Available via license: CC BY-ND 4.0
Content may be subject to copyright.
Behavioral state influences encoding mechanisms of color
and odor integration in a model insect, the bumblebee
Bombus impatiens.
Katelyn Graver1, Jessica Sommer1, Vijay Rao1, Giovanni Tafuri1, and Jordanna D. H. Sprayberry1* (ORC
id 0000-0002-6974-5735)
1Muhlenberg College, Allentown PA, USA
* Corresponding author
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
Abstract
Bumblebees rely on diverse sensory information to locate flowers while foraging. The majority
of research exploring the relationship between visual and olfactory floral cues is performed at
local spatial scales and applicable to understanding floral selection. Floral cue-use during search
remains underexplored. This study investigates how the bumblebee Bombus impatiens uses
visual versus olfactory information from flowers across behavioral states and spatial scales. At
local spatial scales, non-flying animals in an associative learning paradigm will access learned
multimodal cues in an elemental fashion - where a single component of the cue is capable of
eliciting responses. However, bumblebees flying in a windtunnel shift cue-use strategy
depending on the spatiotemporal scale of cue encounter. When both color and odor cues mimic
local/ within patch spatial scale, bumblebees exhibit a gradient of elemental responses - with
highest responses to the complete multimodal color+odor cue, followed by intact color, then
intact odor. When cues mimic an intermediate/ between patch spatial scale, bumblebees exhibit
configural responses to the learned cue - with high responses to only the learned color+odor cue.
Thus the underlying physiological state, sensitive to both activity level and spatiotemporal scale
of sensory information, is modulating encoding and utilization of multimodal floral cues.
Introduction
Bumblebees are essential pollinators in multiple ecosystems[1–4]. However, human
activity has negatively impacted bumblebee populations through multiple pathways, including
but not limited to habitat loss, disease, and pesticide exposure (as reviewed by Goulson et al
2015 [5]). In addition, anthropogenic activity can indirectly impact sensory signals that are
utilized by bumblebees while foraging for floral resources. For example, negative effects of
human activity on floral-scent signaling and reception include a reduction of distance traveled by
floral scent [6,7], increased foraging times [8], decreases in floral scent recognition [9–12], and
alterations of flower visitation rates [13,14]. These negative effects are potentially exacerbated
by asymmetric impacts of sublethal pesticide exposure with odor-learning,but not color-learning,
being impaired [15].
However, bumblebees do not exclusively rely on olfactory cues while foraging. Flowers
provide a diversity of sensory information to pollinators (e.g. color, shape, patterning,
electrostatic fields) (reviewed by Sommer et al. 2022 [16], Rands et al 2023 [17]). These signals
are information-rich; improving detection, discrimination, and learning [18–21]. For example,
bumblebees have been shown to recognize paired visual and olfactory patterns more efficiently
than either cue on its own [20,22]. Multimodal signaling can also be used in a redundant manner
by experienced foragers who must recognize rewarding cues across sensory modalities in a range
of environmental conditions [22]. Learning multimodal floral signals likely improves foraging
success for bumblebees, as access to richer floral information reduces uncertainty [20] and
improves detection in variable sensory environments [23,24]. Understanding the ‘rules’ of
multimodal integration of floral-information is critical to determining the potential impacts of
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
odor pollution, as these could be amplified or attenuated depending on how scent is integrated
with other floral signals. The existing body of work strongly indicates that bumblebees utilize
odor information in conjunction with visual cues such as color, but the rules of integration are
less concrete. For example, work by Lawson et al shows that olfactory pattern learning can be
transferred to visual patterns - implying that neural encoding of floral ‘objects’ flexibly
integrates multiple modalities - while other studies have implied that color information
dominates over scent [20,25]. To some extent varying conclusions across these studies can be
explained by different methodologies - but what about methodological difference modulates
sensory integration by bumblebees? It is possible that the ‘rules’ of sensory integration are
dynamic and dependent on the behavioral-mode of a foraging bumblebee. Recent work proposed
a framework for identifying and understanding foraging phases in terms of state and
state-transitions (search (S), acquisition (A), navigation (N), S→A, N→A, etc.),
forager-background (naive (n), experienced (e), primed (p)), and spatial scale (local (l),
intermediate (i), distant (d)), with the understanding that different types of sensory information
are relevant to different foraging-phases [16]. To start, floral-information is variable depending
on distance to a flower/ patch, with olfactory information typically resolvable at greater distances
