One similarity between natural selection and genetic drift is that both events

To promote evolution literacy, it is important to teach evolutionary principles in introductory biology classes with an “active learning” approach in which students pose and answer questions, solve problems, and discuss and explain issues (Good 1992, McComas 1994, Linhart 1997, Smith 2000). Nonetheless, there is not much literature on the use of active learning to present evolutionary principles. Most students begin their introductory biology classes in college having heard of natural selection, yet few appreciate that there are other significant mechanisms of evolution, namely, genetic drift, mutation, and gene flow. In fact, many students start off equating natural selection with evolution. Students need a solid framework of learning (i.e., a well-organized class with a defined curriculum and stated objectives to help students build their own framework of knowledge) to correct such misconceptions about evolution, and offering them memorable exercises can help provide that framework. This article describes one such classroom exercise that effectively teaches the mechanism of genetic drift to undergraduates. It illustrates, by sampling M&M's milk chocolate candies, a number of concepts that are critical in developing evolution literacy.

Although natural selection is significant because it is the only evolutionary mechanism to produce adaptation, other forces—genetic drift, mutation, and gene flow—can change allele frequencies in a population over time and thus result in evolution. Students are often surprised to learn that even Darwin (1858) recognized that mechanisms other than natural selection were at work: “I am convinced that Natural Selection has been the most important, but not the exclusive, means of modification” (p. 30). The neutral theory of molecular evolution postulates that genetic drift is responsible for the great majority of evolutionary changes at the molecular level (Kimura 1989). Kimura argues that this mechanism of evolution is so important that an apt description of the evolutionary process is “survival of the luckiest.” Of course, some molecular variation can be explained by fitness differences (Messier and Stewart 1997), but the important point is to introduce to students the idea that not all evolutionary change is adaptive, and thus not all change involves natural selection.

Genetic drift is defined as random changes in allele frequencies in a population. The mechanism is so named because the pattern shows the drift of allele frequencies, up and down over time—there is no predictable directional component to change from generation to generation. Genetic drift occurs in all populations that are not infinitely large. It has especially strong effects when populations are small over several generations. The founder effect and the bottleneck effect are special cases of genetic drift. The bottleneck effect occurs when a large population undergoes a drastic reduction in size. For example, natural disasters such as hurricanes, fires, or floods can wipe out large areas of habitat but, by chance, leave small areas undisturbed. The surviving organisms in these undisturbed areas are the lucky founders (their traits have no relation to their survival) of the subsequent populations. The founder effect (which is simulated in the M&M's exercise) is similar to the bottleneck effect, except that the founding individuals are in a geographic location different from that of the original population (for example, a few individuals are blown to an uncolonized island and breed there). These individuals that founded the new population carry just a fraction of the genetic variation of the ancestral population. Random changes in the allele frequencies of the new population, or genetic drift, is the consequence of this sampling of the total genetic variation.

Sewall Wright (1932, 1982) developed the theory and established the importance of genetic drift. He described genetic drift as a powerful “trial and error” mechanism for exploring different fitnesses. Genetic drift shifts demes (local subpopulations in which random mating occurs) around on the fitness landscape and occasionally places one within the selection gradient of a local fitness peak. Natural selection then moves the deme up the peak. Wright also pointed out that these accidents of sampling could at times overpower the forces of natural selection (Wright 1982). (See Futuyma [1998] for a discussion of Wright's fitness landscape model.)

Accidents of sampling (sampling error) are based on sample size. Smaller samples are subject to more chance variation than are larger ones. As with flipping a coin, large deviations from the 50:50 heads-to-tails ratio are not surprising when small sample sizes are involved. As the sample size increases—that is, the more times the coin is tossed—the frequencies tend to approximate the 50:50 theoretical expectation. In a gene pool, the alleles of the adults are a sample of alleles from all the gametes of the previous generation, and thus they are affected by sampling error (Futuyma 1998). The smaller the sample of adults is relative to the original population, the larger the possible deviation, simply by chance. This sampling process starts at conception—not all gametes make it to the zygote stage—and continues throughout life (Ridley 1996).

Because genetic drift is based on a random sampling process rather than a deterministic process, students often have a difficult time understanding and appreciating its role in evolution. Although there are creative classroom exercises depicting evolutionary timelines with toilet paper (O'Brien 2000), demonstrating types of natural selection with jelly beans (Lauer 2000), and showing how complex adaptations can be built via natural selection using the Yahtzee® game (Dickinson 1998), few exercises have focused on genetic drift. The exercise described in the next section simulates one type of genetic drift, the founder effect, by sampling M&M's randomly.

