1....

. .A _
...t1:..:.. . ........\.
r. A ,

.:.....~....L..
..V. ‘.

 

.: 1.. xi:

 

 

”THESIS

T), «r

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

thesis entitled

THE ROLE OF OPTIC VERSUS METABOLIC ODOMETRY
IN HONEYBEE FORAGING ENERGETICS

presented by

Dina Leslie Grayson

has been accepted towards fulfillment
of the requirements for

M.S. Zoology

degree in

 

 

    

 

I

Major professor

Date T/Z/l‘b/O/
7 /

0—7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c-JCIRCIDatoouepas-ms

 

THE ROLE OF OPTIC VERSUS METABOLIC ODOMETRY IN HONEYBEE
FORAGING EN ERGETICS

By

Dina Leslie Grayson

A THESIS

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

MASTER OF SCIENCE

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

2001

ABSTRACT

THE ROLE OF OPTIC VERSUS METABOLIC ODOMETRY IN HONEYBEE
FORAGIN G ENERGETICS

By

Dina Leslie Grayson

Many animals adjust their foraging decisions in response to changes in the
energetic profitability of food sources. However it remains unclear what mechanisms
animals are using to estimate energetic profitability. This study investigates the
mechanism that honeybees use to estimate the foraging costs associated with flight
distance. The two possibilities tested in this study are the optic odometer hypothesis and
the metabolic odometer hypothesis. I trained bees to fly down tunnels of identical length
but differing optical patterns. Thus, I was able to keep the metabolic odometer’s measure
of flight distance equal between the two tunnels, while producing differing perceptions of
flight distance as measured by the optic odometer. Analyzing the dance responses of
bees to these two tunnels indicated a preference for the optically shorter tunnel, which
implies that they that they were using an optic odometer to asses foraging costs
associated with flight. The implications of this study for honeybee neuroethology and

foraging ecology are discussed.

Copyright by
Dina Leslie Grayson
2001

ACKNOWLEDGMENTS

I would like to acknowledge the help of several individuals during different stages
of this project including my main field assistant Leslie VanLangevelde; our lab
technician Micah Gill; my advisor Dr. Fred C. Dyer; my other committee members Dr.
Tom Getty, Dr. Jim Miller, and Dr. Cathy Bristow; Dr. Alan Tessier for providing
statistical guidance; Dr. Richard Hill for help with the metabolic equations; my fellow
graduate students Cynthia Wei, Puja, Batra, Frances Bartlett, and Kevin Guse; our lab’s
postdoctoral student, Matthew Collett; all of the undergraduates whom helped in the
collection of field data and the transfer of that data to the computers Denise Malkowski,
Doug Dinero, Jeanette McGuire, Kelly Atchinson, Rebecca Oas, Jonathan Blythe, Imran
Musaji; Dr. Tom Seeley for taking the time to discuss the project with me; and most
especially my parents, family, and friends for their enduring support Tom and Leslie
Sullivan, Scott Grayson, Andrew Sullivan, Bret and Elizabeth Cutshall, Yale and Barbara
Grayson, Tom Sullivan 11., J oellen Cabot, Felix Albuerne, Arpita Choudhury, and Debbie

Pierce.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. vi
LIST OF FIGURES ................................................................................. vii
KEY TO SYMBOLS ................................................................................ ix
INTRODUCTION ................................................................................... 1

METHODS ........................................................................................... 6

RESULTS ............................................................................................ 19
DISCUSSION ....................................................................................... 35
LITERATURE CITED ............................................................................. 39

LIST OF TABLES

Table 1 - Test period conditions. This table shows the date, time, sugar concentration and
bee group used for each test period ................................................................ 14

Table 2 — Flight speeds. This table shows the average flight speeds and the minimum and
maximum values of flight speed for the two tunnels. Average flight speeds were
calculated by dividing the distance of the entire flight from the hive entrance (12.1m) by
the average times calculated in Table 3, so values of N are the same as in Table 3 ........ 18

Table 3 — Overall tunnel differences. This table shows average flight times and stomach
volumes for the optically shorter horizontal tunnel and the optically longer julesz tunnel.
Flight times were calculated by averaging each bee over all of her foraging trips and then
averaging all the bees in each tunnel. Stomach volume was measured on a different
group of bees which only made one foraging tn'p each. Shapiro-Wilkes test revealed that
all of the data were non-normal, so Wilcoxon rank sums tests were used to compare
tunnels ................................................................................................ 20

Table 4 — Tunnel differences at different levels of sugar concentration. This table show
average flight times and stomach volumes obtained at different sugar concentrations.
Flight times were obtained by averaging each bee over all of her foraging trips at a
particular sugar concentration and then by averaging all of those bee averages in each
tunnel. Stomach volume data on bees that made only one foraging trip to a feeder of
0.50mol/1 solution was averaged. Shapiro-Wilkes tests revealed that all data were not
normally distributed so Wilcoxon rank sums tests were performed to compare the two
tunnels. Stomach volume was only measured at 0.05mol/l and was normally distributed
so a one-way AN OVA was used to compare the two tunnels .................................. 21

Table 5 — Effect of tunnel and sugar concentration on unloaded mass. This table shows
the results of the two-way ANOVA unloaded mass (mg) = tunnel + sugar concentration +
tunnel * sugar concentration ....................................................................... 24

Table 6 - Predicted relative dance response of the horizontal (H) and julesz (J) tunnels
under the metabolic and optic odometer hypotheses ........................................... 30

vi

LIST OF FIGURES

Figure 1 — Tunnel vision. This figure shows the view from the start to the end of the
horizontal tunnel including the cue card on the front of the tunnel ........................... 9

Figure 2 — Hoop-House setup. This is a complete view of the section of the hoop-house
used for these experiments showing the location of the hive and both tunnels ............. 10

Figure 3 - Tunnel feeder design. This shows one of the tunnel feeders consisting of two

vial caps, which were painted, with wire mesh inside them sitting upon a plastic spill tray
with a cross of colored tape. This picture was taken from above the tunnel and shows the
mesh that covers the tunnel to prevent bee escape .............................................. 11

Figure 4 - Effect of tunnel on stomach volume. This figure shows the mean, SEM, and N
of stomach volume in each tunnel ................................................................ 23

Figure 5 — Effect sugar concentration on unloaded mass. This graph shows the mean,
SEM, and N of unloaded mass at each of the three sugar concentrations tested ............ 25

Figure 6 — Effect of tunnel within each test period on estimated metabolic Efficiency.
This figure shows the average, SEM, and N of Efficiency in each tunnel in each test
period. Average values were obtained by first averaging each bee and then averaging the
bee averages. A Shapiro-Wilkes test for normality on the differences between the two
tunnels generated a probability < W of 0.6427 indicating the distribution was normal. A
paired t-test generated a two-sided p > 0.05 indicating there is no significant difference in
Efficiency between the two tunnels ............................................................... 27

Figure 7 - Effect of tunnel within each test period on Net. This figure shows the average,
SEM, and N of Net in each tunnel in each test period. Average values were obtained by
first averaging each bee and then averaging the bee averages. A Shapiro-Wilkes test for
normality on the difference between the two tunnels generated a probability < W of
0.7517 indicating the distribution was normal. A paired t-test generated two-sided p <
0.0001, indicating Net is greater in the optically longer julesz tunnel ...................... 28

Figure 8 — Effect of tunnel within each test period on Rate. This figure shows the
average, SEM, and N of Rate in each tunnel in each test period. Average values were
obtained by first averaging each bee and then averaging the bee averages. A Shapiro-
Wilkes test for normality on the differences between the two tunnels generated a
probability < W of 0.5739, indicating the distribution was normal. A paired t-test
generated a two-sided p > 0.85, indicating there is no significant difference in Rate
between the two tunnels ............................................................................. 29

vii

Figure 9 - Effect of tunnel within test period on observed probability of dancing. This
figure shows the average, SEM, and N of observed probability of dancing in each tunnel
in each test period. Average values were obtained by first averaging each bee and then
averaging the bee averages. A Shapiro-Wilkes test for normality on the arcsin
transformed differences between the two tunnel revealed a probability < W of .4051,
indicating the distribution was normal. A paired t-test on the arrcsin transformed data
generated a two-sided p < 0.05, indicating observed probability of dancing was
significantly greater in the optically shorter horizontal tunnel ................................ 31

