Assessing grey squirrel GUD between peanuts and sunflower seeds

in the presence of red-tailed hawk calls

December 6, 2021

Alysa Royak, Eddie Rice, Michael Rose, and Natalie Sebunia

1
Abstract
When foraging, organisms adhere to the marginal value theorem, attempting to maximize
foraging profits given associated costs of continued foraging. Foraging behavior can be impacted
by environmental factors, including perceived predation risk. This study aimed to investigate
foraging behavior of grey squirrels at Tuttle Park in Columbus, Ohio. We tested the effects of
sound and food type on squirrel foraging behavior by calculating giving-up-density (GUD). We
hypothesized that grey squirrels use auditory cues to determine predation risk when foraging and
predicted squirrels would have a higher GUD when a predatory sound was present. We found
this hypothesis was supported (p-value<0.05). Furthermore, we hypothesized that grey squirrels
would minimize handling time. When given the choice between larger, shelled peanuts and
sunflower seeds, we predicted that squirrels would have a higher GUD for peanuts. This
hypothesis was not supported by our data (p-value=0.252). Therefore, we conclude that GUD
was higher in the presence of hawk screeches but not affected by food type. These results
indicate that squirrels utilize auditory cues to determine predation risk, using perceived risk to
determine whether continued foraging outweighs costs. Further research into squirrels’
perceptions of auditory cues will allow for a more robust understanding of foraging behavior.

Introduction
A growing body of research seeks to investigate how animals optimize their foraging
behavior to mitigate their chances of predation and increase their chances of survival. According
to the marginal value theorem, an animal should leave a given patch of food when the costs of
continued foraging outweigh the benefits of food acquisition (Venkataraman et al. 2017). The
model assumes that discrete patches exist in the environment and takes into account travel time
between patches, energy costs, and resource intake to determine the optimal foraging behavior of
an individual (Venkataraman et al. 2017). A simple way to analyze optimal foraging behavior is
by measuring an organism’s giving-up density (GUD) under different environmental conditions.
GUD describes the percentage of food an animal chooses to abandon or leave behind when
foraging within a given patch (Schmidt and Ostfeld 2008, Bowers et al. 1993, Bowers and
Breland 1996). There are several reasons why an animal may choose to abandon a food resource,
and these can include predation risk (Schmidt and Ostfeld 2008, Orrock et al. 2004, Bowers et al.
1993), energy costs with continued foraging (Newman et al. 1988), and availability of other
nearby resources (Kotler et al. 1999). Overall, however, the animal perceives that the costs of
continuing to forage outweigh the benefits, and therefore abandons the food resource.
Animals often use cost and benefit analysis when foraging, especially when they fear a
predator may be nearby. Previous research has shown that perceived predation risk can greatly
influence the GUD of an individual foraging at a given patch. Schmidt and Ostfeld (2008) found
that squirrels had higher GUDs in the presence of blue jay calls, indicating that squirrels may
listen for predatory noises nearby while foraging, reducing the amount of food they collect when
there is a perceived risk. GUD may also be higher when less protection is available for the
forager (such as an open site versus a forest canopy), there are more foragers present, and the
time of day of the foraging and weather conditions (Schmidt and Ostfeld 2008, Orrock et al.
2004, Bowers and Breland 1996; Newman et al. 1988). These cues may indicate to the forager
that a predator is nearby and that the forager should attempt to escape or hide rather than
continue to forage.
GUD also appears to be correlated with indirect predator cues more frequently than with
direct predator cues (Orrock et al. 2004). Indirect predator cues are those not directly produced