from flowers than their floral displays [26]. Therefore a bumblebee searching for a novel floral
resource might utilize floral-information differently than a bumblebee selecting a resource within
a patch. The majority of multimodal-integration research has been performed at local spatial
scales, typically less than or equal to 3.6 meters (reviewed by Sommer et al 2022 [16]). These
studies provide valuable data about patch exploitation ( S→A), but their findings may not
directly translate to bumblebees in the search (S) or navigation (N) phases. Given that
bumblebees forage large distances, up to 1 kilometer [27–29], an understanding of how spatial
scale and behavioral state influence multimodal integration is crucial to furthering our
understanding of foraging behavior. This phenomenon is not isolated to bumblebees, but applies
to many species of central place foragers {c Sommer}. Our study investigates how bumblebees
integrate color and odor information across behavioral modes via associative learning and
wind-tunnel paradigms (Tables 1, 2). Specifically we use the free moving proboscis extension
reflex paradigm [30,31]to ask if bumblebees trained to a multimodal color+odor cue will
generalize to the learned color paired with a dissonant odor and vice versa. We use a wind tunnel
assay to ask if foraging bumblebees trained to a multimodal color+odor cue will land on flowers
with the learned color paired with a dissonant odor and vice versa. In addition, we manipulate
access to visual and olfactory cues to simulate changes in spatial scale: tests where bees only
have access to odor cues at the launch point are intended to mimic a farther distance from a
flower, while those where both color and odor cues are available represent local spatial scales
[26].
Methods
Animals
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
This study used 13 colonies of Bombus impatiens (Koppert Biological Systems or
BioBest) between 2018 and 2023. The temperature of the colonies was maintained at 75–85◦ F
by wrapping a seedling heat map with an attached thermostat around each. All colonies had
continued access to ad-libitum pollen throughout the studies. Colonies used for FMPER
experiments had ad-libitum access to 30% sucrose solution, while those used in wind-tunnel
experiments only had access to sucrose during feeding windows of 2-3 hours to increase
experimenter access to foraging bees for experimental-trials.
Stimulus Modalities
All experiments tested color and scent sensory modalities. The colors used were yellow and blue
[32]; scents were lily of the valley (LoV) and juniper berry (JB) [31]. Conditioning for FMPER
experiments is rapid enough that multiple associative color+scent conditions could be tested,
while all wind tunnel experiments were based on association to blue+LoV-scented flowers
(diameter 3.5”). Responses to combinations of novel and congruent combinations of trained
modalities were then tested with different experimental paradigms to determine if behavioral
state impacts the generalization of learned multimodal information (FMPER, Table 1; wind
tunnel, Table 2 ).
Localized Associative Reward Learning: FMPER
Associative learning was assessed using Free Moving Proboscis Extension Reflex
(FMPER) [30,31]. Active individual B. impatiens were selected from lab colonies and placed
into screen backed vials to acclimate for 2 hours. During experiments the vials were placed into a
ventilating FMPER apparatus (Fig 1). The ventilation system drew air in through two small holes
in the vial lid and out of the back screens, with flow rates ranging between 0.1 to 0.3 m/s
(VWR-21800-024 hot wire anemometer). During testing bees were able to walk back and forth
in the vials, but could not fly and had a relatively small range of motion.
During conditioning, individual bees were offered a single drop of 50% sucrose on either
a blue or yellow plastic strip that was inserted through one of the holes in the vial lid
(hole-selection was randomized). Absorbent adhesive bandage tape (Cover Roll) was placed on
the back of the strips in order to hold an odor-stimulus - 1 uL of either Lily of the Valley (LoV)
or Juniper Berry (JB), depending on the condition being tested. Bumblebees would undergo four
association trials, during which they were presented with these rewarding color+scent strips at 5
minute intertrial intervals. Individuals that actively participated in the four association trials
would then undergo a test trial after an additional 5 minute wait. During tests bees were
presented with two unrewarding color+scent strips; the sensory values were variable based on
trial condition. Trials were designed to answer four different questions: 1) is a multimodal cue
learned in this paradigm?; 2) does color or odor exhibit dominance in this paradigm?; 3) is
accurate odor with dissonant color generalized?; 4) is accurate color with dissonant odor
generalized? (Table 1). Proboscis extension onto a strip was considered a choice and individuals
that approached the strips three times or went 45 seconds without exhibiting PER were
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
considered no-choice. All tested bees were tagged following experiments to prevent re-testing
and ensure statistical independence of data points.
