Abstract
I have used an exercise involving domestic cats in the General Genetics course at the University of NebraskaLincoln for the past 5 years. Using a coherent set of traits in an organism familiar to the students makes it easy to illustrate principles of transmission and population genetics. The onesemester course consists primarily of sophomores and juniors who have either taken a onesemester introductory biology course, a onesemester cell biology course, or have a strong high school biology background. The students are given a handout and asked to determine the genotype at seven unlinked loci of at least one cat. To fill out the form, the students have to grasp such concepts as dominance, incomplete dominance, temperaturesensitive mutations, epistatic interactions, sex linkage, and variable expressivity. Completing the form reinforces these concepts as they observe the cat's phenotype and fill in the genotype. I then analyze the collected data and use it in my lectures on population genetics to illustrate the HardyWeinberg equilibrium, calculate allele frequencies, and use statistics. This allows the students to look at population genetics in a very positive light and provides concrete examples of some often misunderstood principles.
GENETICS is both more quantitative and more analytical than other areas of biology. The mathematical descriptions of Mendelian inheritance and population genetics are sometimes daunting, and students frequently have powerful misconceptions about population genetics. Good examples to illustrate principles of incomplete dominance, epistasis, and sexlinkage abound, yet these examples are often from organisms outside of the students' experience or interest (fruit flies, snapdragons, summer squash, roosters, etc.). Furthermore, if each example is from a different organism, the students can be distracted from the principles by the details. Common misconceptions by students about population genetics are that dominant alleles will take over a population, that the most abundant phenotype in a population represents the dominant trait, and that deleterious alleles will be eliminated quickly. One would expect that a clear understanding of the HardyWeinberg law would eliminate these misconceptions, but students are often unable to apply the principles to actual examples. Even though they may know how to calculate allele frequencies and test populations for HardyWeinberg equilibrium, their intuitions have not caught up with this knowledge. For example, many students can correctly answer question 1 in Table 1 but will incorrectly answer question 2.
I have attempted to find a coherent set of examples of Mendelian principles and to attack their misconceptions about populations. I have designed an exercise in cat genetics that helps me achieve both goals. The students are interested and engaged, and they also take home an increased knowledge of genetics. Many students have told of how they have interested family members and friends in genetics by discussing a pet cat. The process is simple and catfriendly (the most intrusive part is determining the cat's gender). It is easy to carry out; many students use a family pet, others go to the Humane Society, pet shops, or use a friend's cat. Some have genotyped stray cats they have seen. Some students get so involved that they cannot seem to stop, even though they get no additional credit for additional cats—in the Spring of 1998 one student turned in the genotypes of 31 cats! Many students have given additional information or used a known pedigree to infer more detailed genotypes (i.e., distinguishing LL from Ll). The data obtained by the class can be compared to published allele frequencies for cats in a variety of geographic locations, and the students seem to find this interesting. When their data are used to teach population genetics they can focus on the underlying principles because they are already familiar with the genotypes and phenotypes, and the symbols for the loci and alleles are more meaningful.
THE EXERCISE
The exercise is defined very early in the semester, although the details and the handout are not distributed until after I have covered the topics of transmission genetics necessary for completing the assignment. I try to include examples of cat coat color genes as often as possible when covering Mendelian genetics. A good description of feline genes and phenotypes can be found in Robinson (1991). I have also collected a number of photographs of various cats and include them on the course web page (http://bsbiosci.unl.edu/GCMB/Christensen/cats.html) to illustrate various genotypes. The genes and associated concepts that need to be explained before the students are ready to collect data are as follows.
Long hair: Long hair is recessive, and is indicated by ll. Short hair is either Ll or LL. An explanation of this is appropriate at the very beginning of Mendelian genetics.
Agouti: An explanation of the agouti gene and its effect on hair pigmentation patterns is appropriate in the context of epistasis. Many genetics texts include examples of agouti and black in mice as an example of epistasis. In cats the explanation is complicated by the Tabby gene, which controls the pattern of expression of agouti. This concept is one of the more difficult ones for students. I generally use the example of mice to explain the pattern of pigmentation of each hair caused by the agouti gene, then introduce the idea that in cats the Tabby gene causes the agouti gene to be expressed for a different period of time during the hair growth cycle in the stripes, leading to the Tabby pattern. We do not score alleles of Tabby in the exercise because T^{b} (Tabbyblotched) and T^{a} (Tabbyabyssinian) are very rare in this area. In general it simplifies things for the students to use the Tabby pattern as an indicator of whether the cat is agouti (A_) or nonagouti (aa). If they can see a Tabby pattern in the fur, then the cat must be agouti, whereas if the colors are solid then the cat is nonagouti.
