z: E,

 

 

LIBRARY
Michigan State
- University

 

 

 

This is to certify that the
dissertation entitled

DECISION-MAKING IN A CHANGING ENVIRONMENT: A

LOOK AT THE FORAGING BEHAVIOR OF HONEYBEES

AND BUMBLEBEES AS THEY RESPOND TO SHIFTS IN
RESOURCE AVAILIBILITY

presented by

JOHN M. TOWNSEND-MEHLER

has been accepted towards fulfillment
of the requirements for the

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

 

 

. %/w/\

Major Professor’s/Signature
{/1 0 / 1W (/

Date

 

MSU is an Affirmative Action/Equal Opportunity Employer

_._._.-,_-.-.-.-.‘.-v-.-.-.-.-.—.-.-

 

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.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KIProj/Acc8PrelelRC/DateDue.indd

 

DECISION-MAKING IN A CHANGING ENVIRONMENT: A LOOK AT THE
FORAGING BEHAVIOR OF HONEYBEES AND BUMBLEBEES AS THEY
RESPOND TO SHIFTS IN RESOURCE AVAILIBILITY

By

John M. Townsend-Mehler

A DISSERTATION

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

DOCTOR OF PHILOSPHY

Zoology
Ecology, Evolutionary Biology, and Behavior

2010

ABSTRACT
DECISION-MAKING IN A CHANGING ENVIRONMENT: A LOOK AT THE
FORAGING BEHAVIOR OF HONEYBEES AND BUMBLEBEES AS THEY
RESPOND TO SHIFTS IN RESOURCE AVAILIBILITY
By

John M. Townsend-Mehler

AS any animal moves through its environment seeking resources (whether in the form of
food or mates or information) it is continually faced with the challenge of how best to
allocate its time and energy among available options. In natural environments, animals
must continually choose among a suite of behavioral options available to them in order to
respond effectively to shifts in resources. Historically these behavioral options have been
studied in isolation despite the fact that, in nature, animals must coordinate these
behaviors over time as part of their response to Shifting resource availability. Chapter 1
describes an integrated approach to examining these behaviors by looking simultaneously
at the behavioral processes of extinction, negative contrast effects, and exploration as
bees (both honeybees and bumblebees) respond to a dramatic downshift in available food
resources. Additionally, Chapter 1 explores how the integration of these processes is
influenced by such social factors as colony size and recruitment mechanisms as well as
individual biological factors, particularly body size. The results from this assay suggest
that extinction, contrast effects, and the decision to explore, occur in nature in a much
more integrated fashion than has been previously acknowledged. Additionally, these
results support the notion that while honeybees are much more dependent upon social

information for foraging decisions, bumblebees are, in contrast, much more reliant upon

gathering information on resources directly from the environment. Chapter 2 examines
species differences in more detail, focusing on the question of how honeybees and
bumblebees differ in terms of not only their sensitivity to shifts in reward magnitude, but
also the decision processes subsequent to abandoning a past-profitable site. The results
suggest that bumblebees are much more sensitive to shifts in reward, relative to
honeybees. The results also strongly suggest that bumblebees are much more likely to
explore the environment, and more likely to discover a novel food source subsequent to
departing a past-profitable site. These results are very consistent with the findings from
Chapter 1, and give strong support for notion that bumblebees are comparatively more
reliant than honeybees upon the direct acquisition of information from the environment.
Finally, Chapter 3 examines the long-term effects of honeybees’ tendency to forgo
feeding at a low quality food source when no other food is available, also referred to as
negative contrast effects, and looks at how this behavioral phenomenon might enhance
fitness among honeybees. This was a follow-up on the finding from Chapter 1 that
honeybees are generally very reluctant to resume feeding at a low quality feeder,
following experience at a high quality food source. Chapter 3 tests the hypothesis that
by not exploiting available but low quality food when there is an indication that better
food will be available in the future actually increases their lifetime contribution of food
resources to the colony. The results Show that honeybees that forgo exploiting available
low quality food, having had previous experience with a high quality feeder, are more
likely to live longer and ultimately contribute more food to the hive, than bees that forage

continually at a low quality food source.

This work is dedicated to my parents for their unwavering support and encouragement
and to my children Maggie and Grace for keeping me sane during this processes, and for
always making it fun to come home and play no matter how many hours I spent in the
hoop-house or how many times I got stung. This work is also dedicated to my wife,
Mary Ellen. Her hard work, her patience, her willingness to make sacrifices, and her love
and support throughout this period are ultimately what have made this whole project

possible.

iv

 

ACKNOWLEDGEMENTS

There are many people without whose help this work could not have been completed.
First, I would like to thank my advisor, Dr. Fred Dyer for his patience with the editing
process and his continued enthusiasm for this project. I would also like to thank my
committee members, Dr. Rufus Isaacs, Dr. Erik Altmann, and Dr. Thomas Getty, for their
ongoing involvement in this research, and willingness to provide feedback. I would also
particularly like to thank Dr. Zachary Huang, who, although was not an official member
of my committee, was exceedingly helpful both in terms of encouragement as well as
general feedback on manuscripts. I also thank my lab mates; Cindy Wei for showing me
the ropes in the hoop-house, Frank Bartlett for his help with experimental design and his
input on my manuscript, and Katie Wharton for her feedback and enthusiasm. Also,
there are many people without whom this work would literally have been impossible. I
am very indebted to the hard work and perseverance of our field assistants Jared Ruddick,
Lora Bramlett, Kourtney Trudgen, Mara Trudgen, and Elizabeth Dean. I would also
especially like to thank Kim Maida for her tireless efforts on multiple aspects of this

work both in the field and in the lab.

TABLE OF CONTENTS

LIST OF FIGURES ......................................................................................................... viii

CHAPTER 1 ....................................................................................................................... 1
AN INTEGRATED LOOK AT DECISION-MKING IN HONEYBEES AND

BUMBLEBEES AS THEY SEARCH F OR FOOD ........................................................... 1

INTRODUCTION .................................................................................................. 1

MATERIALS AND METHODS ............................................................................ 6

General Setup .............................................................................................. 6

Marking and Scent-Plugging Bees .............................................................. 7

Experimental Approach .............................................................................. 8

RESULTS ............................................................................................................. 10

State Transitions ........................................................................................ 13

DISCUSSION ....................................................................................................... 28

CHAPTER 2 ..................................................................................................................... 46

DECDING WHEN TO EXPLORE AND WHEN TO PERSIST: A COMPARISON OF
HONEYBEES AND BUMBLEBEES IN THEIR RESPONSE TO DOWNSHIFTS IN

REWARD ......................................................................................................................... 46
INTRODUCTION ................................................................................................ 46
MATERIALS AND METHODS .......................................................................... 49

General Setup ............................................................................................ 49
Marking and scent-plugging honeybees ..................... 5]
Experiment I: Latency to resume feeding following a downshift in reward
................................................................................................................... 51
Experiment II: The resumption of feeding versus the search for novel
alternatives ................................................................................................ 53
RESULTS ............................................................................................................. 54
Experiment I: Latency to resume feeding following a downshift in reward
................................................................................................................... 55
Experiment 2: Search behavior following a downshift in reward ........... 56
DISCUSSION ....................................................................................................... 60

vi

 

CHAPTER 3 ..................................................................................................................... 66
A FUNCTIONAL ACCOUNT OF NEGATIVE INCENTIVE CONTRAST EFFECTS

IN HONEYBEES .............................................................................................................. 66

INTRODUCTION ................................................................................................ 66

MATERIALS AND METHODS .......................................................................... 70

RESULTS ............................................................................................................. 73

Negative Contrast Effects ......................................................................... 73

Foraging activity and lifetime foraging success ....................................... 79

Distribution of Foraging Activity among Test Bees ................................. 79

DISCUSION ......................................................................................................... 81

REFERENCES ................................................................................................................. 86

vii

 

 

LIST OF FIGURES

Figure 1.1. The organization of elements in the array. All experiments were performed
within the same flight cage. After all honeybee trials were completed, the honeybee
observation hive was removed and replaced with a bumblebee hive box .................... 6

Figure 1.2. Time Budgets. Throughout the observation period, bees consistently moved
among the locations in the array, and spent some period of time out of sight, which we
refer to here as exploration. Bars depict the average amount of time bees spent at each of
the locations or to exploration. Here the hour-long observation period is subdivided into
6 10-minute time periods ......................................................................... 14

Figure 1.3. State transitions. Each transition is type is illustrated in the order that they
occurred for each bee. Only the first 15 transitions for each group as the pattern becomes
'noisy thereafter. No bees are shown to transition initially to the test feeder because the
test feeder was the point of origin for all bees at the beginning of the observation period.
Transitions into exploration are not shown because this state was inferred

indirectly ........................................................................................... 15

Figure 1.4, A & B. Contacts with the test feeder are shown both in terms of mean
contacts over time (A), and mean cumulative contacts, over the same span of time (B).
For each group, the first 10 minutes of the testing period is subdivided into 20 30-second
time periods ........................................................................................ 17

Figure 1.5. All contacts with both feeders and trips to the nest. Contacts with both
feeders are shown in terms of the average number of contacts for each experimental
group, for each 5-minute time period over the entire hour of observation. The small and
large arrows represent the latency for 50% and 100% of bees, respectively, to make their
first entry into the hive. The open and filled triangles represent the latency for 50% and
100% of bees, respectively, to make their first contact with the constant feeder. Among
bumblebees, the large arrows are missing because not all bumblebees made return trips to
the hive by the end of the observation period. Similarly not all honeybees returned to the
constant feeder during the observation period ................................................. 19

Figure 1.6. First return trip to the constant feeder. Survivorship curves Show the latency
in minutes for bees to make their first return trip to the constant feeder (where all bees
had been foraging prior) following their initial contacts with the non-rewarding test
feeder ............................................................................................... 21

Figure 1.7. First return trip to the hive. Survivorship curves indicating the latency in
minutes for bees’ to make their first return trip to the hive. The curves indicate that nine
of the ten honeybees from each training group made their first entry to the hive within the
first 25 minutes of testing, and that all honeybees made at least one trip home.
Bumblebees were generally slower to enter the hive and, six bumblebees failed to make a
Single trip home .................................................................................... 23

Figure 1.8. Total time spent within the hive. The average cumulative time spent within
the hive for bees in each group, shown here in minutes i 1 SE ............................ 23

viii

Figure 1.9. Total time Spent exploring (time spent out of sight) (+/- 1 SE). Bumblebees
spent approximately twice as much time exploring on average than honeybees .......... 26

Figure 1.10. Average time spent exploring from each location in the array. This figure
shows how the total time spent exploring (+/- 1 SE) was allocated among the elements in
the array. Honeybees were generally more likely to begin exploring after departing the
test feeder, whereas bumblebees tended to dedicate more time to exploring after
departing the constant feeder ...................................................................... 27

Figure 1.11. Distribution of time spent exploring. Lines represent average total amount
of time devoted to exploration, per bee, for each ten-minute time-period of the hour-long
observation period...............................— .................................................... 28

Figure 2.1. The organization of elements in the array. Note that the novel feeder was
only present during the observation period of Experiment 2 .................................. 51

Figure 2.2. The resumption of feeding. Latencies to resume feeding are shown in
minutes i l 8.6. For each level of downshift (from the 2M solution bees encountered
first), bumblebees took longer to reaccept the feeder. In many cases bumblebees failed to
resume feeding within our experimental time-frame ........................................... 55

Figure 2.3A & 2.3B. Latency to resume feeding and latency to discover novel food
source after downshift to 0.5M. Although all honeybees resumed feeding within the time
period, only a small percentage of bumblebees did so. Additionally, none of the
honeybees discovered the novel food source, presumably because honeybees did not
generally abandon the downshifted feeder, while nearly half of the bumblebees
discovered the novel feeder ........................................................................ 58

Figure 2.4A & 2.4B. Latency to resume feeding and latency to discover novel food
source after downshift to 0.25M. Only 1 bee resumed feeding (a honeybee) following the
downshift to 0.25M sucrose. Subsequent to the downshift, few honeybees went on to
discover the novel feeder, although roughly half of the bumblebees did so ................ 59

Figure 2.5. Probing behavior. Honeybees were approximately twice as likely to
repeatedly probe the test feeder following the downshift than were bumblebees. Graph
indicates mean number of total probes, per bee, made over the entire observation period :1:
ls.e .................................................................................................... 60

Figure 3.1. The organization of the elements in the array. The dark line represents the
boundary of the flight cage ........................................................................ 72

Figure 3.2. Total number of visits per hour for each bee on Day 2 and 3. Following a
bout of foraging at the high quality test feeder between 9 and 10 am, test bees were less
likely to forage at the constant feeder than they were the previous day over the same time
period, and less likely than control bees that did not experience a downshift in rewar. . ..75

ix

Figure 3.3. All foraging trips made by test bees to the constant feeder. Each bee is
represented by a single column, with the identifying pattern of paint dots indicated at the
top of each column. Time is indicated in hourly intervals along the horizontal for each
column. The constant feeder was available for 8 hours each day and the lines represent
the total number of visits (in which bees fed) to the constant feeder for each hour that the
feeder was available. The vertical on the left side indicates the total number of trips to
the constant feeder each hour. The vertical on the right side indicates day of foraging for
each bee. This does not necessarily correspond to the day of the experiment because not
all test bees began foraging on the same day. The gray cells at the top indicate the bees
foraging at the constant feeder prior to their introduction to the test feeder. On
subsequent days, bees began with an average of six trips to the high-quality test feeder,
and then chose whether to visit the constant feeder or abandon foraging for the day. Bees
were observed until they disappeared and were presumed dead (blank cells at the bottom
of each column) ..................................................................................... 77

Figure 3.4 A, B & C. Foraging success and lifespan for test and control bees. All bars
represent means :t 1 SE. Test bees averaged significantly fewer trips each day than bees
in the control group, and bees in the test group tended to live at least two days longer than
bees in the control group (average lifespan for control bees = 7.8 days i: 0.75, average
lifespan for test bees = 10.8 days i 1.01). Test bees also contributed nearly twice as
much food, in terms of millimoles of sucrose, than did control bees, over their

lifetimes ............................................................ ‘ .................................. 80

Figure 3.5 A & B. Distributions of intake for test bees based on where food was
acquired and when. A) Test bees harvested approximately the same amount of food, over
their lifetime, from the test feeder as from the constant feeder. Dotted line represents
total lifetime harvest of control bees. B) Within the first 8 days of foraging, test bees’
total harvest was not significantly different from that of control bees. Because test bees
tended to live approximately 2 days longer, on average, the total lifetime harvest of test
bees exceeded that of control bees .............................................................. 82

CHAPTER 1
AN INTEGRATED LOOK AT DECISION-MAKING IN HONEYBEES AND

BUMBLEBEES AS THEY SEARCH FOR FOOD

INTRODUCTION

A fundamental problem faced by many animals is to choose a course of action that will
lead to a beneficial outcome in the future. In most natural contexts, the selection of
actions occurs in a sequence of steps along the way to an eventual goal. At each step, the
animal takes in information (via sensory organs or from memory) and then selects an
action depending upon some innate or learned evaluation of payoffs, which may or may
not be immediately apparent at the moment of the choice (Stephens and Krebs 1986;
Sutton andBarto 1998). Like other authors (Glimcher 2002; Schuck-Paim et a1. 2004;
Stephens 2008), we refer to this process of action selection or choice behavior as
“decision-making,” without implying the kind of high level cognitive processes that the
term connotes in everyday language. The challenge deciding upon an appropriate course
of action is further complicated by the fact that animals are likely to have only limited
knowledge of the environment and the fact that those payoffs associated with different
options may change over time. In the search for food, as with the search for other
resources, some resources may persist over time while others become depleted and still
others may deplete only to subsequently renew. The animal, therefore, may have to shift
rapidly among alternative courses of action, again without complete information about

the long-term consequences of its shifts.

As important as the sequential context of decision-making is, much research on
animal decision-making focuses on choice behavior in contexts isolated from the
behavioral sequences in which the choices normally take place. An important exception
is the growing body of work showing that food-motivated choice behavior is sensitive not
only to the immediate payoffs associated with known options, but also to energetic state
resulting from previous choices, and to information about the availability and payoffs
associated with upcoming options (Schuck-Paim et a1. 2004; Stephens 2008; McNamara
et al. 1997). Here we extend this emphasis on the sequential context of decision-making
in three ways. First, we examine how choice behavior unfolds over a longer period of
time than is usually studied. Second, we focus on behaviors involved in the search for
acceptable foraging options rather than the exploitation of known options. Third, we
study the interactions among behavioral processes that are typically studied in complete
isolation from one another, but that are presumably deployed in common behavioral
contexts in nature, hence have the opportunity to interact.

To expand on this third point, our goal in this paper is to understand the
interaction of multiple behavioral processes that participate in an animal’s response to
changes in the distribution of resources. Working with foraging bees, we developed an
open-ended assay that enabled us to examine a full suite of behaviors by foragers as they
searched for food within a naturalistic environment. Additionally, to gain insights into
the evolutionary design of these processes, we studied both honeybees (Apis mellifera)
and bumblebees (Bombus impatiens) which share a similar foraging ecology, but are
likely subject to a different suite of constraints related to individual biological

characteristics as well as social ones.

We examined behavioral processes that are widely considered fundamental to
animals’ search for food in nature. These processes have been the subject of intensive
study in both psychology and behavioral ecology, but rarely in a coordinated manner.
The processes that are central to the present study include the following: (1) Extinction,
or the decay of performance when a conditioned behavior is no longer paired with its
reinforcer (Bouton 2002), is generally thought to underlie animals’ ability to abandon a
formerly rewarding food site (Adams-Hunt and Jacobs 2007; Bouton 2007), and may be
akin to the “patch-leaving” decisions studied by behavioral ecologists (Stephens and
Dunlap 2009); (2) Negative incentive contrast effects, or a suppression of motivation to
feed following a downshift in reward, (see F laherty 1996 for a thorough review) which
may enable animals to seek out resources comparable to what they had been exploiting
prior to the downshift (F reidin et al. 2009; Pecoraro et al. 1999); (3) Exploitation of
social information to detect the presence or location of food (Galef and Yarkovsky 2009;
Giraldeau and Caraco 2000; Noble et a1. 2001; Seeley 1985; Domhaus and Chittka 2001);
and (4) Exploration for novel resources (Biesmeijer and de Vries 2001; Wajnberg et al.
2006).

To study how these processes are coordinated as foragers respond to shifts in
resources, we presented free-flying bees foraging within a closed economy (a large flight-
cage) with a problem that animals are likely to confront routinely in nature: what to do
when a formerly profitable foraging site ceases to provide a reward, but other foraging
options remain available. We reasoned that upon arriving at a depleted Site a foraging
animal must choose from among at least four possible responses: persist without reward

at the site, presumably to search for the “missing” food; return to a familiar foraging site

of lower quality; explore the environment for novel alternatives; or, for animals such as
bees that forage from a central place, return home. We mimicked this scenario by
training bees that had had considerable experience at a low-quality feeder with a better
source of food (concentrated sucrose solution) in a different location, while keeping the
low-quality feeder in place. After giving the bees a brief period of experience (either 1 or
10 rewarded trips) at this new feeder, we replaced the high quality food with water, and
then observed bees over the course of the next hour. Through these observations we were
able to quantify a wide range of responses, including the gradual disengagement from the
formerly high—quality feeder (typical of an extinction response); a strong tendency to
revisit the low quality feeder but a reluctance to feed there (reminiscent of negative
contrast effects); visits to the hive, presumably connected with a search for social
information about foraging options; and exploratory flights, likely representing an effort
to search directly for novel food sources.

Although our approach allowed less experimental control than would be the case
in lab-based studies of individual behavioral processes, it affords a unique picture of how
these processes interact with one another as they appear during an extended behavioral
sequence determined by the animal. This picture is enriched by two additional features.
First, by varying the amount of experience with the high-quality feeder, we sought to
manipulate the bees’ perception of one option—persistence at this feeder—and thereby to
explore how this influenced the selection of other options. Second, to gain insights into
the evolutionary pressures that shaped these behavioral processes, we compared the

behavior of honeybees and bumblebees engaged in a similar search task.

