page and Metcalf
1984; Seeley and Mikheyev 2003). How do colony members perceive information about
these conditions? Which conditions are assessed by the queen, and which are assessed by
the workers? Additionally, environmental conditions that favor drone production might
change after the queen has laid drone eggs. Will workers modify the previous egg-laying
decisions of their queen if environmental conditions suddenly become unfavorable for
rearing drones? These questions, which resemble questions ﬂom supply chain
management, must await further work.

In conclusion, our study identiﬁes a role for the honeybee queen in the regulation
of colony drone production. This demonstrates that a social insect queen can inﬂuence
colony male production even when conﬂict with the workers is absent. In order to better
understand the functional signiﬁcance of queen and worker decisions, future work should

attempt to distinguish between actions that arise ﬂom conﬂict over sex allocation and

49

those that arise ﬂom cooperation and shared control over the colony’s investment in

reproduction.

50

CHAPTER 3
MALE ELIMINATION IN THE HONEYBEE

Wharton KE, Dyer FC, Getty T (In press) Male elimination in the honeybee. Behavioral
Ecology. Oxford University Press.

51

Introduction

Cooperation and conﬂict between subtmits of a group are pervasive forces at
multiple levels of biological organization, ﬂom genomes to organisms to animal
societies. Understanding how cooperation arises and is maintained in spite of the
potential for intragroup conﬂicts is a central quest in evolutionary biology (Queller and
Strassmann 1998; Reeve and Hblldobler 2007). In animal societies, individuals often
have competing interests that can lead to potential conﬂicts over the division of resources
or reproduction among society members (Ratnieks et al. 2006). However, it can be tricky
to determine the extent to which group member interactions are shaped by conﬂict
because an observed behavior might be inﬂuenced by multiple environmental, genetic,
and evolutionary factors (Chapuisat and Keller 1999). In this paper we examine a
behavior that is often assumed to spring ﬂom genetic conﬂict between the queen and
female workers in a social insect colony, the elimination of immature males by the
workers. Most studies of male elimination have focused on species in which queen-
worker conﬂict is expected to be high. To determine whether genetic conﬂict is a
necessary precondition for male elimination, we studied honeybees (Apis mellifera), a
species where such conﬂict is expected to be minimal, and in which any evidence of male
elimination is likely to reﬂect cooperative resource allocation to male reproductive
function.

In the social Hyrnenoptera (ants, bees, wasps), the queen and workers are
potentially in conﬂict over their colony’s sex allocation, or investment in males versus
females. Although there are several types of queen-worker conﬂict (Ratnieks et al. 2006),

we consider only conﬂict over sex allocation in this paper. The potential for this conﬂict

52

arises ﬂom haplodiploid sex determination (females are diploid; males are haploid),
which causes workers to be, on average, more closely related to their sisters than to
brothers, whereas the queen is equally related to her daughters and sons (Hamilton 1964).
These patterns of relatedness lead to the prediction that workers should generally prefer a
less male-biased sex allocation than the queen (Trivers and Hare 1976).

Workers might bias sex allocation in their favor by eliminating a portion of the
colony’s immature males. Male elimination has been inferred to occur in several ant
species, based on observations that the secondary sex ratio was less male-biased than the
egg sex ratio (Aron et al. 1994; Aron et al. 1995; Keller et al. 1996a; Passera and Aron
1996; Sundstrbm et al. 1996; Chapuisat et al. 1997; Helms et al. 2000; Rosset and
Chapuisat 2006). Since male elimination allows workers to bias sex allocation in their
favor, these examples are, perhaps not surprisingly, consistently cited as evidence of a
manifest queen-worker conﬂict (Mehdiabadi et al. 2003; Ratnieks et al. 2006).

Although male elimination is consistent with the predictions of queen-worker
genetic conﬂict, it might also serve a cooperative purpose. Workers have an opportunity
to modify their colony’s investment in reproduction when they rear brood; these brood-
rearing decisions might allow the colony to adaptively tailor reproductive effort to
resource availability or other environmental conditions (Wharton et al. 2007). For
example, workers might eliminate some immature males when colony resources are
limited or when the colony is already well-supplied with males. By eliminating males
before they extract full deve10pmental costs and begin extracting high adult maintenance
costs, workers could enhance overall colony efﬁciency and success (Chapuisat et al.

1997)

53

So far, all reported cases of male elimination have been in species where the
queen is singly mated and so the potential for conﬂict is high. Is genetic queen-worker
conﬂict therefore a necessary precondition for male elimination? To determine this, it is
important to investigate whether there is selective male elimination in species in which
the queen mates multiple times. Multiple mating by the queen reduces the average
relatedness among workers and lowers the potential for conﬂict (Moritz 1985; Pamilo
1991b; Pamilo 1991a). If male elimination were solely a consequence of queen-worker
conﬂict, then we should not see it in such species. Only one study examined a casein
which conﬂict was expected to be low (in wood ant colonies headed by a multiply mated
queen); it found no evidence for selective male elimination (Sundstrdm et al. 1996).
Clearly, however, there is a need for data ﬂom other species.

Honeybees are a good candidate species for testing whether male elimination
might spring ﬂom cooperation. Honeybee queens are highly polyandrous (Tarpy and
Nielsen 2002), so queen-worker conﬂict is expected to be very low or absent (Moritz
1985; Ratnieks et al. 2006). Additionally, colonies do not rear as many drones (males) as
possible but rather regulate their investment in drones (Free and Williams 1975),
presumably because drones are costly to produce (Seeley 2002; Seeley and Mikheyev
2003; Hrassnigg and Crailsheim 2005). Conveniently, good estimates of brood survival
are possible in honeybees because researchers can reliably and nondestructively
determine the sex of eggs: drone eggs are found in large “drone” cells whereas worker
eggs are found in small “worker” cells (Ratnieks and Keller 1998). This difference in cell
size also lends itself to experimental manipulations; researchers can limit or manipulate a

colony’s production of drones by controlling the abundance of drone cells in the colony.

54

In this study, we compared the survival of young immature drone and worker
honeybees under two different experimental conditions: one in which colonies had an
excess of older drone brood (D+ colonies) and one in which colonies had virtually no
additional drone brood (D- colonies). See Figure l for the experimental design. This
design was chosen because the presence of drone brood (immature drones) inhibits a
colony’s future production of drones (Free and Williams 1975). The treatments were
induced after the queen had laid both drone and worker eggs in order to isolate the
inﬂuence of worker actions on brood survival. We predicted that if workers adaptively
inﬂuence the colony’s investment in drones by eliminating immature drones, the survival
of drone but not worker eggs would be lower in colonies that already were rearing large

numbers of drones.

Methods

Honeybee colonies

We conducted this study on the Michigan State University campus in East
Lansing, Michigan in May and June, 2007. Twenty colonies of Italian honeybees, Apis
mellifera ligustica, were used. Each colony contained the mother queen and roughly
15,000 workers, and was housed in a standard Langstroth hive consisting of one deep
hive body with ten ﬂames of worker comb. By initially providing the colonies only with

worker comb, we ensured that drone production occurred mainly during the experiment.

55

General methods

Our general strategy was as follows. First, in each of the study colonies we
allowed the queen to lay drone and worker eggs in the presence of older drone brood
(immature drones). Then, we divided the colonies into two groups, one ﬂom which drone
brood was removed and the other to which drone brood was added. Finally, we compared
the survival of the new worker and drone brood during the period in which the colony’s
workers rear the larvae. The experimental setup is presented in Figure 1.

To set up the experiment, we placed one ﬂame of drone comb and one ﬂame of
worker comb (the “treatment” ﬂames) into each colony. These ﬂames were used to
manipulate the abundance of drone brood for the experimental treatments. After ten days,
these ﬂames were largely ﬁlled with brood consisting of older larvae and some pupae.