Wind Tunnel Tests
Association: During daily timed feeding sessions, wind-tunnel testing colonies were given
access to a glass feeder with an LoV-scented blue flower/s (one or two) containing 30% sucrose.
Flowers were 3-D printed with non-toxic PLA filament (eSun blue). Odor stimuli were added by
pipetting 5 uL of the LoV essential oil onto a 2 cm x 2 cm piece of absorbent tape (Cover
Bandage) on the back of each flower. After three consecutive days of flower color + scent
exposure, colonies were eligible for wind tunnel trials. Occasionally throughout a colony’s
lifespan, the foraging activity within the timed feeding window was not robust enough to
maintain adequate levels of honey for the hive. In these cases ad libitum feeding was offered
without floral stimuli for two - three days. Following this, colonies would have an additional two
days to re-associate with the color and scent stimuli. Following timed feeding sessions,
3D-printed flowers were washed and aired out overnight.
Wind tunnel structure:The wind tunnel dimensions was 5.625 ft x 1.625 ft x 1.541 ft; the
walls and floor were white corrugated plastic and the ceiling was transparent plexi-glass. Green
tape ran lengthwise on the walls, floor, and ceiling to provide visual cues of the wall locations
without providing widefield field motion cues for distance measurement by flying bees [33]. Air
flow through the tunnel is provided by a box fan, with insulation and grid-sheets to even out
flow. Panels were placed at ⅓ and ⅔ of the way through the tunnel on opposite sides, making a
serpentine-maze (Figure 2). Air flow through the tunnel is not laminar, but glycerine
smoke-plumes were visualized to confirm that scent plumes are carried all the way through the
tunnel. Average air flow sampled in the center of the tunnel in each section was 0.37 +- 0.11 m/s.
Test bees were introduced into the tunnel at a consistent location, via a launch platform that
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
accepts bee-vials. Test flowers were placed at the opposite end of the tunnel, 1.5m (59in) from
the launch platform.
Testing: Bees that were seen actively feeding from the training flowers in the colony were
removed and placed in plastic vials for wind tunnel experimentation, waiting on a heating mat
(80.6-82.4 F) for 15-40 minutes before being placed on the launch platform of the wind tunnel.
The cap to the bee’s vial was removed so that the bee could exit the vial, and this cap removal
officially began a trial. Bees that did not exit their vial within five minutes were excluded from
analysis. Bees that exited the vial were given 10 minutes to locate and/or land on the test flower.
Test flowers were unrewarded 3D-printed copies of the training flowers (either blue or yellow),
scented with 10 uL of odor (1:100 LoV or 1:100 JB; diluted in unscented carrier), depending on
the experimental condition being tested (Table 2). Bees that landed on the test flower were given
a “land” designation and the time in which they landed on the flower was recorded. Bees that did
not choose to land after 10 minutes were removed from the tunnel and marked as “no-land.” In
addition, we recorded if bees navigated upwind to the third section of the wind tunnel. All tested
bees were tagged and returned to their colony following experimentation to prevent re-selection
and testing, ensuring statistical independence of data points.
Testing effect of cue encounter: Previous work indicates that searching bumblebees most
likely encounter floral cues as either odor first, or odor and visual cues simultaneously
(Sprayberry, 2018). In an effort to push bumblebees into a searching behavioral state (Sommer
2022), we placed panels at ⅓ and ⅔ of the way through the wind tunnel on opposite sides to
create a serpentine-maze path. These tests endeavor to investigate search behavior in bumblebees
at both a local spatial-scale, with clear panels allowing access to both visual and odor cues from
the launch point, and an intermediate spatial-scale, with opaque panels obscuring visual cues
(Figure 3). We used the number of tested individuals that traveled to the third section to calculate
the percentage of ‘searching’ bees. We then used the number of individuals who landed divided
by the number of searching bees to calculate percent landing.
Statistical Analysis
FMPER Analysis: For each experimental question analysis of trial data proceeded as follows:
1) Full response distributions (including no choice) were tested against random (33%) with a
fisher’s exact test. 2) Choice data were tested against random (50%) with a binomial distribution
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
test. 3) All no choice data across trial conditions within an experimental question were tested
with pairwise Fisher comparisons (using the pairwise.fisher.test function in R). Alpha values for
the first two tests were Holm-Bonferroni corrected. P values for the third test were corrected with
the Benjamani-Hockberg method. All analyses are summarized in Table 3.