Figure 10 — Effect of tunnel within test period on number of waggle runs performed.
This figure shows the average, SEM, and N of number of waggle runs performed in each
tunnel in each test period. Average values were obtained by first averaging each bee and
the averaging the bee averages. A Shapiro-Wilkes test for normality on the differences
between the two tunnels generated a revealed a probability < W of 0.0181, indicating the
distribution was non-normal. A Wilcoxon signed-rank test for paired values generated a
two-sided p < 0.005, indicating the number of waggle runs was significantly greater in
the optically shorter horizontal tunnel ............................................................ 32

Figure 11 - Comparison of horizontal to julesz tunnel ratio of number of waggle runs
performed to estimated metabolic Efficiency. Values were obtained by dividing the
average horizontal tunnel value for each test period by the average julesz tunnel value
(given in Tables 6 and 10 for estimated metabolic Efficiency and number of waggle runs
performed respectively). Since ratios are non-normal, a Wilcoxon signed rank test for
paired values was performed and generated a p < 0.03 ........................................ 34

viii

KEY TO SYMBOLS

G = Gain (J) = SV (ul) 0 SC (umol/u1)0 5.8 (J/umol)
SV = Stomach Volume (111) = amount of sucrose solution imbibed

SC = Sugar Concentration (umol/ul)= concentration of the sucrose solution imbibed

4
C = COSI (J) = ngi (S) 0 MRi (J/S)

T, = Time (s) of each of the four segments of a foraging trip

T1 = Time (s) of the flight from the hive to the food source

T2 = Time (s) spent foraging at the food source

T3 = Time (s) spent flying from the food source to the hive

T4 = Time (s) spent at the hive upon returning from the food source

MR, = Metabolic Rate (J/s) for each of the four segments of a foraging trip

MR1(J/s) = 0.003015 0 UM (mg) 0'6” = metabolic rate of bee flying to the food

MR2 (J/s) = 0.0026044 0 UM (mg) 0'4” = metabolic rate of a bee sitting at the food

MR3 (J/s) = 0.003015 0 LM (mg) 0629 = metabolic rate of a bee flying from the food

MR4 (J/s) = 0.0026044 0 LM (mg) 0'4” = metabolic rate of a bee sitting or walking in the
hive

UM = Unloaded Mass (mg) = weight of a bee before it has imbibed any sucrose solution

LM = Loaded Mass (mg) = weight of a bee and the sucrose solution it has imbibed

ix

INTRODUCTION

Since the inception of optimal foraging models in the 1960’s, research into the
foraging decisions of animals has flourished (for review see Stephens and Krebs, 1986).
Many species have been found to be sensitive to differences in the profitability of
alternative food sources. Profitability can be defined as the quality of a food source in
terms its energetic gains and cost (Stephens and Krebs, 1986; Cuthill and Houston, 1997).
Since animals respond to differences in profitability between food sources, it is clear they
must be measuring profitability, but the mechanisms they use remain largely unknown
and inaccessible to researchers. Honeybees provide a rare opportunity to elucidate a
mechanism by which energetic costs associated with flight are perceived.

Honeybees are useful for addressing this question because foragers indicate their
measure of food source profitability in the waggle dance. The waggle dance is used by
foragers to communicate the direction and distance traveled to a food source (for review
see von Frisch, 1967; Dyer, 2002). When a honeybee returns to the hive from a
profitable food source she dances in a figure-eight shaped circuit. The midline between
the two circles of the eight is emphasized by the honeybee ‘waggling’ her abdomen. The
orientation of each waggling run indicates the direction to the food, and the duration of
the waggling run signals the distance to the food.

Honeybees indicate their measure of food source profitability by modulating the
number of waggling tuns performed in the hive after each foraging trip (von Frisch, 1967;
Seeley and Towne, 1992; Seeley, 1995; Seeley et al., 2000). A greater number of
waggling runs are performed for more profitable food sources. Although several aspects

of the dance are positively correlated with profitability (for review see Seeley, 1995;

Stabentheiner, 1996; Wainselboim and Farina, 2000a; Wainselboim and Farina, 2000b;
Seeley etal., 2000) there are only two which honeybees have been shown to utilize.
These are the probability of performing any dances and the number of waggle runs
performed per foraging trip (Seeley and Towne, 1992). These variables determine the
rate at which new bees arrive to exploit a food source through a complex set of individual
and social decisions (Seeley et a1. 1991; Seeley and Towne, 1992; de Vries and
Biesmeijer, 1998). The process commences when bees that are ready to forage but have
not yet found a food source (recruits) randomly choose a dancer to follow. They will
follow a dancer for a few circuits and then leave to search for the site being indicated
(Seeley et al., 1991; Seeley and Towne, 1992). Since recruits only follow a dancer for a
few circuits, they do not have an opportunity to assess the profitability signal directly
(Seeley and Towne, 1992). Yet, since the choice of which dancer to follow is random
and new recruits are readily available to replace those that leave to forage, the number of
recruits that are sent to a particular food source can be predicted by the proportion of
dances in the hive that indicate that food resource (Seeley and Towne, 1992). Thus a
colony allocates the majority of its foraging effort to the more profitable food sources
even though no one bee has knowledge about the profitability of any food source except
her own.

The question of how each forager decides the profitability of her food source can
be broken down into three parts: (1) what variables affect gain? (2) what variables affect
cost? (3) what combination of gain and cost, or currency, do bees use? Gains can be
defined as the gross energy gained from consuming food. In nectar-foraging honeybees

this is a function of the amount and concentration of sucrose solution imbibed, as is

shown by studies demonstrating a correlation between these variables and dance
variables associated with profitability (von Frisch, 1967; Seeley, 1986; Seeley et al.,
1991; Seeley and Towne, 1992; Seeley, 1994; Seeley, 1997; Seeley etal., 2000;
Waddington, 1982, Waddington, 1985, Waddington and Gottlieb, 1990).

Costs can be defined as the amount of energy expended to obtain and consume the
food. In honeybees cost is influenced by two main factors: the distance of the food from
the hive, and the handling time, or the amount of time it takes a bee to remove the nectar
from the flower and to fly between flowers. Several studies have shown that bees judge a
food source that is closer to the hive as being more profitable (von Frisch, 1967; Seeley,
1986; Seeley and Levine, 1987; Seeley et al., 1991; Seeley, 1994; Seeley, 1995). A
decrease in handling time, induced by an increase in flow rate of the sucrose solution, has
been shown to lead to an increase in dance duration (a variable strongly correlated with
number of waggle runs) and probability of dancing (Farina, 1996). Other components of
handling time, such as distance between flowers, have also been shown to affect dance
response (Waddington, 1982).

Once honeybees have assessed the components of gain and cost, they must
combine them into a measure of profitability or currency which maximizes honeybee
fitness (Stephens and Krebs, 1986; Cheverton et al., 1985; Bateson and Kacelnik, 1998).
Three currencies have been examined as the basis of honeybee foraging decisions
(Kacelnik and Houston, 1984; Seeley, 1994; Schmid-Hempel et al., 1985). The most
basic is net gains (Net), where Net is defined as total gains (G) minus total costs (C) of a
foraging trip. The second possible currency is rate of net gain (Rate), where Rate is

defined as [(G-C)/T], and T stands for the total trip time. The third possible currency is

net efficiency (Efficiency), where Efficiency is defined as [(G-C)/C]. The majority of
studies on this topic agree that efficiency is the currency honeybees maximize (Schmid-
Hempel et al., 1985; Kacelnik et. al., 1986; Houston etal., 1988; Wolf and Schmid-
Hempel, 1990; Seeley, 1994; but see Waddington and Holden, 1979; Waddington and
Heinrich, 1979). However, the possibility that honeybees may use different currencies
under different conditions can not be discounted (Seeley, 1994; Houston et al., 1988;
Fewell et al., 1991).

The goal of my research is to discover the nature of the odometer by which bees
measure the costs associated with foraging flight. Until recently it was reasonable to
assume that honeybees measured the energy expenditure during flight between the hive
and the food by measuring the calories consumed (for review see Esch & Burns, 1996).
This assumption followed from von Frisch’s (1967) suggestion that bees record flight
distance for the dance by monitoring energy consumption during flight. However more
recent evidence has shown that bees use optic flow to measure the distance reported in
the waggle dance (Esch & Burns 1995, Esch & Burns 1996, Srinivasan et al., 1996;
Srinivasan etal., 1997; Srinivasan et al., 1998; Srinivasan et al., 2000; Esch et al., 2001).
So, are honeybees evaluating the energetic cost of flight by using an optic odometer or a
metabolic odometer?