 

These species are, in many ways ideal for such a comparison because they are
closely related genera, and are both generalist, nectar-foraging social insects. They
differ, however, in several important respects that might influence the costs and benefits
associated with certain behavioral choices. Bumblebees, for example, tend to be
significantly larger than honeybees (Heinrich 1979a; Goulson et al. 2002), which should
produce a disparity in the metabolic cost of foraging, with bumblebees having a lower
mass-specific metabolic rate. We predicted that, with all else being equal, bumblebees
would be less affected by the metabolic cost of flight, and hence be more likely to invest
time into exploration. In addition, bumblebee colonies are typically much smaller than
honeybee colonies, consisting of between 50 and 400 workers (Heinrich 1979a),
compared to tens of thousands of individuals that make up honeybee colonies (Seeley
1995). Thus, each individual foraging bumblebee is responsible for a much larger
portion of the total energy intake into the hive than is an individual honeybee. Again,
with all else being equal, we reasoned that this cost would impose a greater penalty on
bumblebees foraging at a low reward level and would render bumblebees more selective
about accepting low-quality rewards when they have come to expect higher-quality
rewards. Finally, mechanisms of recruitment to food also differ markedly between the
species in that honeybees are able to communicate both distance and direction to a new
found food source through the waggle dance (Seeley 1995; Dyer 2002; Frisch 1993),
whereas bumblebees employ a more generalized “food alert,” in the form of both excited
runs on the nest (Domhaus and Chittka 2001), and the distribution of pheromones within
the nest (Domhaus et al. 2003). With their greater dependence upon first-hand

acquisition of information, as opposed to acquiring it socially, we predicted that

bumblebees would be quicker to switch among foraging options, including both familiar

and novel locations, and would be less likely to return to the nest.

MATERIALS AND METHODS

General Setup

All experiments took place during the summer of 2005 on the Michigan State University
campus, East Lansing, MI. Bees were maintained, and all experiments were performed,
inside a large outdoor flight-cage measuring 35 m (l) X 5.6 m (w) X 2.3 m (h) (Figure

1.1).

 

5 '6 m Test Feeder . l
Hive

E] l

Constant Feeder

 

 

 

 

< 35 m —>
Figure 1.1. The organization of elements in the array. All experiments were performed

within the same flight cage. After all honeybee trials were completed, the honeybee
observation hive was removed and replaced with a bumblebee hive box.

The flight cage provided the bees with a naturalistic environment (sunlight, 24-hour
light/dark cycles, natural fluctuations in weather etc.) yet allowed us to limit the food
resources to which bees had access, ensuring high motivation. The flight cage was
covered with a mesh fabric shade-cloth, which blocked 30% of incident sunlight. Given

the visual spatial resolution that honeybees possess, even in dim light (Warrant et al.

6

1996), and given too that all experiments were performed during daylight hours, there is
no reason to assume that our use of the shade-cloth would significantly influence
foraging behavior.

Honeybees were housed inside a two-frame observation hive. Honeybee
experiments were carried out during June and July, and after concluding the honeybee
experiments, the colony was removed and replaced with a colony of bumblebees. The
bumblebee colony was purchased from Koppert Biological Systems (Romulus,
Michigan), and consisted, initially, of approximately 100 workers. All bumblebee
experiments took place during August and September. Both species were periodically
provided with pollen by placing it directly into the hive. To maintain an active and
accessible pool of foragers, all bees were given ad libitum access to a low quality feeder
(hereafter referred to as the constant feeder), which offered 0.25 M sucrose dispensed

from an inverted jar on top of a Plexiglas plate.

Marking and Scent-Plugging Honeybees

All focal bees were labeled individually. In addition, to minimize recruitment of
unwanted foragers, focal honeybees were scent-plugged at least one day prior to testing.
This entailed applying a mixture of rosin and beeswax to cover the Nasanov gland, the
source of recruitment pheromone at the tip of the bees’ abdomen. The mixture is melted
and then applied in a thin layer with the aid of a soldering iron. This is a common
practice when working with honeybees (Wei et a1. 2002; Towne and Gould 1988; Wei
and Dyer 2009) and there is no evidence to suggest that this practice influences foraging

behavior, other than limiting the recruitment of nestmates. Bumblebees were not scent-

plugged as they lack a Nasanov gland and do not recruit nest mates as efficiently to

specific feeding Sites.

Experimental Approach

To ensure that all test subjects had roughly similar foraging experience, focal bees were
given a minimum of two days foraging at the constant feeder (the only food source
available in the flight cage) before they could be recruited as test subjects. Bees selected
for experiments were recruited, one at a time, to a high quality feeder (the test feeder), a
clear 0.6 ml microcentrifuge tube inserted through a 14 cm square of blue and yellow
paper, dispensing 2.5M sucrose. The test feeder and constant feeder were located
approximately 15m, in opposite directions, from the hive (north and south respectively).
To recruit bees from the constant feeder to the test feeder, the subject was carried by hand
with the aid of a pipette containing with 2.5M sucrose solution, and then placed at the test
feeder. This process was repeated until the focal bee returned to the test feeder
unassisted, usually within two or three assisted trips. If a bee failed to find the test feeder
within five assisted trips, she was excluded. The test feeder was replaced after each trip
as an additional control for odors. To manipulate reward history, bees were given either
one or ten unassisted training trips (with training levels being assigned randomly).

After a bee completed her allotted number of training trips, the sucrose solution in
the test feeder was replaced with water and the testing period began as the focal bee made
her first contact with the now non-rewarding feeder, and continued for one hour beyond
this initial contact. Throughout the testing period, both feeders and the hive were

videotaped and monitored visually by observers. We recorded all unrewarded contacts

 

with the test feeder as well as all contacts with the constant feeder. The decay of the
initially strong tendency to revisit the test feeder was interpreted as evidence of
extinction, although our assay differs from typical studies of extinction (Bouton 1993,
2004; Couvillon et al. 2001; Abramson and Aquino 2002; Couvillon and Bitterman 1980)
in that our bees could choose among a variety of other behavioral options between
“trials” (visits to the test feeder), and could also control the time interval between trials.

Nearly all bees of both species visited the constant feeder at some point during the
test period, but there was a wide range in terms of the level of “commitment”; ranging
from an abrupt probe (or a series of such probes) with extended proboscis, to periods of
more protracted contact. These short probes, presumably a form of sampling behavior,
seemed to be an indication that bees were reluctant to feed at the constant feeder after
having had experience with the high-quality food at the test feeder (a Sign of negative
contrast effects), although a few bees did eventually come to reaccept the constant feeder.
For this assay “feeding” is defined as any instance in which a bee’s proboscis was seen
making unbroken contact with the sucrose solution for a minimum of 30 seconds, which
is approximately half the time required to fill to repletion. Once a bee was observed
feeding, it was dropped from subsequent analysis, and any contacts with the feeder less
than 30 seconds in duration were regarded simply as probing behavior.

To assess the importance of the hive for each test subject during the search period,
we monitored entries into the hive and subsequent exits, as well as time spent within the
hive on each visit. In the case of honeybees, we also observed foragers as they interacted
with nestmates inside the hive, including bees that were continuing to forage at the

constant feeder and were therefore occasionally performing dances.

 

To measure bees’ tendency to explore for novel options, we determined how
much time bees spent out of sight during the testing period (i.e. time Spent not in view of
the observers posted at the hive and feeders). Due to the relatively small size of our flight
cage, we assumed that time Spent out of sight and in transit between locations in the array
would be minimal. Thus, if a bee spent a substantial amount of time not near a feeder or
the hive, then we assumed her to be exploring.

For each experimental group (e. g. l-trip honeybees, lO-trip honeybees, 1-trip
bumblebees, and lO-trip bumblebees) N=10. All statistical analyses were performed
using SPSS, and all videotapes were coded for subsequent analysis using J Watcher

(Blumstein and Daniel 2007).

RESULTS

Our experiments were designed to explore how two species of nectar-feeding bees would
shift among alternative courses of action during an extended period following the
elimination of a highly rewarding source of food. We observed clear differences between
the two species in their allocation of time among options, and in the details of their
pursuit of other options. We explore these details in the following sections. At the
outset, however, it is worth describing the broad patterns that encompass fundamental
similarities in terms of how both species dealt with the problem they faced.

Following the replacement of sugar solution with water (at the test feeder), all
subjects initially persisted in making multiple visits to the feeder even though it provided

no reward, and all subjects exhibited a decrease in this response over time, the classic

10

behavioral phenomenon referred to as “extinction” which has long history in psychology
going back to Pavlov (1927). As the response to the test feeder waned, all bees went
eventually to either the constant feeder or the hive; the preference for one option or the
other differed markedly between Species, as we discuss below.

Nearly all bees made return trips to the constant feeder and subsequently sampled
from it, although few bees resumed feeding there. Instead, this initial “revisiting” was
characterized by bees alighting on the feeder, briefly probing the sucrose solution, often
multiple times, and then flying away, returning either immediately or some minutes later.
This reluctance to reaccept what had just previously been an acceptable resource (all
focal bees had been recruited from the constant feeder to the test feeder) after an
intervening experience on higher quality food were interpreted as an example of negative
contrast effects (Flaherty 1982). Meanwhile our observations of other bees at the
constant feeder in this study indicated that the reduced interest in this feeder was only
present in bees that had had intervening experience at the test feeder. (This phenomenon
is explored at length in Chapter 3.) This effect was very pronounced in both species in
that only 2 out of 20 bumblebees and l of 20 honeybees actually resumed feeding (i.e.
met our criterion for feeding) during the hour-long observation period.

Both species showed some tendency (although stronger in honeybees than in
bumblebees) to visit the hive. This was to be expected given that both are social insects
that rely, to varying degrees, on social information at the nest regarding the availability of
food in the environment (Domhaus and Chittka 2001; Domhaus et al. 2003; Seeley

1986).

ll

In addition to visiting the constant feeder and the hive, all bees dedicated some
amount of time to exploration for novel food sources. This tendency continued at a low
level throughout the observation period, and was interspersed with return visits to the two
feeders and the hive. One striking pattern overall was that bees switched frequently and
repeatedly among the four behavioral options that we observed (visiting the test feeder,
constant feeder, and the hive, as well as exploring). We expected that bees would
generally be likely to first extinguish at the test feeder, before investigating other options,
and would then be more likely to explore having exhausted known forage options. We
observed instead that bees repeatedly shifted very fluidly among all the possible states.

We discuss this in greater depth below.

Time Allocations Among Behavioral Options

In assessing how much time bees spent within view at each of the three locations, we see
both fundamental similarities as well as evidence of species differences (Figure 1.2).
Over the span of the observation period, both species exhibited a decrease in time spent at
the test feeder, and a gradual increase in time spent within the hive. In contrast, both
species also appeared to dedicate a relatively consistent amount of time to the
investigating the constant feeder as well as exploring the environment. We had
anticipated that bees would turn increasingly to exploration as the observation period
progressed, this did not seem to be the case. In terms of species differences, it appeared
that bumblebees consistently dedicated more time to exploration than honeybees, while
honeybees tended to spend more time within the hive. Between training groups (i.e. the

one and ten-trip groups), the differences appeared to be somewhat more subtle. Although

12

number of training trips did not appear to influence honeybee time budgets, there did
appear to be a training effect on bumblebees. Bumblebees in the lO-trip group spent
more time exploring, but less time at the constant feeder, than did the I-trip bumblebees.
The remaining analyses examine all of these behavioral trends in greater detail,
looking at the frequency of contacts with both feeders, total aggregate time spent both in

the hive and exploring, as well as measuring bees’ tendency to Shift among these options.

State Transitions

We observed many differences between species in terms of the overall pattern of state
transitions made by the focal bees. Here we use the phrase state transitions to refers to a
bee’s shifting between the three locations in the environment (test feeder, constant feeder,
and hive), in which bees departed from one location and subsequently arrived at (i.e. was
seen to make contact with) a different location in the environment. Although exploration
would surely qualify as a behavioral state, we found it difficult to define objectively
when a bee began or terminated a period of exploration. We consider patterns of
exploration in later analyses. In addition, self-transitions, or departures from and
immediate returns to the same location, are not analyzed here.

We found no significant differences between species in terms of the total number
of transitions made, (two-way analysis of variance; F (1 ,37) = 0.76, P =0.39), or between
the two training groups within each Species, (two-way analysis of variance ; F(1,37) =
1.65, P =0.21. There were marked differences between species, however, in terms of the

specific nature of those transitions (Figure 1.3). Bumblebees, in both training groups,

13

 

exhibited a strong tendency to shuttle between the two feeders, making relatively few

return trips to the hive.

Mean Total Time Spent
in Each State (minutes)

Mean Total Time Spent
in Each State (minutes)

 

Bumb leb ees

Horgybees

 

 

 

 

 

 

 

 

-W

 

6.0‘

 

 

4.0—
2.0—
0.0‘

Hive
1:] Exploration

I C onstant- Feeder
I Test-Feeder

 

  

 

 

 

 

 

 

 

 

 

 

 

10 20 3O 40 50 60

Time (minutes)

 

 

 

 

10 20 30 40 50 60

Time (minutes)

 

0'1

Sdlli

001

Figure 1.2. Time Budgets. Throughout the observation period, bees consistently moved
among the locations in the array, and spent some period of time out of sight, which we
refer to here as exploration. Bars depict the average amount of time bees spent at each of
the locations or to exploration. Here the hour-long observation period is subdivided into
6, 10-minute time periods.

14

 

Bumbleb ees Honeybees

 

  

Number of Bees

 

 

 

 

 

 

   

u—i
:.
”U
(0
V1
8
CO
“5 _
b S
g .
Z V V
' \ g \\\\ t\\ \ ‘ 23:2; 322:: %
ENV- $§Sm§§ i\\\
2 4 6 8 10 12 14 2 4 6 8 10 12 I4
Transition Number Transition Number
Transition Type

I Test Feeder to Constant Feeder
D Constant Feeder to Test Feeder
Test Feeder to Hive

I Hive to Test Feeder

Hive to Constant Feeder

I Constant Feeder to Hive

Figure 1.3. State transitions. Each transition is type is illustrated in the order that they
occurred for each bee. Only the first 15 transitions for each group as the pattern becomes
noisy thereafter. No bees are shown to transition initially to the test feeder because the
test feeder was the point of origin for all bees at the beginning of the observation period.
Transitions into exploration are not shown because this state was inferred indirectly.

15

At the same time, honeybees were much more likely to make their first transition from
the depleted test feeder to the hive, and as the testing period progressed, became much
more likely, as a group, to distribute their search among all the locations in the array.
Within each species, comparisons between the different training groups do not yield
similarly strong contrasts. There is a subtle difference in terms of the variety of
transitions performed within each species in that bees in each ten-trip group appear more

likely to perform a wider variety of transitions than bees in the respective one-trip groups.

Extinction (test feeder)

Following their initial contact with the non-rewarding test-feeder, all bees continued to
revisit and probe the feeder without reward, and this tendency decreased over time, a
manifestation of the behavioral phenomenon of extinction. As in previous studies of
extinction in honeybees (Couvillon et al. 2001; Couvillon and Bitterman 1980, 1984), we
observed that the frequency of these unrewarded trips waned significantly over the course
of the first 10 minutes (Figure 1.4, A & B). Prior reward history influenced extinction
more dramatically in bumblebees than in honeybees (two-way, repeated measures
analysis of variance indicated a significant three-way interaction between species,
training level, and time since initial contact with the non-rewarding feeder; (F (2,792) =
7.40, P <0.001). These results differ from previous work (e. g. Couvillon and Bitterman
1980; Gil et al. 2007) in that honeybees in our assay appeared not to be Significantly
influenced by the number of previous rewarded trips, although it may be that with

different reward schedules we would have noticed significant effects of previous reward

16

 

W

 

 

 

 

 

 

 

5n A
3'5
1:
8
LL.
59
Q
E
O
U ’e'
E: i’
G) I
2 X ’ v"

1‘1 \ “ ‘

\\ ' ‘| I
‘~ —Burnblebees (1 Trip)
0 r l 7 l T I ' l T —Bumb1ebees (10 Trips)
2 4 6 8 10 “”- Honeybees (1 Trip)
Time (minutes) "" Honeybees (10 Trips)
B

33 40‘
-o
8
E
19
U
E
o
D
O.)
E
U
.5;
2

 

 

 

Time (minutes)

Figure 1.4, A & B. Contacts with the test feeder are shown both in terms of mean
contacts over time (A), and mean cumulative contacts, over the same Span of time (B).
For each group, the first 10 minutes of the testing period is subdivided into 20, 30-second
time periods.

17

history (see Couvillon and Bitterman 1984 for an anlysis of the over-learning extinction
effect in honeybees). What Figure 1.4 fails to show, and what most studies of extinction
are not designed to capture, is that bees begin to search elsewhere in the environment
while undergoing an extinction process at the test feeder. Figure 1.5 illustrates how the
pursuit of other behavioral options interacts with extinction at the test feeder. This figure
reveals that the rate of responding to the test feeder continues to wane well after the end
of the lO-min period shown in Figure 1.4. It also shows that within a few minutes of
their initial contact with the test feeder, i.e. while continuing to persist at the test feeder,
bees are beginning to visit other locations; bumblebees are likelier to visit the constant
feeder, honeybees to visit the hive. These patterns illustrate that extinction clearly is not
a unitary process that must be completed before bees initiate the exploration for other
options.

Overall, honeybees exhibited a much higher degree of persistence than honeybees
at the test feeder (higher resistance to extinction) as measured by the number of contacts
with the feeder throughout the observation period, (two-way analysis of variance; F (1,36)
= 17.20, P <0.001). Additionally, performance in extinction over the hour was
significantly influenced by training level in bumblebees, but not in honeybees (two-way,
repeated measures analysis of variance indicated a significant interaction between Species
and training level; F (1 ,465) = 4.408, P = 0.036). There was no difference between
bumblebee groups in terms of the time at which bees had completed 95% of their

contacts with the test-feeder (t = 0.433, df= 18, P = 0.67).

18

 

Bumblebees Honeybees

 

N
LII

—— Constant Feeder
----- Test Feeder

; N
l T

,-
01

Mean Number of Contacts
with Feeders
T

 

 

 

Sdlll

T
0'01

Mean Number of Contacts
with Feeders
LII
L

 

 

 

 

 

I
I
\
I
l
I
I
I

 

M

‘ 1431 E I l l

10 20 3O 40 50 60 10 20 30 40 50 60
Time Time

Figure 1.5. All contacts with both feeders and trips to the nest. Contacts with both
feeders are Shown in terms of the average number of contacts for each experimental
group, for each 5-minute time period over the entire hour of observation. The small and
large arrows represent the latency for 50% and 100% of bees, respectively, to make their
first entry into the hive. The open and filled triangles represent the latency for 50% and
100% of bees, respectively, to make their first contact with the constant feeder. Among
bumblebees, the large arrows are missing because not all bumblebees made return trips to

the hive by the end of the observation period. Similarly not all honeybees returned to the
constant feeder during the observation period.

19

On average, bumblebees in the l-trip group had completed 95% of their contacts within
22.1 minutes (S.E. i 3.3 minutes), and bumblebees in the 10-trip group had done so

within 24.3 minutes (SE. :1: 4.0 minutes).

Negative Contrast Effects (Constant feeder)
As bees began to withdraw from the test feeder, nearly all made at least one visit to the
constant feeder; this tendency was strongest in bumblebees and was most apparent at the
beginning of the observation period (Figure 1.6); (Wilcoxon’s test of pair-wise
comparisons indicates that in‘both training groups, bumblebees were significantly quicker
to return to the constant feeder; in both cases P < 0.001, Bonferroni adjusted at = 0.0125).
As with extinction, there was no detectable difference between training groups for
honeybees, honeybees in both groups exhibited similar latencies for their first return trip
to the constant feeder. Interestingly, while training level significantly influenced
bumblebees’ resistance to extinction at the test feeder, there was no effect on their
subsequent tendency to make their first contact with the constant feeder.