2
by the predator but indicative of potential predation. An example of an indirect predator cue
could be an environmental cue such as the weather, which may indicate to a forager that
predators are not particularly active at a given time. On the other hand, direct predator cues are
directly produced by a predator and allow the forager to identify that a predator is in the area.
Examples of direct predator cues include natural smells and sounds of the predator. Orrock et al.
(2004) found that mice had lower GUDs in the presence of indirect predator cues such as when
cover was present, during night, during rain, and when moon illumination was low. As a result,
the mice felt safer foraging at those times; hence, the lower GUDs. Furthermore, the differences
in GUD in these conditions were more significant than when direct scent cues could be used by
the mice (Orrock et al. 2004). In other words, habitat conditions also appear to play a role in the
perceived threat of predation by foragers.
However, other studies suggest that GUD may affected by the handling rate of the food
resource in addition to indirect predator cues. Newman et al. (1988) found that the same quantity
of food was eaten at two locations with differing distances from a canopy, but the amount of time
spent in each patch differed, with a shorter period of time spent in the patch farther from cover.
This indicates that animals may alter their foraging behaviors in ways that allow for a similar
GUD despite potential predation risks. Additionally, it appears as though handling time is an
important factor in foraging behavior. Fox squirrels have been shown to consider both the
immediate benefit of food resources, as well as the future value of the resource if it were to be
cached, with higher GUDs being present when only non-cacheable food resources are present
(Kotler et al. 1999). When food is cacheable, it appears as though the cost of continuing to forage
is decreased because the food can be used later, decreasing the GUD value at which cost
outweighs the benefit of continued foraging.
We seek to investigate whether the GUD of grey squirrels (Sciurus carolinensis) is
affected by the presence of predator sounds. Grey squirrels are an easily accessible organism to
study and previous research on foraging has focused on squirrel behavior, allowing avenues for
comparison to previous research. We hypothesize that grey squirrels will use auditory cues to
determine predation risk when foraging. Therefore, grey squirrels will reduce the time spent
foraging at a site if they hear sounds of a common predator nearby because the risk of getting
eaten is greater than the risk of losing a portion of food they can potentially collect; in turn, GUD
will be higher in the presence of predator sounds. Red-tailed hawks are a common predator of
squirrels and would be a common perceived risk, making red-tailed hawk sounds an adequate
direct predator cue (Fitch et al. 1946). This study also investigates whether the GUD of grey
squirrels will be affected by the size of their food options, which in turn impacts handling time.
We hypothesize that grey squirrels will minimize handling time when foraging. Therefore, when
given the choice between shelled peanuts and sunflower seeds, grey squirrels will have a lower
GUD for sunflower seeds because they are easier to handle as a result of their smaller size,
despite the increased ability of the peanuts to be cached by the grey squirrels.

Methods
In order to measure whether the GUD of squirrels was affected by the presence of
predator calls and whether GUD was affected by the type of food squirrels were foraging for,
trays with food were set out over periods of time for squirrels to forage from and GUD was
calculated based on the mass of food leftover in the tray (Schmidt and Ostfeld 2008, Kotler et al.
1999, Bowers et al. 1993). Five trials were conducted over the span of five weeks from
September to October. Each trial ran for two hours, with one hour consisting of sound and one

3
hour without sound. Each trial also contained two trays: one with sunflower seeds and one with
peanuts. Trials 1 and 3 were conducted with one hour of sound followed by one hour of no
sound, while Trials 2, 4, and 5 were conducted with one hour of no sound followed by one hour
of sound. Altering the order of sound-no sound helped to account for the potential confounding
variable of sound initially scaring away squirrels. Since trials took place relatively close to
human recreation, interference from human activities was limited as much as possible by
conducting the trials in a secluded area that mostly simulated a natural setting.
Trials took place at Tuttle Park in Columbus, Ohio. The 48.293-acre park contains
walking trails with woods and wildlife areas, as well as several other public recreation amenities
(The City of Columbus 2021). Tuttle Park served as an acceptable location for the trials due to
the substantial, secluded, wooded areas which reduced the potential confounding variable of
human presence. Each trial took place near a fallen tree in a wooded area slightly off the
Olentangy River Trail within the park. Before each trial, temperature and weather conditions
were recorded, as well as the time of day and date of the trial. Recording these variables allowed
for records of potential confounding variables within the study.
For the sound portion of the trials, a speaker was placed in a tree above the trays to better
replicate the location of an actual avian predator. This resulted in a more accurate experiment by
eliminating the confounding factor of placing the audio next to the ground, where the predators
wouldn’t normally be. Red-tailed hawks are a common predator of squirrels and would be a
common perceived risk (Fitch et al. 1946) which is why the audio of a red-tailed hawk
screeching was chosen for the study (Appendix 1). Before each trial, 100 grams of peanuts were
weighed with a small scale and placed into a tray filled with sand. In the second tray, 100 grams
of weighted sunflower seeds were placed in sand as well. The two trays were then placed on the
ground with a short distance between them (about 1 meter) for an hour with one sound condition
(either sound or no sound). The red-tailed hawk call was played on a loop. The call lasted about
6 seconds; it would briefly stop, and then play again. The location of the trial was in isolation
from human activity to prevent disturbances to the experiment.
After the first hour of each trial, the contents of the two trays were reweighed and the
GUD for each was calculated by dividing the initial weight (100 grams) by the final weight.
Trays were then refilled to equal 100 grams of food resource and left for another hour with the
opposite sound condition (sound or no sound) to complete the trial. After the second hour, the
contents of each tray were again reweighed and GUD was calculated. A summary of the five trial
conditions can be seen in Table 1 below, as well as time-stamped photos for each trial (Figure 1).