The optic odometer functions by using optic flow. Optic flow can be described
as movement of visual contours over the visual field. The movement may result from
movement of objects relative to a stationary animal, or by movement of the animal
through the environment. Much of the work done to prove the optic flow hypothesis was

been carried out in tunnels, into which bees are trained to find food (Srinivasan et al.,

1996; Srinivasan et al., 1997; Srinivasan et al., 1998; Srinivasan etal., 2000; Esch et al.,
2001). By changing the width, height, and wall and floor patterns in the tunnels it is easy
to manipulate a honeybee’s perception of optic flow in order to make her estimate that
she has traveled much farther than the actual flight distance in the tunnel. It has also been
verified that recruits watching the dances can’t tell that the dancer had been in a tunnel
and would search for food in the field at the distance that was optically simulated in the
tunnel (Esch etal., 2001).

To address the question of whether bees use an optic or metabolic odometer in
their profitability assessment I constructed two tunnels equal in length but different in the
optical pattern presented on the walls and floor. The optical pattern of one tunnel
simulated 300 m of distance, and the other tunnel’s optical pattern simulated 100 m of
distance even though both tunnels were only 10.2 m long. I trained separate groups of
bees to find food at the ends of these two tunnels and recorded their dances while keeping
the sugar concentration equal in both tunnels. With the sugar concentration and lengths
of the tunnels equal the metabolic odometer’s measure of profitability should have also
been equal. On the other hand, if bees were using an optic odometer they should have

judged the optically shorter tunnel to be more profitable.

METHODS

All of the experiments included in this study were conducted at the Inland Lakes
Research and Study Center on the campus of Michigan State University. The
experiments were conducted within a section of a hoop-house measuring approximately
24.4m long, 5.7m wide at the ground, and 2.4m high in the middle. The hoop-house
consists of semicircular metal hoops covered by a grass-green 30% shade cloth. The
hoop-house ends were covered by wooden walls allowing researchers access by doors.

For honeybees to adapt to life in an enclosed space, it is necessary that they have
no outside flight experience. Therefore, at the start of the study, brood were taken from
hives containing Italian-derived bees (Apis mellifera ligustica) and raised in an incubator
until they emerged. Upon emergence the honeybees were placed in a four-frame
observation hive with a queen and extra brood and pollen from other colonies. The
observation hive was placed in the middle of the hoop-house against the East side, so that
the entrance faced West. Both sides of the observation hive consisted of wooden doors
opening onto glass panes to allow easy viewing and recording of honeybee activity. The
entrance to the observation hive was fitted with a shunt that forced all of the honeybees to
enter onto one side of the comb, so that all dances could be observed through one
window.

Throughout the experiments we attempted to keep the observation hive condition
as constant as possible. Specifically, colony population, stored honey, and stored pollen
were regularly visually checked to see if adjustment was required. Colony population
was adjusted through the addition of honeybees raised in the incubator, or the addition of

a frame of brood. Stored honey was removed if excess, and the colony was fed with a

stock feeder in the hoop-house if their supply was inadequate. Pollen was added through
the addition of pollen frames and via a pollen feeder in the hoop-house. The pollen
feeder consisted of a petri dish filled with a pollen and sugar water mixture, which
foragers ingested and carried back to the hive. All of these adjustments to colony state
were carried out on days prior to experiments.

During the experiments honeybees were trained to reach food by flying down one
of two tunnels. The walls and floor of the tunnels were made of wood and lined with an
optical pattern. The tunnel ceiling was covered with black nylon mesh which allowed a
view of the sky, but prevented bees from escaping. Each tunnel was 10.2 m long and
consisted of 3 tunnel sections 3.4 m long each. The tunnels had an inside height of 20 cm
and a width of 11 cm.

The two tunnels differed in the optical pattern presented on the walls and floor.
The optical patterns were chosen to provide the honeybees with differing impressions of
how far they had flown to reach the food. In the ‘julesz’ tunnel a random J ulesz pattern
of 1 cm2 black and white squares covered the entire length. According to previous
studies in tunnels of the same width and height, this should optically simulate 300 m of
flight distance (Srinivasan et al., 2000). In the ‘horizontal’ tunnel, the same Julesz
pattern covered its first 3.4 m, but the remaining 6.8 m were covered with a pattern of
alternating 1cm black and white horizontal stripes oriented axially along the length of the
tunnel (Figure 1, p. 9). Horizontal stripes provide little or no optic flow (Srinivasan et al.,
1997; Srinivasan et al., 1998). Therefore, although it was the same actual length as the
julesz tunnel, the horizontal tunnel optically simulated only 100 m of distance (100 m

from the julesz section and 0 m from horizontal stripe section).

There were two reasons for providing a julesz pattern in the first segment of the
horizontal tunnel. First, if bees measure their distance from the hive to be short, as would
occur in a tunnel containing only horiaontal stripes, they perform round dances instead of
waggle dances (von Frisch, 1967). Covering the first part of the horizontal tunnel with a
Julesz pattern ensured that the bees would perform waggle dances, although ones that
were to a shorter distance than those for the julesz tunnel. Second, we wanted the initial
portion of both tunnels to look similar so that during training we were not preferentially
recruiting different bees to different tunnels based on the appearance of the entrance.

The tunnels were aligned in opposite directions along the North-South axis of the
hoop-house (see Figure 2, p. 10). The tunnels ran lengthwise under the center-line of the
hoop-house in order to prevent honeybees from using different landmarks outside of the
tunnels that might have influenced distance perception. The orientations of the two
tunnels were reversed every time new bees were trained to avoid possible direction
biases. The tunnel entrances were 1.5 m apart from each other. The hive was placed
equidistant from the two tunnel entrances and slightly back towards the East edge of the
hoop-house. Tunnel entrances were marked with cue cards to facilitate the bees finding
them. A blue cue card was always used for the julesz tunnel and a yellow one was
always used for the horizontal tunnel. Feeders used in the tunnels consisted of two
painted vial caps covered with a wire mesh screen and placed upon a plastic spill tray that
had a colored tape cross (see Figure 3, p. 11). Again, blue was used for the julesz tunnel,
and yellow for horizontal tunnel. The tunnels were set on sawhorses 0.8 m above the

ground.

I... , .33

VH.III.,_.,_.I,..LL.._...H_.{1

. , _ .. 1...... .H
.t’t‘., . .1 . U .51.”
“fidflafififiitvrqfivfiafl.efi.. .

new

thatnitbemeewatsnax. nevus.

. .p
a .3 ,5 ., .5

iaiesfiviaarfisqweaéa

~.¢.§.¢d‘1’,5$.®0§1fike16.Q\ xii:

Git 3k¢‘.¥....t# M$fi¥fit 1.. 2.? «Wane :1.» .vunmm.h.m..

Q...

 

Figure 1 — Tunnel vision. This figure shows the view from the start to the end of the

horizontal tunnel including the cue card on the front of the tunnel.

I 5,
\‘h'
.

' II 9 ‘ ,. .
,1. iii" 135‘?“ . .

      
   

A‘.‘ ‘.__.c,_._._-1__ -

Figure 2 - Hoop-House setup. This is a complete view of the section of the hoop—house
used for these experiments showing the location of the hive and both tunnels.

10

 

Figure 3 — Tunnel feeder design. This shows one of the tunnel feeders consisting of two
vial caps, which were painted, with wire mesh inside them sitting upon a plastic spill tray
with a cross of colored tape. This picture was taken from above the tunnel so you can
also see the mesh that covers the tunnel to prevent bee escape.

All experiments followed the same general protocol. On Day 1 the tunnels were
set up by approximately 9:30 am, and separate groups of bees were trained in steps to
find food at the end of each tunnel. Once the bees were flying back and forth smoothly,
by around 1 pm, they were captured in vials and anaesthetized by chilling them in ice. A
50/50 mixture of rosin and beeswax was melted onto their abdomen to cover their
N asanov gland, a process known as scent-plugging (Towne and Gould, 1988). The
N asanov gland produces a pheromone that honeybees release at a food source they find
profitable (von Frisch, 1967). Scaling the gland eliminates the possibility that a bee’s
measure of profitability could be influenced by the amount of Nasanov pheromone in the
air surrounding the food. After scent-plugging the bees’ thorax and abdomen were
marked using unique patterns of colored paint dots (Testor’s Gloss Enamel). Marked
bees were then released onto the feeder tray at the end of the tunnel and allowed to gain
additional experience in the tunnels until approximately 5 pm. Excess bees were
captured.

