Lab 7 Mimicry¹

Spring, 1999

¹Adapted from "Mimicry", by Ernest Williams, Hamilton College. This lab excersize is an Ecological Society of America Lab, printed in Experiments to Teach Ecology, 1993, edited by J. M Beiswenger, and published by the Ecological Society of America

Email me

Back to:

Introduction:

Mimicry is a resemblance in appearance between two organisms. There are many types of mimicry. Here, we will focus on two forms of mimicry: Batesian and Mullerian. In Batesian mimicry, the model is the inedible or unpalatable prey, and the mimic is an edible or palatable prey organism that resembles the model. Most cases of Batesian mimicry involve mimic and model that are different species, but there is no reason to expect that intraspecific mimicry does not also occur, although this might require linkage among defense and advertisement loci. Prey (or plants) that are potentially harmful to their predators (or herbivores) through chemical or other defenses must advertise this ability to the prey. If they do not, the defense may not stop the predator from harming or killing the prey before the defense is discovered by the predator. This would obviously reduce the value of the defense to the prey. The form that the advertisement takes will depend on the means by which the predator detects its prey: distinctive color for visual predators, odors for predators that use chemical detection, etc. As is the case with advertising in broadcast or print media, the success of honest advertisements makes it profitable to engage in false advertising, where one gets the benefit without bearing the cost (= the materials and energy invested in the defense, in the case of prey or plants). False advertising in the natural world is called Batesian mimicry.

There are several factors which affect the value of mimicry as a strategy for avoiding predation. First, the predator must be able to learn to recognize the advertisement of the defended prey. If this is not so, then the model will not be successful and the mimic has no reason to be a mimic, as there is no protection from predation to be had. When the predator is capable of learning to avoid the prey, mimicry may be possible. However, the value of mimicry will depend on:

1. the proportion of mimic and model
2. the degree of distastefulness of the model
3. the degree to which the mimic resembles the model.

In order to imagine how these factors might effect the situation, one should imagine that the predator must occasionally interact with models in order to learn and then remember that they are to be avoided. An encounter between a predator and mimic will encourage the predator to eat organisms displaying the advertisement, which is counter-productive for both models and mimics. When the proportion of mimics rises, fewer predators should avoid both models and mimics. If the model is highly toxic or distasteful, fewer encounters should cause the predator to avoid both mimic an prey for longer periods. When the model and mimic appear different to the predator, it may be able to avoid the model and eat the mimic.

Mullerian mimicry is not really a case of false advertising, as all of the mimics are inedible or distasteful. Here, the mimics use one another as models, and all come to resemble one another, although they may be of different species. The value of mimicry in the case of Mullerian mimicry comes from the behavior of the predator. If the inedible prey are rare, predators may not learn that the prey's distinctive color or odor means that they are inedible. If the predators do not know to avoid the well-defended prey, his makes both the defense and the advertising useless. The advertising may even be harmful when it makes the prey more conspicuous to the predator. However, if the problem is that each inedible or unpalatable species is too rare, then the problem may be overcome if more than one inedible or unpalatable species look alike, which leads to Mullerian mimicry.

In this laboratory exercise, we will test the ability of local birds to feed on mimics and models as their frequencies vary, while keeping palatability and similarity of mimic and model constant. This experiment was first performed by O'Donald and Pilecki (1970). The prey will be artificial food that we will expose to the birds on standard feeding trays (which we will construct). Prey of three colors will be presented: one color will be all palatable prey, one will be a mixture of 3/4 palatable and 1/4 unpalatable prey (many mimics, few models), and the last will be only 1/4 palatable prey (few mimics, many models). We will see if these situations affect the feeding of the birds in the expected ways.

Objectives:

1. Learn about Batesian mimicry.
2. Observe the feeding behavior of the birds and determine if they learn about their prey.
3. Measure the effect of the ratio of mimics to models on feeding behavior.
4. Learn to conduct an experiment and make inferences from the data collected.
5. Learn to use graphs to support the inferences made.
6. Use a statistical test to evaluate the results of an experiment.