To obtain eggs so that we could monitor the brood-rearing decisions of the
workers, we inserted into each colony a “focal frame” consisting of drone and worker
comb in blocks of equal area. On day 0 of the experiment, we caged the queen on this
comb and allowed her to lay eggs for 24 hours. The cage was made of queen excluder
material, which consists of small openings that prevent the queen but not the workers
ﬂom passing through. At the end of this 24-hour egg-laying period, we placed the queen

elsewhere in the colony and took a series of photographs that facilitated counting of

56

Drone brood removed (D-)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Drone brood added (D+)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

oi in; A

 

 

 

 

 

 

 

 

 

 

 

Figure 3.1. Experimental design, ﬂom a bird’s eye view of the moveable ﬂames of comb
in a colony. Dark gray bar = focal ﬂame (composed of equal areas of drone and worker
comb). Light gray bars = treatment ﬂames (D = drone, W = worker). White bars = other
ﬂames in colony (contained mostly worker brood, pollen, honey, and empty worker
cells). The dashed line indicates the queen excluder cage and the triangle ([1) indicates
the position of the queen. At the beginning of experiment (left), the queen was placed on
the focal ﬂame in the queen excluder cage and permitted to lay drone and worker eggs.
Next, the queen was placed elsewhere in the colony and each colony was randomly
assigned to one of two treatment groups (right). The treatment was applied by exchanging
ﬂames between the two groups of colonies. In the drone brood removed (D-) group, the
colony lost its drone treatment ﬂame, and a worker treatment ﬂame was inserted in its
place. In the drone brood added (D+) group, the colony received an additional drone
treatment ﬂame in the place of a worker treatment ﬂame.

57

numbers of worker and drone eggs laid by the queen. We then returned each of the focal
ﬂames to the cage in its colony, allowing worker access but this time with the queen
excluded so that no further eggs could be laid on this comb.

To test whether the workers would adaptively modify their colony’s investment in
drones, we randomly assigned the colonies to one of two treatments, drone brood added
(D+) or drone brood removed (D-). Each D+ colony received a drone treatment ﬂame
ﬂom a D- colony and gave a worker treatment ﬂame to that D- colony in return. Thus,
after this treatment was applied, the D+ colonies had two ﬂames of older drone brood
whereas the D- colonies had none. We applied this treatment over the course of days 2
and 3.

To quantify larvae for estimating brood survival, we took photographs of the focal
ﬂame when the brood had reached the late larval stage and were just prior to capping
(day 8; 8-9 days after oviposition). In order to quantify the pupae on the treatment
ﬂames, we took photographs of the treatment ﬂames twice: once on day 3 (shortly after
the treatment had been applied), and once at the conclusion of the experiment, on day 8
(for the worker treatment ﬂames) and day 9 (for the drone treatment ﬂames).

Before the start of the experiment, our colonies had built small patches of drone
comb in gaps within the hive. Consequently, our colonies contained a small background
number of adult drones during the experiment. Since adult drones might inﬂuence drone
production (Rinderer et al. 1985), we recorded the number of drones embarking on
mating ﬂights to determine whether the D+ and D- colonies had similar numbers of adult
drones (see Seeley 2002). On day 8, an observer sat near the entrance of a hive, recorded

the number of drones that left the colony during a 5 minute interval, and visited each hive

58

in a ﬁxed order until each was visited for 10 minutes total. We used these counts to

calculate a rate of drone departures per minute.

Survival of immature drones and workers

To calculate survival of brood on the focal ﬂame, we used photographs to
quantify (blindly) both the eggs initially present and the larvae that remained after our
experimental manipulation. We used photographs because direct counts would have
taken too long and increased the likelihood that the eggs would die ﬂom exposure to the
summer heat. The larvae we counted were at a late (5-6 day) larval stage; nearly all
larvae that old survive to eclosion (W oyke 1977; Schmickl and Crailsheim 2001;
Schmickl and Crailsheim 2002).

Counting eggs ﬂom photographs is prone to error because of the small size of the
eggs, their lack of color contrast with the wax cells, their position at the bottom of a cell,
and various irregularities of the cells themselves. We expected to undercount the actual
number of eggs on the focal ﬂame. However, because the tendency to undercount
depends mostly on wax comb properties it should not be systematically biased across
treatment groups.

Drone or worker survival on the focal ﬂame was estimated ﬂom the proportion of
eggs in each type of cell that reached the late larval stage. These counts did reveal that we
may have undercounted eggs at the beginning. We counted more worker larvae than
worker eggs in 3 of our colonies (467 larvae, 443 eggs; 593 larvae, 588 eggs; 235 larvae,

74 eggs), and we counted more drone larvae than drone eggs in 2 of our colonies (728

59

larvae, 634 eggs; 584 larvae, 573 eggs). For these cases we assigned a worker or drone

survival value of 1.

Statistical analysis

There were no drone eggs on the focal ﬂame in 2 of our 20 colonies, so we
excluded those colonies ﬂom the analyses. Within each treatment group, all dependent
variables were checked for normality using a Lillefors test. For any dependent variable
that did not meet the criterion for normality, the appropriate nonparametric test was used
for hypothesis-testing. All analyses were performed using MATLAB 7.0, and results are

reported as mean i 1 standard deviation.

Results

Comparisons of the number of eggs on the focal frame and adult drones in the colony

Since the experimental treatment was randomly assigned after the queen had laid
eggs on the focal ﬂame, we did not expect any differences between the treatment groups
in the number of eggs laid. Consistent with this, D- and D+ colonies did not differ in the
number of drone eggs (D- colonies: 456.2 :1: 318.8, D+ colonies: 333.7 :1: 277.5; t = 0.868,
df = 16, P = 0.398) or in the number of worker eggs (D- colonies: 797.8 :1: 473.9, D+

colonies: 619.1 2t 275.8; t = 0.978, df= 16, P = 0.343) on the focal ﬂame. Furthermore,

60

there was no difference in the egg sex ratio (proportion of drone eggs, D- colonies: 0.38 i
0.24, D+ colonies: 0.30 :l: 0.21, Mann-Whitney U test, P = 0.489).

We also found that the abundance of adult drones (produced prior to the
beginning of the experiment) did not differ between treatment groups (drone departures
per minute, D- colonies: 0.51 d: 0.60 , D+ colonies: 0.60 i 0.83; Mann-Whitney U test, P
= 0.774).

Because colonies did not differ signiﬁcantly in the production of drone or worker
eggs on the focal ﬂame, or in the abundance of adult drones, any treatment differences in

colony brood-rearing patterns cannot be attributed to those factors.

Comparisons of brood survival

D- colonies and D+ colonies differed with respect to brood survival (consisting of
2 variables: drone survival, worker survival) on the focal ﬂame (MAN OVA, Wilks’
Aw, = 0.566, P = 0.014) (Fig. 2). Speciﬁcally, a greater proportion of drone eggs were
reared to the late larval stage in D- colonies than in D+ colonies (drone survival, D-
colonies: 0.83 i 0.18, D+ colonies: 0.43 i 0.34; Fm, = 9.29, P = 0.008). D- and D+
colonies did not differ in the proportion of worker eggs that were reared to the late larval
stage (worker survival, D- colonies: 0.78 i 0.16, D+ colonies: 0.72 i 0.25; FL”, = 0.22, P
= 0.642). This indicates that our treatment affected the survival of immature drones but
not workers.

To further investigate how our colonies reared worker brood versus drone brood

on the choice ﬂame, we examined the relationship between drone survival and worker

61

 

O
O

D

(.0
O

I

O
CO
0%

 

 

 

 

to
35
N
iii
'5
Z
2
O
E
B O
E or . O ‘
E 0.5- "
3 0.5- ‘
g 'I
3 o 4 - b '
g I.

.L
-'a 0.3- ‘
‘3 C
.E 0.2 - I -
E- 01 _ O Drone brood removed (D-) . ..
E O Drone brood added (0+)
1 U 1 1 I I ‘

 

 

 

 

l l J I _.
0 0.1 0.2 0.3 0.4 0.5 0.5 0.7 0.8 0.9 1
Proportion of worker eggs surviving to late larval stage

Figure 3.2. The survival of worker eggs and drone eggs to the larval stage on the focal
ﬂame in colonies that had drone brood removed (open circles) and drone brood added
(ﬁlled circles). The overall mean (:1: standard error) survival for each treatment group is
indicated by the correspondingly shaded squares and error bars. The survival patterns
differed signiﬁcantly between the two treatment groups (Wilks’ A1,” = 0.566, P = 0.014).
Speciﬁcally, the survival of worker eggs to the larval stage did not differ between
treatment groups (Fmo = 0.22, P = 0.642), but the survival of drone eggs to the larval
stage was signiﬁcantly lower in the colonies to which drone brood was added than in
colonies ﬂom which drone brood was removed (Fun = 9.29, P = 0.008).