Wind Tunnel Analysis: Both search and land data were analyzed as no-choice data above. For
each, data for all stimulus conditions within a set of wind tunnel trials (opaque or clear
maze-dividers) were tested with pairwise Fisher comparisons (using the pairwise.fisher.test
function in R). P values were corrected with the Benjamani-Hockberg method. All analyses are
summarized in Table 4.
Results
Non-foraging bumblebees will utilize components of a learned multimodal cue
elementally.
To test bumblebees’ ability to learn a color+odor cue, we trained bees to associate a
scented-color strip with a sucrose reward using the free-moving proboscis extension reflex
paradigm (FMPER), which tests for proboscis extension reflex (PER) after conditioning using
two unrewarding strips: one matching their training condition, and one with dissonant
color+odor (Table 1, Fig 1). In both training conditions (blue+lily of the valley (LoV) and
yellow+LoV) the trained cue was selected more frequently than the dissonant cue (p<0.01, Table
3). We then tested if either the color or odor component of a learned color+odor cue was
dominant by testing trained bees with a choice of learned-color + novel-odor versus novel-color
+ learned-odor stimuli. In these tests bumblebees were equally likely to select the learned-color
or learned-odor (Fig 1d, Table 3). We also tested if bumblebees would generalize to individual
learned components of a color+odor cue when compared against an entirely novel cue. This was
performed for each sensory modality, where bumblebees trained to a color+odor cue were tested
against the learned-color+novel-odor in one set of experiments, and against the
learned-odor+novel-color in another set (Figure 2). Bumblebees did generalize to a learned
cue-component for at least one associative condition in each modality: bees trained to blue+LoV
selected yellow+LoV at a higher frequency than yellow+juniper berry (JB); and bees trained to
yellow-LoV selected yellow+JB at a higher frequency than blue+JB (Figure 2, Table 3).
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
Color and odor identity of learned multimodal cues influences the likelihood that
non-foraging bumblebees influence will access sensory information in an
elemental fashion
The responsiveness of bumblebees
within this FMPER paradigm was
variable based on the color and odor
values of multimodal cues (Table 3).
Bumblebees associated to blue-strips
demonstrated a higher percentage of
no-choice responses than those
associated to yellow-strips in tests of
multimodal learning ability (p=0.009,
Table 3). Experiments on bumblebee
likelihood of generalizing to a learned
color + novel odor show similar trends;
all tests utilizing blue associative strips
have a higher percentage of no-choice
responses (Table 3). These tests also
show an increase in no-choice when
the associative odor is juniper berry.
Indeed, color generalization trials
where bumblebees were associated to
blue+JB not only showed a high
percentage of no-choice responses,
bees did not generalize to the learned
color. Rather they chose yellow+LoV
in their test phase. Likewise,bees
trained on yellow+LoV had the highest
percentage of generalization-responses; and bees trained to yellow+JB showed a significant
increase in no-choice responses (Figure 2).
Foraging bumblebees access multimodal information configurally
To test how actively foraging bumblebees respond to components of learned multimodal cues,
we trained colonies to forage on blue-LoV scented flower feeders. Foraging bees were then
collected from feeders and flown in a wind tunnel divided into a three section maze with offset
panels; the final section held an unrewarding flower. We used the number of tested individuals
that traveled to the third section to calculate the percentage of navigating bees. We then used the
number of individuals who landed on the test flower divided by the number of navigating bees to
calculate percent landing (Figure 3). Wind tunnel experiments measured responses of actively
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
foraging bees to various novel or congruent combinations of color and odor (Table 2). The
percentage of navigating bees was consistent across all treatments; unaffected by panel-status,
color, or scent (p>0.3, Figure 4, Table 4). However, cue-value had a substantial effect on landing
frequency, with the blue-LoV flowers having the highest landing frequency, and flowers with any
dissonant modalities showing reduced frequency in most conditions (Figure 4, Table 4).
Spatiotemporal scale of sensory encounter can partially recover elemental cue use
in foraging bumblebees
Wind tunnel tests endeavor to investigate behavior in bumblebees across spatial scales; with
clear panels allowing access to both visual and odor cues from the launch point simulating a
local-spatial scale; and opaque panels obscuring visual cues but allowing access to odor cues
simulating an intermediate spatial scale (Figure 3) (Sommer et al). Bees tested at ‘intermediate’
scales, where odor encounter precedes visual, showed bimodal landing responses, responding
robustly to only the learned multimodal cue combination blue+LoV (Figure 4). Bees tested at
‘local’ scales, where visual cues are encountered in synchrony with or ahead of odor, show a
gradient of landing frequencies; with the highest response to learned-color + learned-odor cues,
followed by learned-color + novel-odor, then novel-color + learned-odor, and no response to a
fully novel cue (Figure 4).