As mentioned above, these visits to the constant feeder, as indicated by Figures
1.5 and 1.6, represent bees’ return to and sampling from the feeder, but not a resumption
of feeding. This apparent reluctance to feed at the constant feeder, following the
intervening experience at the high quality test feeder, is akin to a negative contrast effect
(Flaherty 1982). (Strictly speaking, this study does not contain the appropriate controls
to conclude that this is a negative incentive contrast effect, but the results of subsequent
work (i.e. Chapter 3) support this conclusion; Townsend-Mehler & Dyer in prep.) This

effect is especially dramatic given that throughout this period the constant-feeder

20

remained the only available source of food in the experimental environment, was being
continually visited by the nest-mates of the focal bees, and had formerly been acceptable

to the focal bees themselves.

 

 

 

 

 

 

 

 

 

1.0A
5 ----- Honeybees (1 Trip)
3.). 0.8— "-'Honeybees (10 Trips)
E — Bumblebees (1 Trip)
3 — Bumblebees (10 Trips)
6 0.6—
>-
8
<1)
59 0.41
0
2:3”
5 0.2—
8
Q)
o. 1‘
0.0— "
I l I I l l I
0 10 20 30 40 50 60

Latency to First Hive Entry (minutes)

Figure 1.6. First return trip to the constant feeder. Survivorship curves Show the latency
in minutes for bees to make their first return trip to the constant feeder (where all bees
had been foraging prior) following their initial contacts with the non-rewarding test
feeder.

This effect may also explain the changes in the rate of visitation at the constant feeder
over time, a pattern that is especially pronounced in bumblebees: the number of contacts
was initially high and subsequently decayed over time, reminiscent of the behavioral
phenomenon of extinction. Thus, our data may provide evidence of two processes that
are occasionally conflated in studies of contrast: the tendency to approach a formerly
acceptable food source (which was initially strong, following the first foray away from

the test feeder) and the willingness to reaccept the reward. It seems noteworthy that

21

while both species exhibit similar reluctance to reaccept the food at the constant feeder,

bumblebees maintain a comparatively high approach tendency.

Returns to the hive

Following their first departure from the test feeder, honeybees were significantly quicker
to return to the nest than were bumblebees (Wilcoxon’s test of pair-wise comparisons
indicates that for bees in the l-trip group, honeybees were quicker to make their first
return trip to the hive; P = 0.003, and that honeybees were also quicker among the lO-trip
bees; P = 0.008, Bonferroni adjusted a = 0.0125; Figure 1.7). While there appears to be
an effect on training for bumblebees, neither species exhibited significant training effects
on hive latencies. For honeybees, this comparatively strong tendency to return to the hive
resulted in honeybees spending much more total time within the hive than bumblebees,
with no significant effect of training level (two-way analysis of variance; F (l ,38) =
4.608, P = 0.038; Figure 1.8).

It is clear that for honeybees, the processes that guide their search following the
depletion of the test feeder are much more centered on the hive than for bumblebees.
This is certainly consistent with the notion that, for honeybees, the hive functions as an
important location for the exchange of social information via the dance language as well

as other social cues (Seeley 1985, 1986; Martinez and Farina 2008; Reinhard et al. 2004).

22

 

m" Honeybees (1 Trip)
---' Honeybees (10 Trips)
-— Bumblebees (1 Trip)
-- Bumblebees (10 Trips)

0
ho
J

o
bx
L

.o
4:.
I

 

 

 

O

N

L
I“

Percentage of Subjects Yet to Make First
Visrt to Constant Feeder

 

 

o
'o
J

 

l l I l I l l
0 10 20 30 40 50 60

Latency to First Visit to Constant Feeder (minutes)

Figure 1.7. First return trip to the hive. Survivorship curves indicating the latency in
minutes for bees’ to make their first return trip to the hive. The curves indicate that nine
of the ten honeybees from each training group made their first entry to the hive within the
first 25 minutes of testing, and that all honeybees made at least one trip home.
Bumblebees were generally slower to enter the hive and, six bumblebees failed to make a
single trip home.

 

 

 

 

 

g 404
83
a.
.2 30— I
3 :7.
.E‘ 2
BE A I
e .. 2" I T
:3
o _L_
2 10... ._l_
8
2 0 Bumblebees Honeybees
I I l l
1 Trip 10 Trips 1 Trip 10 Trips

Figure 1.8. Total time spent exploring (time spent out of sight) (+/- 1 SE). Bumblebees
Spent approximately twice as much time exploring on average than honeybees.

23

 

As mentioned above, other bees from the nest continued to come and go from the
constant feeder throughout the day, so there was typically a low level of dance activity
inside the hive. There was no indication among target bees, however, that they were in
search of dances. Across training groups, honeybees spent, on average, a total of 25.6
minutes (SE. = :t 2.6) in the hive, during which there was almost always at least one non-
focal bee performing a waggle dance to the constant feeder at any given time. Although
only five focal honeybees made direct contact with a dancer, this elicited no behavior
characteristic of dance-following among those focal bees. It is possible that focal bees
were instead in search of other social information such as odors or social cues involved
with trophallaxis, but for practical reasons we did not attempt to quantify these. For
honeybees, the vast majority of the time spent within the hive appeared to be spent in a
state of relative inactivity. In the Discussion, we consider other possible reasons why the
honeybees may benefit from going home if not to seek out social information about

foraging opportunities.

Exploration

While we were not able to directly measure bees’ tendency to initiate bouts of
exploration, we were able to assess this indirectly by calculating the amount of time each
bee spent out of sight of the observers (stationed at all three discrete locations in the
array). The assumption is that time spent away from these three locations constitutes
exploration of the environment. We observed that, for both species, the time needed for a
direct flight between any two elements in the array was generally no more than a few

seconds, meaning that time spent in transit would not contribute significantly to this

24

measure of exploration. Furthermore we frequently observed test bees (particularly
bumblebees) repeatedly flying against the mesh of our flight-cage, in an apparent effort to
explore the field immediately surrounding our flight-cage.

We found that bumblebees dedicated significantly more time to exploring the
environment than did honeybees (two-way analysis of variance indicated significant
species differences F (1 ,36) = 8.86, P < 0.001, with no effect of training level)(Figure
1.9). We also found significant species differences with regard to where in the array bees
launched their bouts of exploring. Bumblebees were more likely to do so from the
constant feeder, honeybees more likely to do so from the test feeder or the hive (two-way
analysis of variance on a log transformed ratio of exploring times from each location
indicates significant species differences; F (1 , 37) = 41.16, P <0.0001, and effects of
training level; F (l , 37) = 7.04, P = 0.01), with no significant interaction (Figure 1.10).
We also wanted to test our initial assumption that bumblebees would initiate exploration
sooner than honeybees, following the depletion of the test feeder, and to assess how this
tendency to explore shifted over time. While we found no significant difference between
species in terms of latency to begin exploring, we did find a species difference in terms of
how experience affected the pattern of exploration. Bumblebees in the l-trip condition
tended to initiate exploring much earlier, following the depletion of the test-feeder, than
did bees in the lO-trip condition (repeated measures one-way analysis of variance
indicated a significant three-way interaction between species, training level, and time
since first contact with the non-rewarding feeder; F (l ,232) =6.691, P=0.010)(Figure

1.11).

25

 

N
O
1

G
l

 

f

Kl]
l

I I
Bumblebees Honeybees

I I I fl.
lTrip lOTrips lTrip lOTrIpS

Mean Total Time Spent Exploring per Bee
(minutes)
I
I

ci

 

 

 

Figure 1.9. Total time spent exploring (time spent out of sight), (+/- 1 SE). Bumblebees
spent approximately twice as much time exploring, on average than honeybees.

26

 

Location Prior to Bout of

 

25 '0_ Exploration

I Constant Feeder
20.0- I Hive

1:] Test Feeder

p—A

p

O
1

Total Average Time Spent Exploring From
Each Location in the Array (min)
u. G
'o o
L l

 

1 Trip 10 Trips 1 Trip 10 Trips

 

     

o
'o
I

Bumblebees Honeybees

Figure 1.10. Average time spent exploring from each location in the array. This figure
shows how the total time spent exploring (+/— 1 SE) was allocated among the elements in
the array. Honeybees were generally more likely to begin exploring after departing the
test feeder, whereas bumblebees tended to dedicate more time to exploring after
departing the constant feeder.

27

 

Training Trips
----- 1

$25???

 

'\

v \

r \

"

—‘--' \‘
\-

Average Time Spent Exploring
Per Time Period (minutes)

  

 

 

 

 

Bumblebees

Honeybees

I l l I T I l I | I I T
10 20 30 40 50 60 10 20 30 40 50 60

Time (minutes) Time (minutes)

 

.9 r"
o o
L

 

Figure 1.11. Distribution of time spent exploring. Lines represent average total amount
of time devoted to exploration, per bee, for each ten-minute time period of the hour-long
observation period

DISCUSSION

While we found several similarities between honeybee and bumblebee foragers in their
responses to the depletion of a high payoff resource, we also found striking contrasts.
The time budget for bees in each condition (Figure 1.2) summarizes many of these
Similarities and differences. In both species, foragers initially persisted at past-profitable
site, presumably in search of the reward, and the frequency of unrewarded visits
decreased over time, in a pattern characteristic of extinction (Bouton 2007).
Subsequently, bees began shuttling among the depleted site, the previously experienced
(but lower quality) food source that other members of their colony continued to exploit,
and the nest. Strikingly, when visiting the lower quality food source, both species were

extremely reluctant to resume feeding. This is presumably a manifestation of negative

28

contrast effects (F laherty 1996) in that the intervening experience of the high-quality food
appeared to significantly reduce their tolerance for the low magnitude reward where all
focal bees had been foraging just prior to testing, and where their nestmates continued to
forage without interruption. Rather than feeding, the majority of focal bees engaged in a
prolonged period of unrewarded search among the locations available to them.

In spite of these similarities, we found that individual honeybees and bumblebees
employ strongly divergent strategies when confronted with a dramatic shift in resource
availability. The bumblebees seemed to rely more heavily than the honeybees on first-
hand investigation of the experimental environment following the depletion of the test
feeder. Bumblebees were quicker to extinguish at the depleted test feeder, they were
quicker to investigate the constant feeder (but not significantly more likely to feed), they
were Slower to return to the hive, they spent less time in the hive, and they dedicated
more time to exploring the environment for novel food sources. Many of these contrasts
fall in line with what one would expect, given the difference between the Species in terms
of their reliance on social information in finding new food sources. Lacking a means of
communicating specific spatial information, bumblebee foragers bear the responsibility
individually for locating their next foraging site. Honeybees, by contrast, seemed less
likely to search for novel opportunities (even if known foraging sites did not provide
acceptable rewards), and to return home, the focal point for any social information about
the environment.

Another difference is that bumblebees were more strongly influenced than
honeybees by the amount of foraging experience they received at the test feeder. While

honeybees in both training groups extinguished at the test feeder in a similar fashion,

29

were equally likely to visit the low quality feeder, and equally likely to visit the hive,
low-experience bumblebees abandoned the test feeder more quickly than high-experience
bumblebees, were more likely to visit the low quality feeder, and were quicker to begin
exploring for novel food sources. Interestingly too, this time spent exploring was, for
bumblebees, strongly influenced by the amount of experience at the high reward feeder.
Thus, not only are bumblebees more sensitive to changes in the distribution of rewards,
but this sensitivity to indicators of change is more affected by prior rewarded experience.
In addition to suggesting a broad pattern of differences in the decision-making
strategies employed by honeybees and bumblebees, our results also Shed new light on
individual behavioral processes which, although well studied in isolation, have never
been studied collectively and in a coordinated way. In the following sections, we discuss

each of these four processes in turn.

Extinction

Extinction is considered to be one of the most fundamental of all behavioral phenomena
(Bouton 2007) and has been documented across an wide variety of taxa and experimental
paradigms. As widely studied as it is, there are still considerable gaps in our
understanding of this phenomena. Extinction curves, for example, are a common symbol
of this process, yet “our understanding of the processes underlying those curves remains
very primitive,” (Rescorla 2001). In addition, there has been little exploration of this
process in functional or evolutionary terms. Why is it, for example, that foraging

animals, as tested in operant paradigms, typically do not extinguish immediately at a site

30

that is no longer rewarding? Our results provide some new insights into these issues,
insights that emerged because of our naturalistic research paradigm.

Extinction is typically expressed as the waning of a learned response to a stimulus
when the consequences (positive or aversive) associated with that stimulus no longer
occur. Both during the conditioning phase and during the extinction phase, the stimuli
are presented in a series of trials, and both the timing of the trials and the animal’s
opportunities to engage in other behaviors are controlled by the experimenters.
Responses of the animal to stimuli other than the extinguishing one are not observed, or
at least are not typically reported. In our paradigm, by contrast, the bee decides the
interval between visits both during conditioning and during training, is free to engage in
other activities during the extinction phase, and is monitored continuously throughout.
While this approach sacrifices experimental control, it may more closely resemble the
natural context of extinction.

This approach reveals that extinction of a bee’s response to the formerly
rewarding test feeder is embedded in a sequence of decisions in a more complicated way
than is usually described. It is generally assumed that this attenuation of a conditioned
response is fundamentally what allows foraging animals to abandon a non-profitable site
and initiate searching elsewhere for food (Adams-Hunt and Jacobs 2007; Bouton 2007).
We observed, however, that bees begin investigating other sites while continuing to visit
the depleted site, albeit with diminishing intensity. All focal bumblebees alternately
shifted their efforts between feeders while extinguishing at the test feeder (Figure 1.5);
honeybees visited the other feeder to a lesser extent, but the majority of honeybees made

several trips to the hive while also extinguishing at the test feeder. Thus, it seems that in

31

complex environments, while the experience of non-reward will elicit at least temporary
abandonment of a past profitable site, it is not necessary to extinguish the approach
tendency to a non-rewarding feeder before the forager begins looking for alternatives
(Pecoraro et al. 1999).

Another notewOrthy observation to emerge from our approach is that the bees
exhibited extinction on two time scales (Figures 1.4 & 1.5). First, they exhibited classic
extinction behavior during the first few minutes immediately following their initial
discovery of the depleted feeder, and then they exhibited a more prolonged extinction
across repeated visits that were interspersed by transitions to other sites. The repeated
visits might be regarded as manifestations of spontaneous recovery of the extinguished
response (Rescorla 2004), although this term does not capture the waning intensity of the
repeated visits. This pattern may suggest that there are multiple control processes
underlying the behavioral phenomenon of extinction to a particular location.

Returning to the question of why animals typically exhibit resistance to
extinction, i.e., why the withdrawal from a non-rewarding Site is not immediate, one
mechanistic hypothesis is that animals build up an expectation of reward level and must
effectively relearn the appropriate response to the new stimulus (i.e. non-reward) (Bouton
2004; Gil et al. 2007). Additionally, one functional hypothesis is that this transient
tendency to search for an absent reward is the result of a balance between a cost of
abandoning a site too abruptly, lest the animal miss a reward nearby, versus a cost of
persisting too long and hence missing out on foraging opportunities elsewhere (Stephens
and Krebs 1986). Presumably, then, animals that differ in how they experience these

costs would differ in. their resistance to extinction.

32

Our data Show that compared to honeybees with similar reward histories,
bumblebees are much less resistant to extinction (i.e., they have a lower threshold for
abandoning the depleted site). This, in conjunction with the fact that bumblebees were
much more apt to resume visiting the constant feeder, Spent more time exploring the
environment, and were much more likely to shuttle back and forth between feeders,
strongly suggest it is more incumbent upon bumblebees to minimize the costs of lost
opportunity that it is for honeybees. This difference may exist because bumblebees,
unlike honeybees, are unable to recruit nest-mates to Specific forage sites hence are more
reliant upon the first-hand acquisition of environmental information (Seeley 1985;
Domhaus and Chittka 2001; Domhaus et al. 2003). Having a relatively low threshold for
abandoning a past-profitable site could potentially be an effective means of increasing the
probability that an individual bumblebee will explore and therefore mitigate the cost of
overrun errors.

This interpretation is incomplete, however. Although it explains why honeybees
are more likely to go home to where they might encounter information via nestmates
about alternative feeding opportunities, it does not explain why they are also more likely
to continue to monitor the past profitable site. One possibility is that while individual
honeybees will incur a cost by doing so, these costs would be significantly minimized
due to colony-wide exploration and subsequent communication of novel forage sites. In
addition because honeybee individuals are comparatively small, and colonies are
significantly larger than bumblebee colonies, a single bee foraging without reward, as
occurs during extinction, would result in comparably small energetic costs on a colony-

wide basis. Finally, these persisting foragers may serve as low cost sentinels that are in a

33

position to monitor a past profitable site until it renews, at which point this information
could be communicated to other nest-mates.

Negative Incentive Contrast Effects

Research in psychology has shown that following a downshift in reward, subjects across
a wide range of taxa exhibit a disruption in consurnmatory behavior relative to unshifted
subjects (Couvillon and Bitterman I984; Bergvall and Balogh 2009; Crespi 1942; Papini
et al. 1988; Wiegmann et al. 2003a). Both bee species exhibited negative contrast
effects: following even a short period of experience with the high concentration sucrose,
both species exhibited almost a complete cessation of feeding behavior at the constant
feeder for the duration of the testing period. Nearly all bees (38 of the 40) made at least
one return trip to the low quality feeder and sampled the solution, but only three bees
were observed reaccepting the low quality (0.25M) sucrose solution. This refusal to
reaccept the constant food source occurred despite the continuous availability Of the low
concentration sucrose, despite the fact that all focal bees had been feeding there
immediately prior to the beginning of training, and despite the fact that their nest-mates
continued to come and go from the constant feeder (and in the case of honeybees,
performed recruitment dances to it).

Our assay differs somewhat from many of the paradigms used to explore this
phenomenon. Typically, animals will be presented with a single food source that is
downshifted in reward magnitude or presented alternately with two feeders that vary in
reward magnitude. As in studies of extinction, the temporal and spatial structure of this
task are typically under tight experimental control. Our more naturalistic approach

allowed bees to decide when successive “trials” at the constant feeder began, and what to

34

 

do between trials. As in the case of extinction, our approach revealed insights that might
not have been apparent with a more controlled approach, but which might be informative
about the way that negative contrast effects occur in nature.

One additional striking pattern was that the negative contrast effects were
observed at a different location from where the bees had initially experienced the
downgrade in food quality. This result suggests that the experience of high quality food
effectively shifted the bees’ expectations for the environment as a whole. While previous
studies have shown that bees do respond to disparity between their expected and realized
reward, whether the disparity occurs with respect to the compound stimulus of the flower
as a whole (Couvillon and Bitterman 1984), or just in terms of the payoff associated with
a given color (Gil et al. 2007; Wiegmann et al. 2003b), none have demonstrated the
formation of expectations of the environment on this Spatial scale.

Another interesting insight comes from observing how negative contrast effects at
the constant feeder interact with extinction at the test feeder. The first approach to the
constant feeder occurred only after the initial phase of extinction at the test feeder, but
subsequently the bees alternated between sampling , but not feeding, at the constant
feeder while investigating other options (the test feeder or the hive). A significant
species difference was observed in the patterns of behavioral transition. After the
depletion of the test feeder all bumblebees made multiple trips to the constant feeder,
sampled from it, and typically proceeded to fly away (often back to the test feeder). In
comparison, honeybees spent much of the testing period either within the hive or probing
the depleted test feeder, making comparatively few trips to the constant feeder. This

indicates a fundamental species difference in how bees respond during extinction.

35

Bumblebees were comparatively quick to abandon the depleted feeder and were
subsequently much more apt to explore the environment, including the constant feeder.

Still another significant observation, which was especially apparent among the
bumblebees, is that bees made a strong approach to the constant feeder following the
initial phase of extinction to the test feeder, and then, over a series of visits during which
they refused to accept the constant feeder, they approached it less frequently and then not
at all, almost as if the initial approach response underwent an extinction of its own. Yet
extinction is defined as “the loss of an associatively based behavior when the associative
relationships that generated the original learning have been changed,” (Lovibond 2004).
The fact that a change in reward at one location could elicit what appears to be extinction
at a different location that remains static cannot be readily explained by the current theory
on this phenomenon, and warrants further investigation.

