Foraging Ecology of Grey Squirrels (Sciurus carolinensis)

Developed by Chris Paradise

• Return to Table of Contents
• Return to Biology Department Home Page

INTRODUCTION

A behavior is a response to an external or internal stimulus.  All behaviors are dependent upon nerve impulses, hormones, and other physiological mechanisms that you may have encountered in this course.  Behavior is intimately linked to the environment because it is the way that organisms interact with their surroundings.  The environment consists of everything an organism comes in contact with, including food that it has to find and/or capture, natural enemies that want to use it as food, competitors for food and shelter, and climatic conditions such as temperature and precipitation.

            In every one of these cases, you can identify a behavior that some animal exhibits in response to a stimulus.  For example, on a cold, sunny morning a butterfly will rotate its body so that its wings are perpendicular to the position of the sun.  In this way it absorbs more heat and can commence its daily activities earlier in the day than if it hadn't rotated.  What other examples of behaviors in response to environmental stimuli can you think of?

            Ecology is the study of organisms and their relationships with the environment, be it the abiotic environment or all the other organisms with which they interact.  More specifically, ecologists attempt to understand why organisms are found in specific locations at various frequencies (the distribution and abundance of organisms).  These two key factors of ecology are in part determined by the climate and how well the organism is adapted to living under those climatic conditions.  Distribution and abundance are also determined by the presence or absence of other organisms, and whether they are prey, predators, competitors, potential mutualists or mates of the organism in question.

            Behavioral ecology is the synthesis of these two scientific disciplines, and researchers working in this field are mostly interested in why, or if, it is adaptive for an animal to exhibit certain behaviors, considering the ecological backdrop of its environment.  Behavioral ecologists generally speak in terms of maximizing fitness, or reproductive success, which is the number of offspring that survive to reproduce.  Behavioral and other adaptations are the result of natural selection on the species in question and are integral for success because they allowed its ancestors to have higher reproductive success.  In reality, fitness cannot always be maximized because the environment is constantly changing, and natural selection of adaptations lags behind environmental changes.  Therefore, behavioral ecologists seek to identify the environmental agents that caused the evolution of particular behaviors and why those behaviors assist in maximizing the reproductive success of the individual.  In a nutshell, behavioral ecology seeks to understand how behavior allows an organism to make the most of what its got in a changing environment.

            In this exercise we will study grey squirrel (Sciurus carolinensis) behavior as it relates to finding food in an energy-efficient way and how foraging strategy is affected by the risk of getting caught by predators or the groups size of foraging squirrels.  All animals, squirrels and humans included, must expend a certain amount of energy and time searching for food.  The net balance of this food acquisition must be positive; they must gain more energy from the food they eat than they expend in its search.  This idea encapsulates foraging efficiency or optimal foraging.  This relates to a squirrel's reproductive success because the more positive the energy balance is, the more energy the squirrel will have when mating (so they can copulate with more than one partner, or spend more time searching for the best partner to mate with).

            All animals, except for those at the very top of the food chain, such as humans and top carnivores, have predators for which they must be wary.  The risk of getting caught by these predators depends upon how vigilant the prey is and where they are in relation to shelter from the predators.  It also depends on how many predators are in the area, how large of a group the individual prey is in, and how vigilant the entire group is.

            Shelter affords protection to a prey animal.  For squirrels, shelter is usually the cover of trees.  Because they often have to venture out from the woods to look for food, they increase their predation risk from predators such as hawks, foxes, cats, dogs, and other carnivores.  A squirrel is much better able to evade a predator when it is in a tree; some predators can't climb, and others aren't as fast as squirrels in the trees.  So as squirrels venture further away from shelter they must become more vigilant (i.e., they must keep a sharper eye out for predators).  In addition, if there is a large supply of food available, squirrels will often forage in groups.  This increases vigilance because more eyes can look in more directions for potential threats.  It also decreases individual risk, because the larger the group, the lower the probability that any individual will get attacked.

            Recall that maximizing fitness, or reproductive success, might be correlated with the amount of food an animal eats.  This might be true for a squirrel as it travels around its home range searching for food.  It seems pretty obvious what might happen to a squirrel's fitness if it should happen to be caught by a predator.  However, what might happen to a squirrel’s foraging efficiency if it has to forage with other squirrels?  Or conversely, eat out in the open by itself?  Now we have identified some problems, called trade-offs, and squirrels must make decisions about where to eat and with whom.  If a squirrel, or any forager for that matter, didn't have to worry about predators, then food could be eaten wherever it is found.  This makes more sense than carrying it to safety, which would cost energy, or avoiding that food source altogether.  Consider a squirrel that finds a large supply of food in an open field.  A trade-off might be for it to decrease its energy gain by spending time being vigilant, or looking for predators, thereby increasing its safety.  Another way to increase safety is to forage in a group.  This still presents a cost to the forager, as aggressive interactions over food increase, and individuals still must be vigilant. The latter cost decreases, though, as foragers can take turns being vigilant.  Small energy costs add up over time, but consider the alternative of being caught by a predator!