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
Discussion
The spatiotemporal scale of cue encounter modulates encoding of multimodal
information
Previous multimodal studies on bumblebees have shown support for the ‘efficacy backup
hypothesis’, where access to information across multiple sensory modalities improves learning
and recall in noisy environments [20,22,23,34]. The data presented here also support this
hypothesis; with our findings that unimodal components from a learned multimodal cue are
capable of influencing behavioral choices in our associative learning FMPER paradigm (figs
1,2). In addition a substantial percentage of bees flown in the clear paneled wind tunnel were
responsive to the learned color component of trained floral cues (fig 4). From a neural circuitry
perspective this implies elemental access to the unimodal components of learned multimodal
cues during activation of behavioral responses. However, bumblebee responses to unimodal
cue-components paired with dissonant values are not uniform across experimental paradigms.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
Bumblebees flown in an opaque paneled wind-tunnel, intended to simulate an experienced
forager navigating at an intermediate spatial scale (foraging state N(i,e)), showed highest landing
frequencies on flowers with intact multimodal cues. From a neural circuitry perspective this
implies that behavioral responses are activated in response to a synthesized color+odor percept
(i.e. configurally). The principle difference between wind tunnel paradigms is the temporal
access to unimodal cue components; where opaque panels allow access to odor cues before
color. Indeed this phased cue encounter is how ‘intermediate’ spatial scales in central place
foragers has been defined [16]. Clear panels facilitate either simultaneous color+odor, or color
cue encounter first. The neural mechanisms behind how timing of multimodal cues can modulate
the relationship between unimodal cues has received little attention. A recent study utilizing
conditioning of the proboscis extension reflex in harnessed bumblebees investigated the impact
of color and odor synchronicity on bumblebee learning [25]. In their experimental context they
also found evidence of elemental cue access, with readily learned conditions showing stronger
responses to odor alone. Interestingly bees trained with odor preceding color showed the lowest
responses to unimodal cues, providing some support for hypothesizing that the temporal
structure of odor versus color encounter could shift bumble bees from elemental towards
configural responses. However, that treatment was also poorly learned in their experimental
paradigm; thus the mechanisms behind spatiotemporal scales of multimodal cues and behavioral
activation remain an intriguing site for future study.
The idea that ‘state’ influences sensory processing in insects is not novel. Studies on fruit flies,
hawk moths, and mosquitoes have explored how shifts in state can underlie behavioral
flexibility [35–38]. Flight activity is known to modulate sensory processing - enhancing visual
processing of landing in fruit flies [37] and olfactory processing in hawk moths [38]. Hunger can
modulate responses to odor stimuli - turning aversion to attraction - in fruit flies [36]. Similarly,
in mosquitoes one sensory cue (CO2) can influence responses to an additional cue (host odors)
[35]. In the current body of literature, the presence of one sensory cue (internal or external) is
modulating the responses to an additional cue. In the results presented here, internal activity state
could certainly be influencing some of the differences we see between FMPER and wind tunnel
experiments. However, the shift from configural to elemental cue integration within wind tunnel
treatments represents a more complicated instance of state modulation of sensory processing.
Not all stimuli are created equal: color and odor values influence how bumblebees
respond to learned multimodal cue information
As reviewed by Leonard and Masek [21], the results of multimodal experiments can be shaped
by the salience of chosen stimuli. While existing literature supports that bumblebees can learn all
of the stimulus values used in these experiments [31,32,39,40], our results indicate that in our
FMPER experimental context, yellow and lily of the valley were more appetitive than their
counterparts (blue and juniper berry) (figures 1,2). The FMPER experiments, while excellent for
probing cognitive capability, are far from a free foraging environment; thus direct translation of
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
this work into field preferences is premature. However, these experiments do highlight that
bumblebee responses to multimodal stimuli are driven by neural circuitry that is performing
more complex operations than simply matching potential resources to a learned search image.
Learning clearly has a strong influence on responses, but is operating in conversation with
additional factors such as salience and/ or valence (how appetitive or aversive a stimulus is). This
hints at the kind of cognitive (and thus behavioral) flexibility that makes bumblebees effective
generalist pollinators.