Dominant white: An allwhite cat with nonpink eyes is likely to be W_ rather than albino. Albino cats are fairly rare; out of 1519 cats, my students have never reported one. Dominant white is an example of dominant epistasis as it is impossible to score any of the other color genes in the presence of W. Dominant white kittens often have one or more small pigmented spots on the forehead that usually fade by adulthood.
Siamese and Burmese: The c gene has two different temperaturesensitive alleles, c^{s} (Siamese) and c^{b} (Burmese). Both are recessive to C. The Burmese has more pigmentation in the warm parts of the body than the Siamese. I try to discourage students from genotyping Siamese and Burmese as the breeding of these cats is generally under close control by humans, thus the HardyWeinberg assumption of random mating is unlikely to be true. If students turn in the genotype of a Siamese or Burmese, I count the number of C and c alleles and calculate the allele frequencies, but do not include the other color genes from these cats in the population totals. If the class size were smaller, I would strongly discourage use of any fancy breeds.
Piebald spotting: This is a good example of incomplete dominance and variable expressivity. Simplified, ss cats have no white fur, Ss cats generally have <50% white fur, and SS cats generally have >50% white fur (cats who are 100% white are scored as dominant white and can't be scored for piebald spotting). The S allele is incompletely dominant, but variably expressed, so there is a moreorless continuous gradient of white pigmentation in populations. The standard in the literature is to grade cats for piebald spotting on a scale of 1–10 (Robinson 1991); however, I have found that the 50% guideline works very well. Very few cats are close to the borderline, so the difficulties and inaccuracies caused by this approximation are few.
Orange: This is a good example to illustrate sexlinkage and dosage compensation. Students need to be told that the dominant O allele only changes the black pigment to orange, while the yellow pigments in the fur are unchanged. A Tabby pattern is still visible in an agouti, orange cat. Because the O gene is on the X chromosome, males are hemizygous and are either O or o. Females can be OO and completely orange, oo and have no orange, or are heterozygous, Oo, and are mosaics of orange and nonorange fur due to Xinactivation. The patch size of the mosaic can vary quite a bit, although it correlates with the size of the white spots produced by the piebald spotting locus (Norby and Thuline 1965). Students often ask about male calico or tortioseshell cats; these are rare and are usually XXY, the feline equivalent of Klinefelter's syndrome (Centerwall and Benirschke 1973; Moranet al. 1984).
Dilute: The final gene scored is dilute (d). This is a recessive modifier of pigmentation, which causes the black pigment to become gray (often with a bluish tint) and the orange pigment to be diluted to cream. Dilute can be scored in the presence of any of the other colors except dominant white, and is a good example of an epistatic interaction.
Color genes that are not scored include brown (b), which is found in the Havana brown or chocolate point Siamese and Burmese, but is otherwise very rare in the Midwest (only one Havana brown has been reported out of 1519 cats in five years), and silver (I), which alters the yellow pigments visible in the agouti cat to white. A few students have noticed the silvertipped fur, but if they are instructed to only look at the darker pigments it will not cause confusion. Manx cats are are a good example of a single gene with two phenotypes: dominant tailless and recessive lethal. I use Manx cats as an example of a recessive lethal gene, and leave students space to describe it on the form, but it is rare in this area. Occasionally students report polydactyly, another example of variable expressivity and incomplete penetrance. In areas where these genes are common they could easily be added to the checklist.
Figure 1 shows the checklist that the students are given. A handout briefly describing the purpose of the exercise and reminding them of the genotypes and phenotypes is also distributed. One source of potential error is that it may not be possible for the student to accurately determine the cat's gender. It must be emphasized that they should not guess the gender, but just write unknown. Cats of unknown gender must not be used in the calculation of Orange allele frequencies. The first year that I did this exercise the results with Orange were not in HardyWeinberg equilibrium. In subsequent years I instructed them not to guess the gender if they were not completely sure, and the results with Orange have been in HardyWeinberg equilibrium since then. It is also important to emphasize to the students that they must record their data clearly, particularly taking pains to distinguish O from o and S from s. On the checklist I use a font for the lower case letters that helps distinguish them from upper case (see Figure 1) and I tell the students to write clearly.
