The Red Queen’s Race: An Experimental Card Game to Teach Coevolution

Our game is effectively a simulation in which students take on the roles of host and
parasite populations. This role-playing is the key to student engagement and comprehension.
Students playing the host not only see the oscillations in genotype frequency in the
data they generate: they personally dread the adaptation of the parasite population
to their host hand and suffer the inevitable crash of the host population. Likewise,
students playing the parasite take satisfaction from the growing success of the parasite
population as it adapts to the host population and lament its failure as the composition
of the host hand rapidly shifts away in response. We present the game to the students
as a means to test the key prediction of the RQH: host-parasite coevolution maintains
genetic variation. The students use the game to see for themselves if this prediction
is upheld. In other words, the students are engaged in hypothesis testing.

Our goal for this exercise is to convey four general concepts (Figure 2a): (1) coevolution occurs rapidly; (2) that which is most fit now can become the
least fit in just a few generations; (3) rare advantage, or negative frequency-dependent
selection, can maintain genetic variation over time; and (4) we can use simple games
to represent complex processes and to test hypotheses. Prior to beginning the game,
we encourage instructors to pose four questions to their students that will emphasize
these concepts. Presented in Figure 2b, these “warm-up” questions ask students to reflect upon their initial understanding
of coevolution and how we might study it. They will struggle to answer these questions
prior to playing the game. By guessing and discussing answers with their peers, however,
students will be thinking about the key concepts of the game as they begin playing.

Figure 2. Concepts and questions for classroom discussion. We outline a four central concepts of the game, b warm-up questions to emphasize these concepts, c wrap-up questions in which students revisit and revise their responses to the warm-up
questions, and d questions for discussing the metapopulation level of the game.

Materials

The only required materials for the game are two decks of playing cards per group
of students, one deck for the host population and one for the parasite population.
It is best to use two distinct decks to facilitate separation of the host and parasite
populations (e.g. blue-backed decks for host, red-backed for parasite). We have made
additional resources available online at http://www.indiana.edu/~curtweb/EvolutionLabs/ (Lively 2015]) and in Additional files 1 and 2 (a worksheet with directions and a spreadsheet for data entry and calculations).
We provide the spreadsheet to students as a Google sheet and give all students access
and editing privileges. Students enter their data directly into the Google sheet,
and we project the results of the game live. This set-up facilitates sharing of the
data between all groups. It requires that each group of students have access to a
computer for data entry.

Game Set-Up

We present brief introductory material to the students and then instruct them to split
into groups, collect card decks, and read the directions thoroughly. We conduct this
exercise with students in groups of two: one student playing the host population and
one playing the parasite population. Students are advised to switch halfway through
so that they can personally experience both roles. We suggest students play for 15
generations, which requires a minimum of 45 min.

The sequence of the game is simple, and students catch on after 1–2 generations of
independent play. The game is amenable to modification according to the level and
size of the class and to the desired learning goals. Here, we outline the specific
approach that we have used for playing the game with 24 undergraduate students in
CM Lively’s Evolution course at Indiana University.

Students begin by establishing their starting host and parasite populations. They
shuffle their respective decks and randomly select 12 cards. The remaining cards become
the reserve deck. Each suit is a genotype—clubs, spades, hearts, and diamonds. Students
count the number of individuals of each “genotype” in their hand and record these
data in the generation 0 row of their data sheet (spreadsheet provided online and
in Additional file 2). Students should ensure that these counts sum to 12 for both the host and parasite
populations: the provided spreadsheet includes a column for this purpose. We additionally
constructed the spreadsheet such that counts are automatically translated into frequencies
and plotted. For example, if half of the cards held by the “host” are spades, then
the frequency of the spade genotype is 0.5.

Four basic steps constitute a single “generation” of this game (Figure 3):

Figure 3. Schematic of the set-up and four basic steps of the Red Queen game. These four steps
constitute a single generation of play. We propose 15 generations of play (i.e. 15
repetitions of these steps) for a classroom exercise.

Step 1: host–parasite contact Host and parasite shuffle their populations. They then work together to randomly
pair each host and parasite card, resulting in 12 host-parasite pairs. Students tend
to find their own way of efficiently performing this step. For example, the host student
could lay out her cards, and then the parasite student can lay out his cards next
to the host cards until 12 pairs are formed. This step requires that each group have
enough work space to lay out their pairs. It is also helpful for host and parasite
decks to be distinct in some way (e.g. different backs, say red and blue), so that
host and parasite individuals can be separated following the selection step.

Step 2: infection and selection According to the matching-alleles model for infection genetics, a parasite successfully
infects a host when its genotype matches that of its host (Frank 1993]). Therefore, infection results for those pairs in which the host and parasite genotype
match (e.g. host and parasite are both spades) (Figure 1b). The host individual is sterilized or killed, so the host student discards that
card by placing it in her reserve deck. The parasite student retains the successful
parasite card for subsequent reproduction. For pairs in which the host and parasite
genotype do not match, the host resists infection. The parasite does not survive the
failed infection (as in Salathé et al. 2008]; King et al. 2011]), so the parasite student discards it by placing it in his reserve deck. The host
student retains the successful host card for reproduction. In this formulation of
the game, each parasite individual has only this single chance to infect. Students
will often find that matches (successful infections) are rare in the initial generations
of the game and increase through time as the parasite population adapts.