Ecological implications and limitations
Understanding how the multimodal information flowers provide is utilized by bumblebees in
different foraging phases and across gradients of sensory disruption is critical to conservation
efforts. For example, if a learned-color paired with a novel-odor is generalized to the learned
multimodal-cue by a substantial portion of within-patch foragers - as indicated by our clear-panel
wind tunnel tests - then local disruption of floral odors via anthropogenic activities may be
ameliorated. However, generalization tendencies are likely impacted by the valence and foraging
state of any odor-disruptions. For example Ryalls et al found that diesel pollution on a local level
substantially reduced visitation of a broad spectrum of pollinators, including bumblebees [13]. A
subsequent study found variable effects of diesel-driven odor pollution dependent on floral color
[14]. Interestingly diesel exhaust contains sulfurous components - an attribute shared by several
fungicides that have also been shown to disrupt bumblebee behavior on a local scale [11,12].
Thus our ability to fully interpret these findings is limited by our incomplete understanding of an
odor-valence spectrum for bumblebees.
When considering bumblebees searching for resources at a larger spatial scale, the impacts of
anthropogenic disruption are potentially more dire. Bumblebees travel large distances from their
nest during foraging bouts [27–29], presumably searching for resources if they are not navigating
directly to a previously learned patch. Recent work showed that bumblebees prefer to fly upwind
[41], which primes them to utilize olfactory navigation if they encounter an ‘acceptable’
odor-plume [42,43]. If, as our opaque-panel tests indicate, experienced foraging bees are less
flexible in terms of accepting disrupted color+odor combinations they are more likely to pass
over discovered patches whose odors have been modulated by human activity. Given that habitat
loss and fragmentation both reduce total resources available to bumblebees and increase distance
between viable resource-patches (reviewed by Goulson [5]), anthropogenic-odor disruption
could result in consequential reductions in foraging efficiency. Indeed, understanding the impact
of spatial scale on odor generalization within multimodal cues is a critical next step in
determining the conservation impacts of odor pollution.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
Acknowledgements
The authors would like to thank Muhlenberg College for financial support of this work through summer
research stipends. We thank N. Defino, K. Esbenshade, N. Yousry, P. Henderson, and G. Tafuri for
contributions to equipment construction and data collection.
Funding Statement
This work was internally funded by Muhlenberg College
Data Availability
Data will be made available for download after peer review and publication.
References
1. Heemert C van, Ruijter A de, Eijnde J van den, Steen J van der. 2015 Year-Round Production
of Bumble Bee Colonies for Crop Pollination. Bee World 71, 54–56.
(doi:10.1080/0005772x.1990.11099036)
2. Hegland SJ, Totland Ø. 2008 Is the magnitude of pollen limitation in a plant community
affected by pollinator visitation and plant species specialisation levels? Oikos 117, 883–891.
(doi:10.1111/j.2008.0030-1299.16561.x)
3. Klein AM, Vaissiere BE, Cane JH, Steffan-Dewenter I, Cunningham SA, Kremen C,
Tscharntke T. 2008 Importance of pollinators in changing landscapes for world crops.
Proceedings of the Royal Society B: Biological Sciences 274, 303–313.
(doi:10.1126/science.265.5176.1170)
4. Motten AF. 1986 Pollination Ecology of the Spring Wildflower Community of a Temperate
Deciduous Forest. Ecol Monogr 56, 21–42. (doi:10.2307/2937269)
5. Goulson D, Nicholls E, Botías C, Rotheray EL. 2015 Bee declines driven by combined stress
from parasites, pesticides, and lack of flowers. Science 347, 1255957–1255957.
(doi:10.1126/science.1255957)
6. Farré-Armengol G, Peñuelas J, Li T, Yli-Pirilä P, Filella I, Llusia J, Blande JD. 2015 Ozone
degrades floral scent and reduces pollinator attraction to flowers. New Phytologist 209, 152–160.
(doi:10.1111/nph.13620)
7. McFrederick QS, Kathilankal JC, Fuentes JD. 2008 Air pollution modifies floral scent trails.
Atmospheric Environment 42, 2336–2348. (doi:10.1016/j.atmosenv.2007.12.033)
8. Fuentes JD, Chamecki M, Roulston T, Chen B, Pratt KR. 2016 Air pollutants degrade floral
scents and increase insect foraging times. Atmospheric Environment 141, 361–374.
(doi:10.1016/j.atmosenv.2016.07.002)
9. Girling RD, Lusebrink I, Farthing E, Newman TA, Poppy GM. 2013 Diesel exhaust rapidly
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
degrades floral odours used by honeybees. Scientific Reports 3. (doi:10.1038/srep02779)
10. Yousry N, Henderson P, Sprayberry JD. In press. Fungicide scent pollution disrupts floral-
search and -selection in the bumblebee Bombus impatiens. Agrochemicals 2, 181–192.