It should be noted here too that almost immediately after the depletion of the test
feeder, all bumblebees began visiting and probing the constant feeder and that
bumblebees in the l-trip condition visited and sampled from the constant feeder
significantly more often than bees in the 10-trip condition (Figure 1.5). Following their
first contact with the non—rewarding feeder there iS a period of time in which bumblebees
appear to exhibit a brief period of heightened activity, with regards to visiting the
constant feeder, which subsequently decays. This behavior may be analogous to
frustration effects, a phenomenon first reported in rats where subjects trained to a
sequence of two feeders, exhibited a heightened level of activity towards the second
feeder if they found the first one to be unrewarding (Amsel and Roussel 1952). It is

unclear, however, why the l-trip bees would exhibit greater “frustration” than the bees in

36

the 10-trip condition. It seems possible that bees in both groups experience comparable
levels of agitation upon discovering the non-rewarding test feeder, but with less
experience and therefore less associative strength towards the test feeder, the l-trip bees
are more easily drawn elsewhere. While the function of this heightened activity level is
not entirely clear, there is evidence in rats indicating that negative contrast effects often
accompany an increase in search behavior and orientation (Pecoraro et al. 1999), which
presumably serve to aid in the rediscovery of the “lost” food.

Our results may shed light on the functional significance of the negative contrast
effect, a question that is typically not explored in the psychological research on this
phenomenon. Obviously, this behavior does not maximize the bees’ short-term rate of
energy gain (Stephens and Krebs 1986), given that they rejected the foraging option of
which they had knowledge (which was in fact their only option within our experimental
environment). Nor can we attribute this phenomenon to a memory constraint that
resulted in bees forgetting the location of the constant feeder—nearly all bees visited it
many times and yet still refused to reaccept it. Why would it benefit a forager to elevate
its threshold of acceptance of food in such a dramatic way, such that it would refuse to
accept the only option they know about and which had previously been acceptable?
Perhaps the answer is to be found by considering that the intervening experience on the
high-quality test feeder has signaled to the animals either that the entire environment has
improved in quality, or at least that there may be other options better than what the
constant feeder provides. This view is supported by the fact that most plant Species that
provide nectar tend to produce the reward in patterns that are correlated among members

of the same population (Pleasants and Zimmerman 1979; Zimmerman 1981). Being too

37

ready to reaccept the low-quality reward provided by the constant feeder could
potentially entail an opportunity cost: the forager would miss out on discovering other
sources of food that are as good as the test feeder. Another important consideration is that
many plants tend to produce nectar for several days in a row, and to renew the reward
each day on a circadian schedule. Thus, rather than incurring energetic costs, the
additional predation risk and foraging on a known poor quality resource, it may pay to

wait until the good resource renews.

Returns to the hive

For central place foragers, the decision to return home is likely to be mediated by
multiple factors in that the nest can be a Site for the acquisition of social information, a
safe haven from predators, and the location of stored food. In social bees the nest is, in
addition, a place to gather information about the environment, either directly or
indirectly, from nest-mates. Honeybee foragers can monitor shifts in resource
availability through monitoring waggle dances (Seeley 1985), or by receiving other cues
such olfactory and gustatory information that can indicate an influx of food into the hive
(Martinez and Farina 2008; Reinhard et al. 2004). Similarly, unemployed bumblebees
can receive information from nest-mates about newly discovered resources via excited
runs on the nest (Domhaus and Chittka 2001 ), and the distribution of pheromones
(Domhaus et al. 2003). For bumblebees, which lack a way to communicate distance and
direction to novel food sources, foragers are dependent upon the direct acquisition of
information, which is likely reflected in the fact that the bumblebees in our assay spent

considerably more time exploring the environment than they did in the nest. In addition,

38

this species difference might also be a reflection, as mentioned above, of the fact that
unemployed honeybee foragers are simply less costly on a colony-wide level than
unemployed bumblebees, given the significant difference in colony sizes. This is
supported by our observation that honeybees spent approximately 25-30 minutes of the
hour-long testing period within the nest in a state of apparent inactivity. We also
observed honeybees (panicularly honeybees in the lO-trip condition) making multiple
returns to the hive, which frequently resulted in bees spending no more than a minute or
two within the hive on any given trip home— just long enough to enter, scurry around
within the hive and promptly depart. Perhaps as honeybees searched within the flight
cage, the sight of the hive elicited a learned approach response and simply drew them in.
This behavior could potentially be a product of the small Size of our flight cage relative to
typical honeybee foraging range. It is important to restate here too that none of the focal
bees exhibited any indication of dance-following, despite the fact that throughout the
observation period there was consistently a low level of dance behavior as other bees
shuttled between the nest and the constant feeder. Of the 20 focal honeybees, 5 made
direct contact with other dancing bees yet none of these target bees exhibited behavior
characteristic of dance-following. For the majority of this time spent within the nest,
honeybees appeared to be relatively inactive. It is well known that, for honeybees,
maintaining a pool of inactive foragers is an adaptive strategy for exploiting resources en
masse when new profitable sites are discovered (Anderson 2001). On the other hand,
previous research has also Shown that honeybees will not typically initiate dance
following behavior for up to five hours after returning home from a depleted one (Seeley

and Towne I992). The fact that honeybees chose to remain in a state of apparent

39

inactivity, despite having clearly demonstrated some knowledge of available food, clearly

warrants further investigation.

Exploration

There has been considerable research investigating the role of “scouts” and “recruits” in
foraging among social insects (Biesmeijer and de Vries 2001; Beekman et al. 2007;
Dreller 1998; Jaffe and Deneubourg 1992). Our observation that honeybees were less
likely than bumblebees to initiate exploration of the environment accords with previous
research indicating that typically only a small minority of honeybees will begin exploring
for novel food sources after abandoning a past profitable site (Seeley 1995). In addition,
there has been no evidence to suggest that bumblebees have foragers that fall into such
categories as “scouts” or “recruits.” This would explain why bumblebees were generally
much more prone to exploration than honeybees. While we expected to find Species
differences with regards the to the probability of exploring, we were surprised that
honeybees typically launched bouts of exploration after departing one of the two feeder
sites, as opposed to after departing the hive. This would seem to suggest that, for
honeybees in this context, the cues that ultimately elicit exploratory behavior are not
stimuli within the hive, but instead likely to be associated with the experience of a non-
rewarding or otherwise unacceptable feeder. This type of behavior appears to correspond
with Beismeijer and de Vries’ concept of “inspector,” referring to otherwise unemployed
foragers that visit former forage sites (Biesmeijer and de Vries 2001). The exact nature

of this exploring, i.e. whether this principally constitutes a local search within the vicinity

40

of the feeder, or if the feeder serves as springboard to search beyond familiar locations,
also seems to warrant further investigation.

Finally, although the species differed significantly in terms of the total amount of
time Spent exploring, there were strong similarities in terms of how this exploration was
distributed over the testing period. There is a general assumption in much of the foraging
literature that animals shift to explore for novel food sources, or depart a patch, following
a decline in quality of the current patch (Stephens and Krebs 1986; Ydenberg et al. 2007).
There is evidence Showing that this shift in search can be elicited by time spent foraging
without reward (Krebs et al. 1974), failure to attain a minimum level of food reward
(Hodges 1985a) or cues indicating that the patch falls below a minimum level of quality
relative to the environment (Charnov 1976). Given this background, we expected to see a
general increase in the time devoted to search as bees gradually abandoned either one or
both of the feeders. Instead, bees in our assay devoted time to search throughout the
testing-period at a relatively consistent rate (see Figure 1.11). This would appear to
indicate that this decision to search for novel resources is best described as a recursive
process, that bees, and bumblebees in particular, will repeatedly shift between searching

among familiar sites of known quality and the search for novel ones.

Sequential Decision-Making

Perhaps our most significant finding, when looking across all of these decisions
collectively, is the degree to which individual decisions appeared to interact with one
another. Throughout the observation period, bees repeatedly shifted among the various

locations in their environment, and did so in a way that suggests a continuous evaluation

41

of multiple behavioral options. Furthermore, our assay brings into sharp relief an
important dimension of search decisions as animals are likely to experience them in
nature. As Figure 1.2 illustrates, time spent dedicated to one behavioral Option,
necessarily means time spent away from other options, hence animals must effectively
budget their time for all these responses, what Gallistel refers to as “allocating behavior,”
(Gallistel and Gibbon 2000). The question then becomes what type of approach Should
we adOpt in order to consider individual decision processes in the context of a network of
other decisions?

Increasingly within the study of artificial intelligence, the approach has been
toward viewing individual decisions as necessarily part of sequences that must be
completed to reach an eventual goal (Sutton and Barto 1998; Stephens 2008), although
this does not mean simply following a specific sequence of actions. Instead, a sequential
decision-making problem is one in which there are many possible sequences that could
lead to a goal, and the decision-maker’s task is to determine which is the most suitable
(according to some optimality criterion). More formally, the sequential decision-making
framework posits a set of states, a set of actions available in each state that mediate
(perhaps probabilistically) transitions to other states, payoffs associated with each action,
and a way to compute (or learn) the value of alternative actions. The decision-maker’s
goal then is to determine the sequence of decisions that would maximize its payoff from
each starting point among the collection of states, (also referred to as “state-space”). This
framework has proven to be a powerful and effective way to solve problems, principally
within artificial intelligence, that entail large state and action spaces, and uncertainty

about the consequences of individual actions (Grefenstette et al. 1990; Rabinovich et al.

42

2006), and is likely to apply to many decision problems faced by animals as well
(Stephens 2008; Mangel and Clark 1988; Sutton et al. 1998).

Applying this framework to our study enables us to regard individual decisions
not as encapsulated processes, but instead as part of an integrated network of decisions.
We therefore consider the locations in the array (e.g. the hive, test feeder, and constant
feeder), in addition to exploration, as the behavioral states available to bees. Within each
of these states bees must decide whether to remain within that state or shift to another
state. The pattern of transitions among states would then be regarded as a strategy that
depends upon both prior reward history (particularly in the case of bumblebees) and
species-specific differences in behavioral priorities.

Viewing the bees’ behavior within this sequential decision-making framework
opens up a fresh perspective on how we might interpret any one of the individual
responses being observed. Extinction at the test feeder, for example, if viewed in
isolation as a steady reduction in the response to that site, could potentially be interpreted
as the test subject continually lowering her assessment of the intrinsic value of that site.
Viewed in the context of other available decisions, however, it may be that there is no
change in the bee’s evaluation of the test feeder per se, following the contact with the
water, but that other locations are increasing in value. There is substantial evidence
suggesting that spatial context influences performance in extinction (Bouton 2004), we
are suggesting that this notion of “context” be extended to what one might consider to
essentially be the landscape of other decisions as well.

In extending this to other decisions, we see that the initially high probability of

remaining at the test feeder, despite the fact that it had been rendered non-rewarding, is

43

overtaken by the tendency to visit the constant feeder (especially in bumblebees), while at
the same time both of these tendencies are gradually replaced by transitions to the hive.
The research challenge is to decide whether this initial approach to the constant feeder,
and its subsequent abandonment, are explained by increases and decreases in the intrinsic
value of that option, or by shifts in the value of other options, or both.

Additionally, it is interesting to consider how a sequential decision-making
framework might accommodate the negative contrast effects we observed. In both
species, brief experience with a high reward at the test feeder appeared to elevate the
threshold of acceptability of the food at the constant feeder, rendering it generally
unacceptable. This may mean that changes in the bees’ experience with the high quality
reward at one location, despite it being relatively brief, could potentially provide a signal
that the environment as a whole has improved. As a result, bees shift their expectations
of the food available at the constant feeder and possibly at other feeding sources that have
yet to be discovered. In a sequential decision-making framework, this would be akin to
recalculating the values of all states and not just the one where a given shift was
experienced. The most important point here is that existing models of decision-making
appear insufficient to explain the behavioral patterns described here, warranting, we
believe, a serious reconsideration of how these processes are thought to play out in nature
or other complex environments.

Conclusion
Our research exposes the potential pitfalls of partitioning processes associated with the
decision-making of foraging into categories that are analyzed in isolation. By

recognizing the behavioral phenotype of any organism as being more than simply a

44

collection of individual components, and instead as a highly integrated system, we gain a
richer and much more ecologically valid sense of how these processes operate. Using
this integrated approach we are able to examine how species Specific behavioral priorities
and the subsequent landscape of available options sculpt these decision-making
processes, and gain insight into not only the true nature of the challenges animals face in
the search for food, but also of complexity of the solutions that animals have evolved to

meet them.

45

CHAPTER 2
DECDING WHEN TO EXPLORE AND WHEN TO PERSIST: A COMPARISON
OF HONEYBEES AND BUMBLEBEES IN THEIR RESPONSE TO

DOWNSHIFTS IN REWARD

INTRODUCTION

Given the dynamic character of natural environments, any animal seeking food (or any
resource that varies in space and time) must be able to abandon a resource that is
declining in quality and search for something better. This ability has been examined
extensively in behavioral ecology and multiple other disciplines (Azoulay-Schwartz et al.
2004; Cohen et al. 2007; Daw et al. 2006; Kaelbling et al. 1996). Much of the focus in
behavioral ecology has been on how individual foragers weigh the costs and benefits of
staying with a current patch versus searching for novel alternatives (Wajnberg et al. 2006;
Krebs et al. 1974; Charnov 1976; McNamara 1982; Waage 1979; Driessen and Bernstein
1999; Maeda and Takabayashi 2001). Less well understood is how animals that forage
socially, such as eusocial insects, solve this problem. In social organisms the payoffs
associated with different decisions are experienced on both an individual and a group
level, hence they may be influenced by the energy demands of individuals and the group
as well as by the availability of communication systems that aid recruitment to food.

We studied how social factors influence individual decisions of honeybee (Apis
mellifera) and bumblebee (Bombus impatiens) foragers when confronted with a food

resource that, after a period of foraging experience, declines in quality. In general,

46

foragers faced with a declining resource must decide between persisting at that resource
and searching for novel food sources elsewhere.

Our previous work strongly suggests that honeybees and bumblebees employ
divergent strategies for responding to shifts in resource distribution (Townsend-Mehler
and Dyer submitted). Specifically, we found that while bumblebees are quicker than
honeybees to abandon a depleted high-quality food source, and then, after abandoning it,
quicker to return to previously visited site and more likely to investigate the environment
for novel foraging options. Meanwhile, honeybees, in addition to persisting longer than
bumblebees at a depleted site before abandoning it, are quicker to return to the nest and
more likely to remain there for an extended period of time. These patterns were
consistent with the conclusion that while bumblebees are more reliant on individually
acquiring information about foraging options directly from the environment, honeybees
tend to be more reliant upon gathering information, via other nestmates about available
foraging options.

Our previous study subjected honeybees and bumblebees to a rather extreme
change in the quality of the food source: a solution of 2.5M sucrose was replaced with
water. Thus while a clear species difference was apparent in the bees’ subsequent search
behavior, it was unclear whether this difference was the result of the bees’ sensitivity to
changing resource quality or a difference in their subsequent choice behavior, or both.
Here we investigate how the species respond to a less dramatic decrease in the
concentration of a sucrose reward. Furthermore, the difference we observed in the bees’
tendency to explore for novel food resources was based on indirect evidence, namely the

‘ duration of time spent out of site of observers at different locations in the environment.

47

Here our goal is to provide direct evidence of species differences in the exploration for
novel feeding sites. Additionally, our goal is to understand the individual search
strategies of foragers in the context of evolved individual and social differences between
Species.

Honeybees and bumblebees are closely related (Michener 2000) and are both
generalist nectar foragers, but they differ in three important respects that are likely to
influence the costs and benefits of decisions by individual foragers. First, bumblebees
are significantly larger in size than honeybees (Heinrich 1979a; Goulson et al. 2002),
meaning that in terms of mass-specific metabolic rate, it is likely less costly for
bumblebees to engage in flight and explore for novel alternatives that it would be for
honeybees. Second, bumblebee colonies have fewer workers (SO-400; Heinrich 2004)
than honeybee colonies, which tend to have upwards of 15,000 workers (Seeley 1995). It
is likely then that the foraging decisions of individual bumblebees—including the choice
to explore or to persist at a site that is decreasing in profitability —would have a greater
and more immediate impact on that the colony’s energy budget than would be the case
for honeybees. This difference in colony number likely means too that there is a disparity
in opportunity costs between the species associated with searching or persisting at a given
site. Given the comparatively small size of bumblebee colonies, each forager bears a
larger share of responsibility for finding new resources as compared with honeybee
colonies. Finally, bumblebees have a generalized “food alert” that signals the presence of
food (Domhaus and Chittka 2001), but they lack honeybees’ ability to communicate the
location of profitable forage sites (Seeley 1995; Dyer 2002; Frisch 1993); this difference

in the sophistication of the communication system means that individual bumblebees

48

ultimately carry more responsibility for finding profitable resources in the environment
than do honeybees which can depend upon nestmates sharing their information with the
colony. All of these differences appear to point in the same direction: that bumblebees
would be less likely to tolerate a decline in resource profitability, more likely to abandon
a declining resource, and subsequently more likely to search for novel alternatives.

This study builds upon a previous study in which both species were observed in a
naturalistic but controlled environment in which all foraging options could be
manipulated and monitored. In the present study bees were trained to an artificial feeder
which, after a fixed number of foraging trips, was downshifted in quality (i.e. the sucrose
concentration was reduced). We began by measuring bees’ latency to resume feeding at
a food source following varying degrees of downshift in food reward. This enabled us to
assess each species’ sensitivity to shifts in reward as well as subsequent foraging
behavior. In a second experiment, the feeder was downshifted again, to a point below the
known threshold for abandonment of both species, at which point a novel feeder was
introduced into the experimental array, enabling us to assess each species’ likelihood of

discovering the novel food source.

MATERIALS AND METHODS

General Setup
All experiments took place during the summers of 2008 and 2009 on the Michigan State
University campus, East Lansing, MI. Bees were maintained, and all experiments were

performed, inside a large outdoor flight-cage (Figure 2.1) measuring 40 m (l) X 5.6 m

49

(w) x 2.3 m (h). The flight cage provided the bees with a naturalistic environment
(sunlight, 24-hour light/dark cycles, natural fluctuations in weather etc.) yet allowed us to
limit the food resources to which the bees had access. The flight cage was covered with a
mesh fabric shade-cloth, typically used in greenhouses, which blocked 30% of incident
sunlight. Given the visual spatial resolution that honeybees possess, even in dim light
(Warrant et al. 1996), and given too that all experiments were performed during daylight
hours, there is no reason to assume that our use of the Shade-cloth would significantly
influence foraging behavior. Honeybees (Apis mellifera) were housed inside a two-frame
observation hive, and bumblebees (Bombus impatiens) were housed within a small hive-
box. The bumblebee colony was obtained from Koppert Biological Systems (Romulus,
Michigan) and consisted of approximately 100 workers. Both species were housed
within the same extended flight cage, but were separated by plywood partition. Both
hives were periodically provided with pollen by placing it directly into the hive. To
maintain an active and motivated pool of foragers, bees were given ad libitum access to a
low quality feeder; an inverted jar on top of a Plexiglas plate that dispensed an unscented
0.25 M sucrose solution. All focal bees were marked with dots of enamel paint to make
them individually discriminable. To provide all test subjects with roughly similar
foraging experience prior to testing, all focal bees were given a minimum of two days
foraging at a this low quality feeder, the only food source available in the flight cage,
before they were recruited as test subjects. The two Species were studied during the same
portion of the summer, hence should have been exposed to the same regime of

temperature, day length, and humidity.

50

Flight Cage N »

C] Novel Feeder Hive

 

 

 

 

 

5.6 m

 

 

 

[1 Test Feeder
30 m

/ A
\ /

 

Figure 2.1. The organization of elements in the array. Note that the novel feeder was
only present during the observation period of Experiment 2

Marking and scent-plugging honeybees

To control recruitment, all focal honeybees were scent-plugged prior to testing. This
entailed covering the bees’ Nasanov gland with a mixture of rosin and beeswax hence
preventing release of recruitment pheromone. This is a common practice when working
with honeybees (Wei et al. 2002; Towne and Gould 1988; Wei and Dyer 2009) and there
is no evidence of this practice significantly influencing foraging behavior other than
making it difficult for recruits to find a food source indicated by dancers. Bumblebees
were not scent plugged, as they do not possess an efficient mechanism for olfactory

recruitment.