62

survival within each treatment group. There was no correlation between drone survival
and worker survival in the D- colonies (r = 0.168, p = 0.666). However, in the D+
colonies, drone survival and worker survival tended to be negatively correlated, although
this tendency was not statistically signiﬁcant (r = -0.556, p = 0.120).

Our D+ colonies varied in the number of drone pupae on the treatment ﬂames.
We tested whether this variation across the D+ colonies (9 colonies) explained any
variation in drone survival on the focal ﬂames in those colonies. We found no correlation
between drone survival and the number of drone pupae on treatment ﬂames at the end of
the experiment (D+ colonies: r = 0.496, p = 0.175).

To test whether the treatment caused a shift in sex ratio ﬂom the egg stage to the
larval stage, we calculated the change in sex ratio for each focal ﬂame by subtracting the
larval sex ratio ﬂom the egg sex ratio. This shift in sex ratio was not signiﬁcantly
affected by the treatment (change in sex ratio; a positive change indicates a less male-
biased larval sex ratio, D- colonies: -0.03 i 0.05, D+ colonies: 0.03 i 0.15, Mann-

Whitney U test, P = 0.063).

Discussion

This study provides evidence for conditional male elimination by honeybee
workers and is one of only a few studies to date that uses an experimental approach to
investigate the expression of this behavior. In colonies with an excess of older drone
brood, the survival of newly produced drone brood was lower than in colonies with an

absence of older drone brood. There was no effect of treatment on the survival of worker

63

brood (Fig. 2). These patterns of brood survival indicate that honeybee colonies, like
individual organisms, regulate investment in male reproductive function.

As for how this regulation occurs, this study points to an important role for the
workers. Previous work has shown that the presence of drone brood suppresses the
further production of drone brood in honeybee colonies (Free and Williams 1975),
suggesting that colonies regulate drone production via a negative feedback mechanism.
Colonies manage to do this through some combination of queen and worker decisions. In
this study, we isolated and examined worker inﬂuences on brood patterns by removing
any direct inﬂuence of the queen: treatment groups were assigned after the queen laid
eggs, and the queen was prevented ﬂom laying additional eggs on the focal ﬂame
following treatment assignments. Furthermore, the treatment difference in drone but not
worker survival indicates that brood rearing decisions by the workers rather than an
overall diﬁerence in brood viability was responsible for the patterns. In short, the
treatment differences in drone survival demonstrate that honeybee workers can modify
their colony’s investment in immature drones after the queen lays eggs.

From the perspective of the colony, brood production should be an efﬁcient
process if both the workers and the queen are able to respond actively and separately to
environmental conditions that convey information about the value of rearing males. The
results of this study, together with a previous study by Wharton et al. (2007), suggest that
both the honeybee queen and workers can inﬂuence their colony’s investment in males.
There is evidence that this shared or joint control over brood production is found in other
species, including the harvester ant Messor pergandei (Ode and Rissing 2002) and the

ﬁre ant Solenopsis invicta (Aron et al. 1995; Passera et al. 2001). Shared control might in

64

some cases reﬂect an evolutionary tug-of-war between the queen and workers for control
over colony sex allocation (Trivers and Hare 1976; Reuter and Keller 2001). This shared
control might arise because neither the queen nor the workers are able to gain complete
control over sex allocation in species in which queen-worker conﬂict is expected. For
example, in the ﬁre ant, it appears that the queen is constrained by the ability of the
workers to eliminate males, and that the workers are in turn constrained by the primary
sex ratio produced by the queen (Aron et al. 1995; Passera et al. 2001). In other cases,
including in species without queen-worker conﬂict, shared control might allow colonies
to respond ﬂexibly and efﬁciently to changing environmental conditions such as
ﬂuctuations in food availability (Ode and Rissing 2002; Wharton et a1. 2007). This
cooperative view might help to explain various phenomena such as why the Linepithema
humile queen continues to produce male eggs outside the period of sexual production
even though workers eliminate the vast majority of those eggs (Aron et al. 1994; Passera
and Aron 1996).

Sibling cannibalism occurs in other animals, including amphibians and predatory
birds, during periods of low food availability. Although this sibling cannibalism is
traditionally considered a form of parent-offspring conﬂict, researchers suspect that it
might serve an adaptive purpose for both parents and offspring in that it promotes a better
distribution of limited resources among offspring (Mock and Forbes 1995). Similarly,
parent-offspring conﬂict between a social insect queen and workers has typically served
as an explanation for the ability of social insect workers to modify their colony’s
investment in males (Mehdiabadi et al. 2003; Ratnieks et al. 2006). Indeed, the results

ﬂom Sundstrom et al. (1996) lend strong support to a role for such conﬂict in some

65

colonies of the wood ant Formica exsecta. Until our study, all known instances of male
elimination in social insects have been reported in species with a high potential for
queen-worker conﬂict (Aron et al. 1994; Aron et al. 1995; Keller et al. 1996a; Passera
and Aron 1996; Sundstrdm et al. 1996; Chapuisat et al. 1997; Hehns et al. 2000; Rosset
and Chapuisat 2006). Our evidence for male elimination in honeybees, a species with
little or no potential for queen-worker conﬂict (Moritz 1985; Ratnieks et al. 2006),
suggest that it might be more appropriate to evaluate its function in the context of overall
colony performance.

Colony performance is shaped in large part by the ways in which resources are
allocated between individuals within the society. The ways in which resources are
distributed to brood are sometimes likely to provide beneﬁts in terms of colony efﬁciency
and performance rather than in terms of direct beneﬁts to kin (Chapuisat et al. 1997; Aron
et al. 2001; Linksvayer 2008). Our results are consistent with the notion that male
elimination in honeybees allows workers to adaptively modify the queen’s egg laying
decisions in a way that increases colony efficiency. By eliminating excess male brood,
the workers might channel resources to more pertinent aspects of colony reproduction or
survival, such as additional females or an increased ability to retain colony food reserves.
Although we did not measure such life history measures of colony performance in our
study, previous studies with honeybees suggest that rearing drones does not trade off with
a colony’s production of workers, but might negatively impact the colony’s ability to
stockpile honey for winter survival (Allen 1963; Seeley 2002).

A major question emerging ﬂom our study is whether an ability of the social

insect queen or workers to inﬂuence colony sex allocation reﬂects within-colony genetic

66

conﬂict or shared interests over the colony’s resource allocation to male reproductive
function. Because a mixture of the two explanations might apply in a species with a high
potential for queen-worker conﬂict, disentangling the relative effects of each might be
challenging. To do so, researchers must consider both environmental and genetic factors
that favor conﬂict or cooperation within groups (Ratnieks and Reeve 1992; Reeve and
Hblldobler 2007; Linksvayer 2008). These environmental and genetic inﬂuences on
colony dynamics should be teased apart and investigated using manipulative experiments
(Sundstrom et al. 1996; Chapuisat and Keller 1999). Finally, researchers should consider
proximate factors, such as the costs of brood manipulation, that might constrain the
ability of the queen and workers to inﬂuence sex allocation (Nonacs and Carlin 1990;
Chapuisat et al. 1997; Chapuisat and Keller 1999; Beekman et al. 2003; Beekman and
Ratnieks 2003; Reuter et a1. 2004).

In conclusion, social insect biologists should exercise caution when interpreting
male elimination as being indicative of a manifest queen-worker conﬂict. We echo the
plea of Chapuisat and Keller (1999) for additional manipulative experiments, which
should lead to a better understanding of the evolutionary factors that promote conﬂict and

cooperation and the ways in which those forces shape social dynamics.