Table 1. This table displays the date, time, and team member(s) present for each trial.
Trial Date Time Team Members Present
1 9/19/21 2:00 PM Eddie Rice, Natalie Sebunia
2 10/1/21 3:25 PM Michael Rose
3 10/3/21 2:30 PM Alysa Royak, Michael Rose, Natalie
Sebunia
4 10/6/21 4:45 PM Natalie Sebunia
5 10/10/21 2:20 PM Alysa Royak, Eddie Rice, Michael Rose

4
Trial 1; 9/19/21:

Trial 2; 10/1/21:

5
Trial 3; 10/3/21:

Trial 4; 10/6/21:

6
Trial 5; 10/10/21:

Figure 1. Time-stamped pictures from all 5 trials, showing trial, showing tray placement and members
present.

After data collection, the data was analyzed in R (Version 4.11) using a two-way
ANOVA analysis to determine overall significance of the presence of sound and the food type. A
corresponding Tukey test was then run to determine which specific variables showed significant
differences in GUD between trials. Significance was determined with a p-value of less than 0.05.
This data was plotted on a box-and-whisker graph with error bars for further analysis.

Results
The raw data results of all five trials, including the weather conditions, initial mass of
food for each treatment group, final mass of food, and calculated GUD for each treatment group
in each trial, can be found in Appendix 2. A shortened summary of the weather conditions for
each trial can be seen in Appendix 3, while the average GUDS for each condition across all five
trials are summarized below in Table 2.

Table 2. Average GUDs across all 5 trials for each condition.

Variable GUD
Peanuts w/ sound 0.91
Sunflower Seeds w/
0.88
sound
Peanuts w/o sound 0.68
Sunflower seeds w/o
0.64
sound

7
 GUD appeared to be higher in the presence of sound than without the presence of sound.
Squirrels had a 0.232 greater GUD when sound was present with peanuts versus when sound was
present without peanuts, a 0.196 greater GUD when sound was present with sunflower seeds
than when no sound was present with peanuts, a 0.276 greater GUD when sound was present
with peanuts than when no sound was present with sunflower seeds, and a 0.240 greater GUD
when sound was present with sunflower seeds than when no sound was present with sunflower
seeds (Table 4). The GUD raw data was further analyzed using a two-way ANOVA test. These
results can be seen in Table 3.

Table 3. Two-Way ANOVA Results.

Variable F statistic p-value
Sound 49.201 2.92*10^-6

Food.Type 1.413 0.252

Sound:Food.Type 0.014 0.907

Sound appeared to have a significant effect on squirrel GUD (p-value = 2.92*10^-6).

However, food type did not have a significant effect on squirrel GUD (p-value = 0.252) and the
two variables in conjunction also did not have a significant affect on squirrel GUD (p-value =
0.907). Figure 2 shows the results of the two-way ANOVA analysis in a box-and-whisker plot.

Figure 2. Box-and-Whisker Plot of GUD Data. “Yes” and “No” variables represent the presence of lack of sound,
while “Peanut” and “Sunflower” variables represent the type of food item present.

8
 The box-and-whisker plot shows significantly lower average GUD values for no sound
conditions. The no sound condition boxes for both peanuts and sunflower seeds are relatively
similar, whereas the sound condition boxes for peanuts and sunflower seeds are also relatively
similar to one another. This further indicates that a significant difference in GUD exists between
sound and no sound conditions. Because significance was recorded for the sound variable, a
Tukey test was run to further analyze significant correlations in the data. Table 4 shows the
results of this analysis.

Table 4. Tukey Test Results.