At the start of each experiment we marked approximately 21 bees in the
horizontal tunnel and 28 bees in the julesz tunnel. We used a smaller number of bees in
the horizontal tunnel because they made faster round trips. Having an overall lower
number of bees in the horizontal tunnel made the average number of bees at each feeder
at any one time (5-10 bees) approximately equal in both tunnels. This was necessary for
two reasons. First, each observer at the end of the tunnel could only keep track of a
certain number of bees feeding simultaneously, and this limit was equal for both
observers. Second, it is possible that bees use the number of other foragers at a food

source as an indicator of profitability.

12

On Day 2 the tunnels were set up in the morning with the first sugar solution we
planned on testing that day. The bees were allowed time to find the food (15 minutes to 4
hours, depending on the air temperature). One assistant was stationed at the end of each
tunnel with a tape recorder and stopwatch. The assistants would record the exact time to
the second of each bee’s arrival and departure from the feeder. They were also
responsible for maintaining abundant sugar solution in the feeder. I remained at the hive
using a video camcorder (Canon E51) to record the arrivals and departures of the painted
bees. A digital clock and thermometer were placed in view of the camera under the
dance floor.

Once all observers were in place and there were at least 5 bees attending each
feeder we would record flight times and dances for approximately 1 hour. Each
recording session at a constant sucrose concentration is called a test period. Between test
periods all observers would take a small break and choose another sucrose solution.
Sucrose solutions were chosen from a set of sucrose solutions ranging from 0.50 mol/l to
1.00 mol/l in 0.25 mol/l increments. Occasionally higher or lower sugar concentrations
were added to the set if the bees were particularly unmotivated or overly excited. This
was necessary to avoid ceiling or floor effects in the bees’ response to the two tunnels
(Waddington & Gottlieb 1990). An assistant would haphazardly choose a solution and
the previous solution was removed from the set of possible solutions to use during the
remaining test periods that day. We continued recording test periods until either the day
was over (approximately 5 pm) or the weather deteriorated. Overall we trained three
groups of bees using this protocol (Bee Groups 1-3 in Table 1, p. 14), and recorded two

to four test periods with each, which came to a total of nine test periods (Table 1, p. 14).

13

 

Table l - Test period conditions. This table shows the date, time, sugar concentration and

bee group used for each test period.

 

 

Test Period Date Time Sugar (mol/l) Bee Group
A 10-Aug-00 13:59 - 14:30 1.00 1
B 10-Aug-00 15:04 - 15:31 0.75 1
C 10-Aug-00 16:13 - 16:40 0.50 1
D l4-Aug-00 11:57 - 12:27 0.50 2
E l4-Aug-00 13:10 - 14:01 0.75 2
F 14-Aug-00 14:38 - 15:06 1.00 2
G 14-Aug-00 15:42 - 16:16 1.25 2
H 1-Sep-00 13:52 - 14:35 0.75 3
I l-Sep-OO 15:10 — 16:30 0.50 3

 

By making the actual lengths of the two tunnels identical and offering identical

sucrose solutions, we hoped to make the bees’ metabolic measure of profitability equal.

In order to check this we needed six variables: the flight time from the hive to the food,

the time at the food, the flight time from the food to the hive, the time at the hive upon

returning from foraging, unloaded bee mass, and stomach volume. We collected data on

the four time segments during our initial experiments, but we performed a second

experiment to obtain an estimate of stomach volume and unloaded mass for bees from the

observation hive. For the second experiment we trained bees as in Day 1 of the previous

protocol but without scent-plugging or marking them. At the end of Day 1 or on Day 2

we measured stomach volume and body mass of bees in each tunnel at three levels of

14

sugar concentration, 0.5, 1.5, and 2.25mol/l. Bees were captured in vials immediately
after they terminated imbibing sugar solution at the feeder. I would remove her head and
draw her stomach contents from the esophagus into a capillary tube (Mettler P-100) as
quickly as possible to avoid food consumption. The amount of solution in the capillary
tube was measured in mm and converted into ul. The bee’s body and head were weighed
to the nearest 5 mg on a torsion balance.

Foraging trip time information, probability of dancing (0 if no waggle runs, 1 if
any waggle runs), and number of waggle runs performed were extracted from the audio
and video tapes and entered into an Excel worksheet. The audio tapes contributed
information on when each bee arrived and departed the feeder. The video tapes recorded
the time each bee left the hive to go to the feeder, the time at which each bee returned to
the hive from the feeder, how many waggle runs each bee completed after a trip to the
feeder, and when each bee left the hive after dancing. If a bee could not be seen during
its entire time in the hive after a foraging trip it was not included. This prevented
underestimating number of waggle runs due to runs performed outside the view of the
camera. Only data on the bees’ activities after the sucrose concentration had been
constant for 30 minutes were used for analyses. This was done to account for the
possibility that bees take time to adjust their dance response to a new sucrose solution
(Richter and Waddington, 1993; von Frisch, 1967).

To test the assumption that metabolically measured profitability was equal
between the two tunnels energetic gain and cost were calculated. The amount of energy
gained in I per foraging trip, gain (G), was calculated as

G = Gain (J) = SV (11]) 0 SC (umol/ul) 0 5.8 (J/umol)

15

where SV is stomach volume in ill and SC is sucrose concentration in umol/ul. In this

equation 5.8 J/umol is the standard conversion for the energy content of sugar (Kleiber,
1961).
The metabolic measure of cost (C) was calculated using the equation

4
C = Cost (J) = 2;;1‘, (s) 0 MR; (J/s)

where T, is the time in s of each of the four segments of a foraging trip, and MR1 is the
metabolic rate in J/s for each of the four segments of a foraging trip. The numbering of
the four segments of a foraging trip is the same for both the time and the metabolic rate:
1) the flight from the hive to the food source, 2) foraging at the food source, 3) the flight
from the food source to the hive, and 4) visit at the hive upon returning from the food
source. To determine the metabolic rate of each of the foraging trip sections I used an
allometric equation for forward flight at 0.5 m/s for segments 1 and 3 and an allometric
equation for walking or sitting bees for segments 2 and 4 (Wolf et al., 1989). I converted
these equations to Us by using a respiratory quotient of 1 for bees, therefore 1 ml 02 =
21.117 J/ml (Nachtigall etal., 1989; Rothe and Nachtigall, 1989). The equations I used
for each segment of the foraging trip are as follows:

MR1 (J/s) = 0.003015 . UM (mg) ”-629

MR2 (NS) = 0.0026044 . UM (mg) ““92

MR3 (J/s) = 0.003015 . LM (mg) 0629

MR4 (J/s) = 0.0026044 . LM (mg) “492
where UM is the unloaded mass of the bee in mg, and LM is the mass of the bee plus the

weight of its stomach contents in mg. To convert stomach volume in ill to the weight of

16

the stomach contents in mg, I used the standard densities of different concentrations of
sucrose solution in the CRC Handbook of Chemistry and Physics (Hodgman et al., 1958).
I decided to use loaded mass for the last two segments of the foraging trip even though
crop load does not directly affect metabolic rate (Balderrama et al., 1992; Moffatt, 2000)
because flow rate does affect metabolic rate (Moffatt, 2000), and the feeders used in these
experiments dispensed food ad libitum.

There are many other factors that could affect metabolic rate, but which do not
affect the present study. Flight speed is known to affect metabolic rate, but not over the
range of 0 to 4.3 m/s, which covers the flight speeds in my experiments (Table 2, p. 18)
(Nachtigall et al., 1989). As mentioned above flow rate affects metabolic rate, but both
tunnel feeders dispensed sugar solution ad libitum (Balderrama et al., 1992). Bees have
also been found to increase their thoracic temperature with increasing sugar concentration
(Waddington, 1990). Yet, in this study both tunnel feeders offered the same sugar
concentration at the same time, so if this made the estimates of energy spent inaccurate it
would have done so equally for both tunnels. Weight —Specific heat production has been
shown to vary with age of the worker bees (Fahrenholz et al., 1992), but there is no
reason to suspect that the foragers in one tunnel were systematically older than in the
other tunnel on all three days testing took place. Malate dehydrogenase phenotype has
been shown to affect flight metabolic rates (Harrison et al., 1996), but there is no reason
to think that the bees in any one tunnel would have a different phenotype. These
possibilities are one of the reasons we had the tunnel entrances designed to look identical

at the start of training. This should have limited any possible bee preference.