RESULTS
During the past 5 years 766 students returned the form out of a total of 789 enrolled in the class (97%). This shows that it is an easy exercise to do, and clearly worth the 10 extra credit points (out of 500 possible for the course) they receive. Students get no additional credit for additional cats but there is room on the form for 5 genotypes and some students ask for additional forms and do large numbers of cats. The record so far is one student who turned in 31 cats, all from her family farm. Another student worked at a kennel and determined the genotype of 15 cats that were brought in for care while she was on duty. Many students bring photos or attach locks of fur to the forms. One student who wondered whether her cat was a male calico brought him into my office! The average number of cats per student is 2.1 (1644 cats reported by the 789 students), a reflection of their enthusiasm for the exercise. The numbers have been rising each year: in 1999 the average was 2.7 cats per student, and the return rate was 98%. This exercise is also a good way to interact with students oneonone. With an enrollment of 120–200, this is the best opportunity for me to interact with students casually about genetics. Recently I started to require that the students meet with me individually when they turn in the form. This allows me to meet each student outside of lecture, correct misunderstandings, and catch errors in the forms.
Student accuracy is high, with <5% of the raw forms needing some form of correction. Discussing the cats with each student helps reduce the error rate. I ask the students where they found or saw the cats and try to be alert for duplicates (they are instructed to use different cats from other students). Once the forms are all in, I make a tally of the genotypes in the sample (I usually have one or two honors students help with the data analysis). We include only nonfancy breeds from our general geographic region. The instructions encourage local nonfancy cats and most of the students comply. Table 2 lists the numbers of cats of each genotype from 1998. Some of the totals do not quite agree due to incomplete forms (for example, a student may have been confused about the A locus and not remember the cat well enough for me to have confidence in the genotype at that locus, but the genotype at the W locus is clear and is therefore used in the table).
INTERPRETING THE DATA IN CLASS
I first introduce the concepts of the HardyWeinberg law in class by presenting the assumptions (infinite population size, random mating, and no migration, mutation or selection), and then the conclusions (allele frequencies do not change over time, and the phenotypic frequencies can be derived from the allele frequencies as p^{2}, 2pq, and q^{2}). Next, I introduce the idea that if a population is in HardyWeinberg equilibrium the assumptions are likely to be true. I illustrate the ideas with the data they have collected on the S locus because the heterozygotes can be scored and alleles can be counted directly. Using the 1998 data, the allele frequency calculation for S is p = ((2 × 18) + 104)/(2 × 231) = 0.30. The allele frequency calculation for s is q = (104 + (2 × 109))/(2 × 231) = 0.70. Note that p + q = 1. We can then ask if the cat population is in HardyWeinberg equilibrium by calculating expected values from p and q, and using a χ^{2} test. The expected values are p^{2} × 231 = 21 for the SS genotype, 2pq × 231 = 98 for the Ss genotype, and q^{2} × 231 = 112 for the ss genotype. A χ^{2} analysis results in a P value of 0.65, thus the hypothesis of HardyWeinberg equilibrium can be accepted. The students have already learned the χ^{2} test, but at this point they need to understand that there is only 1 d.f. in this analysis because the observations have been used once in deriving p and q and a second time to calculate the expected numbers using the total number of cats observed.
We can also directly count the alleles for the Orange locus, but it is complicated by the fact that the males only have one allele and the females have two. For the 1998 data, the calculation of p (for the O allele) is (30 + (2 × 9) + 37)/(107 + (2 × 114)) = 0.25. The allele frequency, q, of the o allele is (77 + 37 + (2 × 68))/(107 + (2 × 114)) = 0.75. Once again, p + q = 1. It must be emphasized to the class that the total number of alleles is the number of males plus twice the number of females, not twice the number of cats. This is an opportunity to explain once again why guessing the gender was discouraged. Calculating the expected values is also complicated by the sexlinkage of Orange. I prefer to use the observations as few times as possible in calculating expected values, so rather than using the actual numbers of males and females, I simply divide the total number of cats by 2 to get the expected numbers of males and females. The expected number of OY males is p × 221/2 = 28, the expected number of oY males is q × 221/2 = 82, the expected number of OO females is p^{2} × 221/2 = 7, the expected number of Oo females is 2pq × 221/2 = 42, and the expected number of oo females is q^{2} × 221/2 = 62. The χ^{2} analysis gives a P value of 0.70 (3 d.f.), so the hypothesis of HardyWeinberg equilibrium can be accepted.