Experiment 1: Latency to resume feeding following a downshift in reward

The goal of this experiment was to test the hypothesis that bumblebees would be more

sensitive than honeybees to a downshift in the reward provided by a foraging site.

51

Specifically, we predicted that it would take a smaller reduction in sucrose concentration
to disrupt their feeding.

Bees selected as test subjects were recruited one at a time from the low quality
feeder to a high quality feeder (hereafter referred to as the test feeder) that dispensed a
2M sucrose solution. This feeder was made from a 0.6 ml microcentrifuge tube inserted
through a 10 cm square of blue paper, mounted on top of a 10cm x 10cm x 4cm block of
wood. To recruit focal bees, they were carried by hand using a pipette, also containing a
2M sucrose solution, and then placed at the test feeder. Typically, bees required two to
three of these assisted trips before they were able to find the test feeder independently.
Bees that failed to find the test feeder independently within four assisted trips were
discarded. Once bees did find the test feeder on their own, they were then allowed five
additional foraging trips. Following each bee’s departure from the feeder, the paper and
microcentrifuge tube were replaced with new ones as a control for odors. After a focal
bee had completed her five foraging trips, we replaced the 2M solution with 1.5M, 1.0M,
or 0.5M sucrose. We then measured the disruption in feeding by recording the bees’
latency to resume feeding at the test feeder. The observation period began once a focal
bee made her first contact with the downshifted feeder, and continued for 45 minutes, or
until the subject resumed feeding at the test feeder. The test feeder was monitored with a
video camera during this period and all contacts with the feeder were recorded. We
considered a bee to be “feeding” if she remained at the feeder with her proboscis
extended into the sucrose solution for a minimum of 30 seconds, this being
approximately half the time required to fill to repletion. Contacts with the feeder lasting

less than 30 seconds in duration were simply recorded as “probes.” If a bee failed to

52

resume feeding within the 45 minute observation period, then her latency was scored as
45 minutes. The different concentrations of downshifted reward were randomized among
subjects. For honeybees the sample size for each treatment group was N = 10. For
bumblebees the sample sizes were N = 16 (for the 0.05M group), N = 15 (for the1.0M

grouP), and N = 23 (for thel .5M group).

Experiment I]: The resumption of feeding versus the search for novel alternatives
Experiment 2 was designed to replicate the conditions of Experiment 1 and then to
examine more closely what bees did during the period of disruption following the
downshift in sucrose concentration of the test feeder. Specifically, we tested the
hypothesis that bumblebees would be more likely to respond to a disruption in feeding by
seeking out, and discovering, novel sources of food. As in Experiment 1, the test feeder
was downshifted after a series of foraging visits. The downshift was to either 0.5M or
0.25M sucrose, levels that were chosen based on results of Experiment 1, because they
were expected to elicit a significant disruption in feeding for both species.

To assess the extent to which bees initiate exploratory behavior following the
downshift, we introduced a novel feeder into the experimental environment at the same
time that the test feeder was downshifted. The novel feeder was identical in appearance
and sucrose concentration to the pre-downshift test feeder: it consisted of a
microcentrifuge tube, containing 2M sucrose, inserted through a 10 cm square of blue
paper, mounted onto a small block of wood. This novel feeder was placed approximately
10 m from the test feeder, a location where bees had never received a food reward, and it

was recessed into the ground to the extent that it was not visible from the test feeder. The

53

novel feeder was visible to human observers within a radius of approximately five
meters; given their visual resolution and the presence of obstructing vegetation, bees
would have had to be within 2 m to see it. This ensured that the bees could not find the
novel feeder through local search in the vicinity of the test feeder of the hive, but would
have to travel in a new direction and distance and search actively for a new foraging
opportunity. Following the downshift bees were observed for 15 minutes, or until they
were seen feeding at the novel feeder, whichever came first. This was a shorter
observation period than in Experiment 1, chosen to provide a more stringent test of the
idea that the bees encountered the novel feeder because they were actively looking for it.
In each experimental group of honeybees N = 10. For the bumblebees downshifted to
0.5M and 0.25M sucrose, N= 12 and 10, respectively.

All video tapes for both experiments were coded initially using .I Watcher
(Blumstein and Daniel 2007), and all subsequent statistical analyses were performed

using SPSS.

RESULTS

Following a downshift in the reward at the test feeder, bumblebees were much more
likely than honeybees either to delay feeding at the downshifted feeder or to abandon it
altogether (Experiment 1). Additionally bumblebees were found to be much more likely
to explore the environment and subsequently discover a novel food Site following their

abandonment of the downshifted food source (Experiment 2).

54

Experiment 1.: Latency to resume feeding following a downshift in reward

For each level of downshift from the original level of 2.0M (to 1.5 M, 1.0M or 0.5M
sucrose), bumblebees exhibited a greater disruption in feeding, which was manifested as
either a greater latency to resume feeding or a failure to resume feeding within the 45
minutes of observation (Figure 2.2). The difference between the species with regards to
disruption in feeding is strongly dependent upon the degree of downshift. Two-way
analysis of variance on log-transformed latency to resume feeding, indicates a significant
interaction between species and downshifted reward-level (F (2,78) = 4.59, P = 0.013).
Additionally, there was no significant difference in latencies between bumblebees

downshifted to 1.5M and honeybees downshifted to 0.5M (t = 0.436, df= 31, P = 0.667).

 

.rs
.0
O
T

I Bumblebees
1:1 Honeybees

   
 

30.00“l

20.00m

10.00-

 

Mean latency to Resume Feeding (min)

 

 

 

 

 

 

.0
o

1.5 1.0 0.5

Sucrose Concentration (in moles) Following

Figure 2.2. The resumption of feeding. Latencies to resume feeding are shown in
minutes :1: l s.e. For each level of downshift (from the 2M solution bees encountered
first), bumblebees took longer to reaccept the feeder. In many cases bumblebees failed to
resume feeding within our experimental time-frame.

55

Experiment 2.: Search behavior following a downshifi in reward

This experiment built upon our results in Experiment 1 by increasing the degree of
downshift enough to cause most bees of both species to abandon the test feeder altogether
and to test whether the bees engaged in exploration following their abandonment of the
test feeder, by providing a novel feeder for bees to discover if they began searching after
the downshift.

Following the downshift to 0.5M sucrose, both species exhibited some degree of
disruption in feeding, although this disruption was much more pronounced in bumblebees
(similar to the results of Experiment 1). While half of the focal bumblebees resumed
feeding following the downshift to 0.5M, all 10 honeybees resumed feeding at some
point during the observation period (Figure 2.3A). Survivorship analysis indicated
significant species differences in the time to resume feeding (Wilcoxon’s statistic on
species differences = 7.00, df= l, P =0.008). Also, subsequent to the 0.5M downshift,
bumblebees were much more likely to discover the novel feeder: 6 of the 15 bumblebees
began feeding at the novel feeder after abandoning the test feeder, whereas no honeybees
discovered the novel feeder (Figure 2.38). Survivorship analysis indicated significant
Species differences in terms of the latency to discover the novel feeder (Wilcoxon’s
statistic on species differences = 3.84, df= l, P =0.050).

A qualitatively Similar, but more exaggerated, pattern was seen subsequent
to the downshift to 0.25M. Both species effectively ceased feeding at the test feeder;
only 1 honeybee resumed feeding while none of the focal bumblebees did so (Figure
2.4A). There was therefore no significant difference between species in terms of their

tendency to resume feeding at the test feeder (Wilcoxon’s statistic = 1.99, df= 1, P =

56

0.317). Although honeybees and bumblebees demonstrated a comparable reluctance to
reaccept the test feeder, their specific behavioral response (i.e. what they did in lieu of
feeding), differed markedly. While 5 of the 10 bumblebees discovered the novel feeder
within the observation period, only one of the honeybees did so (Figure 2.48)
(Wilcoxon’s statistic = 4.397, df= l, P =0.036). Also, while honeybees were just as
reluctant as bumblebees to feed at the test feeder, they had a much stronger tendency to
continue to alight on the test feeder and probe the sucrose solution without feeding
(Figure 2.5) (t = 3.40, df= 12.93, P = 0.005). This finding closely aligns with our
previous work in which we found honeybees to be significantly more resistant to
extinction at a past-profitable feeder, relative to bumblebees (Townsend-Mehler and Dyer
submitted). Based on our previous work we also presume that following a downshift in

reward, honeybees made more frequent return trips to the hive.

57

 

 

O 1—-4

00 O
1 1
”#

.9
ox
I

-------

.o
.1;
L

p
N
l
I
1
1
1
1
1
1
1
1
1
1
1
’J

"" Honeybees ‘1

-— Bumblebees ‘

llllllll
02468101214

Latency to Resume Feeding at Test
Feeder (min)

Percentage of Bees Yet to
Resume Feeding at Test Feeder

.o
T

 

 

 

 

S
1

o
'oo
1

 

.0 .o
«D- O\
1 1

o
1O
1

Percentage of Bees Yet to
Discover Novel Feeder

u“ Honeybees
— Bumblebees

 

 

 

 

o
'o
I

11111111’
02468101214

Latency to Discover Novel Feeder
(mm)

Figure 2.3A & 2.3B. Latency to resume feeding and latency to discover novel food
source after downshift to 0.5M. Although all honeybees resumed feeding within the time
period, only a small percentage of bumblebees did so. Additionally, none of the
honeybees discovered the novel food source, presumably because honeybees did not
generally abandon the downshifted feeder, while nearly half of the bumblebees
discovered the novel feeder.

58

 

 

fl
o
I

o
be
I

.9
Ox
1

0
Its
l

o
lo
I

---- Honeybees
— Bumblebees

Percentage of Bees Yet to
Resume Feeding at Test Feeder

 

 

_o
o

 

IllIlIll
02468101214

Latency to Resume Feeding at Test
Feeder (min)

 

o
be
1

o
'ox
L

 

 

o
21:.
P

Percentage of Bees Yet to
Discover Novel Feeder

o
Eu
I

"" Honeybees
— Bumblebees

 

o
'o
l

 

 

 

llllllll
02468101214

Latency to Discover Novel Feeder
(min)

Figure 2.4A & 2.4B. Latency to resume feeding and latency to discover novel food
source after downshift to 0.25M. Only 1 bee resumed feeding (a honeybee) following the
downshift to 0.25M sucrose. Subsequent to the downshift, few honeybees went on to
discover the novel feeder, although roughly half of the bumblebees did so.

59

 

 

 

 

 

 

 

 

 

 

 

53 40"1

1:

Q)

O)

u. fl.—
g 30-4

If;

2”;

e 20-l

E

*3

1—

§ 10‘

2

l
Bumblebees Honeybees

Figure 2.5. Probing behavior. Honeybees were approximately twice as likely to
repeatedly probe the test feeder following the downshift than were bumblebees. Graph
indicates mean number of total probes, per bee, made over the entire observation period :t
1 s.e.

DISCUSSION

For foraging animals, the decision to abandon one Site and subsequently explore the
environment for a new one can be viewed as the product of two distinct processes. First,
the forager must detect that the current patch has fallen below some minimum threshold
of profitability. Then, once this threshold is passed, the animal must decide what to do
next, e. g., whether to persist in seeking acceptable food at the subthreshold foraging site,
to go home, or to search for resources elsewhere. We found that honeybees and
bumblebees differed markedly in how they expressed these processes in the face of
downshifted rewards. Bumblebees were much more sensitive to shifts in reward

magnitude in that they exhibited some degree of disruption in feeding following even a

60

25% reduction in food reward (Figure 2.2), which scarcely affected honeybees.
Honeybees showed no appreciable change in consummatory behavior unless the feeder
was downshifted by 75% of its original sucrose concentration. As for the bees’ responses
following a downshift to below the threshold of acceptability bumblebees exhibited a
strong tendency to seek out and discover the novel feeder, while honeybees were much
more likely to remain in the vicinity of the downshifted feeder and continue probing
without actually feeding (Figure 2.5).

These observations build upon our earlier findings (Townsend-Mehler and Dyer
submitted) that when a rewarding food source became depleted (or in the case of this
earlier experiment, converted to water), honeybees were slower to abandon it and more
likely to return to the nest, while bumblebees quickly abandoned the feeder and spent
more time engaged in what appeared to be exploration for alternative food sources.
While that study used an indirect measure of exploratory behavior (time spent out of Sight
of observers), here we provide direct evidence that bumblebees are in fact more likely
than honeybees to discover novel food sources pursuant to a downshift in reward.

How can we account for these species differences in terms of their response to
this foraging problem? To consider this from a mechanistic perspective first, the greater
sensitivity of bumblebees to a downshift (Figure 2.2) could simply reflect differences in
sensory thresholds (see Scheiner et al. 2004 for a review on response thresholds in bees).
While differences in perception may certainly underlie observed behavioral differences,
this cannot fully explain the bees’ responses in our experiments. A downshift from 2.0M
to 0.25M sucrose is enough to inhibit the resumption of feeding in both species, but once

this threshold is reached, bumblebees seek out, and tend to discover, novel food sources

61

while honeybees are much more likely to continue probing the downshifted site, (i.e. are
more resistant to extinction), or return to the nest (Townsend-Mehler et al. submitted).
Thus, the patterns we see imply not only a difference in sensitivity to change, but also
divergent behavioral strategies for responding to a reduction in reward once it is detected.
To turn to an evolutionary explanation for this contrast, as mentioned at the outset
of this chapter, that bumblebees and honeybees differ in a number of traits that could lead
to differing payoffs associated with the behavioral options available to foragers. Here we
address what seem the most relevant differences between the species, namely individual
body Size, colony size, and mechanisms for recruitment. The fact that bumblebee
foragers tend to be significantly larger than honeybees, with Bombus impatiens foragers
averaging almost three times the mass opris mellifera (Cnaani et al. 2002; Schaffer et
al. 1979), is very likely to affect the energetic consequences. of individual foraging
decisions. Previous work shows that bumblebees have a lower mass-specific metabolic
rate than honeybees (Wolf et al. 1989; Suarez et al. 1996; Wolf et al. 1999), meaning that
bumblebees incur a lower cost, per unit mass, for sustained flight. Given too that
energetic requirements have been shown to scale directly with hive biomass (Schaffer et
al. 1979), it is tempting to suppose that this lower cost of flight, coupled with lower
energetic requirements would explain the greater propensity of bumblebees to explore for
novel food sources as the food reward is downshifted. These energetic differences alone,
however, are insufficient to explain the behavioral patterns described here given that in
this study, as well as in our previous work (Townsend-Mehler and Dyer submitted), there
is no apparent species difference in terms of the tendency to engage in flight in the 15-60

minutes following the downshift. Instead, what we see is that while bumblebees are

62

much more likely to abandon the downshifted feeder and go on to explore elsewhere,
honeybees tend to continue to revisit and probe the feeder without actually feeding.
Clearly, both species have an incentive to explore, following the downshift in reward, but
because honeybees and bumblebees appear equally likely to engage in unrewarded
search, it seems unlikely that these energetic differences, either in terms of metabolic cost
and colony wide energetic requirements could explain these behavioral differences.

In addition to body Size, the species differ significantly in terms of colony size.
Honeybee colonies typically have one to two orders of magnitude more workers than
bumblebee colonies, and total colony biomass differs to a similar degree (Seeley 1995;
Heinrich 2004). Each individual honeybee forager is therefore effectively responsible for
a much smaller proportion of the daily food intake for the hive, compared to a foraging
bumblebee. This means that individual honeybees foraging without reward would be
proportionately less costly for the colony as a whole, the relevant unit of fitness for a
eusocial insect) than would be the case for bumblebees. This could potentially explain
why honeybees are much more likely to continue to monitor a past-profitable resource,
once it has fallen below the threshold of acceptability. If we consider each foraging
honeybee to be a sensory unit for the colony (Seeley 1994), then the apparent reluctance
of individual foragers to abandon a past profitable site could be considered a means of
maintaining low cost sentinels in the field, monitoring a site until it resumes profitability.
Considering too that honeybees will likely have a multitude of other foragers in the field,
this strategy allows honeybees to explore on a colony-wide basis, while also monitoring
sites that are likely to become profitable. In bumblebees, by contrast, the small size of

the foraging force could mean that the cost of dedicating individual bees to the

63

monitoring past-profitable Sites would likely be too great because there is not a large pool
of scout bees to discover other alternative currently profitable sites. Our results here are
certainly consistent with the notion that while bumblebee individuals shift readily
between exploration and the exploitation of known resources, as with majoring and
minoring (Heinrich 1979b), honeybees are more apt to distribute these tasks across
foragers.

The importance of this “sentinel strategy” among honeybees comes into sharp
relief when we consider the third marked contrast between the species, which is
mechanism by which foragers communicate to their nestmates about food sources.
Although there has been extensive research on the communication systems of both
species (Dyer 2002; Beekman and Bin Lew 2008; Domhaus and Chittka 2004; Domhaus
et al. 2006), their role in influencing the behavioral strategies of individual foragers in the
field remains unclear. Still, it is usually assumed that because the honeybee dances
provide Specific spatial information about the distribution of resources, individual
foragers are less reliant than are bumblebees upon gathering information about foraging
options through individual exploration. Additionally, because honeybees have the
capability of recruiting en masse to a Specific location, this would likely have the effect
of dramatically reducing the potential cost of continuing to monitor a past profitable site,
given the high probability in nature that depleted nectaries will refill over time (N icolson
et al. 2007). That honeybees appear to be much more tolerant of Shifts in reward, could
be a strategy to maximize the likelihood of large groups of foragers converging on the
best food source out of those available on a single food source, given the natural variation

that is likely to exist among flowers within any large patch or flowering tree. This

64

corresponds with previous work suggesting that honeybees are particularly well adapted
for the massive exploitation of large flowering trees Occurring at relatively low densities
(Domhaus and Chittka 2004). For bumblebees, by contrast, our results support the long-
held assumption that the comparatively high degree of flexibility demonstrated by
bumblebees in terms of their tendency toward shifting among possible forage options
plays a significant role in their discovery of novel foraging sites (Heinrich 1979b; Waser
1986)

Bees have been the focus of extensive research on foraging going back several
decades. Historically this research has tended to focus on either the individual decisions
of foragers in the field (Greggers and Menzel 1993; Hodges 1985b; Pyke 1978), or the
collective decision-making of colonies as they shift among available food sources (Seeley
1986; Seeley et al. 1991). Although many studies have investigated the relationship
between the decision-making of foraging at the individual level and group-level
characteristics (Cartar and Abrahams 1996; Schmid-Hempel et al. 1985; Seeley 1989),
ours is one of the few studies to explicitly compare the foraging strategies of different
Species, faced with the same search task. By integrating our investigation of individual
level behavior with information on group-level characteristics, this species comparison
provides clearer evidence about how the moment-to-moment decision-making processes
of individuals are Shaped by social factors. By investigating this relationship between
individual behavior and group level characteristics, we can continue broaden

understanding of the context in which these processes have evolved.

65

CHAPTER 3
A FUNCTIONAL ACCOUNT OF NEGATIVE INCENTIVE CONTRAST

EFFECTS IN HONEYBEES

INTRODUCTION

Extensive research in psychology has shown that across a wide taxonomic range, animal
subjects including humans will exhibit a marked shift in performance when encountering
a food reward that differs Significantly from an expected reward, a behavioral
phenomenon referred to as contrast effects (see F laherty 1996 for a thorough review). It
has been shown that animals that experience a downshift in reward, for example, will
often exhibit a strong disruption in consummatory behavior, relative to unshifted animals
that consistently receive only a low level of reward (negative contrast effects). In rats
this disruption in feeding can be linked to slowed running speed (Crespi 1942) reduced
lick rate (Leszczuk and Flaherty 2000), or a shift toward search behavior (Pecoraro et al.
1999). There is evidence of negative contrast effects across a wide range of taxa,
including rats (Pellegrini and Mustaca 2000), deer (Bergvall et al. 2007) marsupials
(Papini et a1. 1988) monkeys (Roma et al. 2006) and bees (Couvillon and Bitterman
1984; Wiegmann et al. 2003b). Studies have also shown that subjects exhibit an
analogous increase in performance pursuant to an upshift in reward (positive contrast

e fects) although this phenomenon tends to be somewhat less robust experimentally

(Flaherty 1982).