17

 

Table 2 — Flight speeds. This table shows the average flight speeds and the minimum and
maximum values of flight speed for the two tunnels. Average flight speeds were
calculated by dividing the distance of the entire flight from the hive entrance (12.1m) by
the average times calculated in Table 3, so values of N are the same as in Table 3.

Horizontal Tunnel J ulesz Tunnel

Average Min. and Max. Average Min. and Max.

 

Flight Speed to the 0.20 0.02 - 0.81 0.07 0.02 — 1.10
Food (m/s)
Flight Speed from the
Food (m/s) 0.16 0.003 - 0.48 0.12 0,01 _ 0,53

 

18

RESULTS

Two lines of evidence suggest that the bees were interpreting the different optical
patterns in the two tunnels as intended. First, I observed slower flight times in the
optically longer julesz tunnel (Table 3, p. 20). This is consistent with previous studies
showing that bees prefer to keep a constant rate of optic flow when they are flying, and
will slow down or speed up to attain that preferred rate of optic flow (Srinivasan et al.,
1996; Esch and Burns, 1996). The flight speed differences that I observed indicate that
honeybees are noticing the difference in optic patterns. Second, I observed that bees
imbibed more sugar solution in the optically longer julesz tunnel (Table 3). Previous
studies suggest that honeybees tend to imbibe greater amounts of sugar solution with
increasing distance (Nunez, 1982). Thus, my data suggest that bees interpreted the julesz
tunnel as longer. To ensure that these differences were consistent across changes in sugar
concentration I carried out this comparison at all sugar levels used in the experiments,
and found the same general result (Tables 4, p. 21). The one exception to the trend was
that at 1.00 mol/l bees flew from the food to the hive significantly faster in the optically
longer julesz tunnel. Yet, the overall pattern of flight times and stomach volumes in
Tables 4 indicates that bees were interpreting the tunnels as intended; so, further
confirmation by investigating the distance signal of the waggle dance was unnecessary.

Having confirmed the tunnels are in fact interpreted differently by the bees, the
next step was to calculate the predictions of the metabolic odometer hypothesis. To
complete this, I first had to use the equations described in the methods to determine
values for gain and cost, and to do that I needed to determine correct estimates of

stomach volume and unloaded mass for bees from the observation hive.

19

 

Table 3 - Tunnel flight time and stomach volume differences. This table shows average
flight times and stomach volumes for the optically shorter horizontal tunnel and the
optically longer julesz tunnel. Flight times were calculated by averaging each bee over
all of her foraging trips and then averaging all the bees in each tunnel. Stomach volume
was measured on a different group of bees that only made one foraging trip each.
Shapiro-Wilkes tests revealed that all of the data were non-normal, so Wilcoxon rank
sums tests were used to compare tunnels.

Horizontal Tunnel J ulesz Tunnel

MeaniSEM N Mean-:SEM N P-Value

 

Flight Time from the Hive to the 60.3 i 6.1 40 182.1 1 31.1 49 < 0.0001
Tunnel Feeder (5)

Flight Time from the Tunnel

Feeder to the Hive (s) 75.8 i 14.2 40 97.63: 11.4 50 < 0.0001

Stomach Volume (pl) 39.4 i 2.8 34 50.8 t 2.6 22 < 0.01

 

20

 

Table 4 — Tunnel flight time and stomach volume differences at different levels of sugar
concentration. Flight times were obtained by averaging each bee over all of her foraging
trips at a particular sugar concentration and then by averaging all of those bee averages in
each tunnel. Stomach volume data taken on bees that made one foraging trip to a feeder
of 0.50mol/l solution were averaged. Shapiro—Wilkes tests revealed that all data were not
normally distributed, except for stomach volume at 0.5 mol/l. Wilcoxon rank sums tests
were performed to compare flight times in the two tunnels. Stomach volume was

analyzed using a one-way ANOVA.

 

 

 

Sugar Horizontal Tunnel Julesz Tunnel
Concentation
(mol/l) Mean 1 SEM N Mean :1: SEM N P-Value
0.50 60.9 i 16.4 29 152.7 i 24.2 28 < 0.0001
0.75 62.4 i 7.5 37 139.2 3; 26.3 35 < 0.001
Foodward Flight
(5) 1.00 71.5 :I: 10.0 25 183.8 i 44.9 31 < 0.0001
1.25 67.7 i 6.0 6 195.0 i 40.0 11 < 0.007
0.50 64.4 i 9.1 33 87.4 i 19.0 30 < 0.02
0.75 65.2 i 10.5 38 87.9 x 12.1 35 0.002
Homeward Flight <
(S) 1.00 92.9 i 31.7 27 88.0 i 9.8 35 < 0.0003
1.25 62.0 i- 19.2 7 1170 i 22.6 13 < 0.02
Storm“ V°lume 0.50 34.7 i 3.8 15 50.9 i 3.5 6 < 0.03

(#1)

21

To choose an appropriate estimate of stomach volume, I needed to determine
whether the factors of tunnel and sugar concentration affected stomach volume. A
Shapiro-Wilkes test for normality revealed that stomach volume was not normally
distributed, so I used a Kruskal-Wallis rank sums tests to investigate the independent
effects of sugar concentration and tunnel on stomach volume. These tests revealed that
tunnel did significantly affect stomach volume (p < 0.01), but sugar concentration did not
(p > 0.07) (Figure 4, p. 23). Previous studies have shown an increase in stomach volume
with sugar concentration (Waddington, 1990); the fact that it was not quite significant in
these experiments is probably due to small sample size in some of the sugar
concentrations (0.50 mol/l N: 21, 1.50 mol/l N: 14, 2.25 mol/l N=21). Since I did not
have estimates of stomach volume at every level of sugar concentration and the effect of
sugar concentration was not significant, I used each tunnel’s average of stomach volume
for the energetic calculations of gain and cost.

To determine the appropriate estimate of unloaded mass for bees from the
observation hive I needed to determine the effect of tunnel and sugar concentration. The
data on unloaded mass were normally distributed (Shapiro-Wilkes probability < W =
0.1183), so I performed a two-way AN OVA to test for the effect of tunnel, sugar and
their interaction on unloaded mass (Table 5, p. 24). There was no significant effect of
tunnel or of the interaction between tunnel and sugar concentration, but there was a
significant effect of sugar (Figure 5, p. 25). The cause of this significant effect of sugar
could be due to an interaction effect of bee mass and feeder sugar concentration, so that

bees of greater mass decided to stay in the hive and stop foraging after discovering that

22

22

on
8:

34

H-i

Jim
:3?

Stomach Volume (111)
N r.»
a 8?

 

 

 

 

 

H»
‘3‘?

Horizontal Julesz

Figure 4 — Effect of tunnel on stomach volume. This figure shows the mean, SEM, and N
of stomach volume in each tunnel.

23

the feeder was too low in concentration. Regardless of the cause, the best estimate of
unloaded mass, given this effect, would be a separate estimate of weight for each of the
sugar concentrations used. However, I did not have measurements of weight at all of the
sugar concentration levels used; so, I decided that the least biased estimate of unloaded
mass for bees from the observation hive was an overall average of unloaded mass.

This should not have biased the results for two reasons. First, all of my
conclusions are drawn from comparisons of the two tunnels, and since there was no
significant interaction between sugar concentration and tunnel, using an overall estimate
of unloaded mass instead of a different value for each sugar concentration, would affect
both tunnels equally. Second, the effect of sugar concentration is likely to be less in my
experiments because the range of sugar concentrations over which this effect was
discovered (0.5 to 2.25mol/l) was greater than the range of sugar concentrations used in
the experiments (0.5 to 1.25molll). Therefore the overall estimate of unloaded mass is

the best available estimate of unloaded mass and was used in the energetic equations.