67

CHAPTER 4
THERE IS MORE TO SEX ALLOCATION THAN THE SEX RATIO

Introduction

Darwin (1874) was puzzled by how natural selection could shape the ability of an
organism to adjust its offspring sex ratio, or the proportion of its offspring that are male.
Over 130 years later, work on sex ratios has greatly advanced our understanding of how
organisms respond to variable reproductive opportunities, including competition with
relatives for mates (Hamilton 1967) or resources (Clark 1978), environmental conditions
(Trivers and Willard 1973), and within-family conﬂicts (Trivers and Hare 1976). Sex
ratio theory is successful due largely to the development of models that generate testable
qualitative and quantitative predictions for behavior (West et a1. 2000). However, sex
ratio models collapse two dimensions of information [number of males, number of
females] to one dimension [number of males/(number of males + number of females)],
discarding information about a potentially important aspect of reproductive investment:
clutch size. In this paper we assert that clutch size is often an important dimension of
reproductive effort and that biologists can gain a better understanding of their study
system by focusing on two-dimensional sex allocation decisions rather than on one-
dimensional sex ratio decisions.

Clutch size, or the sum total of male and female offspring, is an important ﬁtness
component of animals (Lack 1947). Life history theory predicts that an iteroparous

organism might increase its ﬁtness by producing more offspring during a better than

68

average reproductive opportunity (Steams 1992). This theory has been supported by
empirical work in a wide range of taxa including birds, insects, and mammals (West and
Sheldon 2002; Holand et al. 2006). Recent empirical evidence suggests also that females
in some species have the ability to adjust both their offspring clutch size and sex ratio
(Olsson et al. 2005; Dubois et al. 2006). Despite the potential importance of clutch size,
it is often ignored or treated only as a nuisance covariate in sex ratio research (Krackow
and Tkadlec 2001; Hardy 2002). In fact, many researchers statistically control for
differences in clutch sizes rather than treat clutch size as potentially important.
Increasingly, however, researchers are recognizing that clutch size is not irrelevant, and
that in many cases sex ratio and clutch size are interdependent factors that should be
examined simultaneously (N agelkerke 1994; Greeff 1997).

The theories of optimal offspring sex ratio and clutch size are uniﬁed in sex
allocation theory. Sex allocation is deﬁned as the allocation of resources to male versus
female reproductive function (Chamov 1982). Although sex allocation covers many
topics in biology, in this paper we focus on the area of sex allocation research that refers
to an organism’s total investment in male versus female offspring. An organism’s sex
allocation can be characterized by two simple variables: the number of male [m] and
number of female [f] offspring that it produces. Note that these two variables ﬁrlly
determine both sex ratio [m/(m+f)] and clutch size [m+f]. In contrast to sex ratio theory,
sex allocation theory therefore asserts that a clutch of 3 males and 6 females is not
equivalent to a clutch with 4 males and 8 females. Despite the potential usefulness of
examining the raw numbers of male and female offspring, publication titles reveal that

biologists tend to think primarily in terms of sex ratios rather than sex allocation (Figure

69

4.1). We suggest that our understanding of sex allocation will be aided by applying
models and analyses that use the raw numbers of males and females, rather than the ratio
of males to females, as the primary variables of interest.

Our goal in this paper is to convince readers that analyses of reproductive
decisions should consider both allocation to sex ratio and clutch size (or equivalently the
total nmnber of males and females) before reducing reproductive decisions to the single
dimension of sex ratio. To illustrate this point, we use existing datasets documenting
offspring production by two species of haplodiploid insects: honeybee queens (Wharton
et al. 2007) and parasitoid wasps (Shuker et al. 2004). The results ﬂom both examples
demonstrate that analyzing sex allocation behavior leads to fundamentally different

biological conclusions than those drawn ﬂom analyzing sex ratio behavior.

Example 1: Offspring sex allocation by honeybee queens (Apis mellifera)

Overview of study

Social insect colonies, comprised of many individuals, must allocate resources to
life history traits like growth, survival, and reproduction. How colonies achieve this
through the actions and interactions of colony members remains a puzzle (Bourke and
Franks 1995). In a recent study, Wharton et al. (2007) showed that the honeybee queen is
able to inﬂuence her colony’s sex allocation by inﬂuencing her colony’s regulation of
male (drone) production. Here we brieﬂy review the details of that study and highlight
the differences that arise ﬂom analyzing the queen’s ability to inﬂuence her colony’s sex

allocation versus sex ratio.

70

250 . .
- Behav Ecol Sociobiol
Anim Behav

I: Behav Ecol

 

 

     

 

200

 

 

150

100

Number of Articles

50

 

 

Sex Ratio Sex Allocation
Term in Article Title

Figure 4.1. Number of published articles (between the years 1965 and March 2008)
listed in the ISI Web of Science that include either the term “sex ratio(s)” or “sex
allocation” in the article title. The data in this ﬁgure are limited to three journals that
specialize in behavioral ecology: Behavioral Ecology and Sociobiology, Animal
Behavior, and Behavioral Ecology.

71

Wharton et al. (2007) tested whether the honeybee queen is able to use previous
egg-laying experience to regulate her production of male and female eggs. To do this,
the researchers either allowed queens to lay male eggs, or prevented them ﬂom doing so,
for a period of time, and examined their subsequent tendency to lay male or female eggs.

Refer to Wharton et al. (2007) for speciﬁc methodological details.

Results

To test whether the queens in the two treatment groups differed in their egg sex
ratio (number of male eggs divided by the total number of eggs), we used an ANOVA.
Queens that had previously been prevented ﬂom laying drone eggs (WC queens)
produced an egg sex ratio (mean :I: 1 standard deviation) of 0.59 i 0.40, whereas queens
that had previously laid drone eggs (DC queens) produced an egg sex ratio of 0.35 i
0.36. These egg sex ratios were not signiﬁcantly different (F132 = 2.340, P = 0.140;
Figure 4.2).

To test whether the queens in our two treatment groups differed in their egg sex
allocation (number of female eggs, number of male eggs), we used a multivariate analysis
of variance (MANOVA). A MANOVA tests for differences in the centroid locations of
the dependent variables among the treatment groups. We then used post hoc analyses to
provide further insights into the decisions of the queens.

The egg sex allocation produced by the queens was signiﬁcantly affected by the
experimental treatment (Wilks’ A132 = 0.717, P = 0.030; Figure 4.3). Speciﬁcally, WC

queens laid a greater number of drone eggs than did DC queens (number of drone eggs,

72

 

0.9 _

'5' 0.8 ..

.O .0
O) “-sl
T I
-—I
l I

 

Q
.1:

 

O

(,0
l-————l

1

Sex ratio (proportion mal
C)
01

o 0
'_~. ix)
I T
I I

 

 

 

 

 

 

O

 

WC queens DC queens

Figure 4.2. Mean (3: SE) sex ratio of honeybee queens that had previously been
prevented ﬂom laying drone eggs (WC queens) and queens that had previously been

allowed to lay drone eggs (DC queens). The egg sex ratios are not signiﬁcantly different
(F132 = 2.340, P = 0.140).

73

1200

 

 

I3 WC queens .
I DC queens

 

 

 

1000

I

r“:
Lr'

 

800 - .
600 ~ -

400 ~ -

Eggs on drone comb

200 - -

 

 

 

0 200 400 600 800 1000 1200
Eggs on worker comb

Figure 4.3. Mean (t SE) sex allocation of honeybee queens that had previously been
prevented ﬂom laying drone eggs (WC queens) and queens that had previously been
allowed to lay drone eggs (DC queens). The egg sex allocation patterns are signiﬁcantly
affected by the treatment (Wilks’ A132 = 0.717, P = 0.030). WC queens laid more drones
eggs but the same amount of worker eggs as DC queens (see text).