Condition 1 Condition 2 Diff Lower Upper P-Value
Limit Limit
Sound:Peanut NoSound:Peanut 0.232 0.09586836 0.36813164 0.0008725
NoSound:Sunflower NoSound:Peanut -0.044 0.18013164 0.09213164 0.7921865
Sound:Sunflower NoSound:Peanut 0.196 0.05986836 0.33213164 0.0040186
NoSound:Sunflower Sound:Peanut -0.276 -0.41213164 -0.13986836 0.0001442
Sound:Sunflower Sound:Peanut 0.036 -0.17213164 0.10013164 0.8725385
Sound:Sunflower NoSound:Sunflower 0.240 0.10386836 0.37613164 0.0006248

The Tukey test showed that significant differences in GUD were found for the following
treatment groups: sound with peanuts and no sound with peanuts (p-value = 0.0008725), sound
with sunflower seeds and no sounds with peanuts (p-value = 0.0040186), no sound with
sunflower seeds and sound with peanuts (p-value = 0.0001442), and sound with sunflower seeds
and no sound with sunflower seeds (p-value = 0.0006248). The treatments of sound with both
sunflower seeds and peanuts (p-value = 0.8725385) and no sound with both sunflower seeds and
peanuts (p-value = 0.7921865) showed no statistically significant differences in GUD values.

Discussion
The results of the GUD data analysis revealed that the presence of red-tailed hawk sounds
appear to affect the GUD of squirrels (p-value = 2.9*10^-6), whereas the type of food present in
the tray does not significantly affect the GUD of squirrels (p-value = 0.2518344). The Tukey test
revealed that this significant difference in GUD for sound was present for all possible
combinations of treatment groups with sound and no sound, regardless of the type of food
present. This includes peanuts with sound and peanuts without sound (p-value = 0.0008725),
sunflower seeds with sound and peanuts without sound (p-value = 0.0040186), sunflower seeds
without sound and peanuts with sound (p-value = 0.0001442), and sunflower seeds with sound
and sunflower seeds without sound (p-value = 0.0006248). The treatments of sound with both
sunflower seeds and peanuts and no sound with both sunflower seeds and peanuts showed no
statistically significant differences in GUD values. Because the overall effect of food type had no
significant effect on GUD, we can conclude that any differences in GUD between groups with
sound and no sound and two different food variables are a result of the presence of sound rather
than the differing food type. Additionally, in all treatments with statistically significant GUD
difference, the data shows that GUD is higher when sound is present compared to when sound is
not present.
Because squirrels gave up a statistically significant greater amount of food in the presence
of red-tailed hawk sounds, we can infer that squirrels use the presence of predator sounds as an
indicator of a predation threat. Therefore, they choose to sacrifice food when they detect a threat