            Our objectives are to test the foraging efficiency of grey squirrels under conditions of varying predation risk and group size.  We will use sunflower seeds placed at different distances from trees, and we will vary patch size by varying the number of piles of seeds.  We will then record the distribution of squirrels with respect to patch size and distance, and measure aggressive interactions and number of seeds eaten per minute.  You should make predictions as to your expectations regarding aggressive interactions and foraging efficiency as squirrel density, distance to cover, or patch size increase, close to cover.  Foraging efficiency will be measured as seeds eaten per minute of haphazardly selected squirrels.

WEEK 1 – MATERIALS:

• Clipboards and data sheets                                                                  
• Sunflower seeds
• Small pans glued to plywood boards                                                    
• Tape measures
• Timers (2 per group)

WEEK 1 – PROCEDURES:

1.         Your class will break up into groups of two to three, each group being responsible for collecting data at one patch and one distance.  See Figure 1 for a schematic of all the treatments.  If your class has more than four groups, you should repeat some of the treatments at various locations on campus, as determined by your instructor.  One person in each group will be responsible for measuring and relaying the feeding rates to the data recorder, who will have the clipboard and data sheet, and the third person (or second, if only two) will record group size every minute and any other observations of squirrel behavior, such as aggressive behavior.

2.         The main qualitative predictions that we will discuss turn out to be fairly apparent and intuitive. Newman and Caraco (1987) performed the original research on the problem we are considering here, and a detailed account of the experiment and the mathematics involved can be found there.  What we want to think about and discuss are the following four variables:

            a.         Squirrel feeding rate:  We will use sunflower seeds as the food source, because squirrels should eat them as they find them, rather than caching or carrying them away.  Our observations will be recorded as the number of seeds eaten per minute by any one squirrel.  For that squirrel, also note the group size while that squirrel was being recorded.  If the squirrel leaves the patch before a minute is up, record the number eaten and the number of seconds so you can later convert it to seeds/minute.  If group size changes during the minute, as squirrels come and go, record the average group size for that minute.  Observations can be taken more than once for a single squirrel; if a squirrel stays for five minutes, you can record up to five measurements of foraging rate.  If there is more than one squirrel at your designated patch, select one focal animal to watch, and then others, collecting as much data as you can.

            b.         Group size:  The size of the group is also an important variable.  The squirrels would most likely not want to be in a group size that is larger than the number of feeding sites within a patch, although that expectation may change as distance to cover changes.  How do you think that might be so?  How might group size relate to both patch size and distance to cover?

            c.         Distance to cover:  Even though the predation risk increases with distance to cover, the energetic and time costs to the squirrel in vigilance and travel time also increase.  We will vary the distance by placing small pans with the sunflower seeds in them either 5 or 15 m from tree cover (Figure 1).  Given food items of the same size, then, would you expect the feeding rate to increase or decrease as distance to cover increases?  Remember that we are manipulating this variable; it is not being measured.

            d.         Patch size:  The number of small pans at each location is our patch size, which may be 1 or 4 pans.  This is another variable that we are manipulating, and your group will be responsible for one patch size.  As patch size increases from 1 to 4 pans, consider how that might affect group size, regardless of distance to cover.  Then consider how patch size and distance to cover may interact to affect group size and feeding rate.

            e.         We can quantify the effects of distance to cover, patch size, and group size on the feeding rate of squirrels by varying two factors and letting the squirrels vary the other.

Figure 1.  Schematic of field set up.  The placement of patches is random with respect to placement of other patches, and should be more spread out than depicted here.

3.         During the week of this lab, we will not meet at our regularly scheduled time.  Instead, to take advantage of when squirrels are most active, we will meet at 7 a.m.  You have the option of when you will attend, but you must attend one day during the week, from 7 to 8:30 a.m.  When the class goes outside, set up your patch at the appropriate distance and put sunflower seeds in each pan.  Then move back behind cover and watch for squirrels.  Take observations for as long as you can.  At 8:30, pack up your patch, bringing your seeds with you.  We don’t want to leave seeds behind, as that may bias observations of your classmates coming after you.

            a.         Measure the distance from cover with a tape measure.  These distances are to the nearest cover (= tree).  Each group will have to set up their pan, and after that all observers should back off a safe distance from the pans.  Don't crowd the tree line where squirrels may come from.  You may have to wait 15 minutes or more for the first squirrels to show up, or given the nature of fieldwork, they may not show up at all.

            b.         Do not disturb the squirrels!  You may find that squirrels on campus will walk right up to you.  Don't feed them.  Ignore them, as they may not go over to the pans if you disturb them.  Once they find the pans, they will most likely stay near there.  Above all, you need to exercise patience while conducting this experiment.

4.         After you tabulate your data, calculate means and standard deviations of your data as instructed in Table 2, and turn your data sheets in to your instructor.  The Biology 112 faculty will compile all data together for analysis in the second week.  Prior to that, you will receive a sheet with all data for you to examine.  The section at the end of this chapter explains the data analysis.

WEEK 2 – OBJECTIVES:

1.         Review the statistical analysis to be used with your instructor.  Analyze the class data, using the appropriate test statistic.