The exercise

The goal is to measure the allelic diversity of one marker gene in numerous small founder populations compared to a large ancestral population. The material needed is one large bag (1 pound of M&M's works well for a class of 35–40). The bag of candies represents a gene pool. The different colors of M&M's represent different alleles of one gene. Thus two M&M's compose one individual. Individual organisms, however, are not explicitly represented in this exercise because its focus is on allele frequencies.

The exercise involves passing the bag of M&M's around the classroom and having students pour a relatively small sample on their desk. Students are simulating the founder effect, what happens genetically (to the allele frequencies of our one marker gene) when a few individuals are blown in a storm from the mainland to an island, for example. An assumption of this exercise is that sampling the bag without replacement does not affect the results—in other words, the bag of M&M's is behaving genetically as if it were infinitely large.

Have students tally their sample size of individuals (number of sampled alleles/2 alleles per individual = number of individuals) and the color frequencies within the sample. Either instruct students to choose an even number of M&M's from the bag (which slows down the sampling time) or provide a strategy if an odd number of alleles are sampled (although dealing with half an individual is not a problem because this is theoretical sampling, students could round up or down for the number of individuals or simply eat one randomly chosen M&M). Students do not sort the alleles into homozygote or heterozygote individuals; they simply compute the overall allele frequencies in their new small gene pool.

While students are sampling M&M's and determining the allele frequencies, write on the board the frequencies of the original population. Determine this before the exercise for your specific bag or use the expected frequencies in a bag of M&M's: 30 percent brown, 20 percent each of yellow and red, 10 percent each of orange, green, and blue (Eleanor Eggert [Mars, Inc.], personal communication, 2001). Table 1 provides an example of color frequencies for a one-pound bag and a format for recording results. (A new color [purple, aqua, or pink] will be added to bags after consumer preference is determined. The expected color frequencies of these bags will thus be 20% each of red, yellow, and the new color and 10% each of orange, green, blue, and brown [Eleanor Eggert (Mars, Inc.), personal communication, 2002].)

Randomly call on students to provide results for your data sheet. Also ask for results from the extremes—the largest and the smallest samples. The results should illustrate the points below.

• The allele frequencies of each student's sample are virtually always different from the original population's frequencies (evolution via genetic drift).

• The allele frequencies of each student's sample are different from frequencies of other samples (genetic drift increases variation among populations). This point is important for understanding processes such as population divergence within a species.

• Alleles with the lowest frequency in the original population have the lowest probability of becoming fixed (reaching 100 percent frequency) and the highest probability of getting lost (reaching 0 percent frequency). Alleles with the highest frequency in the original population have the highest probability of becoming fixed and the lowest probability of getting lost. The probability of fixation equals the frequency of the allele's occurrence in the population (Futuyma 1998). Extrapolating over time, there is a march to homozygosity for all neutral genes occurring in all populations (Ridley 1996). This march is counteracted by mutation and gene flow. In short, genetic drift decreases variation within a population over time.

• By chance, a small population may have an exception ally high frequency of a rare allele. This point can be observed in many human populations that were founded by relatively few individuals. The Amish population of Lancaster County, Pennsylvania, has a very high frequency (0.07) of the Ellis-van Creveld syndrome (polydactylous dwarfism) compared with that in most populations (0.001) because the allele was present in the founding population (Freeman and Herron 1998).

• Genetic drift has a stronger effect in small populations than in large ones. The differences from the ancestral population will be greatest, on average, in the smallest samples. Have students predict the allele frequencies if their sample from the M&M's bag had been 300 candies instead of 13.

• Remember that this exercise follows only one genetic locus—genetic drift affects allele frequencies at all loci simultaneously.

Follow-up homework assignments

In-class work and homework assignments enable students to process the material and construct a framework of knowledge. The classic experiments of Peter Buri (1956), a student of Sewall Wright, are useful in this context and have the added benefit of helping students with graph interpretation. Buri studied the effects of genetic drift in numerous small populations over time. He started with 107 lines of Drosophila, each one starting with a 0.5 frequency of two different mutant alleles, bw and bw75. Homozygotes of each type and the heterozygotes have distinct body coloration. The bw75/bw75 homozygote is bright red-orange (in young flies), heterozygotes are light orange, and the bw/bw homozygotes are white. Eight males and eight females (32 alleles), chosen at random, were used to start each subsequent generation. Buri followed his Drosophila lines for 19 generations. Allele frequencies became more evenly distributed from 0 to 1, with many populations losing or fixing the bw75 allele (and vice versa for the other allele).

Typically, students are simply presented with all the results from Buri's (1956) experiments. Instead, give students the starting frequency histogram of percentage of flies with different alleles and ask them to predict, using frequency histograms, what will happen in 5, 10, and 20 generations. In the next class, have students compare their predictions with Buri's results. Many evolutionary texts provide clear summary graphs of the results (e.g., Freeman and Herron 1998, Strickberger 2000).