74

WC queens: 925.0 :E 665.4, DC queens: 337.9 i 390.3; F122 = 6.950, P = 0.015).
However, WC and DC queens did not differ in the number of worker eggs they produced
(number of worker eggs, WC queens: 558.7 i 534.0, DC queens: 762.3 d: 466.1; F132 =
0.995, p = 0.329). Overall, WC queens laid a signiﬁcantly larger number of total eggs
than did DC queens (total eggs, WC queens: 1483.8 i 337.4, DC queens: 1100.2 d: 439.7;

F132 = 5.747, P = 0.025)

Discussion

The conclusions that we draw ﬂom this study are fundamentally different
depending on whether we analyze the reproductive decisions with a sex allocation or sex
ratio approach. The offspring sex allocation of honeybee queens was inﬂuenced by
previous egg-laying experience. Speciﬁcally, queens who initially laid drone eggs
subsequently laid fewer drone eggs than the queens who were initially prevented ﬂom
producing drone eggs. This striking pattern was revealed by our test for an effect of
treatment on egg sex allocation (number of worker eggs, number of drone eggs).
However, when we discarded information and collapsed our data to the single dimension
of sex ratio (drone eggs/total eggs), our test failed to ﬁnd any signiﬁcant differences
between the treatment groups.

There are biological reasons to suspect a discemable treatment difference in sex
allocation but not sex ratio behavior in honeybees. Colonies that are not allowed to
invest in drones do not display an increased investment in workers (Allen 1963; Page and
Metcalf 1984), suggesting that there is not a colony-level tradeoff between producing
drones and workers. Instead, there appears to be a tradeoff between a colony’s

investment in drones and the size of its honey reserve, a key determinant of overwinter

75

survival (Seeley 2002). Since a colony’s investment in drones likely does not directly
trade off with its investment in workers, examining a colony’s sex ratio might not be
biologically appropriate. Instead, we suggest that researchers should examine a colony’s
joint investment in drones and workers (sex allocation), which provides a more complete

picture of the colony’s joint investment in reproduction and growth.

Example 2: Offspring sex allocation by female parasitoid wasps (Nasonia

vitripennis)
Overview of study

Local mate competition (LMC) theory predicts that when populations are
structured so that only female offspring disperse, a mother should produce a female-
biased clutch of offspring (Hamilton 1967). This female-biased clutch serves to reduce
the local competition between brothers for mates. When two or more females share the
same patch, this theory predicts that the optimal offspring sex ratio will depend on the
extent to which the females are related. Speciﬁcally, if females are able to assess their
relatedness, females that are more closely related to each other are predicted to lay a more
female-biased clutch of offspring than are unrelated females (Frank 1985).

Shuker et. al (2004) assessed the effect of patch partner relatedness on sex ratio
behavior in the gregarious parasitoidwasp Nasonia vitripennis. The researchers assigned
mated females to one of three experimental treatments: A) sisters that were ﬂom the
same host and therefore had environmental cues about relatedness, B) sisters that were

ﬂom different hosts and therefore had no environmental cues about relatedness, and C)

76

unrelated females. The researchers replicated these experiments using wasps ﬂom two
strains: HV236 and HV3 07. They recorded the number and sex of all offspring that
emerged ﬂom the hosts (and therefore both wasps combined), and converted these
numbers into sex ratios (number of male offspring/total number of offspring) for
statistical analyses. Refer to Shuker et. al (2004, Experiment 1), for the detailed
experimental methods and statistical analyses.

Here we compare the sex ratio analyses of Shuker and his collagues with our
reanalysis of the data in that study. Our analysis focuses on the sex allocation behavior

of the wasps, i.e. the numbers of male and female offspring that were produced.

Results

Shuker et al. examined how patch partner relatedness affected offspring sex
ratios. The main question of interest in the study was whether related females (those in
treatments A and B combined) produced a more female-biased sex ratio than did
unrelated females (those in treatment C). However, general linear models revealed that
relatedness had no effect on the brood sex ratios (sex ratio of related: 0.260 i 0.007; sex
ratio of unrelated: 0.236 1 0.013; F1339 = 2.587, P = 0.109). The results ﬂom Shuker et
al. (2004) are presented in Figure 4.4.

We reanalyzed these data using a MANOVA to test for the effect of treatment on
two response variables: the number of male offspring produced and the number of

female offspring produced. If a MANOVA revealed a signiﬁcant difference between

77

 

 

 

  

 

 

 

 

 

 

 

 

.8. _
r: -
é
a i i -
E
.9; _
o
E _
s
or , _
236 SH 236 DH 307 SH 307 DH Unrelated
Oviposltlon partner

Figure 4.4. Data ﬂom Shuker et. al. 2004. Female sex ratio behavior and oviposition
partner. Females were either with a female ﬂom the same strain (black bars: HV236;
white bars: HV307) or were unrelated (gray bars: one HV236 female with one HV307
female). Additionally, females ﬂom the same strain were either raised on the same host
(SH) or different hosts (DH). Error bars show standard errors.

78

groups with respect to a particular factor, we continued with univariate post-hoe analyses
to determine which combinations of dependent variables were responsible for inter-group
differences. Speciﬁcally, we tested for treatment effects on the dependent variables
themselves (number of male offspring produced, number of female offspring produced)
and on two combinations of the dependent variables (total number of offspring produced,
sex ratio of the offspring). The sex ratios were arcsine-transfonned for the analyses. All
analyses were conducted using MATLAB 7.0. Means are presented i SE.

In both wasp strains, treatment had a signiﬁcant effect on sex allocation behavior
(Wilks’ A2333 = 0.915, P<0.0001; Figure 4.5). Since this MANOVA indicated a
signiﬁcant effect of treatment, we proceeded with additional MANOVAs that tested for
differences between each pair of treatments (Scheiner and Gurevitch 2001). There was
no signiﬁcant difference between treatments A and B, indicating that wasps in treatments
A and B did not differ in their sex allocation decisions (Wilks’ A1373 = 0.990, P=0.242).
Thus, related wasps made similar reproductive decisions when ovipositing, independent
of whether they were raised on the same or on a different host. Wasps in treatment C
made different reproductive decisions than did wasps in either treatment A (Wilks’ Ai,i99
= 0.882, P<0.0001) or treatment B (Wilks’ A1204 = 0.900, P<0.0001).

To test the hypothesis that related females make different sex allocation decisions
when ovipositing together than do unrelated females, we compared the sex allocation
decisions of females in treatment C with those of females in both strains across the
combined treatments A and B. There was a signiﬁcant effect of relatedness on sex
allocation decisions (Wilks’ A1339 = 0.923, P<0.001; Figure 4.5). Unrelated females

produced a signiﬁcantly smaller combined clutch than did related females (combined

79

 

50

I I I I I I I I

 

HV236 Same host
HV236 Different host

HV307 Same host ‘
HV307 Different host
Unrelated

20~ _a4%_. __{}__ _

40-

P00...

 

 

 

i

 

 

0 l I I l I J l

l l
0 10 20 30 40 50 60 70 80 90 100
Number of Females

 

Figure 4.5. Wasp sex allocation behavior in relation to oviposition partner (mean i SE
number of female and male offspring produced). Females were either with a female ﬂom
the same strain or were unrelated (one HV236 female with one HV 307 female). There
was a signiﬁcant effect of relatedness on sex allocation decisions (Wilks’ A1339 = 0.923,

P<0.001).

80

clutch size of related: 111.386 :1: 2.752; combined clutch size of unrelated: 78.773 5:
5.103; F1339 = 28.123, P < 0.0001). This smaller combined clutch size of unrelated
females was due to their lower production of both male and female offspring compared to
the related females (males: F1339 = 14.494, P < 0.001; females: F1339 = 23.841, P <
0.0001 ). There was no signiﬁcant difference in sex ratio between related and unrelated
females (sex ratio of related: 0.260 i 0.007; sex ratio of unrelated: 0.236 :t 0.013; F1339 =

2.587, P = 0.109).

Discussion

The conclusions drawn ﬂom this study depend on whether the data are analyzed
using a sex allocation or a sex ratio approach. Related and unrelated female wasps made
different offspring sex allocation decisions. Related females produced more female and
male offspring than unrelated females, and in total laid nearly 50 percent more eggs than
unrelated females. Why this might be remains a puzzling biological pattern that is
deserving of future work.. Yet these results raise the possibility that female Nasonia
vitripennis wasps have the ability to assess kinship with oviposition partners.