66

That animals commonly evaluate a reward based upon the temporal context in
which it is experienced (i.e. relative to previous rewards) is significant from an ecological
perspective in that it can lead to an apparent deviation from optimality (Stephens and
Krebs 1986). It has been shown, for example, that a disruption in consummatory
behavior in the context of negative contrast effects can lead to a Si gnificant reduction in
body mass in rats (Valle 1990). Similarly, following a downshift, honeybees exhibit a
marked decrease in willingness to exploit available food and this disruption in feeding
can be transient, persisting for a few minutes (Couvillon and Bitterman 1984), or in some
cases persist for up to an hour (Townsend-Mehler and Dyer in prep).

Since it was first described (Crespi 1942; Tinklepaugh 1928), this phenomenon
has been mainly in the realm of psychological research with the principal focus being on
underlying mechanisms (see Flaherty 1996; Amsel 1992). While more recently studies
have begun to illuminate this phenomenon in functional terms, (most notably Freidin et
al. 2009; Pecoraro et al. 1999; Bergvall and Balogh 2009; Bergvall et al. 2007), the
adaptive value of negative contrast effects has yet to be tested explicitly.

Here we set out to examine the negative contrast effects in honeybees as a
potential strategy for maximizing foraging success over the long term, specifically over
the lifetime of the foraging bee. We tested the hypotheses that the reduced foraging
activity that typically accompanies a downshift in reward, and has been Shown to entail
the non-exploitation of available food in bees (Townsend-Mehler and Dyer in prep),
would over the long term, result in greater longevity, relative to non-shifted bees, and

therefore confer greater lifetime foraging success.

67

Previous studies strongly suggest that the shift away from consummatory
behavior that characterizes negative contrast is effectively a Shift toward search oriented
behavior, and presumably a means to increase the probability of rediscovering the “lost”
food (F reidin et al. 2009; Pecoraro et al. 1999). It has been Shown too that a downshift in
reward will elicit a shift in flower-choice in bumblebees (Wiegmann et al. 2003b), and is
thought to analogously guide diet choice in deer (Bergvall et al. 2007). No study to date,
however, has explicitly shown that this tendency to resist available food, following a
downshift, is in any way tied to either greater reproduction or greater foraging success,
typically used as a proxy measure of reproductive success (Stephens and Krebs 1986).

In recent work, we found that after a brief experience with a high quality reward,
honeybees were very unlikely to resume foraging at a familiar low quality site, for
approximately a hour following the downshift, despite the fact that the low quality feeder
was the only available food source (Townsend-Mehler & Dyer, submitted). We set out to
examine this tendency to refrain from exploiting available food following experience
with high quality reward (i.e. successive negative contrast effects) over a long-term
period. Presumably, if honeybees experience a significant downshift in reward on
multiple successive days of foraging, this would result in a comparatively lower level of
foraging activity relative to non-shifted bees. In the present study we test the hypothesis
that this repeated disruption in feeding behavior, following a downshift on successive
days of foraging, is a strategy for maximizing the life-time foraging success of
honeybees, by effectively reducing the amount of time and energy bees invest in known
poor quality food rewards, and instead diverting their more energy toward a known high

payoff reward.

68

The logic for this hypothesis is two-fold. First, the life-Span of a given bee
starting from emergence, is determined to a large extent by the onset of foraging
(Lindauer 1963; Neukirch 1982). Presumably then, limiting foraging activity during this
period in a bees’ life would likely lead to greater longevity and would possibly affect
foraging success over the lifetime of the individual. Second, for many floral species,
depleted patches will replenish over time (typically on a 24 hour cycle) (N icolson et al.
2007). Thu,s bees that abandon a flower or patch that has decreased in quality or become
depleted, will typically return the following day. We reasoned that, for honeybees, a
dramatic reduction in the exploitation of a floral resource following a downshift, (i.e.
negative contrast effects) is likely to be a strategy for minimizing the energetic cost of
pursuing poor quality resources while maximizing the likelihood of being alive to exploit
high quality resources on subsequent days. It is important to point out here too that
because bees forage, not for themselves, but for the colony, that an individual bee’s
foraging activity is not determined by her daily nutritional needs but by the needs of the
colony.

ln our assay, each day test bees were given brief access to a high quality food
source and nearly constant access to a low quality food source. During this period,
control bees within the same hive, and foraging alongside the test bees, were given daily
access to only the low quality feeder, with no other available foraging options. This
allowed us to examine the tendency of test bees, having experienced a high quality
reward, to resume foraging at a known low quality feeder when there were no other
known feeding options. This also allowed us to determine what, if any, effect a reduction

in foraging activity would then have on lifespan and subsequent foraging success.

69

We predicted that following experience at the high payoff feeder, test bees would be
much less likely to resume foraging at a low quality site, relative to control bees which
experience no shift in reward quality, and would therefore live longer and ultimately

contribute more food to the hive than control bees.

MATERIALS AND METHODS

All experiments took place during the summer of 2008 on the Michigan State University
campus, East Lansing, MI. Bees were maintained, and all experiments were performed,
inside a large outdoor flight-cage (Figure 3.1) measuring 20 m (1) x 5 .6 m (w) x 2.3 m
(h). The flight cage provided the bees with a naturalistic environment (sunlight, 24-hour
light/dark cycles, natural fluctuations in weather etc.) yet allowed us completely control
the food resources to which the bees had access. The flight-cage was composed of a
mesh fabric shade cloth, typically used in greenhouses, which blocked 30% of incident
sunlight. Honeybees (Apis mellifera) were housed inside a two-frame observation hive
and were periodically provided with pollen by placing it directly into the hive. To ensure
that all of our subject bees were of known age, bees were hatched within an incubator and
marked with enamel paint dots within 24 hours of emergence. Immediately following the
paint-dotting process bees were added to the hive. Each day for eight hours (9:00 —
17:00), bees were given access to an artificial feeder, an inverted jar on top of a Plexiglas
plate that dispensed a 0.25M sucrose solution (hereafter referred to as the constant
feeder). Because this feeder was the only readily available food source within the flight

cage, shortly after bees began foraging they would begin visiting the constant feeder.

70

Upon their first arrival at the constant feeder they were marked again with additional
paint-dots to make them individually discriminable. Test bees were chosen at random
amongst the pool of marked foragers visiting the constant feeder. To ensure they had
initial knowledge of the constant feeder, only bees that had a minimum of approximately
two consecutive days of foraging experience were chosen to be test bees. Following this
initial period of foraging consistently at the constant feeder, test bees were trained to
visually distinct highly rewarding feeder which dispensed a 2M sucrose solution (Figure
3.1). This feeder (hereafter referred to as the test feeder) was an inverted micropipette
tubule inserted through a piece of blue and yellow paper. The training process entailed
transporting bees from the constant feeder to the test feeder with the use of a pipette also
containing 2M sucrose. Test bees were given 4-5 assisted trips in this fashion after which
time they were typically able to find the test feeder each morning without assistance.
Bees that failed to find the test feeder independently after this period of training were
discarded. The test feeder was made available every morning to test bees only for as long
as it took the test bees to complete a group average of six foraging trips, after which time
the test feeder was replaced with an identical empty feeder. This typically took between
15 and 20 minutes. Control bees seen foraging at the test feeder were immediately
discarded. The constant feeder was monitored throughout the day to record foraging trips

for bees in both control and test groups.

71

Flight Cage N "*

 

 

 

 

 

 

 

 

 

 

/\
Hive
5'6 m 1:] Test Feeder I
Constant
Feeder
v
' 4 30 m \

Figure 3.1. The organization of the elements in the array. The dark line represents
the boundary of the flight cage.

It has been Shown in previous work that honeybees will typically not fill their
crop to capacity when foraging and that the volume of nectar acquired is a function of
sucrose concentration (Schmid-Hempel et al. 1985). We determined, by capturing and
dissecting bees as they departed from feeders dispensing both 0.25M and 2M sucrose,
that when feeding on 2M sucrose bees would often fill their crop, on average to
approximately 40 pl (honeybees have a maximum crop volume of 50 pl), and that when
feeding on 0.25M sucrose bees will typically depart the feeder with, on average, 30 ul of
sucrose solution. From this, we calculated that test bees, exploiting a 2M on 6
consecutive trips would, in the long term, acquire less sucrose than bees that foraged
constantly, for 8 hours, on a 0.25M sucrose solution. This required that test bees could
benefit from fidelity to the high reward test feeder only if they forage for a greater
number of days (i.e. lived longer) than control bees.

We had also determined from previous work that following a downshift in
reward, honeybees will often visit but not necessarily feed at a low quality food source
(Townsend-Mehler and Dyer in prep). To distinguish between these two behaviors

experimentally, we defined “feeding” as any unbroken contact with the sucrose solution,

72

with proboscis extended, for a minimum of 30 seconds, this being approximately half the
time needed for a bee to feed to repletion under normal circumstances.

This process of giving test bees brief access to a high quality reward, and
subsequent access to the low reward feeder for the remainder of the day, was repeated
each day for the lifespan of all focal bees (test bees N = 13, and bees in the control group
N = 27). The entire experiment lasted for approximately 3 weeks, during which time the
weather remained reasonably constant with the average temperature during the day
remaining near 80° F with no precipitation.

Our goal here was three-fold: to assess whether downshifted bees were less likely
to forage at the constant feeder than non-shifted control bees; to determine if this
decrease in foraging activity would translate into a longer lifespan; and finally to
determine if this longer lifespan would ultimately result in a greater lifetime foraging

success, relative to bees that foraged continually at the same low rate of payoff.

RESULTS

Negative Contrast Effects

Each morning, test bees foraged for a brief period of time at the high quality test feeder,
which was then replaced with a non-rewarding feeder. Subsequent to their bout of
foraging at the test feeder, test bees were significantly less likely return to and resume
foraging at the constant feeder, relative to control bees. This reluctance to visit the
constant feeder is evident both over a period of several hours (Figure 3.2) as well as over

their lifetime as foragers (Figure 3.3). Figure 3.2 illustrates that, on their second full day

73

of foraging, which is prior to the introduction of the test feeder, both test bees and control
bees tended to visit the constant feeder at roughly the same rate (t-test on total visits over
this three hour period for bees in each group Show no significant difference between
groups; t = 1.1 1, df= 37, P = 0.272). This indicates that both groups initially found the
constant feeder equally acceptable. On the third day of foraging, test bees were given
brief access to the high reward, which was subsequently removed. Figure 3.2 also
illustrates that following this exposure to the high reward (which typically lasted for
approximately 20 minutes, and an average of six trips), test bees were significantly less
likely to forage at the constant feeder than they were on the previous day (before
exposure to the high reward), and significantly less likely than control bees to forage at
the constant feeder over the same time period (repeated measures analysis of variance
indicates a Si gnificant interaction between day of foraging and whether bees were in the
test or control groups (F (1,76) = 13.56, P <0.001)). The results suggest that the foraging
of the control bees at the constant feeder was also suppressed on Day 3 following the
introduction of the test bees to the high-quality feeder, although not to the same degree as
the foraging of the test bees was. This suppression of the control bees’ behavior may be
the result of direct or, more likely, indirect interactions within the colony between the two

foraging groups (Seeley 1995).

74

 

0
“5:55 20‘ Time Period
._5 E [11011
5g; 15 Q Ill-12
:[_. 10—4 3 .12-1
=3 on .0.
8'50
2 53 5*

O

1.1..

 

 

 

 

CD

 

N
O
l

C?
1

Mean number of
Foraging Tn'ps Per Bcc
‘i’
tsej

m
I

 

 

 

 

CD

   

Day 2 Day 3

Figure 3.2. Total number of visits per hour for each bee on Day 2 and 3. Following a
bout of foraging at the high quality test feeder between 9 and 10 am, test bees were less
likely to forage at the constant feeder than they were the previous day over the same time
period, and less likely than control bees that did not experience a downshift in reward.

75

Figure 3.3. All foraging trips made by test bees to the constant feeder. Each bee is
represented by a single column, with the identifying pattern of paint dots indicated at the
top of each column. Time is indicated in hourly intervals along the horizontal for each
column. The constant feeder was available for 8 hours each day and the lines represent
the total number of visits (in which bees fed) to the constant feeder for each hour that the
feeder was available. The vertical on the left side indicates the total number of tripsto
the constant feeder each hour. The vertical on the right side indicates day of foraging for
each bee. This does not necessarily correspond to the day of the experiment because not
all test bees began foraging on the same day. The gray cells at the top indicate the bees
foraging at the constant feeder prior to their introduction to the test feeder. On
subsequent days, bees began with an average of six trips to the high-quality test feeder,
and then chose whether to visit the constant feeder or abandon foraging for the day. Bees
were observed until they disappeared and were presumed dead (blank cells at the bottom
of each column).

76

 

412illlllllEllll‘ll”lllll

 

 

 

 

Slll l l l lllll ll l l l l l l l l

 

 

.llél ll l Ell l l

l l l l l l l l l

 

 

YWG YWR YWW YYY

 

 

 

 

 

 

 

l l l l l l l l

 

 

fill lll lll lll l l l

 

 

 

 

alll llélllllllllllll l l l l l l l l

 

 

tlél \l ll lll ll lllllll l llll lll l l l l l l

 

 

{liltltllél l l l l ll l l l l ll l

 

 

a1l l l lll lll ll Ill l l ill III I‘

lllll ll l<l<

 

 

 

 

 

 

OA-

 

 

afltl l lll lll 1|le l l l l l l l l l l

 

 

lltl || lZ lll lll lll Sl<l ' l l l l l l l l

 

 

 

 

 

 

 

lll l<l ll llll l l l l l l l l l l l

 

 

 

allll l lll lll l l l l l l l l ill ll

 

 

 

 

 

.4144le l. l l l l l l 6 l Mllllll

 

 

 

The suppression of feeding at the constant feeder seen on the first day after the
test bees were introduced to the high-quality feeder persisted when we observed these
bees on subsequent days, although the bees varied considerably from one another and
over time in how they responded to the constant feeder. Some general patterns are as
follows. First, on the first few days following their introduction to the test feeder, the
response to the constant feeder was very low for most bees (see also Fig. 2) especially
early in the day. Second, although some bees on some days (or for some portion ofa
day) foraged very actively at the constant feeder, their overall level level of interest was
lower than before they had discovered the test feeder, and lower than that of control bees
(see below).

Superimposed on these general patterns is considerable variation among bees in
how they responded to the constant feeder after the test feeder was Shut down each day.
Bee YWG represents what is perhaps the most extreme example of fidelity to the test
feeder in that after her initial experience at the test feeder she made only a few return
trips to the constant feeder in l 1 days of foraging. At the other extreme were bees BBB
and YSW, which made relatively frequent visits to the constant feeder throughout their
lives. An interesting intermediate pattern was Shown by bee YOO. She initially was
very reluctant to return to the constant feeder, making essentially no trips for 10 days,
after which she steeply increased her frequency of trips to the constant feeder. The
pattern shown by YOO may be a general trend. Overall there was a general tendency for
test bees to visit the constant feeder at an increasing rate over their lifetimes (paired t-test
indicates significant difference in total number of visits between bees’ first and last day

of foraging; (t = 5.371, df= 25, P < 0.001).

78

 

Foraging activity and lifetime foraging success

This general reluctance of bees to resume foraging following their brief period of
experience at the high quality feeder resulted in a lower level of daily foraging activity
for test bees, relative to control bees. Test bees made significantly fewer total foraging
trips each day, to both feeders, than did bees in the control group (I = 3.14, df= 38, P =
0.003) (Figure 3.4A). Additionally test bees tended to live longer than control bees (t =
2.72, df= 38, P = 0.043) (Figure 3.48), and had a higher lifetime contribution to the nest
in terms of total millimoles of sucrose delivered (t = 3.59, df= 38, P = 0.001) (Figure
3.4C). It should be pointed out here too that although not all bees began foraging on the
same day, (both control and test bees were recruited into the experiment over time) there
was no significant difference in age between test bees and control bees on their first day

of foraging.

Distribution of Foraging Activity among Test Bees

While some test bees exhibited a remarkably high degree of fidelity to the test
feeder, most of the test bees foraged periodically at the constant feeder. When
considering total amount of sucrose harvested over their lifetime, test bees acquired the
same amount of food from the test feeder as from the constant feeder (t-test: t= 0.45, df=
37, P = 0.96) (Figure 3.5A). Additionally, test bees lived on average 2 days longer than
the 8 days that control bees lived. Figure 3.5B illustrates how much food was harvested
by test bees within their first 8 days of foraging as well as any food harvested beyond

their first 8 days.

79

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5‘ A _ B
Q ‘1— *5 12
E I I “L- 8..
.. 60-0- a 10—
g Dar
(-5 c0 8* . T ..
a 4 ‘. ..a .1.
a ”at: 6 .
a 6 4-
l— on
o 20.0-l g
a: a ..
s < 2 _
0.0 l O 1
Control Test Control Test

'0 C
£33 8.001
E
Q)
Q J
93 6.00
e H
c...
o T

o 4 0 . .. , .
E
2 2-00- . .5
H .

 

 

 

 

 

 

 

 

 

0.00
Control Test

Figure 3.4 A, B & C. Foraging success and lifespan for test and control bees. All bars
represent means i 1 SE. Test bees averaged significantly fewer trips each day than bees
in the control group, and bees in the test group tended to live at least two days longer than
bees in the control group (average lifespan for control bees = 7.8 days i 0.75 , average
lifespan for test bees = 10.8 days :1: 1.01 ). Test bees also contributed nearly twice as
much food, in terms of millimoles of sucrose, than did control bees, over their lifetimes.

80

Over the first 8 days there is no significant difference between test and control bees in
terms of total harvest of sucrose (t-test: t = 1.44, df= 37, P = 0.16). Had test bees not
lived an additional 2 days on average, there would be no difference between groups with
regards to total harvest. This shows that the test bees’ pattern of restricting their foraging
activity at the constant feeder, effectively saving themselves for the test feeder, translated

into higher lifetime foraging success than they would have had.

DISCUSSION

The results support our the hypothesis that the disruption in consummatory behavior
pursuant to a downshift in reward (i.e. negative contrast effects) represents a strategy for
suppressing the foraging activity, and increasing the probability of surviving to exploit
the reward when it renews on subsequent days. Additionally, because of this extension in
their lifespan, bees are able to gather more food in the long term, compared to bees that
forage continuously for a single low level of reward.

Following their experience at the high reward feeder, test bees in our assay
exhibited a significant decrease in their tendency to forage at the low quality feeder,
relative to bees that experienced only the low level of reward. While the results upheld
our prediction that test bees would live longer than control bees, and that this increase in
lifespan would increase their lifetime foraging success, we were surprised that test bees
would visit the constant feeder to the extent that they did. However, the results suggest
that this mixed strategy of foraging from both feeders, as opposed to foraging exclusively

from the test feeder, was essential to boosting their overall foraging success.

81

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A
6.00—
"U
8
a
g 5.00~
:1: __
a l ............................................
e 4.00—
8
V)
g __
3 3.00—
g
E 2.00—
E
*3
1— 1.00—
0.0 1
Test Feeder Constant Feeder
Amormt Harvested from Each Feeder
B
6.00-
_U __
a5-00- .44..
E ----- ------------------------
a 4.00—l _E;E;E§Eg£;§g£55§355553
e eisgsgsgsgsgsgsgsgsgsgs
8 ll
£93004
O .§:§:§:§:§:§:§:§:§:§:§:
U) I-Z-Z-Z-Z-Z-I-Z-Z-I-Z-I
3
glow 3
E l
1004 l
000 §:§:§:§:§:§:§:§:§:§:§:

 

I
First 8 Days All Subsequent Foraing

Total Amount of Sucrose Harvested within First 8
Days and Any Subsquent Foraging

Figure 3.5 A & B. Distributions of intake for test bees based on where food was
acquired and when. A) Test bees harvested approximately the same amount of food, over
their lifetime, from the test feeder as from the constant feeder. Dotted line represents
total lifetime harvest of control bees. B) Within the first 8 days of foraging, test bees’
total harvest was not significantly different from that of control bees. Because test bees
tended to live approximately 2 days longer, on average, the total lifetime harvest of test
bees exceeded that of control bees.