(doi:10.3390/agrochemicals2020013)
11. David NF, Henry TJ, Sprayberry JDH. 2022 Odor-Pollution From Fungicides Disrupts
Learning and Recognition of a Common Floral Scent in Bumblebees (Bombus impatiens).
Frontiers Ecol Evol 10, 765388. (doi:10.3389/fevo.2022.765388)
12. Sprayberry JDH, Ritter KA, Riffell JA. 2013 The effect of olfactory exposure to
non-insecticidal agrochemicals on bumblebee foraging behavior. PLoS ONE 8, e76273.
(doi:10.1371/journal.pone.0076273.s001)
13. Ryalls JMW, Langford B, Mullinger NJ, Bromfield LM, Nemitz E, Pfrang C, Girling RD.
2022 Anthropogenic air pollutants reduce insect-mediated pollination services. Environ Pollut ,
118847. (doi:10.1016/j.envpol.2022.118847)
14. Lusebrink I, Girling RD, Dobrindt L, Jackson CW, Newman TA, Poppy GM. 2023
Investigating the effects of diesel exhaust and flower color on flower visitation by free-flying
honey bees. Arthropod-plant Inte 17, 11–17. (doi:10.1007/s11829-022-09941-w)
15. Muth F, Francis JS, Leonard AS. 2019 Modality-specific impairment of learning by a
neonicotinoid pesticide. Biology Letters 15, 20190359–5. (doi:10.1098/rsbl.2019.0359)
16. Sommer J, Rao V, Sprayberry J. 2022 Deconstructing and contextualizing foraging behavior
in bumble bees and other central place foragers. Apidologie 53, 32.
(doi:10.1007/s13592-022-00944-3)
17. Rands SA, Whitney HM, Ibarra NH de. 2023 Multimodal floral recognition by bumblebees.
Curr. Opin. Insect Sci. 59, 101086. (doi:10.1016/j.cois.2023.101086)
18. Kunze J, Gumbert A. 2001 The combined effect of color and odor on flower choice behavior
of bumble bees in flower mimicry systems. Behavioral Ecology 12, 447–456.
19. Kulahci IG, Dornhaus A, Papaj DR. 2008 Multimodal signals enhance decision making in
foraging bumble-bees. Proceedings of the Royal Society B: Biological Sciences 275, 797–802.
(doi:10.1098/rspb.2007.1176)
20. Leonard AS, Dornhaus A, Papaj DR. 2011 Flowers help bees cope with uncertainty: signal
detection and the function of floral complexity. Journal of Experimental Biology 214, 113–121.
(doi:10.1242/jeb.047407)
21. Leonard AS, Masek P. 2014 Multisensory integration of colors and scents: insights from bees
and flowers. Journal of Comparative Physiology 200, 463–474.
(doi:10.1007/s00359-014-0904-4)
22. Lawson DA, Whitney HM, Rands SA. 2017 Colour as a backup for scent in the presence of
olfactory noise: testing the efficacy backup hypothesis using bumblebees ( Bombus terrestris).
Royal Society Open Science 4, 170996–13. (doi:10.1098/rsos.170996)
23. Kaczorowski RL, Leonard AS, Dornhaus A, Papaj DR. 2012 Floral signal complexity as a
possible adaptation to environmental variability: a test using nectar-foraging bumblebees,
Bombus impatiens. YANBE 83, 905–913. (doi:10.1016/j.anbehav.2012.01.007)
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
24. Jordan KA, Sprayberry JDH, Joiner WM, Combes SA. 2023 Multimodal processing of noisy
cues in bumblebees. iScience , 108587. (doi:10.1016/j.isci.2023.108587)
25. Riveros AJ. 2023 Temporal configuration and modality of components determine the
performance of bumble bees during the learning of a multimodal signal. J Exp Biol 226.
(doi:10.1242/jeb.245233)
26. Sprayberry JDH. 2018 The prevalence of olfactory- versus visual-signal encounter by
searching bumblebees. Scientific Reports 8, 1–10. (doi:10.1038/s41598-018-32897-y)
27. Osborne JL, Clark SJ, Morris RJ, Williams IH, Riley JR, Smith AD, Reynolds DR, Edwards
AS. 1999 A landscape‐scale study of bumble bee foraging range and constancy, using harmonic
radar. Journal of Applied Ecology 36, 519–533.