Additional concepts for discussion

After this introduction to genetic drift using the M&M's exercise, the following concepts can help reinforce and elaborate your points.

Time

The sampling done for the M&M's exercise represents the starting population on a new island. What will happen over time to this population? Here you can discuss the effects of small population sizes over long periods of time, the population dynamics of founder–flush cycles, the extinction of most founding populations, and the increase in additive genetic variance in founding populations (or save this last point for your upper-level evolution class).

Random versus nonrandom processes: Natural selection compared with genetic drift

This exercise can be used to reinforce the difference between random and nonrandom processes by contrasting genetic drift and natural selection. It helps to point out that both genetic drift and natural selection are sampling processes; genetic drift is a random sampling process, and natural selection is a biased sampling process. While natural selection ultimately depends on the random variation that mutation produces, the process of natural selection itself is nonrandom. With natural selection, individuals differentially reproduce and survive as determined by the interaction between their phenotype and the environment—this is not random. Describe to the students a genetic situation in which the different alleles, instead of being neutral in effect, have different fitnesses because they influence a significant phenotypic trait. For example, if a student (acting as “the environment”) purposely chose all red M&M's—because individuals with those alleles were observed most readily or tasted the best or ran the slowest or had the showiest courtship display—instead of taking a random sample, it would be a case of very strong natural selection, not genetic drift. In this example, the red M&M's were chosen deterministically, not randomly.

Effective population size

Not all individuals in a population reproduce. Some are too young or too old or lack the opportunity. The reproductive individuals are the only ones, obviously, that contribute genes to the next generation. The number of reproductive individuals is generally much lower than the absolute number of observed adults in the population. The number of reproducing individuals is one factor that determines how the population behaves genetically over time (how strong the effects of genetic drift are). A genetically meaningful representation of this smaller population size is called the effective population size. It is the number of individuals in a theoretical population (random mating among adults) in which the amount of genetic drift equals that of the actual population. The effective population size is reduced by a number of factors—a skewed sex ratio, fluctuations in population size, small breeding groups, overlapping generations, and variable fertility, for example (Futuyma 1998, Ridley 1998).

The difference between the absolute population size and the effective population size can be demonstrated using M&M's. Use two bags to represent two populations having different sex ratios of breeding adults. For one population with a sex ratio of breeding individuals of 1:1 (one male to one female), 50 percent of the M&M's represent potential alleles from the breeding male population and 50 percent represent the female alleles. For a population with a skewed sex ratio of breeding individuals—for example, a polygynous system with a 1:9 sex ratio (one male to 9 females)—the male pool of alleles would be just 10 percent of the bag and the female pool of alleles the remaining 90 percent. For each population, 50 percent of the alleles for the next generation will come from the male gene pool and 50 percent will come from the female gene pool. Have students take founding populations from the different populations and compare the results. For the population with the skewed sex ratio, the 50 percent contribution to the next generation from the males will come from the relatively small sample of alleles (10 percent of the bag). In contrast, for the population with an even sex ratio, the male contribution to the next generation will come from a relatively large sample (50 percent of the bag). The population with the skewed sex ratio will be influenced more dramatically by genetic drift.

Real examples

The history of the northern elephant seal (Mirounga angustirostris) provides a classic example of the consequences of genetic drift through the bottleneck effect. In the 1890s, overhunting purportedly reduced this species to about 20 animals. The effective population size at this time was fewer than 20 because the northern elephant seal has a polygynous mating system. The current population size is approximately 30,000. Of 24 enzymatic loci analyzed, none shows variation—an extreme situation that suggests a history of genetic drift (Futuyma 1998). More recent analyses of mtDNA confirm the extremely low levels of genetic variation and provide estimates of population sizes for the bottleneck event (Hoelzel et al. 1993, Hedrick 1995, Weber et al. 2000). Interestingly, the lack of genetic variation at allozyme loci is not fully explained by the estimated bottleneck size and duration (Hedrick 1995).

The high number of endemic species of Drosophila (more than 100 in the picture-winged group, more than 800 in the family Drosophilidae) in the Hawaiian Islands is a result of numerous founder events. Analysis of salivary chromosome banding patterns reveals the repeated pattern of colonization between older and younger islands (Strickberger 2000).

Darwin's finches (Geospiza magnirostris) in the Galápagos provide an example of the founder effect and resultant nonselected morphological change. The population of finches created by the founder effect had larger bills than the source population. Song type, a nongenetic culturally inherited trait, was affected as well (Grant and Grant 1995).

Collared lizards (Crotaphytus collaris) in the Ozark Mountains live in small, isolated remnants of desert habitat. Population sizes are small, and 11 of the 14 populations examined were fixed for a specific multilocus genotype (Freeman and Herron 1998).