In contrast, an earlier analysis of sex ratio decisions in this dataset led Shuker et
al. (2004) to conclude that wasps cannot assess kinship. Their analysis assumed that sex
ratios were the variable of interest. Indeed, the predictions of LMC were originally
ﬂamed in a sex ratio context, as Hamilton (1967) assumed that females could vary sex
ratio but not clutch size. Based on the results of this study, future LMC theorists should
build models incorporating the ability of females to inﬂuence both the sex and the

number of their offspring. More generally, this reanalysis highlights the importance of

81

considering both sex ratio and clutch size as ways by which parents can adjust their sex

allocation.

Conclusions

Sex allocation and sex ratio are not identical terms. Whereas sex allocation
theory examines an organism’s total joint investment in male and female offspring
[males, females], sex ratio theory collapses these two dimensions of information into the
sole dimension of sex ratio [males/(males + females)]. In the examples we presented
above, tests of sex ratios failed to reveal a signiﬁcant effect of experimental treatment on
the reproductive decisions of honeybee queens and female parasitoid wasps. However,
tests of sex allocation revealed signiﬁcant and important inﬂuences of experimental
treatment on reproductive decisions. The examples in this paper illustrate that the
ﬂaming of questions asked by biologists can greatly affect the conclusions they draw
ﬂom their studies.

The examples in this paper focused on sex allocation decisions by haplodiploid
insects. In haplodiploid species, haploid (unfertilized) eggs develop into males and
diploid (fertilized) eggs develop into females. Females in these species can therefore
precisely control their egg sex allocation by controlling whether or not they fertilize their
. eggs (Ratnieks and Keller 1998). Recent evidence suggests the ability to adaptively
adjust offspring sex allocation isn’t limited to haplodiploid species, but potentially also
occurs in birds and other vertebrates with chromosomal sex determination (West and

Sheldon 2002; Holand et al. 2006). Therefore, females across a wide variety of taxa are

82

able to adjust their offspring sex allocation.

As for how females actually adjust offspring sex allocation, there are several
potential strategies. For example, to produce a greater investment in male offspring, a
female might increase her offspring sex ratio, increase her clutch size, or use some
combination of these strategies (Figure 4.6). Each of these strategies makes implicit
assumptions about the biological constraints faced by the organism (West and Sheldon
2002). If females are constrained by the number of offspring they can produce, they
might increase their investment in males by increasing their offspring sex ratio. This
assumption was used by Hamilton (1967) when he formulated LMC theory. If females
instead are constrained in altering the sex of their offspring, they might increase their
investment in males by increasing their clutch size. In some cases, females might have
the ability to concurrently alter both sex ratio and clutch size.

A subtle shift in both sex ratio and clutch size can affect the number of males
produced, but might not be captured by traditional statistical methods that test for the
effects of treatment on sex ratio or clutch size. An excellent example of this can be found
in the wasp dataset ﬂom above. Within the HV3 07 strain of wasps, females ﬂom the
same or different hosts did not demonstrate statistically signiﬁcant differences in either
their offspring sex ratio (F19130 = 0.571, P = 0.451) or clutch size (F1,13o = 0.976, P =
0.325). However, by using the number of males and females in a multivariate analysis of
variance (MANOVA), we found that these wasps made different sex allocation decisions
(MANOVA, Wilks’ AU.“ = 0.955, P = 0.052; Figure 4.7). This difference was due to a
signiﬁcant difference in the number of male offspring (FU30 = 5.324, P = 0.023). This

example suggests that by treating the numbers of male and female offspring as dependent

83

100

' ‘ 0.50::
Alter sex ratio / g =

/
10°. /

    

025 / Alter sex ratio and clutch size

—..—.._.._.._p

, D... 200

O

Alter clutch size”!

Figure 4.6. Three ways that a female can increase (i.e., double) her production of males.
A female might a) alter her sex ratio (dashed line), b) alter her clutch size (dotted line), or
c) slightly alter both sex ratio and clutch size (dashed-dotted line). By any one of these
strategies, a female can double her production of males (i.e. switch ﬂom producing 25
males to producing 50 males). Each pie represents a female’s investment in male
offspring (gray areas) and female offspring (white areas).

84

3o - .

 

 

 

 

 

 

 

 

 

 

El HV307 Same host
0 HV307 Different host f
25 .
O
O
'5
2
'6
h 20 h
i
3
15 - -
5'1: . . . .
0 75 80 85 90 95

Number of Females

Figure 4.7. Wasp sex allocation behavior within the strain HV307 (mean i SE number of
female and male offspring produced). The wasp sisters were either ﬂom the same host or
ﬂom different hosts. The slopes of the dashed lines indicate the mean sex ratio
(males/total) for each group (long dashes: same host; short dashes: different host).

85

variables, it is possible to detect biological patterns that might not be captured by
traditional analyses that use either sex ratio or clutch size as the dependent variable.
Many researchers use sex ratios as their variable of interest because it is not
feasible to collect data on the total numbers of males and females in a clutch. Indeed, in
species that produce clutches consisting of thousands of eggs, it is often most practical to
sample. Sampling usually involves recording the number of males and females of some
predetermined number of offspring, which allows researchers to estimate sex ratios but
not on the number of males and females in the entire clutch. Based on the ﬁndings in this
paper, we suggest that researchers attempt to collect, whenever possible, the total number
of males and females produced in a clutch. By doing so, they might be able to draw more

insightful conclusions about sex allocation than they would by focusing on sex ratios.

We suggest that an exclusive focus on sex ratios can lead to a number of potential
problems. First, a focus on sex ratios makes some limiting assumptions about the biology
of the species and the constraints that they face. Indeed, species are believed to display a
great variation in the degree to which females can adjust their offspring sex ratio (West et
a1. 2002). Second, a focus on sex ratios leads to statistical tests that might overlook other
important components of sex allocation behavior, such as clutch size. Many tests
statistically control for differences in clutch sizes rather than treat clutch size as a
potentially important variable (Hardy 2002). Third, a focus on sex ratios can
fundamentally change the conclusions that are drawn ﬂom the study, as demonstrated in
the examples in this paper.

In conclusion, biologists might gain more insightful interpretations of their system

of study by focusing on two-dimensional sex allocation decisions rather than one-

86

dimensional sex ratio decisions. Sex allocation theory provides a powerful way to
determine the inﬂuence of environmental factors on animal behavior (West et al. 2000).
We encourage researchers in behavioral ecology to examine sex allocation decisions, of

which sex ratios are one, but not the sole, component.

87

REFERENCES

Allen MD (1958) Drone brood in honey bee colonies. Journal of Economic Entomology
5 1 246-48

Allen MD (1963) Drone production in honey-bee colonies (Apis mellifera 1.). Nature
199:4895-4896

Allen MD (1965a) The effect of a plentiful supply of drone comb on colonies of honey
bees. Journal of Apicultural Research 4:109-119

Allen MD (1965b) The production of queen cups and queen cells in relation to the
general development of honeybee colonies and its connection with swarming and
supersedure. Journal of Apicultural Research 4: 1 21-141

Aron 8, Keller L, Passera L (2001) Role of resource availability on sex, caste and
reproductive allocation ratios in the Argentine ant Linepithema humile. Journal of
Animal Ecology 70:831-839

Aron S, Passera L, Keller L (1994) Queen-worker conﬂict over sex ratio: A comparison
of primary and secondary sex ratios in the Argentine ant, Iridomyrmex humilis.
Journal of Evolutionary Biology 7:403-418

Aron S, Vargo EL, Passera L (1995) Primary and secondary sex ratios in monogyne
colonies of the ﬁre ant. Animal Behaviour 49:749-757

Beekman M, Komdeur J, Ratnieks FLW (2003) Reproductive conﬂicts in social animals:
who has power? Trends in Ecology & Evolution 18:277-282

Beekman M, Ratnieks FLW (2003) Power over reproduction in social Hymenoptera.
Philosophical Transactions of the Royal Society of London Series B-Biological
Sciences 358:1741-1753

Bourke AF G, Franks NR (1995) Social Evolution in Ants. Princeton University Press,
Princeton, NJ.

Carnazine S, Deneubourg J-L, Franks NR, Sneyd J, Theraulaz G, Bonabeau E (2001)
Self-organization in Biological Systems. Princeton University Press, Princeton,
NJ.