Materials:

Cookie Presses	Wax Paper	Labeling Tape
White Flour (5 lb)	Sunflower Seeds	Rulers
Lard (2 lb)	Petri Dishes	Thermometer
Food Coloring	Top Loading Balance	Marking Pens
Plastic Spoons	Weighing Trays	Paper for Feeding Array Charts
Plastic Knives	Disposable Gloves
Quinine Sulfate	5 plastic mixing bowls (1 lg, 4 small)

Overview of the Experiment:

We will assess the effect of mimicry by allowing birds to choose among three colors of artificial prey. Unknown (at first) to the birds, two colors will be a mixture of unpalatable (model) and palatable (mimic) prey. We will measure the bird's preferences by putting out prey twice a day (once in the early morning and once in the early afternoon), allowing the birds to feed unimpeded, and recording the number of prey items within each category eaten prior to putting out the next batch of prey.

Preparing the Prey:

Choice of color:

We have a limited choice of colors: yellow, blue, green, red, orange (red + yellow), and purple (red + blue). Two things should be kept in mind when assigning the colors to be used by the class (we will all have to use the same). First, the colors should contrast so that the birds will have no difficulty in choosing among the prey. Second, the colors should be coordinated with the range of colors normally found in the area at the particlular time of year during which the laboratory is conducted. Colors are used in nature as signals (warning colors, like the orange of monarch butterflies, or attractive colors, like the red of ripened dogwood berries) or as a part of a cryptic strategy (green often blends into a background of leaves in the Spring and Fall). Think of the things that birds might eat when you are doing the experiment and what colors they are and make the choice of prey colors from among the possibilities above.

In the spirit of truth in experimentation, you should be aware that we will have confounding factors in this experiment. Note that each of the colors will have a different proportion of unpalatable prey (0%, 25%, and 75%). Suppose that you make the yellow prey the completely palatable prey. If you find that they are preferred by the birds, was it because they were not bitter tasting, or do birds simply have a strong preference for yellow prey? You can't tell because prey color and prey palatability are confounded. If we had time, we could eliminate this problem. We could, for instance, perform the experiment more than once and switch the color used for each type of prey. Alternatively, we could conduct a separate experiment to test the color preferences (if any) of birds for prey and perhaps eliminate color preference as a potential factor. We have less time so we will assume here that the local birds have no preferences for the colors we are using.

Making the Prey:

1. Mix the dough. Before mixing the dough, put on disposable rubber gloves. There are two reasons for this. First, it will keep you from dyeing your hands. Second, it will prevent you from mixing whatever you have on your hands into the dough. Keep the gloves on throughout the prey-making process. Put 2 lbs of lard into the large mixing bowl and add 1.6 times the weight of the lard in white flour. Mix thoroughly.
2. Divide and color the dough. Divide the dough into three equal portions and place each in a different bowl (the large bowl can be one of these). Add food coloring to each (depending on the color choices made earlier) until you reach a bright, strong color in each. Mix until each dough is evenly colored throughout.
3. Make some of the dough unpalatable. We will use quinine sulfate, a bitter flavoring, to make the dough unpalatable. It will not harm the birds (it may even protect them from avian malaria). Ask me why and I will tell you why the gin and tonic was the favorite drink of the British Raj. Separate out approximately 1/4 of the dough colored for the 1/4 model to 3/4 mimic treatment and 3/4s of the dough colored for the 3/4 model / 1/4 mimic treatment and place these into separate bowls. Weigh each of the lumps slated to become unpalatable. Add 0.75% of the weight of the dough in quinine sulfate and mix thoroughly. Record the weights of the lumps and the amount of quinine added. Use the labeling tape to label each bowl (now numbering 5) with what it contains.
4. Make strands of dough. Clean and wash a cookie press. This is a necessary step whenever going to a different category of dough. Place some dough into a cookie press. Choose a disk with small, round holes in it and press out spaghetti-like strands (approx. 3 mm in diameter) of dough onto a piece of wax paper. Clean out the press when you are done with it.
5. Cut the strands into caterpillars. Using the plastic knives and rulers, cut the strands into 15 mm pieces. Do this only for one dough type at a time and clean the knife before going to another dough type.
6. Store the caterpillars for each feeding separately. Each feeding will require 5 petrie plates, labeled with the dough type it contains. Place the following number of prey in each:
   • Dish 1 -- 32 pieces of palatable prey
   • Dish 2 -- 8 pieces of the model prey from the 1/4 model/ 3/4 mimic dough
   • Dish 3 -- 24 pieces of the mimic prey from the 1/4 model/ 3/4 mimic dough
   • Dish 4 -- 24 pieces of the model prey from the 3/4 model/ 1/4 mimic dough
   • Dish 5 -- 8 pieces of the mimic prey from the 3/4 model/ 1/4 mimic dough