82

Previous studies strongly suggest that the suppression of consummatory behavior
that characterizes negative contrast effects is, in effect, the product of a shift from
exploitation to the exploration for the “lost” resources, (Pecoraro et al. 1999; Fernandez-
Juricic et al. 2004). As mentioned earlier, this hypothesis that negative contrast effects
promote search for currently available alternatives of sufficient quality is not a likely
explanation for what we observed in honeybees. Previous studies have shown that,
following a downshift in reward, individual honeybees are generally unlikely to initiate
exploratory behavior (Townsend-Mehler and Dyer submitted). Following a steep
reduction in profitability, honeybees will typically persist in searching known food
sources and then return home, foraging at poor-quality resources at a much lower level
than bees that have yet to learn about the availability of high quality food. The
honeybee’s strategy makes sense given that their nectar resources tend to replenish over
time (N icolson et al. 2007), The honeybees’ reluctance to resume foraging at the constant
feeder allows them to continue to monitor a past-profitable site over time, and to increase
the chance that they will live to exploit such a site. Meanwhile, being part of a large
social group means that the bee can rely on both food stores and nest mates to supply her
with food while she waits for a past profitable site to renew its profitability.

This suppression of foraging that we observed was highly variable among test
bees. Some test bees exhibited a strong fidelity to the test feeder, accepting only the high
reward food, and essentially forgoing the constant feeder for several days at a time,
despite having demonstrated knowledge of its availability. This wide range in fidelity

could be due, at least in part, to natural variation in sucrose response thresholds. It has

83

 

been shown that sucrose response thresholds are both highly heritable and that multiply
mated queens give rise to worker bees with a range of response thresholds (Page et al.
1998). It seems likely then, that some bees may be less apt to perceive the sugar
concentration of the constant feeder after experience with the high reward, and therefore
are much more likely to avoid foraging there. In addition, there is the general trend that
test bees appear to be more likely to regularly visit the constant feeder as they age. This
is likely analogous to the phenomenon, described in multiple studies on risk-sensitive
foraging, that foragers on an negative energy budget will often become more risk-prone
in their foraging decisions (Kacelnik and Bateson 1996). In functional terms it seems
likely that, for the present study, as bees reach the end of their lifespan and have an
increasingly likelihood of mortality, they consequently have less to gain by “saving
themselves” for the high reward feeder. This could potentially explain the observation
that bees become more likely to visit the constant feeder toward the end of their lifespan,
although this is a question that clearly merits further investigation.

This tendency of honeybees to limit their time spent pursuing a known but low
quality reward, represents a particularly striking example of animals controlling their own
mortality. Although there are multiple other examples of animals’ decision processes
influencing their mortality, particularly with regard to behavior and immune response
(Ardia et al. 2003; McNamara and Buchanan 2005; Svensson et al. 1998), this is, to our
knowledge, the first time this has been demonstrated in bees.

Although ours is the first study to explicitly demonstrate the fitness consequences
of negative contrast effects, it is important to point out that, for honeybees, one reason

that forgoing the low quality feeder a viable strategy, is that honeybees forage socially.

84

 

Because honeybees that choose to limit their foraging are still able to profit from the
foraging efforts of nest-mates, this phenomenon of negative contrast effects as we
describe it here is likely to have very different fitness consequences for animals that do

not forage socially, although this too clearly merits further investigation.

85

REFERENCES

Abramson, C. I. & Aquino, I. S. (2002). Behavioral studies of learning in the
Africanized honey bee (Apis mellifera L.). Brain Behavior and Evolution, 59, 68-86.

Adams-Hunt, M. M. & Jacobs, L. F. (2007). Cognition in foraging. In:
Foraging: behavior and ecology (Ed. by Stephens, D. W., Brown, J. S. & Ydenberg, R.
C .), pp. 105-138. Chicago: University of Chicago Press.

Albert, R., Jeong, H. & Barabasi, A.-L. (2000). Error and attack tolerance of
complex networks. Nature, 406, 378-382.

Amsel, A. (1992). Frustration theory .' an analysis of dispositional learning and
memory Cambridge: Cambridge University Press.

Anderson, C. (2001). The adaptive value of inactive foragers and the scout-
recruit system in honey bee (Apis mellifera) colonies. Behavioral Ecology, 12, 111-119.

Aparicio, C. F. & Baum, W. M. (1997). Comparing locomotion with lever-press
travel in an operant simulation of foraging. pp. 177-192: Soc Exp Analysis Behavior Inc.

Ardia, D. R., Schat, K. A. & Winkler, D. W. (2003). Reproductive effort
reduces long-term immune function in breeding tree swallows (Tachycineta bicolor).
Proceedings of the Royal Society of London Series B-Biological Sciences, 270, 1679-
1683.

Azoulay-Schwartz, R., Kraus, S. & Wilkenfeld, J. (2004). Exploitation vs.
exploration: choosing a supplier in an environment of incomplete information. Decision
Support Systems, 38, 1-18.

Barker, L. M. (1994). Learning and behavior .' a psychobiological perspective.
New York

Bastian, J. & Esch, H. (1970). Nervous control of indirect flight muscles of
honey bees. Zeitschrift Fur Vergleichende Physiologic, 67, 307-&.

Bateson, M., Healy, S. D. & Hurly, T. A. (2003). Context-dependent foraging
decisions in rufous hummingbirds. Proceedings of the Royal Society of London Series 8-
Biological Sciences, 270, 1271-1276.

Bednekoff, P. A. & Lima, S. L. (1998). Re-examining safety in numbers:
interactions between risk dilution and collective detection depend upon predator targeting
behaviour. Proceedings of the Royal Society of London Series B-Biological Sciences,

265, 2021-2026.

86

Beekman, M. (2005). How long will honey bees (Apis mellifera L.) be stimulated
by scent to revisit past-profitable forage sites? Journal of Comparative Physiology a-
Neuroethology Sensory Neural and Behavioral Physiology, 191, 1 1 15-1120.

Beekman, M. & Bin Lew, J. (2008). Foraging in honeybees - when does it pay to
dance? Behavioral Ecology, 19, 255-262.

Beekman, M., Gilchrist, A. L., Duncan, M. & Sumpter, D. J. T. (2007). What
makes a honeybee scout? Behavioral Ecology and Sociobiology, 61, 985-995.

Bergvall, U. A. & Balogh, A. C. V. (2009). Consummatory simultaneous
positive and negative contrast in fallow deer: Implications for selectivity. Mammalian
Biology, 74, 23 8-241.

Bergvall, U. A., Rautio, P., Luotola, T. & Leimar, O. (2007). A test of
simultaneous and successive negative contrast in fallow deer foraging behaviour. Animal
Behaviour, 74, 395-402.

Biesmeijer, J. C. & de Vries, H. (2001). Exploration and exploitation of food
sources by social insect colonies: a revision of the scout-recruit concept. Behavioral
Ecology and Sociobiology, 49, 89-99.

Bitterman, M. E. (1976). lncentive contrast in honey bees. Science, 192, 380-
382. ‘

Bitterman, M. E., Menzel, R., Fietz, A. & Schafer, S. (1983). Classical-
conditioning of proboscis extension in honeybees (Apis mellifera). Journal of
Comparative Psychology, 97, 107-1 19.

Bjork, E. L. & Bjork, R. A. (2003). Intentional forgetting can increase, not
decrease, residual influences of to-be-forgotten information. Journal of Experimental
Psycho/0g), 29, 524-531.

Blumstein, D. T. & Daniel, J. C. (2007). Quantifying behavior the J Watcher
way. Sunderland, MA.: Sinauer Associates.

Bouton, M. E. (1993). Context, time, and memory retrieval in the interference
paradigms of Pavlovian learning. Psychological Bulletin, 114, 80-99.

Bouton, M. E. (1994a). Conditioning, remembering, and forgetting. Journal of
Experimental Psychology-Animal Behavior Processes, 20, 219-231.

Bouton, M. E. (1994b). Context, ambiguity, and classical-conditioning. Current
Directions in Psychological Science, 3, 49-53.

Bouton, M. E. (2002). Context, ambiguity, and unleaming: Sources of relapse
after behavioral extinction. pp. 976-986: Elsevier Science Inc.

87

Bouton, M. E. (2004). Context and behavioral processes in extinction. Learning
& Memory, 11, 485-494.

Bouton, M. E. (2007). Learning and behavior .' a contemporary synthesis.
Sunderland, Mass: Sinauer Associates, Publishers.

Bouton, M. E. & F anselow, M. S. (1997). Learning, motivation, and cognition .'
the functional behaviorism of Robert C. Bolles. Washington, DC: American
Psychological Association.

Bouton, M. E., Westbrook, R. F., Corcoran, K. A. & Maren, S. (2006).
Contextual and temporal modulation of extinction: Behavioral and biological
mechanisms. 352-360.

Brown, J. H., Gillooly, J. E, Allen, A. P., Savage, V. M. & West, G. B. (2004).
Toward a metabolic theory of ecology. Ecology, 85, 1771-1789.

Bruce, V. & Green, P. R. (1990). Visual perception .' physiology, psychology
and ecology, 2nd edn. Hove ; Hillsdale: Erlbaum.

Caraco, T. (1981). Energy budgets, risk, and foraging preferences in dark-eyed
juncos (Junco hyemalis). Behavioral Ecologz and Sociobiology, 8, 213-217.

Cartar, R. V. & Abrahams, M. V. (1996). Risk-sensitive foraging in a patch
departure context: A test with worker bumble bees. pp. 447-458: Amer Soc Zoologists.

Charnov, E. L. (1976). Optimal Foraging, The Marginal Value Theorem.
Theoretical Population Biology, 9, 129-136.

Chittka, L., Geiger, K. & Kunae, J. (1995). The influence of landmarks on
distance estimation of honeybees. Animal Behaviour, 50, 23-31.

Chittka, L., Thomson, J. D. & Waser, N. M. (1999). Flower constancy, insect
psychology, and plant evolution. Naturwissenschaften, 86, 371-377.

Clark, C. W. & Dukas, R. (1994). Balancing foraging and antipredator demands
- an advantage of sociality. American Naturalist, 144, 542-548.

Cnaani, J., Schmid-Hempel, R. & Schmidt, J. O. (2002). Colony development,
larval development and worker reproduction in Bombus impatiens Cresson. Insectes
Sociaux, 49, 164-170.

Cohen, J. D., McClure, S. M. & Yu, A. J. (2007). Should I stay or should I go?
How the human brain manages the trade-off between exploitation and exploration. pp.
933-942: Royal Society.

Cooper, A. J. (1993). Limitations of bumblebee flight, Cambridge.

88

Couvillon, P. A., Ablan, C. D., F erreira, T. P. & Bitterman, M. E. (2001). The
role of nonreinforcement in the learning of honeybees. Quarterly Journal of

Experimental Psychology Section B-Comparative and Physiological Psychology, 54,
127-144.

Couvillon, P. A. & Bitterman, M. E. (1980). Some Phenomena of Associative
Learning in Honeybees. Journal of Comparative and Physiological Psychology, 94, 878-
885.

Couvillon, P. A. & Bitterman, M. E. (1984). The over-leaming extinction effect
and successive negative contrast in honeybees (Apis mellifera) Journal of Comparative
Psychology, 98, 100-109.

Couvillon, P. A., Ferreira, T. P. & Bitterman, M. E. (2003). Delayed
alternation in honeybees (Apis mellifera). Journal of Comparative Psychology, 117, 31-
35.

Crespi, L. P. (1942). Quantitative variation of incentive and performance in the
white rat. American Journal of Psychology, 55, 467-517.

Cuthill, I. C., Kacelnik, A., Krebs, J. R., Haccou, P. & Iwasa, Y. (1990).
Starlings exploiting patches - the effects of recent experience on foraging decisions
Animal Behaviour, 40, 625-640.

Dam, G. & Kordin, K. (2009). Exploration and Exploitation During Sequential
Search. Cognitive Science, 33, 530-541.

Daw, N. D., O'Doherty, J. P., Dayan, P., Seymour, B. & Dolan, R. J. (2006).
Cortical substrates for exploratory decisions in humans. Nature, 441, 876-879.

Dechaume-Moncharmont, F. X., Domhaus, A., Houston, A. 11., McNamara, J.
M., Collins, E. J. & Franks, N. R. (2005). The hidden cost of information in collective
foraging. Proceedings of the Royal Society B-Biological Sciences, 272, 1689-1695.

Dill, L. M. (1987). Animal decision-making and its ecological consequences - the
future of aquatic ecology and behavior. Canadian Journal of Zoology-Revue Canadienne
De Zoologie, 65, 803-81 1.

Domhaus, A., Brockman, A. & Chittka, L. (2003). Bumble bees alert to food
with pheromone from tergal gland. Journal of Comparative Physiology a-Neuroethology
Sensory Neural and Behavioral Physiology, 189, 47-51.

Domhaus, A. & Chittka, L. (2001). Food alert in bumblebees (Bombus
terrestris): possible mechanisms and evolutionary implications. Behavioral Ecology and
Sociobiology, 50, 570-576.

Domhaus, A. & Chittka, L. (2004). Why do honey bees dance? Behavioral
Ecology and Sociobiology, 55, 395-401.

89

Dornhaus, A., Klugl, F., Oechslein, C., Puppe, F. & Chittka, L. (2006).
Benefits of recruitment in honey bees: effects of ecology and colony size in an
individual-based model. Behavioral Ecology, 17, 336-344.

Dreller, C. (1998). Division of labor between scouts and recruits: Genetic
influence and mechanisms. Behavioral Ecology and Sociobiology, 43, 191-196.

Driessen, G. & Bernstein, C. (1999). Patch departure mechanisms and optimal
host exploitation in an insect parasitoid. Journal of Animal Ecology, 68, 445-459.

Dukas, R. & Real, L. A. (1991). Learning foraging tasks by bees; a comparison
between social and solitary species. Animal Behaviour, 42, 269-276.

Dukas, R. & Visscher, P. K. (1994). Lifetime learning by foraging honeybees.
Animal Behaviour, 48, 1007-1012.

Dyer, F. C. (2002). The biology of the dance language. Annual Review of
Entomology, 47, 917-949.

Ellington, C. P., Machin, K. E. & Casey, T. M. (1990). Oxygen-consumption of
bumblebees in forward flight. Nature, 347, 472-473.

F ernandez-Juricic, E., Erichsen, J. T. & Kacelnik, A. (2004a). Visual
perception and social foraging in birds. Trends in Ecology & Evolution, 19, 25-31.

F ernandez-Juricic, E., Siller, S. & Kacelnik, A. (2004b). Flock density, social
foraging, and scanning: an experiment with starlings. Behavioral Ecology, 15, 371-379.

F laherty, C. F. (1982). Incentive contrast - a review of behavioral changes
followinng shifts in reward. Animal Learning & Behavior, 10, 409-440.

Flaherty, C. F. (1996). Incentive relativity. Cambridge ; New York: Cambridge
University Press.

Flaherty, C. F., Capobianco, S. & Hamilton, L. W. (1973). Effects of septal
lesions on retention of negative contrast. Physiology & Behavior, 11, 625-631.

F reidin, E., Cuello, M. I. & Kacelnik, A. (2009). Successive negative contrast in
a bird: starlings' behaviour after unpredictable negative changes in food quality. Animal
Behaviour, 77, 857-865.

Frisch, K. v. (1993). The dance language and orientation of bees, lst Harvard
University Press pbk. edn. Cambridge, Mass: Harvard University Press.

F ulop, A. & Menzel, R. (2000). Risk-indifferent foraging behaviour in
honeybees. Animal Behaviour, 60, 657-666.

90

Galef, B. G. & Wigmore, S. W. (1983). Transfer of information concerning
distant foods - a laboratory investigation of the inforrnation-center hypothesis. Animal
Behaviour, 31, 748-758.

Galef, B. G. & Yarkovsky, N. (2009). Further studies of reliance on socially
acquired information when foraging in potentially risky situations. Animal Behaviour, 7 7,
1329-1335.

Gallistel, C. R. (1990). The organization of learning. Cambridge, Mass: MIT
Press.

Gallistel, C. R. & Gibbon, J. (2000). Time, rate, and conditioning. Psychological
Review, 107, 289-344.

Gallistel, C. R. & NetLibrary Inc. (1993). The organization of learning. In:
Learning, development, and conceptual change. , pp. viii, 648 p. Cambridge, Mass: MIT
Press.

Gegear, R. J. & Thomson, J. D. (2004). Does the flower constancy of
bumblebees reflect foraging economics? Ethology, 110, 793-805.

Gelman, R. & Gallistel, C. R. (1978). The child's understanding of number.
Cambridge, Mass: Harvard University Press.

Gil, M., De Marco, R. J. & Menzel, R. (2007). Learning reward expectations in
honeybees. Learning & Memory, 14, 491-496.

Gil, M. & Farina, W. M. (2002). Foraging reactivation in the honeybee Apis
mellifera L.: factors affecting the return to known nectar sources. Naturwissenschaften,
89, 322-325.

Giraldeau, L.-A. & Caraco, T. (2000). Social foraging theory. Princeton, NJ:
Princeton University Press. '

Glimcher, P. W. (2002). Decisions, Decisions, Decisions: Choosing a Biological
Science of Choice. Neuron, 36, 323-332.

Gordon, D. M. (1999). Interaction patterns and task allocation in ant colonies. In:
Information processing in social insects (Ed. by Detrain, C., Deneubourg, J. L. &
Pasteels, J. M.), pp. 51-67. Basel ; Boston: Birkhauser Verlag.

Grefenstette, J. J., Ramsey, C. L. & Schultz, A. C. (1990). Learning sequential
decision rules using simulation-models and competition. Machine Learning, 5, 355-381.

Greggers, U. & Menzel, R. (1993). Memory dynamics and foraging strategies of
honeybees. Behavioral Ecology and Sociobiology, 32, 17-29.

91

Hall, C. L. & Kramer, D. L. (2008). The economics of tracking a changing
environment: competition and social information. Animal Behaviour, 76, 1609-1619.

Hammer, M. & Menzel, R. (1995). Learning and memory in the honeybee.
Journal of Neuroscience, 15, 1617-1630.

Heinrich, B. (1975). Thermoregulation in bumblebees. 2. energetics of warm-up
and free flight. Journal of Comparative Physiology, 96, 155-166.

Heinrich, B. (1979a). Majoring and minoring by foraing bumblebees, Bombus
vegans - experimental analysis. Ecology, 60, 245-255.

Heinrich, B. (1979b). Bumblebee economics. Cambridge: Harvard University
Press.

Heinrich, B. (2004). Bumblebee economics. Cambridge, Mass.: Harvard
University Press.

Heyes, C. M. (1994). Social-leaming in animals - categories and mechanisms.
Biological Reviews of the Cambridge Philosophical Society, 69, 207-231.

Hodges, C. M. (1985). Bumblebee foraging - the threshold departure rule.
Ecology, 66, 179-187.

Hiilldobler, B. & Wilson, E. O. (1990). The ants. Cambridge, Mass.: Belknap
Press of Harvard University Press.

Jaffe, K. & Deneubourg, J. L. (1992). On foraging, recruitment systems and
optimum number of scouts in eusocial colonies. Insectes Sociaux, 39, 201-213.

Jander, U. & Jander, R. (2002). Allometry and resolution of bee eyes
(Apoidea). Arthropod Structure & Development, 30, 179-193.

Jandt, J. M. & Domhaus, A. (2009). Spatial organization and division of labour
in the bumblebee Bombus impatiens. Animal Behaviour, 77, 641-651.

Kacelnik, A. & Bateson, M. (1996). Risky theories - The effects of variance on
foraging decisions. American Zoologist, 36, 402-434.

Kaelbling, L. P., Littman, M. L. & Moore, A. W. (1996). Reinforcement
learning: A survey. Journal of Artificial Intelligence Research, 4, 237-285.

Kahneman, D. & Tversky, A. (1984). Choices, values, and frames. American
Psychologist, 39, 341-350.

Kamil, A. C. (1983). Optimal foraging theory and the psychology of learning.
American Zoologist, 23, 291-302.