2.         Discuss the results with the class.  Pay particular attention to deviations from your predictions and why those deviations may exist, and also to variation between squirrels.

WEEK 2 - MATERIALS:

Data summary sheets                                        Computers with statistical analysis program

WEEK 2 - PROCEDURES:

1.         Your instructor will review the statistical methods to be employed.  Perform your statistical analysis as a class or as a group.

2.         Discuss these questions with your group, and address them as they relate to your results:

            a.         Were your results similar to your predictions?

            b.         Did our experiment include a distance from cover at which squirrels did not go?

c.                   How do distance from cover and patch size interact?

d.                  What effect does distance from cover or patch size have on group size?  Did one variable have a greater effect than the other?

e.                   Discuss why there might be deviations from the predictions, including how individual variation of squirrels might play a role in the outcome of the experiment. 

DATA ANALYSIS: REGRESSION AND ANALYSIS OF VARIANCE

            Because there are random fluctuations in sample observations, we must use statistical tests of probability to analyze the data.  We ask the question, "what is the probability that our data fit our null hypothesis (the hypothesis of no difference)?"  If the probability is low (usually 5% by convention, denoted as P < 0.05), we reject the null hypothesis.  We then conclude that our results are statistically significantly, and support our alternative hypothesis, the hypothesis that there is a treatment effect.  We risk making a mistake if we reject the null hypothesis when it is true (called a Type I error), or by accepting the null hypothesis when it is not true (Type II error).  The point is that in statistical analysis, we do not deal with absolute answers of yes or no, but of relative answers of probably, possibly, or probably not.  Because of the way the scientific method works, we can never prove that something is true, and only eliminate other possible causes.

            There are two (or more) possible ways to analyze the squirrel data.  One is with a regression analysis, which examines the relationship between a dependent variable and an independent variable.  The other is analysis of variance (ANOVA), which examines differences among mean responses (and variation around the mean) for all treatment groups.  Let’s examine regression first.

            Regression can easily be done using Excel or JMP, depending on what your instructor specifies.  Linear regression assumes that the data were randomly collected and independent observations.  This may not always be true in our case, if you collected multiple observations of feeding rates of the same squirrel.  However, for our purposes we will assume that each observation is independent.  Linear regression also assumes, or tries to fit, a straight line to the data set in question.  In other words, it specifies a regression equation that has a constant slope, such that feeding rate changes at a constant rate as group size, patch size or distance to cover change (depending on which one you’re looking at).  The best way to determine whether linear regression will work for your data set is to make a scatterplot of your data and visually inspect the graph.  This is often the best way to start any data analysis: make various plots and examine trends in the data.  If you (and your instructor) decide that regression is the way to go, you can perform individual regressions of the dependence of feeding rate on: 1) distance to cover, 2) patch size, and 3) group size.  Regression is a desirable approach to determine whether a relationship exists between feeding rate and group size, but distance to cover and patch size may also be analyzed with analysis of variance.

            If you don’t have a linear relationship between feeding rate and either distance to cover or patch size, you may want to use t-tests or analysis of variance (ANOVA).  The two are related, although t-tests are easier to perform and grasp conceptually.  Basically, these statistical tests determine the probability that two (t-test) or more (ANOVA) means are drawn from the same population (no treatment effect) or different populations (treatment effect).  The null hypothesis is that there is no difference among treatment combinations, and the probability determined from the test is the probability that the two samples are drawn from the same population.  A probability of 0.05 is our cutoff; a probability equal to or less than that indicates that there is 5% or less chance that the two samples are from the same population.  Thus, we conclude a treatment effect. 

            The advantage of ANOVA over t-tests is that we can test more than two means at one time, and we can determine effects of more than one factor and any interactions between them.  For instance, we can test the effects of both patch size and distance to cover simultaneously, as well as determine if there is an interaction between them.  An interaction occurs when Factor A has a strong effect at one level of Factor B and little or no effect at another level of Factor B.  For instance, distance to cover might have a strong negative effect on feeding rate, such that feeding rate declines as distance increases, but only in small patches.  In large patches, there might be no change in feeding rate as distance to cover increases. That’s an interaction.  Whether you use t-tests or ANOVA, both are easy to perform on Excel or JMP, and your instructor will give you a brief tutorial on all statistical methods to be employed.

            There are many advantages and disadvantages of each statistical method, and scientists often exhibit preferences for one over another.  Also, it is wise to incorporate statistical analysis in your experimental design, although inspection of the data after it’s collected might lead us to change our originally specified statistical test.  I usually design my experiments so that they can be analyzed using analysis of variance, because I like to test for interactions among two or more factors.  However, that doesn’t preclude me from using regression analysis after the experiment is completed.  We’ve designed this experiment so that you can use one of these two methods, and not other methods with which you may be familiar.

REFERENCE:

Newman J.A. & Caraco T. (1987) Foraging, predation hazard and patch use in grey squirrels.  Animal Behavior 35:1804-1813.

• Return to Table of Contents
• Return to Biology Department Home Page

© Copyright 2001 Department of Biology, Davidson College, Davidson, NC 28036
Send comments, questions, and suggestions to: chparadise@davidson.edu