   Add a couple of extra pieces of the proper type to each dish in case accidents occur when placing the prey on the feeding tray.

7. Store the dough in a refrigerator. This will inhibit mold growth and will make the prey pieces easier to handle.

Preparing Sampling Arrays:

Overview of Sampling Array:

The birds must be presented with a standardized amount of prey at each feeding so that feedings can be compared with one another. We will use a grid of 100 positions (in a 5 x 20 square array) for this purpose. The grid has already been place in a convenient place on campus and filled with sunflower seeds to condition the local birds to feed there. However, we can't put the same kind of prey at each location in the grid each time. The birds might then be able to tell which prey were mimics and which were models: something they should not be able to do. So, we must randomize the placement of the five kinds of prey on the feeding table. The procedure below describes a method for doing this. We will plan the array for each feeding before starting the experiment.

Overview of Sampling Array:

Prepare the Array Sheets. Those not involved in making the dough will make the sampling arrays used to place the prey onto the feeding tray at each feeding. We will do this twice a day for five days, and so we need 10 sampling arrays. Each array will be a 5 by 20 cell table, so draw a 5 by 20 cell table on a piece of paper.

Record the symbols use for each prey type and the range of each. We will have to standardize the symbols for each prey type, so that whoever reads the sampling array will know what they mean. We will do this as a group for each prey type and you should record the symbols below.

• All palatable prey
• Models from the 1/4 model/ 3/4 mimic prey type
• Mimics from the 1/4 model/ 3/4 mimic prey type
• Models from the 3/4 model/ 1/4 mimic prey type
• Mimics from the 3/4 model/ 1/4 mimic prey type

Use these symbols to fill in the sampling array. We also need to know how many of each prey type there should be. There are 100 cells on the feeding tray. The prey should be in the following ratios:

• One third (4/12ths of total) should be palatable prey.
• One third (4/12ths of total) should be the 1/4 model/ 3/4 mimic dough.
  • One fourth (1/12th of total) of these should be models
  • Three fourths (3/12ths of total) of these should be mimics
• One third (4/12ths) should be the 3/4 model/ 1/4 mimic dough.
  • Three fourths (3/12ths of total) of these should be models
  • One fourth (1/12th of total) of these should be mimics

The only way to divide 100 cells into the right proportions is to make 8 prey 1/12 of the total. This means that the total is 96 cells (=8 x 12) on the feeding tray and four unoccupied cells.

Read the random numbers from the random number table. We need to know where each of the 96 prey should go. We can use a random number table to do this. Your instructor will describe a random number table and hand out a page of random numbers. You will pick (as randomly as possible) a place to start and read out the random numbers in groups of 2. These two digits can be used to decide which type of prey belongs in a particular cell according to the key below:

• 00 to 31 -- Palatable prey
• 32 to 39 -- A model prey from the 1/4 model/ 3/4 mimic dough
• 40 to 63 -- A mimic prey from the 1/4 model/ 3/4 mimic dough
• 64 to 71 -- A model prey from the 3/4 model/ 1/4 mimic dough
• 72 to 95 -- A mimic prey from the 3/4 model/ 1/4 mimic dough
• 96 to 99 -- No prey

Assign each cell of the array a type of prey. As you read a pair of random digits from the table, start at the upper left hand corner of the sampling array sheet and put the symbol corresponding to the prey type (determined by the random numbers) into the cell. Keep a running count of the number of each prey type used. Proceed from left to right, as though you were reading. Do not record the digits, they are used only to make the decision about which prey to put into the cell on the chart. It is OK to use the same pair of digits more than once if they happen to occur more than once in the random number table. When you reach the maximum number of any prey type, it becomes ineligible for future cells and so you just ignore any random digit pairs that fall in that prey type's range and keep reading until all prey types become ineligible. At this point you will have filled in all of the 100 cells.