92

 

Kamil, A. C., Krebs, J. R. & Pulliam, H. R. (1987). Foraging behavior. New
York: Plenum Press.

Kamil, A. C. & Roitblat, H. L. (1985). The ecology of foraging behavior -
implications for animal learning and memory. Annual Review of Psychology, 36, 141-
169.

Kammer, A. E. & Heinrich, B. (1974). Metabolic rates related to muscle activity
in bumblebees. Journal of Experimental Biology, 61, 219-227.

Keasar, T., Rashkovich, E., Cohen, D. & Shmida, A. (2002). Bees in two-
armed bandit situations: foraging choices and possible decision mechanisms. Behavioral
Ecology, 13, 757-765.

Keasar, T. & Shmida, A. (2002). Bees in two-armed bandit situations: foraging
choices and possible decision mechanisms. Behavioral Ecology, 13, 757-765.

Kleiber, M. (1947). Body size and metabolic rate. Physiological Reviews, 27,
511-541.

Krebs, J. R. & Davies, N. B. (1997). Behavioural ecology .' an evolutionary
approach, 4th edn. Cambridge, Mass.: Blackwell Science.

Krebs, J. R, Ryan, J. C. & Charnov, E. L. (1974). Hunting by expectation or
optimal foraging - Study of patch use by chickadees. Animal Behaviour, 22, 953-&.

Laverty, T. M. & Plowright, R. C. (1988). Flower handling by bumblebees: a
comparison of specialists and generalists. Animal Behaviour, 36, 733-740.

Lawson, P. R. (1960). Learning and behavior. New York,: Macmillan.

Leszczuk, M. H. & Flaherty, C. F. (2000). Lesions of nucleus accumbens
reduce instrumental but not consummatory negative contrast in rats. Behavioural Brain
Research, 116, 61-79.

Lima, S. L. & Dill, L. M. (1990). Behavioral decisions made under the risk of
predation - a review and prospectus. Canadian Journal of Zoology-Revue Canadienne De
Zoologie, 68, 619-640.

Lindauer, M. (1952). Ein Beitrag zur Frage der Arbeitsteilung im Bienenstaat.
Zeitschrift Fur Vergleichende Physiologic, 34, 299-345.

Lindauer, M. (1963). Allgemeine Sinnesphysiologie. Orienterung im Raum.
F ortschritte der Zoologie, 16, 58-140.

Macarthur, R. H. & Pianka, E. R. (1966). On Optimal Patch use of a Patchy
Environment American Naturalist, 100, 603-609.

93

Mackintosh, N. J. (1974). The psychology of animal learning. London, New
York,: Academic Press.

Maeda, T. & Takabayashi, J. (2001). Patch-leaving decision of the predatory
mite Amblyseius womersleyi (Acari : Phytoseiidae) based on multiple signals from both
inside and outside a prey patch. Journal of Insect Behavior, 14, 829-839.

Mangel, M. & Clark, C. W. (1988). Dynamic modeling in behavioral ecology.
Princeton, N.J.: Princeton University Press.

Martinez, A. & Farina, W. M. (2008). Honeybees modify gustatory
responsiveness after receiving nectar from foragers within the hive. Behavioral Ecology
and Sociobiology, 62, 529-535.

Mazur, J. E. (1986). Learning and behavior. Englewood Cliffs, N.J.: Prentice-
Hall.

Mazur, J. E. (2001). Learning and behavior, 5th edn. Upper Saddle River, N.J.:
Prentice Hall.

McNamara, J. (1982). Optimal patch use in a stocastic environment. Theoretical
Population Biology, 21, 269-288.

McNamara, J. M. & Buchanan, K. L. (2005). Stress, resource allocation, and
mortality. Behavioral Ecology, 16, 1008-1017.

McNamara, J. M. & Houston, A. I. (1992). Evolutionarily stable levels of
vigilance as a function of group-size. Animal Behaviour, 43, 641-658.

McNamara, J. M., Webb, J. N., Collins, E. J., Szekely, T. & Houston, A. I.
(1997). A general technique for computing evolutionarily stable strategies based on errors
in decision-making. Journal of Theoretical Biology, 189, 211-225.

Mellgren, R. L. & Elsmore, T. F. (1991). Extinction of operant-behavior - an
analysis based on foraging considerations. Animal Learning & Behavior, 19, 317-325.

Menzel, R. (1999). Memory dynamics in the honeybee. Journal of Comparative
Physiology a-Neuroethology Sensory Neural and Behavioral Physiology, 185, 323-340.

Menzel, R. (2001). Searching for the memory trace in a mini-brain, the honeybee.
Learning & Memory, 8, 53-62.

Menzel, R. & Gerber, B. (2000). Contextual Modulation of Memory
Consolidation. Learning and Memory, 7, 151-158.

Menzel, R., Manz, G. & Greggers, U. (2001). Massed and spaced learning in
honeybees: The role of CS, US, the intertrial interval, and the test interval. Learning &
Memory, 8, 198-208.

94

Menzel, R. & Muller, U. (1996). Learning and memory in honeybees: From
behavior to neural substrates. Annual Review of Neuroscience, 19, 379-404.

Michener, C. D. (2000). The Bees of the World. Baltimore, MD: Johns Hopkins
University Press.

Moffatt, L. (2000). Changes in the metabolic rate of the foraging honeybee:

effect of the carried weight or of the reward rate? Journal of Comparative Physiology a-
Sensory Neural and Behavioral Physiology, 186, 299-306.

Molet, M., Chittka, L., Stelzer, R. J., Streit, S. & Raine, N. E. (2008). Colony
nutritional status modulates worker responses to foraging recruitment pheromone in the
bumblebee Bombus terrestris. Behavioral Ecology and Sociobiology, 62, 1919-1926.

Neukirch, A. (1982). Depdendence of the life—span of the honeybee (Apis
mellifera) upon flight performance and energy consumption. Journal of Comparative
Physiology, 146, 35-40.

Nicolson, S. W., Nepi, M. & Pacini, E. (2007). Nectaries and nectar. Dordrecht:
Springer.

Nieh, J. C. (2004). Recruitment communication in stingless bees (Hymenoptera,
Apidae, Meliponini). Apidologie, 35, 159-182.

Noble, J., Todd, P. M. & Tuci, E. (2001). Explaining social learning of food

preferences without aversions: an evolutionary simulation model of Norway rats.
Proceedings of the Royal Society of London Series B-Biological Sciences, 268, 141-149.

Nonacs, P. (2001). State dependent behavior and the Marginal Value Theorem.
Behavioral Ecology, 12, 71-83.

Nunez, J. A. (1982). Honeybee foraging strategies at a food source in relation to
its distance from the hive and the rate of sugar flow. Journal of Apicultural Research, 21,
139-150.

Nunez, J. A. & Giurfa, M. (1996). Motivation and regulation of honey bee
foraging. Bee World, 77, 182-196.

Page, R. E., Erber, J. & Fondrk, M. K. (1998). The effect of genotype on
response thresholds to sucrose and foraging behavior of honey bees (Apis mellifera L.).

Journal of Comparative Physiology a-Neuroethology Sensory Neural and Behavioral
Physiology, 182, 489-500.

Page, R. E. & Pankiw, T. (2000). Response thresholds to sucrose predict

foraging division of labor in honeybees. Behavioral Ecology and Sociobiology, 47, 265-
267.

95

Papini, M. R., Mustaca, A. E. & Bitterman, M. E. (1988). Successive negative
contrast in the consummatory responding of didelphid marsupials. Animal Learning &
Behavior, 16, 53-57.

Pavlov, I. P. (1927). Conditioned reflexes. Oxford, UK: Oxford University Press.

Peat, J. & Goulson, D. (2005). Effects of experience and weather on foraging
rate and pollen versus nectar collection in the bumblebee, Bombus terrestris. Behavioral
Ecology and Sociobiology, 58, 152-156.

Pecoraro, N. C., Timberlake, W. D. & Tinsley, M. (1999). Incentive downshifts
evoke search repertoires in rats. Journal of Experimental Psychology-Animal Behavior
Processes, 25, 153-167.

Pellegrini, S. & Mustaca, A. (2000). Consummatory successive negative contrast
with solid food. Learning and Motivation, 31, 200-209.

Pellegrini, S., Seal, M. F. L. & Papini, M. R. (2008). Scaling relative incentive
value: Different adjustments to incentive downshift in pigeons and rats. Behavioural
Processes, 79, 182-188.

Pellegrini, S., Wood, M., Daniel, A. M. & Papini, M. R. (2005). Opioid
receptors modulate recovery from consummatory successive negative contrast.
Behavioural Brain Research, 164, 239-249.

Pleasants, J. M. & Zimmerman, M. (1979). Patchiness in the dispersion of
nectar resources - evidence for hot and cold spots. Oecologia, 41, 283-288.

Poysa, H. (1987). Feeding-vigilance trade-off in the teal (A nas crecca) - effects
of feeding method and predation risk. Behaviour, 103, 108-122.

Pyke, G. H. (1978). Optimal foraging - movement patterns of bumblebees
between inflorescences Theoretical Population Biology, 13, 72-98.

Rabinovich, M. 1., Huerta, R. & Afraimovich, V. (2006). Dynamics of
sequential decision making. Physical Review Letters, 97 , 4.

Real, L. A. (1981). Uncertainty and pollinator-plant interactions - the foraging
behavior of bees and wasps on artificial flowers. Ecology, 62, 20-26.

Reinhard, J., Srinivasan, M. V., Guez, D. & Zhang, S. W. (2004). Floral scents
induce recall of navigational and visual memories in honeybees. Journal of Experimental
Biology, 207, 4371-4381.

Rescorla, R. (2001). Experimental Extinction. In: Contemporary learning
theories--pavlovian conditioning and the status of traditional learning theory (Ed. by
Klein, S. B. & Mowrer, R. R.), pp. 119-154. Hillsdale, N.J.: Erlbaum Associates.

96

Rescorla, R. A. (2004). Spontaneous recovery. Learning & Memory, 11, 501-
509.

Riveros, A. J. & Gronenberg, W. (2009). Olfactory learning and memory in the
bumblebee Bombus occidentalis. Naturwissenschaften, 96, 851-856.

Roma, P. G., Silberberg, A., Ruggiero, A. M. & Suomi, S. J. (2006). Capuchin
monkeys, inequity aversion, and the frustration effect. Journal of Comparative
Psychology, 120, 67-73.

Schaffer, W. M., Jensen, D. B., Hobbs, D. E., Gurevitch, J., Todd, J. R. &
Schaffer, M. V. (1979). Competition, foraging energetics, and the cost of sociality in 3
speices of bees. Ecology, 60, 976-987.

Scheiner, R., Page, R. E. & Erber, J. (2004). Sucrose responsiveness and
behavioral plasticity in honey bees (Apis mellifera). Apidologie, 35, 133-142.

Schmid-Hempel, P., Kacelnik, A. & Houston, A. I. (1985). Honeybeees
maximize efficiency by not filling their crop. Experientia, 41, 1229-1229.

Schuck-Paim, C. & Kacelnik, A. (2007). Choice processes in multialtemative
decision making. Behavioral Ecology, 1-8, 541-550.

Schuck-Paim, C., Pompilio, L. & Kacelnik, A. (2004). State-dependent
decisions cause apparent violations of rationality in animal choice. Plos Biology, 2, 2305-
2315.

Seeley, T. D. (1983). Division of labor between scouts and recruits in honeybee
foraging. Behavioral Ecology and Sociobiology, 12, 253-259.

Seeley, T. D. (1985). The information-center strategy of honeybee foraging.
F ortschritte der Zoologie, 31, 75-90.

Seeley, T. D. (1986). Social foraging by honeybees - how colonies allocate
foragers among patches of flowers. Behavioral Ecology and Sociobiology, 19, 343-354.

Seeley, T. D. (1989). Social foraging in honey bees - how nectar foragers assess
their colony nutritional status. Behavioral Ecology and Sociobiology, 24, 181-199.

Seeley, T. D. (1994). Honey-bee foragers as sensory units of their colonies.
Behavioral Ecology and Sociobiology, 34, 51-62.

Seeley, T. D. (1995). The wisdom of the hive .' the social physiology of honey bee
colonies. Cambridge, Mass.: Harvard University Press.

Seeley, T. D., Camazine, S. & Sneyd, J. (1991). Collective decision-making in
honey-bees - how colonies choose among nectar sources. Behavioral Ecology and
Sociobiology, 28, 277-290.

97

 

Seeley, T. D. & Towne, W. F. (1992). Tactics of dance choice in honey-bees - do
foragers compare dances. Behavioral Ecology and Sociobiology, 30, 59-69.

Seeley, T. D. & Visscher, P. K. (1988). Assessing the benefits of cooperation in

honeybee foraging - search costs, forage quality, and competitive ability. Behavioral
Ecology and Sociobiology, 22, 229-23 7.

Seligman, M. E. (1970). On generality of laws of learning. Psychological
Review, 77, 406-&.

Shafir, S., Waite, T. A. & Smith, B. H. (2002). Context-dependent violations of
rational choice in honeybees (Apis mellifera) and gray jays (Perisoreus canadensis).
Behavioral Ecology and Sociobiology, 51, 180-187.

Shettleworth, S. J. (1998). Cognition, evolution, and behavior. New York:
Oxford University Press.

Shettleworth, S. J., Krebs, J. R., Stephens, D. W. & Gibbon, J. (1988).
Tracking a fluctuating envrionment - a study of sampling. Animal Behaviour, 36, 87-105.

Sih, A. & Christensen, B. (2001). Optimal diet theory: when does it work, and
when and why does it fail? Animal Behaviour, 61, 379-390.

Sladen, F. W. L. (1912). The Humble-Bee. London: Macmillan.

Staples, J. F., Koen, E. L. & Laverty, T. M. (2004). 'Futile cycle' enzymes in
the flight muscles of North American bumblebees. Journal of Experimental Biology, 207,
749-754.

Stephens, D. W. (2008). Decision ecology: Foraging and the ecology of animal
decision making. Cognitive Affective & Behavioral Neuroscience, 8, 475-484.

Stephens, D. W., Brown, J. S. & Ydenberg, R. C. (2007). Foraging: behavior
and ecology. Chicago: University of Chicago Press.

Stephens, D. W. & Dunlap, A. S. (2009). Why do animals make better choices
in patch-leaving problems? Behavioural Processes, 80, 252-260.

Stephens, D. W. & Krebs, J. R. (1986). Foraging theory. Princeton, N.J.:
Princeton University Press.

Stout, J. C. & Goulson, D. (2002). The influence of nectar secretion rates on the
responses of bumblebees (Bombus spp.) to previously visited flowers. Behavioral
Ecology and Sociobiology, 52, 239-246.

Suarez, R. K., Lighten, J. R. B., Joos, B., Roberts, S. P. & Harrison, J. F.
(1996). Energy metabolism, enzymatic flux capacities, and metabolic flux rates in flying

98

honeybees. Proceedings of the National Academy of Sciences of the United States of
America, 93, 12616-12620.

Sutton, R. S. & Barto, A. G. (1998). Reinforcement learning: an introduction.
Cambridge, MA: MIT Press.

Svensson, E., Raberg, L., Koch, C. & Hasselquist, D. (1998). Energetic stress,
immunosuppression and the costs of an antibody response. Functional Ecology, 12, 912-
919.

Team, R. D. C. (2009). R: A language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing.

Templeton, J. J. & Giraldeau, L. A. (1996). Vicarious sampling: The use of
personal and public information by starlings foraging in a simple patchy environment.
Behavioral Ecology and Sociobiology, 38, 105-114.

Thom, C., Gilley, D. C., Hooper, J. & Esch, H. E. (2007). The scent of the
waggle dance. Plos Biology, 5, 1862-1867.

Timberlake, W. (1993). Animal behavior - a continuing synthesis. Annual
Review of Psychology, 44, 675-708.

Tinklepaugh, O. L. (1928). An experimental study of representative factors in
monkeys. Journal of Comparative Psychology, 8, 197-236.

Toth, A. L., Kantarovich, S., Meisel, A. F. & Robinson, G. E. (2005).
Nutritional status influences socially regulated foraging ontogeny in honey bees. Journal
of Experimental Biology, 208, 4641-4649.

Towne, W. F. & Gould, J. L. (1988). The spatial precisions of the honey bees'
dance communication. Journal of Insect Behavior, 1, 129-155.

Townsend-Mehler, J. M. & Dyer, F. C. submitted. Decision-making in a
dynamic environment.

Townsend-Mehler, J. M., Dyer, F. C. & Maida, K. submitted. Exploration
versus Exploitation and Bumblebees versus Honeybees: A look at the tendency to
explore and the biolgocial and social factors that influence this decision in foraging bees.

Tsuruda, J. M. & Page, R. E. (2009). The effects of foraging role and genotype
on light and sucrose responsiveness in honey bees (Apis mellifera L.). Behavioural Brain
Research, 205, 132-137.

Tversky, A. & Kahneman, D. (1981). The framing of decisions and the
psychology of choice. Science, 211, 453-458.

99

Valle, F. P. (1990). Evidence for a calorically costly negative consummatory
contrast effect in rats. Animal Learning & Behavior, 18, 387-392.

van Alpen, J. J. M., Bernstein, C. & Driessen, G. (2003). Information
acquisition and time allocation in insect parasitoids. Trends in Ecology & Evolution, 18,
81-87.

Waage, J. K. (1979). Foraging for patchily-distributed hosts by the parasitoid,
Nemeritis-canescens Journal of Animal Ecology, 48, 353-371.

Wajnberg, E., Bernhard, P., Hamelin, F. & Boivin, G. (2006). Optimal patch

time allocation for time-limited foragers. Behavioral Ecology and Sociobiology, 60, 1-10.

Wajnberg, E., Fauvergue, X. & Pons, O. (2000). Patch leaving decision rules
and the Marginal Value Theorem: an experimental analysis and a simulation model.
Behavioral Ecology, 1 1, 577-586.

Warrant, E., Porombka, T. & Kirchner, W. H. (1996). Neural image
enhancement allows honeybees to see at night. Proceedings of the Royal Society of
London Series B-Biological Sciences, 263, 1521-1526.

Waser, N. M. (1986). Flower constancy - definition, cause, and measurement.
American Naturalist, 127, 593-603.

Wei, C. A. & Dyer, F. C. (2009). Investing in learning: why do honeybees, Apis
mellifera, vary the durations of learning flights? Animal Behaviour, 77, 1165-1177.

Wei, C. A., Rafalko, S. L. & Dyer, F. C. (2002). Deciding to learn: modulation
of learning flights in honeybees, Apis mellifera. Journal of Comparative Physiology a-
Neuroethology Sensory Neural and Behavioral Physiology, 188, 725-737.

Wiegmann, D. D., Wiegmann, D. A. & Waldron, F. A. (2003). Effects of a
reward downshift on the consummatory behavior and flower choices of bumblebee
foragers. Physiology and Behavior, 79, 561 -566.

Wilson, E. O. & Taylor, R. W. (1967). The ants of Polynesia (Hymenoptera,
F ormicidae). Honolulu: Entomology Dept, Bernice P. Bishop Museum.

Wolf, T., Ellington, C., Davis, S. & Feltham, M. (1996a). Validation of the
doubly labelled water technique for bumblebees Bombus terrestris (L.). J Exp Biol, 199,
959-972.

Wolf, T. J., Ellington, C. P. & Begley, I. S. (1999). Foraging casts in
bumblebees: field conditions cause large individual differences. Insectes Sociaux, 46,
291-295.

100

 

Wolf, T. J., Ellington, C. P., Davis, S. & Feltham, M. J. (1996b). Validation of
the doubly labelled water technique for bumblebees Bombus terrestris (L). Journal of
Experimental Biology, 199, 959-972.

Wolf, T. J., Schmidhempel, P., Ellington, C. P. & Stevenson, R. D. (1989).
Physiological correlates of foraging efforts in honeybees - oxygen-consumption and
nectar load. Functional Ecology, 3, 417-424.

Zimmerman, M. (1981). Patchiness in the dispersion of nectar resources -
probable causes. Oecologia, 49, 154-157.

101

TY

11,: till“

 

”'lllllijlllllllilll