Examples that specifically relate to conservation biology readily capture student interest. The 35-year study on the Illinois prairie chicken (Tympanuchus cupido pinnatus) clearly demonstrates the genetic and fitness consequences of the bottleneck effect, as well as the challenges of conserving small populations (Westmeier et al. 1998). Habitat loss is correlated with a drastic reduction in population size (N < 50) of this once widespread species. Population size and fitness (as measured by the number of fertile incubated eggs per total eggs and number of hatched eggs per total) steadily declined, as did genetic diversity (measured by allelic diversity of fewer microsatellites). Despite conservation efforts that enlarged available habitat, overall population size and fitness decreased. Neighboring populations went extinct and thus genetic exchange via gene flow stopped. Once this happened, the fitness of the focal population decreased dramatically. However, conservation efforts improved the chickens' fitness (increased fertility and hatching success) by introducing individuals from genetically diverse populations from other Midwestern states (Westmeier et al. 1998). This study is remarkable in that it provides data on several of the variables associated with the extinction vortex threatening small populations and shows that inbreeding depression is a real concern (Soulé and Mills 1998).

Speciation

Genetic drift has been thought to play an important role in the formation of species (Carson 1975, Templeton 1996), particularly in the peripheral isolation model proposed by Mayr (1954). In this model, relatively small, isolated populations located on the periphery of an ancestral population diverge relatively rapidly from the ancestral population, because of the influence of random genetic changes. This mechanism of speciation has been somewhat controversial because species differences via genetic drift would be nonadaptive. In general, the mechanism of natural selection is probably important in conjunction with genetic drift in the continued divergence of small populations from ancestral populations (for a general discussion of this topic, see Freeman and Heron 1998 and, especially, Futuyma 1998).

Mutation, gene flow, and natural selection

Other mechanisms of evolution can be incorporated into the discussion of the M&M's exercise. Some of us remember that there used to be light brown M&M's. They have been lost from the population and replaced by blue ones. What is the simplest explanation for this pattern? Mutation of light brown to blue followed by fixation by genetic drift (neutral alleles)? By natural selection on an allele that once had a selective advantage and is now neutral? Selection against light brown and selection for the newly arisen (via mutation) blue (but no bags are known to have ever contained both blue and light brown M&M's)? Selection for blue accompanied by gene flow that spread the blue allele to other populations? Be prepared for potentially lively interactions if you ask students to demonstrate gene flow from a population on one side of the room to a population on the other. The allelic diversity of this genetic locus has increased dramatically recently. Purple, pink, and aqua alleles are now readily found in some bags.

Assessment

Although no formal assessment of this exercise has been done, its effectiveness can be evaluated by assigning homework questions about genetic drift after the standard lecture presentation of the topic and then again after the M&M's sampling exercise. Informal evaluation indicates that this exercise readily captures student interest and helps clarify challenging concepts.

Summary points of this exercise

Genetic drift is a nonadaptive mechanism of evolution that deserves more complete and interactive treatment in the classroom. The following points about genetic drift can be clearly illustrated using the M&M's sampling exercise: Furthermore, this exercise provides an introduction to several other important evolutionary topics (speciation, natural selection, gene flow, mutation).

• Genetic drift has a stronger effect in small populations than in large ones.

• Alleles can be fixed or lost by chance.

• Genetic drift increases variation among populations and decreases variation within a population over time.

The M&M's sampling exercise stimulates critical thinking and develops evolution literacy. You will know that your students have grasped these concepts when they understand that a random sampling process can result in what appears to be a nonrandom assortment of M&M's and when they are more interested in counting M&M's than in eating them.

Acknowledgements

I thank Alison Chubb, Dave Darda, Matthew Greenstone, Shawn Kutcha, David B. Wake, and an anonymous reviewer for helpful comments on the manuscript, and I thank Henry and Barbara Staub for the inspiration to develop this exercise.

References cited

1

.

Gene frequency in small populations of mutant Drosophila

2

.

The genetics of speciation at the diploid level

4

.

Using a popular children's game to explore evolutionary concepts

7

.

Evolution education: An area of needed research

Journal of Research in Science Teaching

8

.

The founding of a new population of Darwin's finches

9

.

Elephant seals and the estimation of a population bottleneck

10

.

Elephant seal genetic variation and the use of simulation models to investigate historical population bottlenecks

11

.

The neutral theory of molecular evolution and the world view of the neutralists

12

.

Jelly Belly® jelly beans and evolutionary principles in the classroom: Appealing to the students' stomachs

What are the similarities between natural selection and genetic drift?

What is one similarity and one difference between genetic drift and natural selection?

What is the similarities between natural selection and evolution?

Can natural selection and genetic drift happen at the same time?