Generating hypotheses:

Because you did not design this experiment, you will have to think about it a bit before you will really understand it. To encourage this, you should stop here and try to predict the outcomes.

1. Rank the color prey in order of the bird's feeding preference.
2. Which palatable prey type will be the least protected from predation by the birds?
3. Which palatable prey type will be the most protected from predation by the birds?
4. How will you decide if mimicry is an effective strategy for avoiding predation?
5. How will you decide if the relative proportion of mimics and models is important to the success of mimicry?

Gathering the Data:

Assingment of personnel for each feeding:

During the lab period, we will assign pairs of students to set out prey and collect data at particular times. Remember that the data collected will be used by everyone in the class.

Feeding the birds:

When it is your turn to feed the birds, come to my laboratory and obtain a sampling array sheet and the prey. Take the materials to the feeding tray and place the prey on the table in the pattern specified by the sampling array. In the morning, you may have to obtain the needed materials from Brian Quarles, whose lab is next to mine (304 Harned Hall).

Data to record:

After a minimum of three hours and a maximum of four hours, return to the feeding tray and simply count the number of each prey type remaining on the tray. Record the results on the sampling array, date the array and not the temperature and weather (clear, light rain, etc.).

Data Analysis:

Gathering the data into a table:

Each day, the instructor will record the data taken on a chart posted on his office door. At the end of the sampling period, you must come by the office and make your copy of the data table.

Summarizing the data to support or refute a hypothesis:

Make a hypothesis for each of the questions below and construct a chart or graph that supports (or refutes) that hypothesis.

1. Does the intensity of predation change throughout the sampling period?
2. Was there any difference in morning and afternoon feeding intensities?
3. Did weather affect overall feeding intensity?
4. Is there a difference in the proportion of each color consumed?
5. Examine the first two and the last two days. Is there any evidence that the birds prey preferences are different at the beginning and end of the experiment (in other words, did the birds learn anything)?
6. Are mimics less likely to be eaten than are palatable prey that are not mimics?
7. Is there any evidence that the proportion of mimics affects the proportion of mimics eaten?

Using statistics to support or refute a hypothesis:

In the section above, you used visual presentations of the data to confirm or refute hypotheses. When there is a clear difference, the graph or chart may be enough to convince your audience that you have made your point. In most cases, this is not true and you have to have a more objective way of making a decision about the correctness of a hypothesis. For this, we most often use statistics. We will do that here for a specific hypothesis. However, statistics are normally designed to make a decision about the null hypothesis, not the alternative hypotheses you generated above. Examine the question and the alternative (H_a) and null (H₀) hypotheses below for illumination about the difference between the two.

If there was no effect of palatability, then we would expect that each of the color of prey would have the same proportion eaten. Make sure you see this point before going on (see me if you do not)! Let's calculate these proportions (remember to use data from the last two days only):

Calculate the proportion eaten by dividing the number eaten by the number set out. Notice the the proportions are not the same. Does this mean that they are really different, or is the difference just due to random chance. If you did the entire experiment over, would you expect to get the same results exactly? No, chance would alter the results. So, what we must ask is this. Is the difference in proportions observed above due to chance alone (an thus the null hypothesis is true), or is there a real difference in the preferences of the birds (and we accept the alternative hypothesis)? We will use a statistic called the Chi-square to decide this. This statistic measures the deviation of the observed results from the result expected if the null hypothesis were true. We have almost calculated the expected outcome. If there is no difference between any of the three prey colors, then the proportion eaten in each type should be the same as the total proportion of prey eaten. We can calculate this by dividing the total number of prey eaten by the total number of prey. Do this now.