Chapuisat M, Keller L (1999) Testing kin selection with sex allocation data in eusocial
Hymenoptera. Heredity 82:473-478

88

Chapuisat M, Sundstrtim L, Keller L (1997) Sex-ratio regulation: the economics of
ﬂatricide in ants. Proceedings of the Royal Society of London Series B-Biological
Sciences 264:1255-1260

Chamov EL (1982) The Theory of Sex Allocation. Princeton University Press, Princeton,
NJ.

Clark AB (1978) Sex-Ratio and Local Resource Competition in a Prosirnian Primate.
Science 201 :163-165

Cremer S, Heinze J (2002) Adaptive production of ﬁghter males: queens of the ant
Cardiocondyla adjust the sex ratio under local mate competition. Proceedings of
the Royal Society of London Series B-Biological Sciences 269:417-422

Darwin C (1874) The Descent of Man, and Selection in Relation to Sex (2nd edition).
John Murray, London

de Menten L, Cremer S, Heinze J, Aron S (2005a) Primary sex ratio adjustment by ant
queens in response to local mate competition. Animal Behaviour 69: 103 1-1 035

de Menten L, Fournier D, Brent C, Passera L, Vargo EL, Aron S (2005b) Dual
mechanism of queen inﬂuence over sex ratio in the ant Pheidole pallidula.
Behavioral Ecology and Sociobiology 58:527-533

Delaplane KS, Harbo JR (1987) Drone production by young versus old worker
honeybees in queenless colonies. Apidologie 18:115-119

Dubois NS, Kennedy ED, Getty T (2006) Surplus nest boxes and the potential for
polygyny affect clutch size and offspring sex ratio in house wrens. Proceedings of
the Royal Society B-Biological Sciences 273:1751-1757

Estoup A, Solignac M, Comuet J M (1994) Precise assessment of the number of patrilines
and of genetic relatedness in honeybee colonies. Proceedings of the Royal Society
of London Series B-Biological Sciences 258:1-7

Frank SA (1985) Hierarchical Selection Theory and Sex-Ratios .2. on Applying the
Theory, and a Test with Fig Wasps. Evolution 39:949-964

Free I B (1957) The food of adult drone honeybees (Apis mellifera). The British Journal
of Animal Behaviour 5:7-11

Free J B (1967) The production of drone comb by honeybee colonies. Journal of
Apicultural Research 6:29-36

Free J B, Williams [H (1975) Factors determining the rearing and rejection of drones by
the honeybee colony. Animal Behaviour 23:650-675

89

Fukuda H, Ohtani T (1977) Survival and life span of drone honeybees. Research in
Population Ecology 19:51-68

Greeff J M (1997) Offspring allocation in externally ovipositing ﬁg wasps with varying
clutch size and sex ratio. Behavioral Ecology 8:500-505

Hamilton WD (1964) The genetical evolution of social behavior. Journal of Theoretical
Biology 7:1-16

Hamilton WD (1967) Extraordinary Sex Ratios. Science 156:477-&

Hardy ICW (2002) Sex Ratios: Concepts and Research Methods. Cambridge University
Press

Helms KR, Fewell J H, Rissing SW (2000) Sex ratio determination by queens and
workers in the ant Pheidole desertorum. Animal Behaviour 59:523-527

Henderson CE (1991) Reproductive Investment in Drones in Honey Bee (Apis mellifera)
Colonies. In: Entomology. Cornell University, Ithaca, N.Y., pp 92

Henderson CE (1994) Inﬂuence of the presence of adult drones on the further production
of drones in honey bee (Apis mellifera L) colonies. Apidologie 25:31-37

Herbers J M, DeHeer CJ, F oitzik S (2001) Conﬂict over sex allocation drives conﬂict
over reproductive allocation in perennial social insect colonies. American
Naturalist 158:178-192

Holand O, Mysterud A, Roed KH, Coulson T, Gjostein H, Weladji RB, Nierninen M
(2006) Adaptive adjustment of offspring sex ratio and maternal reproductive

effort in an iteroparous mammal. Proceedings of the Royal Society B-Biological
Sciences 273:293-299

Hrassnigg N, Crailsheim K (2005) Differences in drone and worker physiology in
honeybees (Apis mellifera). Apidologie 36:255-277

J efﬂee EP (1958) A shaped grid for estimating quantities of brood and pollen in combs.
Bee World 39:115-118

J ohansson TSK, J ohansson MP (1971) Effects of drone comb on brood and honey
production in honey bee Apis mellifera (Hyrnenoptera-Apidae) colonies. Annals
of the Entomological Society of America 64:954-&

Keller L, Aron S, Passera L (1996a) Intemest sex-ratio variation and male brood survival
in the ant Pheidole pallidula. Behavioral Ecology 7:292-298

Keller L, Lhoste G, Balloux F, Plumey O (1996b) Queen number inﬂuences the primary
sex ratio in the Argentine ant, Linepithema humile (= Iridomyrmex humilis).
Animal Behaviour 51 :445-449

90

Koedam D, Contrera FAL, Fidalgo AO, Imperatriz-Fonseca VL (2005) How queen and
workers share in male production in the stingless bee Melipona subnitida Ducke
(Apidae, Meliponini). Insectes Sociaux 52:114-121

Krackow S, Tkadlec E (2001) Analysis of brood sex ratios: implications of offspring
clustering. Behavioral Ecology and Sociobiology 50:293-301

Kronauer DJ C, Schoning C, Pedersen J S, Boomsma JJ, Gadau J (2004) Extreme queen-
mating ﬂequency and colony ﬁssion in Aﬁican army ants. Molecular Ecology
13:2381-2388

Lack D (1947) The signiﬁcance of clutch size Ibis 89:302-252

Lee PC, Winston ML (1985) The effects of swarm size and date of issue on comb
construction in newly founded colonies of honey bees (Apis mellifera L.).
Canadian Journal of Zoology 63:524-527

Lee PC, Winston ML (1987) Effects of reproductive timing and colony size on the
survival, offspring colony size and drone production in the honey bee (Apis
mellifera). Ecological Entomology 12: 1 87-1 95

Linksvayer TA (2008) Queen-worker-brood coadaptation rather than conﬂict may drive
colony resource allocation in the ant T emnothorax curvispinosus. Behavioral
Ecology and Sociobiology 62:647-657

Mehdiabadi NJ, Reeve HK, Mueller UG (2003) Queens versus workers: sex ratio conﬂict
in eusocial Hymenoptera. Trends in Ecology & Evolution 18:88-93

Miller DG, Ratnieks F LW (2001) The timing of worker reproduction and breakdown of
policing behaviour in queenless honey bee (Apis mellifera L.) societies. Insectes
Sociaux 48: 1 78-1 84

Mock DW, Forbes LS (1995) The Evolution of Parental Optimism. Trends in Ecology &
Evolution 102130-134

Moritz RFA (1985) The effects of multiple mating on the worker-queen conﬂict in Apis
mellifera L. Behavioral Ecology and Sociobiology 16:375-377

Morse RA, Strang GE, Nowakowski J (1967) Fall death rates of drone honey bees.
Journal of Economic Entomology 60:1198-1202

Nagelkerke CJ (1994) Simultaneous optimization of egg distribution and sex allocation in
a patch-structured population. American Naturalist 144:262-284

Neurnann P, Moritz RF A (2000) Testing genetic variance hypotheses for the evolution of
polyandry in the honeybee (Apis mellifera L.). Insectes Sociaux 47:271-279

91

Newton DC, Michl DJ (1974) Cannibalism as an indication of pollen insufﬁciency in
honeybees: ingestion or recapping of manually exposed pupae. Journal of
Apicultural Research 132235-241

Nolan WJ (1925) The brood rearing cycle of the honey bee. Bull. US. Dept. Agric.
1349:1-56 -

Nonacs P, Carlin NF (1990) When can ants discriminate the sex of brood? A new aspect
of queen-worker conﬂict. Proceedings of the National Academy of Sciences USA
87:9670-9673

Ode PJ, Rissing SW (2002) Resource abundance and sex allocation by queen and
workers in the harvester ant, Messor pergandei. Behavioral Ecology and
Sociobiology 51:548-556