Step 3: reproduction Each surviving host makes two offspring and dies. Students simulate this process
by adding one card of the matching suit for each surviving host card (for a total
of two cards of the same suit). Each successful parasite makes three offspring and
dies. Students simulate this process by adding two cards of the matching suit for
each surviving parasite card (for a total of three cards). Sometimes, a student’s
reserve deck does not have enough cards of a given suit to give each surviving individual
enough offspring. In this case, the student should randomly select cards from the
reserve deck until all individuals have reproduced. These randomly selected offspring
will not match the genotype of the parent; students can think of this step as mutation.

We feel that the greater offspring number of parasites relative to hosts (3 vs. 2)
is biologically realistic. Computer simulations also demonstrated that this tends
to generate smoother oscillatory dynamics than equivalent offspring numbers: it increases
the probability of matching by facilitating rapid evolution of the parasite (data
not shown).

Step 4: population size regulation Students rarely have exactly 12 individuals at the end of the reproduction step.
The population nonetheless remains fixed at 12. If populations have too few offspring
(common for the parasite population in particular), students should randomly select
offspring from the reserve deck until they have 12 offspring. This step can be thought
of as immigration. If populations have too many offspring (common for the host population),
students should shuffle the offspring and randomly select 12 cards to make the next
generation. They should return the remainder to their reserve deck. This step is consistent
with a carrying capacity for the population. The students then record the number of
individuals of each genotype under generation 1 of the spreadsheet. Repeat steps 1–4
for 14 more generations. We find that 15 generations is sufficient to obtain 3–4 oscillations
(Figure 4).

Figure 4. Red Queen dynamics in sample game data. Seven groups of students each played the Red
Queen game for 15 generations during a class period of Indiana University’s S318 Honors
Evolution course. a–d The oscillations in frequencies of host (bold lines) and parasite (faded lines) genotypes (suits) over 15 generations for four different populations (student groups).
Frequencies are derived from the numbers of host and parasite individuals of each
genotype recorded at each generation (step 4). Matching host and parasite genotype
frequencies are displayed in the same color (e.g. the club genotype is in bold blue for host and faded blue for the parasite). Each group experienced multiple oscillations in genotype frequencies,
and the trajectories differ widely between groups. e Highlights the time-lagged nature of oscillations for a single matching pair of host
(bold) and parasite (faded) genotypes (clubs). Data derived from population 3 above. f The dynamics of the metapopulation: the average frequency of each host and parasite
genotype across seven groups (the four shown above plus three additional). The oscillations
in genotype frequencies are damped, which we would predict based upon the fact that
the game does not include any inherent fitness differences between genotypes. Raw
data in Additional file 3.

Outcome

After 15 generations, the students will have generated host and parasite genotype
frequencies through time. If using a data entry system that allows live updates and
data sharing, each group will also have access to the data and plots of other groups
in the class. In Figure 4, we show sample data generated by our own students (raw data provided in Additional
file 3). Oscillations in host and parasite genotype frequencies, with a time-lag of a few
generations, are obvious (Figure 4a–d). The specific follow-up exercises that an instructor wishes to follow should
be tailored to the level of the students, the prior coverage of these topics in the
class, and the instructor’s specific educational goals. We propose several questions
to return students to the game’s key concepts. Presented in Figure 2c, these “wrap-up” questions ask students to revisit, and perhaps revise, their answers
to the warm-up questions (Figure 2b). In answering them, students use the data they’ve generated to measure time lags
in parasite adaptation, changes in host fitness over time, and genetic variation in
the host and parasite populations.

We also encourage instructors to show students the results at the level of the “metapopulation,”
meaning across all populations. We propose this for several reasons. First, coevolution
leads to divergence between populations: as a host and parasite population reciprocally
adapt, they can adopt distinct evolutionary trajectories from their neighbors, just
by chance alone. Students will see this when they compare allele frequencies at generation
15 in different groups (Figure 4a–d). Secondly, no genotype has an inherent fitness advantage in the game: fitness
is determined solely by the frequency of a genotype’s matching partner. This is an
unusual idea that is obvious in the metapopulation data: each host and parasite genotype
is maintained at ~25% of the metapopulation, and the oscillations are damped (Figure 4f). Finally, we have come to realize that coevolution must be considered at the metapopulation
level (Thompson 2005]): for example, moderate gene flow between populations can promote coevolution by
increasing genetic variation (Gandon et al. 1996]; Lively 1999]; Gandon and Michalakis 2002]; Greischar and Koskella 2007]). This exercise exposes students to this kind of metapopulation thinking. The provided
spreadsheet for data entry (Additional file 2) includes a tab to calculate and plot the average host and parasite genotype frequencies
across all groups of students (Figure 4f). In Figure 2d, we present three discussion questions that highlight these key metapopulation points:
the striking divergence between populations, equal mean fitness of all genotypes,
and migration between populations.