The next step is to calculate the expected number of prey that should have been eaten from each of the types. This is easy. Just multiply the expected proportion times the number of prey set out (the first line above). Put these figures into the first row below and copy the actual number of prey eaten from above. Subtract the number observed from the number expected. This gives you the difference between observed and expected. This should be relatively small if what you observed was what you expected. However, we can't stop here, as we want to total the differences to get an overall difference. Because of the way you calculated the expected, you will see that the differences sum to 0. They always will. This would mean that we always accept the null hypothesis, no matter how large the individual differences are! To avoid this pitfall, we will square the differences, so that they are all positive and can't sum to 0 (unless there are no differences between observed and expected). Do the squares. We need to do another correction next. We need to correct these sum squared terms for the size of the data we are working with. If you do this experiment with only a few prey, say 100 or so, the differences should be in the ones or tens and the square should be no larger than 2 or 3 hundred. What if you had done the experiment over and over and had set out thousands of prey. Now the differences between observed and expected could be much larger (in the hundreds, say), even if random error was still the only source of error. Now, when you square them, the squares could be in the ten thousands, although there is no experimental difference except the larger number of prey in the second experiment. We need to correct for the actual size of the experiment done, This is done by dividing the squared differences by the expected number. Do this and sum the these figures. This sum is the Chi-square statistic for your data.

So, how do we use this number to make a decision about the null hypothesis. Well, think about it a bit. When the null (which lead to the calculation of the expected numbers) is true, there should be little difference between the expected and the observed, leading to a low Chi-square value. When the expectations are wrong, the Chi-square value should be larger. But where do we draw the line and say that the Chi-square value is too large? This is how we do it.

First, you must remember that we are deciding about the null hypothesis of no difference between prey types. If there is no difference among types, then the Chi-square value comes from random, uncontrolled error having nothing to do with the kind of prey. Statisticians have calculated how likely a particular Chi-square value is, given that the null hypothesis is correct and the calculations are published in a table (see below). As the Chi-square value goes up, it is less and less likely that you would get such a high Chi-square value and this is reflected in the table. You compare your Chi-square with the value in the table. However, before you can do this, you must decide how unlikely your Chi-square can be before you decide that it is too unlikely and that you will have to reject the null (remember, that is what produced the expected numbers). The standard cut-off point is 5%. This means that, if I should only get a Chi-square value as large as I actually got less than 5% of the time, then I will reject the null hypothesis (and, by default, accept the alternative).

You can now look up your Chi-square and make the decision to accept or reject the null. The table is at the end of the page. However, the table has two dimensions. Each row corresponds to a different degree of freedom (labeled d. f.) Why this? Well, not all experiments are the same. The Chi-square value will differ when the number of prey types changes (and, therefore, the number of expecteds you calculate). Here, you had three types, and you calculated three expected numbers of prey, but you could have had two or ten types. One would expect a lower Chi-square value when there are fewer types and a larger value when there are ten. So we need to look up the Chi-square value that corresponds to our experimental setup. This number is the degrees of freedom, which is the number of expecteds minus one (or 3 - 1 = 2 for us). Ask the instructor why you must subtract one. There isn't space here to do so.

What to hand in to the instructor:

Hand in your data table, the seven hypotheses you generated and the graphs or charts you made to support them. Tell me, below each chart, whether the data supports the data or not. Hand in your Chi-square calculations and tell me, based on the calculations, whether you accept or reject the null hypothesis. Also, hand in the answers to the problems below.

Thinking about the laboratory:

1. We assumed that all of the prey were either equally palatable or unpalatable. Describe an experiment in which you test for the effect of different degrees of unpalatability.
2. We assumed that models and mimics were impossible to tell apart (visually). Describe an experiment in which you test for the effect of some differences between how mimics and models appear.

References:

O'Donald, P., and C. Pilecki. 1970. Polymorphic mimicry and natural selection. Evolution 24:395-401.

Pilecki, C., and P. O'Donald. 1971. The effects of predation on artificial mimetic polymorphisms with perfect and imperfect mimics at varying frequencies. Evolution 25:365-370.

Wickler, W. 1968. Mimicry in Plants and Animals. McGraw-Hill, New York, 225 pp.

Last updated on January 19, 1999