Every year since I began emphasizing that the students not guess the gender the population has been in HardyWeinberg equilibrium for S and O, with sample sizes of 200–300 cats each year. This suggests that the students have effectively sampled the cat population of Nebraska and some nearby areas (mainly Iowa and Kansas), and that the allele frequencies are relatively constant across this geographical area.
Having established that the collected cat population is in HardyWeinberg equilibrium I then use the HardyWeinberg law to estimate the allele frequencies of the other alleles. For example, the frequency of aa cats is 137/237, and is also equal to q^{2}. Solving for q gives a value of 0.76, and p = 1 − q = 0.24.
The allele frequencies can be compared to published allele frequencies of cats from various geographic locations. Several references include lists of allele frequencies found in a variety of locations in North America and Europe (Halpine and Kerr 1986; Kleinet al. 1986; Dunnet al. 1989; Höger 1994). These comparisons allow the students to understand the forces that change allele frequencies, particularly migration and selection. Some of these references also discuss human migrations and the resulting cat migrations and allele frequencies.
Finally, I discuss some common misconceptions that people generally have about population genetics. All of these misconceptions are shown to be untrue by the HardyWeinberg law, but some practical examples from the class's data help reinforce the ideas. The first misconception is that dominant alleles will take over a population. The second misconception is that if a phenotypic ratio is 3:1, then the more abundant phenotype represents the dominant allele. Having memorized Mendel's ratios, students are apt to conclude that if threequarters of a population shows a phenotype, then that is the dominant one. I use their cat data to show that this is only true for the progeny of a Mendelian monohybrid cross and has nothing to do with populations in equilibrium. A good example to use for both misconceptions is W, which is dominant but has a very low allele frequency of 0.05 and has not changed since the published work of Halpine and Kerr (1986). I try to use their data to show them how their intuitive answer to question 2 in Table 1 is incorrect. This is also an opportunity to challenge the assumptions they are making when they answer the question. The third misconception is that lethal or deleterious alleles will be wiped out quickly. Having seen HardyWeinberg at work, they can understand that although the allele frequency of a recessive lethal will drop quickly at first, the rate of loss declines as more and more of the lethal alleles are found in heterozygotes.
CONCLUSIONS
Students learn several genetic principles from this exercise and they have fun with it. They work with concepts such as dominance, incomplete dominance, variable expressivity, epistasis, sexlinkage, and X chromosome inactivation. They also learn about basic population genetics and the HardyWeinberg law. Most of the students enjoy the project and it increases their enthusiasm for genetics. I have spoken to several former students, who indicated that cat genetics remains one of the things they remember best from the class. The exercise helps tie together the transmission genetics and population genetics sections of the course. Although I use this exercise for a college level introductory genetics course I believe at least parts of it could be used at the highschool level as well. Cats provide excellent examples of several genetic principles at any level, and if the students can collect data from enough cats, the HardyWeinberg law can be illustrated using studentcollected data.
Acknowledgments
This exercise was initiated following a conversation with Larry Harshman who told me that Tim Prout had once used a student exercise involving cat genotypes. Larry Harshman, John Osterman, Aleata Triplett, Patricia Pukkila, and two anonymous reviewers made helpful comments on the manuscript. Tony Joern did not. There were also helpful contributions by a number of teaching assistants through the years including Aleata Triplett, Maryanne Skavdahl, and Bill Shaffer. I also thank my Bios 301 students from 1995–1999 for their enthusiastic participation in this project.
Footnotes

Communicating editor: P. Pukkila
 Received December 7, 1999.
 Accepted March 24, 2000.
 Copyright © 2000 by the Genetics Society of America