 

Table 5 — Effect of tunnel and sugar concentration on unloaded mass. This table shows
the results of the two-way AN OVA unloaded mass (mg) = tunnel + sugar concentration +
tunnel * sugar concentration.

 

Degrees of

 

Source Sum of Squares F Ratio Prob > F
Freedom
Tunnel 1 1.8268 0.0166 > 0.9
sugar . 1 1233.1215 11.23666 < 0.002
Concentration
3|:
Tunnel sugar 1 4.2701 0.0389 > 0.8

Concentration

 

24

18

100— 14
90- 10 '
Mr 80-
701
60‘
50-
40-
30-
20-
10-
0-

mg)

I-H

 

 

Unloaded Mass

 

/

0.50 1.50 2.25

Sugar Concentration (mol/l)

Figure 5 — Effect sugar concentration on unloaded mass. This graph shows the mean,
SEM, and N of unloaded mass at each of the three sugar concentrations tested.

25

Once the estimates of stomach volume and unloaded mass were chosen, I used
them to estimate metabolic Efficiency, as that is the currency bees most likely use to
measure profitability (Schmid-Hempel et al., 1985; Kacelnik et al., 1986; Houston et al.,
1988; Wolf and Schmid-Hempel, 1990; Seeley, 1994). To determine whether estimated
metabolic Efficiency was equal between the two tunnels, I generated average values of
Efficiency for each tunnel in each test period and performed a paired t-test (Figure 6, p.
27). Although the value was not significant, it was very near significance in the direction
indicating the horizontal tunnel would be more profitable. Yet, there was no statistical
difference between the two tunnels, so bees should judge the two tunnels about equal in
profitability if bees use a metabolic odometer to estimate Efficiency.

Even though Efficiency is the most likely currency used by bees, I also examined
what outcomes the metabolic odometer hypothesis would predict if bees were using the
two other currencies. I calculated estimates of metabolic Net and metabolic Rate,
analogous to the way described for Efficiency, and performed paired t-tests to compare te
estimates for the two tunnels (Figure 7, p. 28; Figure 8, p. 29). For the currency Net, the
values for the horizontal tunnel were significantly less than for the julesz tunnel. Thus, if
bees were using a metabolic odometer to estimate this currency they should judge
profitability to be higher in the julesz tunnel. For the currency Rate there was no
significant difference between the horizontal and the julesz values. Thus, if bees were
using a metabolic odometer to measure Rate they should have judged the profitability of
the two tunnels to be equal. Table 6 (p. 30) summarizes these predictions of the

metabolic odometer hypothesis and the predictions of the optic odometer hypothesis.

26

 

Horizontal
[j Julesz

c»
O
o.

 

 

N
LII

 

N
O

r—n
O

Efficiency (J/J)
u. 5’.

 

A B C D E F G H I
1.00 0.75 0.50 0.50 0.75 1.00 1.25 0.75 0.50
1 1 1 2 2 2 2 3 3
Test Period
Sugar Concentration (mol/l)
Bee Group

Figure 6 - Effect of tunnel within each test period on estimated metabolic Efficiency.
This figure shows the average, SEM, and N of Efficiency in each tunnel in each test
period. Average values were obtained by first averaging each bee and then averaging the
bee averages. A Shapiro-Wilkes test for normality on the differences between the two
tunnels generated a probability < W of 0.6427 indicating the distribution was normal. A
paired t-test generated a two-sided p > 0.05 indicating there is no significant difference in
Efficiency between the two tunnels.

27

 

[3 Horizontal 11

3501
300 16 DJulesz 14 s.

250‘ 1614 16 9 5
2001 16 10 9
150— 8 17 3
100- 1I 5'l 8
0

A B C D E F G H I
1.00 0.75 0.50 0.50 0.75 1.00 1.25 0.75 0.50
1 1 1 2 2 2 2 3 3
Test Period
Sugar Concentration (mol/l)
Bee Group

 

 

 

Net (J)

 

 

 

Figure 7 —— Effect of tunnel within each test period on Net. This figure shows the average,
SEM, and N of Net in each tunnel in each test period. Average values were obtained by
first averaging each bee and then averaging the bee averages. A Shapiro-Wilkes test for
normality on the difference between the two tunnels generated a probability < W of
0.7517 indicating the distribution was normal. A paired t-test generated two-sided p <
0.0001, indicating Net is greater in the optically longer julesz tunnel.

28

 

1.2 1 E] Horizontal 6

El Julesz a 11
1‘ 16 9 14

0.8— 14
10
- 16
0'6 s 9 s 3
0 4_ 12 17
0.2 — I I
0 _ ,

A B C D E F G H I
1.00 0.75 0.50 0.50 0.75 1.00 1.25 0.75 0.50
1 1 1 2 2 2 2 3 3
Test Period
Sugar Concentration (mol/l)
Bee Group

Figure 8 — Effect of tunnel within each test period on Rate. This figure shows the
average, SEM, and N of Rate in each tunnel in each test period. Average values were
obtained by first averaging each bee and then averaging the bee averages. A Shapiro-
Wilkes test for normality on the differences between the two tunnels generated a
probability < W of 0.5739, indicating the distribution was normal. A paired t-test
generated a two—sided p > 0.85, indicating there is no significant difference in Rate
between the two tunnels.

 

 

 

Rate (J/s)

 

 

 

29

 

Table 6 — Predicted relative dance response of the horizontal (H) and julesz (J) tunnels
under the metabolic and optic odometer hypotheses.

 

 

Hypothesis Currency Predicted relative dance response
. Both tunnels should be judged equal in
Metabolic . . . . . .
Odometer Efficrency profitability, With a non-s1gn1ficant trend of greater H = J
profitability in the optically shorter horizontal
tunnel (Figure 6, P. 29).
Metabolic Net The optically longer julesz tunnel should be H < J
Odometer judged greater in profitability (Figure 7, p. 30).
1...... 3332;123:1881: :6 “a: m H -.
Odometer p y g ’ p. ' —
Optic The optically shorter horizontal tunnel should be H > J
Odometer N/A judged greater in profitability (Introduction, p. 5).

 

Having clarified the predictions of the metabolic and optic odometer hypotheses, I

could determine how honeybees actually responded to the two tunnels. To capture each

bee’s measure of profitability I assessed two dance variables: number of waggle runs and

probability of dancing (0 if no dances performed, 1 if any dances performed). I then

assessed an average value of these variables for each bee and averaged the bee averages

to obtain a value for each tunnel in each test period (Figure 9, p. 32; Figure 10, p. 33).

Paired value tests revealed that both probability of dancing and number of waggle runs

performed were significantly greater in the optically shorter horizontal tunnel.

30

 

E] Horizontal

E! Julesz

 

 

 

 

 

 

Probability of Dancing
O
"F‘

A B C D E F G H I
1.00 0.75 0.50 0.50 0.75 1.00 1.25 0.75 0.50
1 1 1 2 2 2 2 3 3
Test Period
Sugar Concentration (moi/l)
Bee Group

Figure 9 — Effect of tunnel within test period on observed probability of dancing. This
figure shows the average, SEM, and N of observed probability of dancing in each tunnel
in each test period. Average values were obtained by first averaging each bee and then
averaging the bee averages. A Shapiro-Wilkes test for normality on the arcsin
transformed differences between the two tunnel revealed a probability < W of .4051,
indicating the distribution was normal. A paired t-test on the arrcsin transformed data
generated a two-sided p < 0.05, indicating observed probability of dancing was
significantly greater in the optically shorter horizontal tunnel.

31

 

I Horizoan

 

 

 

 

 

30- D Julesz
16
25- 18 1914
8 20 11 9 7
5 _ .
m 20 ,
O
_. 15- 13
§ 9 11
_ 17 15
3 10 . ’ 16 , s
i 5‘  ‘ I ‘ H i ’
A
Test Period
Sugar Concentration (mol/l)
Bee Group

Figure 10 — Effect of tunnel within test period on number of waggle runs performed.
This figure shows the average, SEM, and N of number of waggle runs performed in each
tunnel in each test period. Average values were obtained by first averaging each bee and
the averaging the bee averages. A Shapiro-Wilkes test for normality on the differences
between the two tunnels generated a revealed a probability < W of 0.0181, indicating the
distribution was non-normal. A Wilcoxon signed-rank test for paired values generated a
two-sided p < 0.005, indicating the number of waggle runs was significantly greater in
the optically shorter horizontal tunnel.