28. Hellwig KW, Frankl R. 2000 Foraging habitats and foraging distances of bumblebees,
Bombus spp. (Hym., Apidae), in an agricultural landscape. Journal of Applied Entomology 124,
299–306. (doi:10.1046/j.1439-0418.2000.00484.x)
29. Lihoreau M, Collett T, Raine NE, Reynolds AM, Stelzer RJ, Lim KS, Smith AD, Osborne
JL, Chittka L. 2012 Radar tracking and motion-sensitive cameras on flowers reveal the
development of pollinator multi-destination routes over large spatial scales. PLoS biology 10,
e1001392. (doi:10.1371/journal.pbio.1001392)
30. Muth F, Carvalheiro L, Cooper TR, Bonilla RF, Leonard AS. 2017 A novel protocol for
studying bee cognition in the wild. Methods in Ecology and Evolution 9, 78–87.
(doi:10.1111/2041-210x.12852)
31. Sprayberry JDH. 2020 Compounds without borders: A mechanism for quantifying complex
odors and responses to scent-pollution in bumblebees. PLoS Computational Biology 16,
e1007765. (doi:10.1371/journal.pcbi.1007765)
32. Ings TC, Raine NE, Chittka L. 2009 A population comparison of the strength and persistence
of innate colour preference and learning speed in the bumblebee Bombus terrestris. Behavioral
Ecology and Sociobiology 63, 1207–1218. (doi:10.1007/s00265-009-0731-8)
33. Baird E, Boeddeker N, Srinivasan MV. 2021 The effect of optic flow cues on honeybee flight
control in wind. Proc Royal Soc B 288, 20203051. (doi:10.1098/rspb.2020.3051)
34. Lawson DA, Chittka L, Whitney HM, Rands SA. 2018 Bumblebees distinguish floral scent
patterns, and can transfer these to corresponding visual patterns. Proceedings. Biological
sciences / The Royal Society 285, 20180661–8. (doi:10.1098/rspb.2018.0661)
35. van Breugel F, Riffell J, Fairhall A, Dickinson MH. 2015 Mosquitoes Use Vision to
Associate Odor Plumes with Thermal Targets. Curr. Biol. 25, 2123–2129.
(doi:10.1016/j.cub.2015.06.046)
36. Vogt K et al. 2021 Internal state configures olfactory behavior and early sensory processing
in Drosophila larvae. Sci. Adv. 7, eabd6900. (doi:10.1126/sciadv.abd6900)
37. Ache JM, Namiki S, Lee A, Branson K, Card GM. 2019 State-dependent decoupling of
sensory and motor circuits underlies behavioral flexibility in Drosophila. Nat. Neurosci. 22,
1132–1139. (doi:10.1038/s41593-019-0413-4)
38. Chapman PD, Burkland R, Bradley SP, Houot B, Bullman V, Dacks AM, Daly KC. 2018
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
Flight motor networks modulate primary olfactory processing in the moth Manduca sexta. Proc.
Natl. Acad. Sci. 115, 5588–5593. (doi:10.1073/pnas.1722379115)
39. Gumbert A. 2000 Color choices by bumble bees (Bombus terrestris): innate preferences and
generalization after learning. Behavioral Ecology and Sociobiology 48, 36–43.
(doi:10.1007/s002650000213)
40. Langridge KV, Wilke C, Riabinina O, Vorobyev M, Ibarra NH de. 2021 Approach Direction
Prior to Landing Explains Patterns of Colour Learning in Bees. Front. Physiol. 12, 697886.
(doi:10.3389/fphys.2021.697886)
41. Combes SA, Gravish N, Gagliardi SF. 2023 Going against the flow: bumblebees prefer to fly
upwind and display more variable kinematics when flying downwind. J Exp Biol 226.
(doi:10.1242/jeb.245374)
42. Carde RT, Willis MA. 2008 Navigational Strategies Used by Insects to Find Distant,
Wind-Borne Sources of Odor. Journal Of Chemical Ecology 34, 854–866.
(doi:10.1007/s10886-008-9484-5)
43. Baker KL, Dickinson M, Findley TM, Gire DH, Louis M, Suver MP, Verhagen JV, Nagel KI,
Smear MC. 2018 Algorithms for Olfactory Search across Species. J. Neurosci. 38, 9383–9389.
(doi:10.1523/jneurosci.1668-18.2018)
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted January 30, 2024. ; https://doi.org/10.1101/2024.01.30.577961doi: bioRxiv preprint