9
while foraging. As a result, we can accept the hypothesis that grey squirrels will use auditory
cues to determine predation risk when foraging, reducing the time spent foraging at a site if they
hear sounds of a common predator nearby because the risk of getting eaten is greater than the
risk of losing a portion of food they can potentially collect. This conclusion supports the findings
of Schmidt and Ostfeld (2008), who found that squirrels listen for potential threats while
foraging, influencing the foraging behavior of the squirrel.
Because auditory cues appear to influence squirrel GUD, these results also indicate that the
direct predator cue of sound has a negative effect on squirrel foraging behavior, with the
presence of sound decreasing the percentage of food foraged by the squirrels. While Orrock et al.
(2004) found that indirect predator cues have a greater effect on GUD than direct predator cues,
our study shows that the direct predator cue of sound does play a significant role in increasing
the GUD of foraging squirrels.
Additional data indicated squirrels did not have a preference for sunflower seeds or peanuts
when foraging regardless of the presence of sound. Therefore, we must reject the hypothesis that
grey squirrels will minimize handling time when foraging and accept the null hypothesis that no
difference in GUD exists between squirrels foraging for sunflower seeds versus squirrels
foraging for peanuts. In other words, there does not appear to be a perceived difference in risk
when squirrels forage for peanuts versus when squirrels forage for sunflower seeds. This result
goes against the Kotler et al.’s (1999) findings that higher GUDs are present when food
resources cannot be cached by the forager because we found no significant effect of food type on
GUD. Therefore, further research should be done on this topic to determine if other aspects of
food type affect the GUD of foragers.
Although steps were taken to reduce confounding variables in this study, some limitations
to the results must be addressed. The most significant limitation of this study was the inability to
isolate squirrels as the foraging species. It is likely some of the food in the trays during trials was
taken by other species, especially songbirds. In fact, a blue jay was seen taking sunflower seeds
on two occasions, indicating that the food taken from the trays was not only taken by squirrels.
While attempts were made to limit the amount of food taken by other species, we were wary of
scaring away potential squirrels and adding human presence near the trays as a confounding
variable to the results and limited our interaction with the trays as much as possible during trials.
Furthermore, while the location of the tray sites was isolated, it was located in a public park,
potentially introducing the confounding variable of human influence on squirrel GUD. However,
the trays were away from walking trails, so this variable was mostly accounted for. It is also
important to note that the speaker placement may have served as a potential limitation to the
study. We were only able to place the speaker slightly high in a nearby tree, but the placement
was much lower than a red-tailed hawk would most likely sit in a tree. This may have influenced
the GUD of the squirrels, especially if the sound came across as close to the trays or very loud. A
final limitation to this study was the differing weather conditions between trials. Most
significantly, Trial 4 was conducted during a rainy, colder day, and while it does not appear to
have affected the results, there is a chance this weather difference served as a confounding
variable.
Future research in this area could focus on the effects of different predator calls on squirrel
GUD to see if squirrels rank sounds of certain predators as being higher risks than other predator
sounds. This study could also be done using unfamiliar predator sounds to see if squirrels
perceive the unfamiliar animal call as a risk, thereby increasing GUD, or choose to ignore the
unfamiliar animal sound. Additionally, future research could focus on comparing the GUD of

10
squirrels in the presence of different direct predator cues, such as sound, smell, and visual
imagery of the predator, to see if certain cues are determined to be greater risks to the squirrels
than other cues. Finally, future research could examine the effect of the frequency or intensity of
a given sound on squirrel GUD. Continuing research in this area will help to further understand
the behavior of squirrels when foraging and how potential perceived risks can influence the
amount of food left behind at a given foraging patch.
Investigating the foraging behavior of other organisms within forest ecosystems would also
help to develop a more robust understanding of community dynamics within these ecosystems.
For example, understanding songbird foraging behavior in addition to squirrel foraging behavior
could provide insights into competition between two species, as well as how detriments to one
species may indirectly benefit or be a detriment to another species in the community. The
methodology used in this study could also be expanded to other ecosystems, aiding in the
continued growth of our understanding of foraging behavior
Furthermore, continuing investigation into foraging behavior can help influence
conservation plans and provide a more thorough understanding of environmental ecosystem
processes. Understanding how energy flows through ecosystems and under what conditions
organisms may be more likely to sacrifice potential energy gains is vital to understanding how a
given community functions. This knowledge can then be used to influence ecosystem
conservation plans, including which species populations should be introduced, monitored, or
kept in check. Because the behavior of organisms in an ecosystem ultimately affects the
functioning of that ecosystem, understanding the foraging behaviors of animals is vital to
preserving the natural world.

Acknowledgements
We would like to thank Layne Gaynor for providing guidance on this project, as well as
helping to develop our hypotheses and methods. Furthermore, we would like to thank Kyle
Sozanski for his help with editing and revisions to our paper, as well as the anonymous peers
who provided feedback. We would also like to thank all team members for attending data
collection trials, helping to analyze and interpret data and results, and contributing to writing and
editing the final paper. Notably, Natalie contributed much to the writing of the final paper, with
help from Michael in both writing and editing, while Eddie and Alysa aided in some writing,
editing, and inserting tables and figures.