32

These results contradict the predictions of the metabolic odometer hypothesis
(Table 6, p. 30). However, Efficiency was very near significantly predicting the optically
shorter horizontal tunnel should be more profitable (Figure 6, p. 27), which would mean
metabolic Efficiency and the optic odometer both predict the observed results. To clarify
this situation, I took advantage of the linear relationship between number of waggle runs
and Efficiency to test if the observed pattern of waggle runs is predicted by the metabolic
odometer (Seeley, 1994). If bees were using metabolic Efficiency to determine the
number of waggle runs performed, the difference between Efficiency measurements in
each tunnel should equal the difference between number of waggle runs performed in
each tunnel. To test this I performed a Wilcoxon signed rank test for paired values on the
horizontal to julesz tunnel ratios of number of waggle runs and metabolic Efficiency in
each test period (Figure 11, p. 34). The results show that the observed profitability
difference between the two tunnels, as estimated by number of waggle runs, is greater

than can be accounted for by metabolic Efficiency.

33

 

E Horizontal to julesz ratio of # of waggle runs
[:1 Horizontal to julesz ratio of Efficiency

 

 

 

Horizontal to Julesz Ratio

 

A B C D E F G H I
1.00 0.75 0.50 0.50 0.75 1.00 1.25 0.75 0.50
1 1 1 2 2 2 2 3 3
Test Period
Sugar Concentration (mol/l)
Bee Group

Figure 11 - Comparison of horizontal to julesz tunnel ratio of number of waggle runs
performed to estimated metabolic Efficiency. Values were obtained by dividing the
average horizontal tunnel value for each test period by the average julesz tunnel value
(given in Tables 6 and 10 for estimated metabolic Efficiency and number of waggle runs
performed respectively). Since ratios are non-normal, a Wilcoxon signed rank test for
paired values was performed and generated a p < 0.03.

34

DISCUSSION

Research in foraging ecology has shown that a diverse array of species are
sensitive to differences in food source profitability. Many studies have strived to
discover which specific measure of profitability, or currency, an animal may use to make
foraging decisions (for review see Stephens and Krebs, 1986). Unfortunately, the
mechanisms by which animals perceive these currencies have remained largely
inaccessible. My study attempted to determine which mechanism honeybees use to
measure the cost associated with flight costs associated with of profitability.

Until recently, honeybees were believed to use a metabolic odometer to assess the
flight distance reported in the waggle dance (von Frish, 1967). Yet, it has now been
shown that honeybees use an optic odometer to measure the distance to a food source (for
review see Esch and Burns, 1996). To determine whether bees use this optic odometer as
part of their calculations for energetic profitability I constructed two tunnels of equal
actual length, but differing in optically simulated length. I then observed bees’ measure
of profitability, indicated in their dance response, while they foraged at feeders of equal
concentration in the two tunnels. The results show that bees measured the optically
shorter tunnel to be more profitable. This result is consistent with the optic odometer
hypothesis, and cannot be accounted for by the hypothesis that bees base their foraging
decisions on a metabolic odometer. Thus, although we still cannot exclude the possibility
that bees monitor metabolic expenditures in some circumstances, my results taken
together with those of Esch and Burns (1996) and Srinivasan et a1. (1996, 2000) suggest
that there is no basis for assuming that bees measure flight distance by monitoring energy

expenditure.

35

Note that my data do not exclude the possibility that bees use a time odometer.
This is because their flight time is shorter in the tunnel deemed more profitable.
However, a time based odometer has been excluded as a means of measuring to a food
source (Srinivasan et al., 1996). Thus, such a mechanism seems unlikely to explain my
results.

Not only do my data illuminate the process by which bees regulate recruitment to
a food source, but they also may shed light on other foraging decisions. For examplem it
is common for foraging honeybees to only partially fill their honey stomach (crop). The
reason behind this has been a puzzle for several decades (Nunez, 1982; Kacelnik et al.,
1986; Varju and Nunez, 1991; Schmid-Hempel; 1993; Schmid-Hempel et a1, 1985;
Moffatt, 2000). There are two main hypotheses as to why honeybees only partially fill
their crops: informational exchange and maximization of Efficiency. The informational
exchange hypothesis is that honeybees leave a food source with a partially filled crop in
order to spread information about the food source to other bees in the hive (Nunez, 1982;
Varju and Nunez, 1991; Moffatt, 2000). The Efficiency hypothesis is that honeybees
partially fill their crops in order to maximize the Efficiency of the foraging trip (Kacelnik
et al., 1986; Schmid-Hempel, 1993; Schmid-Hempel et al., 1995).

One of the arguments that has been used both to support (Wolf etal., 1989) and to
cast doubt (Moffatt, 2000) on the maximizing Efficiency hypothesis is now made
questionable by some of the findings in my study. The first premise of the argument is
that in order for Efficiency [(G-C)/C] to predict variable crop loads, costs must increase
with increasing crop load. The second premise is that costs are measured directly via

metabolic rate. This argument has been used against the maximizing Efficiency

36

hypothesis in studies that found metabolic rate (and hence cost) does not vary with crop
load (Moffatt, 2000), and it has been used to support the maximizing Efficiency
hypothesis in studies that found metabolic rate to vary with crop load (Wolf et al., 1989).

The results of my study undermine the second premise in this argument. In my
experiments I induced bees to take on significantly different crop loads by exposing them
to different amounts of optic flow on the flight to the feeder, but without a substantial
difference in metabolic expenditure (Tables 3, p. 20, and 4, p. 21). This implies that
honeybees are using optic flow instead of metabolic expenditure to judge the costs. If
metabolic rate does not change with crop load it only implies that honeybees are not
maximizing metabolically measured efficiency. They could still be maximizing their
perceived efficiency if they were adding some aspect of crop load to their optical measure
of costs for the flight home. Studies done with lead weights lend some support to this
idea (Schifferer, 1952 as reported by von Frisch, 1967; Schmid—Hempel, 1986). By
taking into account honeybee’s optic measure of costs, future studies could clarify why
honeybees only partially fill their crops.

The partial crop filling debate clearly shows the value of knowing the
mechanisms underlying adaptive behavior. Without knowledge of these mechanisms one
might be misled about decisions based on sources of information available to the animal.
Furthermore, knowing the mechanism allows more accurate control of a variable during
experiments. For example, to determine how precise honeybee profitability estimates
are, one could decrease measurement error by precisely controlling the optic flow of the

foraging flight. Also, my results suggest that studies of the energetic gains and costs

37

involved in foraging decisions may have to deal with animals that use non-energetic
modalities to measure gains and costs.

The discovery that honeybees’ primary measure of distance related foraging costs
is not a metabolic odometer also raises interesting questions about the neural mechanisms
underlying the bees’ estimation of profitability. My data taken together with previous
work suggest that bees base their responses to food on a combination of optical and
gustatory cues (sugar concentration and flow rate) (Waddington, 1990; Nunez, 1982).
This raises the question of where and how in the brain this input is combined with the
other components to modulate the profitability signal of the waggle dance.

In addition to the result that honeybees use an optic odometer to measure the cost
of flight component of profitability, this study highlights the promise of using tunnels to
answer previously inaccessible questions. Tunnels have previously been used almost
exclusively to answer navigational questions (Srinivasan et al., 1996; Srinivasan et al.,
1997; Srinivasan et al., 1998; Srinivasan et al., 2000; Esch etal., 2001). However,
tunnels hold promise for investigating questions such as those raised in the partial crop
filling debate as well as other areas of honeybee foraging ecology and energetics.
Tunnels also show promise due to their logistical appeal. By simulating long distances
within small enclosed areas, it is much easier to control for variables such as the amount
of available forage, and often unnecessary to wait for days for the correct weather and
flowering conditions as has been done previously (von Frisch, 1967; Seeley, 1995).
Finally, as my results suggest, it may now be more feasible to study the sensory and
integrative mechanisms underlying foraging decisions that are expressed in flight over

many hundreds or thousands of meters.

38

LITERATURE CITED

Balderrama, N. M., Almeida de B., L. O. & Nunez, J. A. 1992 Metabolic rate during
foraging in the honeybee. Journal of Comparative Physiology B 162, 440-447.

Bateson, M. & Kacelnik, A. 1998 Risk-Sensitive foraging: Decision making in variable
environments. In Cognitive Ecology (ed. R. Dukas), pp. 297-341. Chicago: The
University of Chicago Press.