Oldroyd BP, Ratnieks FLW (2000) Evolution of worker sterility in honey-bees (Apis
mellifera): how anarchistic workers evade policing by laying eggs that have low
removal rates. Behavioral Ecology and Sociobiology 47:268-273

Olsson M, Wapstra E, Uller T (2005) Differential sex allocation in sand lizards: bright
males induce daughter production in a species with heteromorphic sex
chromosomes. Biology Letters 1:378-380

Page RE (1981) Protandrous reproduction in honey bees. Environmental Entomology
10:359-362

Page RE, Erickson EH (1988) Reproduction by worker honey bees (Apis mellifera L).
Behavioral Ecology and Sociobiology 232117-126

Page RE, Metcalf RA (1984) A population investment sex ratio for the honey bee (Apis
mellifera L). American Naturalist 124:680-702

Pamilo P (1991a) Evolution of colony characteristics in social insects 1. Sex allocation.
American Naturalist 137:83-107

Pamilo P (1991b) Evolution of colony characteristics in social insects 2. Number of
reproductive individuals. American Naturalist 138:412-433

Passera L, Aron S (1996) Early sex discrimination and male brood elimination by
workers of the Argentine ant. Proceedings of the Royal Society of London Series
B-Biological Sciences 263: 1041-1046

Passera L, Aron S, Vargo EL, Keller L (2001) Queen control of sex ratio in ﬁre ants.
Science 293:1308-1310

Pen 1, Taylor PD (2005) Modelling information exchange in worker-queen conﬂict over
sex allocation. Proceedings of the Royal Society of London Series B-Biological
Sciences 272:2403-2408

92

Pratt SC (1998) Decentralized control of drone comb construction in honey bee colonies.
Behavioral Ecology and Sociobiology 42: 193-205

Pratt SC (2004) Collective control of the timing and type of comb construction by honey
bees (Apis mellifera). Apidologie 35: 193-205

Queller DC, Strassmann J E (1998) Kin selection and social insects. Bioscience 48:165-
1 75

Ratnieks FLW, Foster KR, Wenseleers T (2006) Conﬂict resolution in insect societies.
Annual Review of Entomology 51:581-608

Ratnieks FLW, Keller L (1998) Queen control of egg fertilization in the honey bee.
Behavioral Ecology and Sociobiology 44:57-61

Ratnieks FLW, Reeve HK (1992) Conﬂict in single-queen Hymenopteran societies - the
structure of conﬂict and processes that reduce conﬂict in advanced eusocial
species. Journal of Theoretical Biology 158:33-65

Reeve HK, Holldobler B (2007) The emergence of a superorgarrism through intergroup
competition. Proceedings of the National Academy of Sciences of the United
States of America 104:9736-9740

Reuter M, Helms KR, Lehmann L, Keller L (2004) Effects of brood manipulation costs
on optimal sex allocation in social hymenoptera. American Naturalist 164:E73-
E82

Reuter M, Keller L (2001) Sex ratio conﬂict and worker production in eusocial
hymenotera. American Naturalist 158: 1 66-1 77

Rheindt FE, Gadau J, Strehl CP, Hr'illdobler B (2004) Extremely high mating ﬂequency
in the Florida harvester ant (Pogonomyrmex badius). Behavioral Ecology and
Sociobiology 56:472-481

Rinderer TE, Richard L. Hellmich I, Danka RG, Collins AM (1985) Male reproductive
parasitism: a factor in the aﬁicanization of European honey-bee populations.
Science 228:1119-1121

Robinson GE, Page RE, F ondrk MK (1990) Intracolonial behavioral variation in worker
oviposition, oophagy, and larval care in queenless honey bee colonies. Behavioral
Ecology and Sociobiology 26:315-323

Rosset H, Chapuisat M (2006) Sex allocation conﬂict in ants: When the queen rules.
Current Biology 162328-331

Sasaki K, Kitamura H, Obara Y (2004) Discrimination of larval sex and timing of male
brood elimination by workers in honeybees (Apis mellifera L.). Applied
Entomology and Zoology 39:393-399

93

Sasaki K, Obara Y (2001) Nutritional factors affecting the egg sex ratio adjustment by a
honeybee queen. Insectes Sociaux 48:355-359

Sasaki K, Satoh T, Obara Y (1996) The honeybee queen has the potential ability to
regulate the primary sex ratio. Applied Entomology and Zoology 31 :247-254

Scheiner SM, Gurevitch J (2001) Design and Analysis of Ecological Experiments, 2nd
edn. Oxford University Press, Oxford

Schmickl T, Blaschon B, Gurmann B, Crailsheim K (2003) Collective and individual
nursing investment in the queen and in young and old honeybee larvae during
foraging and non-foraging periods. Insectes Sociaux 50:174-184

Schmickl T, Crailsheim K (2001) Cannibalism and early capping: strategy of honeybee
colonies in times of experimental pollen shortages. Journal of Comparative
Physiology a-Sensory Neural and Behavioral Physiology 187:541-547

Schmickl T, Crailsheim K (2002) How honeybees (Apis mellifera L.) change their
broodcare behaviour in response to non-foraging conditions and poor pollen
conditions. Behavioral Ecology and Sociobiology 51 :415-425

Seeley TD (1985) Honeybee Ecology. Princeton University Press, Princeton, NJ.

Seeley TD (2002) The effect of drone comb on a honey bee colony's production of
honey. Apidologie 33:75-86

Seeley TD, Mikheyev AS (2003) Reproductive decisions by honey bee colonies: tuning
investment in male production in relation to success in energy acquisition.
Insectes Sociaux 50:134-138

Seeley TD, Morse RA (1976) The nest of the honey bee (Apis mellifera L). Insectes
Sociaux 23 :495-512

Shuker DM, Reece SE, Taylor J AL, West SA (2004) Wasp sex ratios when females on a
patch are related. Animal Behaviour 68:331-336

Stearns S (1992) The Evolution of Life Histories. Oxford University Press, Oxford,
England

Strassmann J (2001) The rarity of multiple mating by females in the social Hymenoptera.
Insectes Sociaux 48:1-13

Sundstrdm L, Chapuisat M, Keller L (1996) Conditional manipulation of sex ratios by ant
workers: A test of kin selection theory. Science 274:993-995

Taber S, Owens CD (1970) Colony founding and initial nest design of honey bees, Apis
mellifera L. Animal Behaviour 18:625-&

94

Tarpy DR, Nielsen D1 (2002) Sampling error, effective paternity, and estimating the
genetic structure of honey bee colonies (Hymenoptera : Apidae). Annals of the
Entomological Society of America 95:513-528

Trivers RL, Hare H (1976) Haplodiploidy and the evolution of the social insects. Science
191:249-263

Trivers RL, Willard DE (1973) Natural Selection of Parental Ability to Vary the Sex
Ratio of Offspring. Science 179:90-92

Visscher PK (1989) A quantitative study of worker reproduction in honey bee colonies.
Behavioral Ecology and Sociobiology 25:247-254

Webster TC, Peng YS, Duffey SS (1987) Conservation of nutrients in larval tissue by
cannibalizing honey bees. Physiological Entomology 12:225-231

West SA, Herre EA, Sheldon BC (2000) The beneﬁts of allocating sex. Science 290:288-
290

West SA, Reece SE, Sheldon BC (2002) Sex ratios. Heredity 88:117-124

West SA, Sheldon BC (2002) Constraints in the evolution of sex ratio adjustment.
Science 295: 1685-1 688

Wharton KE, Dyer FC, Getty T (in press) Male elimination in the honeybee. Behavioral
Ecology

Wharton KE, Dyer FC, Huang ZY, Getty T (2007) The honeybee queen inﬂuences the
regulation of colony drone production. Behavioral Ecology 18: 1092-1099

Winston ML (1987) The Biology of the Honey Bee. Harvard University Press,
Cambridge, Massachusetts

Woyke J (1977) Cannibalism and brood-rearing efﬁciency in the honeybee. Journal of
Apicultural Research 16:84-94

95

 

 

        

IIIIIII

ii iiiiiiiiiiiiiiiiiiiiiiii

3