Literature Cited
Bowers, M.A. and B. Breland. 1996. Foraging of gray squirrels on an urban-rural gradient: Use
of the GUD to assess anthropogenic impact. Ecological Applications 6:1135-1142.
Bowers, M. A., J. L. Jefferson, M. G. Kuebler. 1993. Variation in giving-up densities of
foraging chipmunks (Tamias striatus) and squirrels (Sciurus carolinensis). Oikos
66:229-236.
Fitch, H.S., F. Swenson, D.F. Tillotson. 1946. Behavior and food habits of the red-tailed hawk.
The Condor 48:205-237.
Kottler, B.P., J.S. Brown, M. Hickey. 1999. Food storability and the foraging behavior of fox
squirrels (Sciurus niger). The American Midland Naturalist 142:77-86/
Newman, J.A., G.M. Recer, S.M. Zwicker, T. Caraco. 1988. Effects of predation hazard on
foraging “constraints”: Patch-use strategies in grey squirrels. Oikos 53:93-97.
Orrock, J.L., B.J. Danielson, and R.J. Brinkerhoff. 2004. Rodent foraging is affected by

11
 indirect, but not by direct, cues of predation risk. Behavioral Ecology 15:433-437.
Schmidt, K.A., and R.S. Ostfeld. 2008. Eavesdropping squirrels reduce their future
value of food under the perceived presence of cache robbers. The American Naturalist
171:386-393.
The City of Columbus. 2021. Tuttle Park. Columbus Recreation and Parks Department.
https://www.columbus.gov/recreationandparks/parks/Tuttle-Park/.
Venkataraman, T.S., T.S. Kraft, N.J. Dominy, K.M. Endicott. 2017. Hunter-gatherer residential
mobility and the marginal value of rainforest patches. National Academy of
Sciences 114:3097-3102.

Appendices

Appendix 1

Red-tailed hawk noise 1:

https://open.spotify.com/track/5UxxVprd515nA72cETcctI?si=vMMjaovNTWOHHdqL-
twFPQ&context=spotify%3Aalbum%3A5PbEY5NamEcNPVHVJmbGaP

Red-tailed hawk noise 2:

https://open.spotify.com/track/41uUHSJdoPUT76lZxiKgeo?
si=PSzmnMLLT5yrLkjnlQ0TZg&context=spotify%3Aalbum
%3A5PbEY5NamEcNPVHVJmbGaP

Appendix 2

Appendix 2 Table. Data collection from all 5 trials.

Treatment Weather Initial Final Mass GUD (final
Group Conditions Mass (g) (g) mass/initial
mass)
Trial 1 Peanuts Temperature: 100 94 0.94
(9/19/21 w/sound 86℉
@2:00pm) Peanuts 100 69.1 0.691
w/out sound Weather:
Sunflower sunny, humid, 100 81 0.81
seeds w/ bright
sound
Sunflower 100 69.6 0.696
seeds w/out
sound
Trial 2 Peanuts Temperature: 100 94 0.94
(10/1/21 w/sound 75℉
@3:25pm) Peanuts Weather: 100 72 0.72

12
 w/out sound sunny, clear
Sunflower skies, low 100 96 0.96
seeds w/ humidity,
sound moderate
Sunflower brightness 100 66 0.66
seeds w/out
sound
Trial 3 Peanuts Temperature: 100 82 0.82
(10/3/21 w/sound 73℉
@2:30pm) Peanuts Weather: 100 67 0.67
w/out sound cloudy, gray,
Sunflower damp, rainy, 100 91 0.91
seeds w/ cool
sound
Sunflower 100 65 0.65
seeds w/out
sound
Trial 4 Peanuts Temperature: 100 92 0.92
(10/6/21 w/sound 66℉
@4:45pm) Peanuts Weather: 100 52 0.52
w/out sound rainy, humid,
Sunflower grey but 100 89 0.89
seeds w/ bright, warm
sound
Sunflower 100 66 0.66
seeds w/out
sound
Trial 5 Peanuts Temperature: 100 95 0.95
(10/10/21 w/sound 81℉
@2:20pm) Peanuts Weather: 100 81 0.81
w/out sound sunny, partly
Sunflower cloudy, low 100 82 0.82
seeds w/ humidity,
sound moderate
Sunflower brightness, 100 52 0.52
seeds w/out windy/breezy
sound

Appendix 3

Appendix 3 Table. Weather conditions for all 5 trials

Trial Weather Conditions
1 Temperature: 86℉
Weather: sunny, humid, bright

13
2 Temperature: 75℉
Weather: sunny, clear skies, low humidity,
moderate brightness
3 Temperature: 73℉
Weather: cloudy, gray, damp, rainy, cool
4 Temperature: 66℉
Weather: rainy, humid, grey but bright, warm
5 Temperature: 81℉
Weather: sunny, partly cloudy, low humidity,
moderate brightness, windy/breezy

14