Cheverton, J ., Kacelnik, A. & Krebs, J. R. 1985 Optimal foraging: constraints and
currencies. In Experimental Behavioral Ecology and Sociobiology (ed. B.
Holldobler & M. Lindauer), pp. 109—126. Stuttgart: Gustav Fischer Verlag.

Cuthill, I. & Houston, A. I. 1997 Managing time and energy. In Beahavioural Ecology:
an evolutionary approach (ed. J. R. Krebs & N. B. Davies), pp. 97-120. Oxford:
Blackwell Science Ltd.

de Vries, H. & Biesmeijer, J. C. 1998 Modeling collective foraging by means of
individual behaviour rules in honey-bees. Behavioral Ecology and Sociobiology
44, 109-124.

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

Esch, H. E. & Burns, J. E. 1995 Honeybees use optic flow to measure the distance of a
food source. Naturwissenschaften 82, 38-40.

Esch, H. E. & Burns, J. E. 1996 Distance estimation by foraging honeybees. Journal of
Experimental Biology 199, 155-162.

Esch, H. B, Zhang, S., Srinivasan, M. V. & Tautz, J. 2001 Honeybee dances
communicate distance measured by optic flow. Nature 411, 581-583.

Fahrenholz, 1L., Lamprecht, I. & Schricker, B. 1992 Calorimetric investigations of the

different castes of honey bees, Apis mellifera camica. Journal of Comparative
Physiology B 162, 1 19-130.

Farina, W. M. 1996 Food-exchange by foragers in the hive - a means of communication
among honey bees? Behavioral Ecology and Sociobiology 38, 59-64.

Fewell, J. H., Ydenberg, R. C. & Winston, M. L. 1991 Individual foraging effort as a

function of colony population in the honey bee, Apis mellifera L. Animal
Behaviour 42, 153-155.

39

Harrison, J. F., Nielsen, D. I. & Page, R. E. J. 1996 Malate dehydrogenase phenotype,
temperature and colony effects on flight metabolic rate in the honey-bee, Apis
mellifera. Functional Ecology 10, 81-88.

Hodgman, C. D., Weast, R. C. & Selby, S. M. 1958 CRC Handbook of Chemistry and
Physics. Clevland: Chemical Rubber Publishing Co.

Houston, A., Schmid-Hempel, P. & Kacelnik, A. 1988 Foraging strategy, worker
mortality, and the growth of the colony in social insects. American Naturalist 131,
107-1 14.

Kacelnik, A. & Houston, A. I. 1984 Some effects of energy costs on foraging strategies.
Animal Behaviour 32, 609-614.

Kacelnik, A., Houston, A. I. & Schmid-Hempel, P. 1986 Central-place foraging in honey

bees: The effect of travel time and nectar flow on crop filling. Behavioral Ecology
And Sociobiology 19, 19-24.

Kleiber, M. 1961 The Fire of Life: An Introduction to Animal Energetics. New York:
Wiley.

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 186,
299-306.

Nachtigall, W., Rothe, U., Feller, P. & Jungmann, R. 1989 Flight of the honeybee HI.
Flight metabolic power calculated from gas analysis, thermoregulation and fuel
consumption. Journal of Comparative Physioogy B 158, 729-737.

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.

Richter, M. R. & Waddington, K. D. 1993 Past foraging experience influences honey bee
dance behaviour. Animal Behaviour 46, 123-128.

Rothe, U. & Nachtigall, W. 1989 Flight of the honeybee IV. Respiratory quotients and
metabolic rates during sitting, walking and flying. Journal of Comparative
Physiology B 158, 739-749.

Schmid-Hempel, P. 1986 Do honeybees get tired? The effect of load weight on patch
departure. Animal Behaviour 34, 1243- 1250.

Schmid-Hempel, P. 1993 On optimality, physiology and honeybees - a reply to Varju and
Nunez. Journal of Comparative Physiology A 172, 251-256.

40

 

Schmid-Hempel, P., Kacelnik, A. & Houston, A. I. 1985 Honeybees maximize efficiency
by not filling their crop. Behavioral Ecology And Sociobiology 17, 61-66.

Seeley, T. D. 1989 Social foraging in honey bees: How nectar foragers assess their
colony's 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: Harvard University Press.

Seeley, T. D. 1997 Honey bee colonies are group-level adaptive units. The American
Naturalist 150, S23-S41.

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.

Seeley, T. D. & Levine, R. A. 1987 Social Foraging by honeybees: How a conlony
tracks rich sources of nectar. Neurobiology and Behavior of Honeybees , 38-53.

Seeley, T. D., Mikheyev, A. S. & Pagano, G. J. 2000 Dancing bees tune both duration
and rate of waggle-run production in relation to nectar-source profitability.
Journal of Comparative Physiology A 186, 813-819.

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.

Srinivasan, M. V., Zhang, S. W., Lehrer, M. & Collett, T. S. 1996 Honeybee navigation
en route to the goal: Visual flight control and odometry. Joumal-of—Experimental
Biology 199, 237-244.

Srinivasan, M. V., Zhang, S. W. & Bidwell, N. J. 1997 Visually mediated odometry in
honeybees. Joumal-ofiExperimental-Biology 200, 2513-2522.

Srinivasan, M. V., Zhang, S. W. & Lehrer, M. 1998 Honeybee navigation: Odometry
with monocular input. Animal-Behaviour 56, 1245-1259.

Srinivasan, M. V., Zhang, S., Altwein, M. & Tautz, J. 2000 Honeybee navigation:
Nature and calibration of the "odometer". Science 287, 851-853.

Stabentheiner, A. 1996 Effect of foraging distance on the thermal behaviour of honeybees
during dancing, walking and trophallaxis. Ethology 102, 360-370.

41

Stephens, D. W. & Krebs, J. R. 1986 Foraging Theory. Princeton: Princeton University
Press.

Towne, W. F. & Gould, J. L. 1988 The spatial precision of the honey bee’s dance
communication. Journal of Insect Behavior 1(2), 129-155.

Varju, D. & Nunez, J. 1991 What do foraging honeybees optimize? Journal of
Comparative Physology A 169, 729-736.

von Frisch, K. 1967 The Dance Language and Orientation of Bees. Cambridge: Harvard
University Press.

Waddington, K. D. 1982 Honey bee foraging profitability and round dance correlates.
Journal of Comparative Physiology A Sensory Neural and Behavioral Physiology
148, 297-302.

Waddington, K. D. 1985 Cost-Intake information used in foraging. Journal of Insect
hysiology 31, 891-897.

Waddington, K. D. 1990 Foraging profits and thoracic temperature of honey bees (Apis
mellifera). Journal Of Comparative Physiology B Biochemical Systemic And
Environmental Physiology 160, 325-330.

Waddington, K. D. & Gottlieb, N. 1990 Actual vs. perceived profitability: A study of
floral choice of honey bees. Journal Of Insect Behavior 3, 429-442.

Waddington, K. D. & Heinrich, B. 1979 The foraging movements of bumblebees on
vertical "inflorescences": An experimental analysis. Journal of Comparative
Physiology A Sensory Neural and Behavioral Physiology 134, 113-118.

Waddington, K. D. & Holden, L. R. 1979 Optimal foraging: On flower selection by bees.
American Naturalist 114, 179-196.

Wainselboim, A. J. & Farina, W. M. 2000a Trophallaxis in the honeybee Apis mellifera
(L.): The interaction between flow of solution ad sucrose concentration of the
exploited sources. Animal Behaviour 59, 1177-1185.

Wainselboim, A. J. & Farina, W. M. 2000b Trophallaxis in filled-crop honeybees (Apis
mellifera L.): Food-loading time affects unloading behaviour.
Naturwissenschaften 87, 280-282.

Wolf, T. J., Schmid-Hempel, P., Ellington, C. P. & Stevenson, R. D. 1989 Physiological

correlates of foraging efforts in honey-bees: Oxygen consumption and nectar
load. Functional Ecology 3, 417-424.

42

Wolf, T. J. & Schmid-Hempel, P. 1990 On the integration of individual foraging
strategies with colony ergonomics in social insects: nectar-collection in
honeybees. Behavioral Ecology and Sociobiology 27, 103-111.

43

 

IIIIIIIIIIIIIIIIIIIII

lllllllljllllllll111113111