A PRIMER ON PROBABILITY AND MENDELIAN GENETICS

By Robert Merritt, Ph.D., professor, Biological Sciences Department, Smith College

The materials in this document were developed for Your Genes, Your Chromosomes: A
Laboratory in Human Genetics, a course offered through the Smith College Summer Science and
Engineering Program (SSEP). Information about that program is at www.smith.edu/ssep.

Basic Rules of Probability

The science of genetics began with Gregor Mendel’s explanation of the transmission of genetic
determinants from one generation to the next, an explanation that relies on two basic rules of
probability. Before considering those two rules, it is important to understand what is meant by
the term probability. The probability of an event occurring is defined as the long-run relative
frequency of the event, in other words, the proportion of trials in which the event occurs in. For
example, a quarter has two sides, one with the depiction of a head and defined as heads (H) and
the other lacking a head and defined as tails (T). Since there are two equally likely outcomes
when tossing a quarter and only one is defined as heads, the probability of heads on any toss is
equal to one-half. A fair and balanced die has six sides with each side having a unique number
of dots ranging one through six. On any roll of the die, there are six equally likely outcomes.
When rolling a die, the probability of rolling a four is one-sixth. Now to the first of the two basic
rules, the rule of products.

The Rule of Products: The rule of products states that the probability of independent events
occurring together is the product of their separate probabilities. For example, if two quarters
are tossed and one caught in each hand, the probability of obtaining a head and a head is equal to
one-half that the right hand holds heads times one-half that the left hand holds heads or one-
fourth (1/2 x 1/2 = 1/4). Using the rule of products, it is possible to easily calculate the
probability of all possible outcomes for the dual coin flip:

Right hand Left hand Probability

H H 1/2 x 1/2 = 1/4
H T 1/2 x 1/2 = 1/4
T H 1/2 x 1/2 = 1/4
T T 1/2 x 1/2 = 1/4

Note that the sum of the probabilities of all possible outcomes is equal to one. What if the hand
that holds the coin is considered irrelevant? The probability of obtaining two heads will remain
one-fourth as will the probability of obtaining two tails because there is only one way to obtain
each of these outcomes. However, calculating the probability of obtaining one head and one tail
requires the use of the second basic rule, the rule of sums.
The Rule of Sums: The rule of sums states that when there are two or more mutually exclusive
ways of obtaining the same outcome, the probability of that outcome is the sum of their (the
different ways) separate probabilities. Since one head and one tail can be obtained in two
mutually exclusive ways (right H and left T or right T and left H) the probability of one head and
one tail is equal to 1/4 + 1/4 = 1/2. As another example, suppose that when rolling a die, the
outcome sought is simply an even number. There are three mutually exclusive ways of obtaining
that outcome - roll a 2 or a 4 or a 6 – and the probability of an even outcome is equal to 1/6 + 1/6
+ 1/6 = 1/2. Using the rule of sums can become tedious as the number of ways of achieving a
particular outcome increases. Taking an example closer to genetics, suppose the question was
the probability of obtaining four girls and four boys in a family of eight. This question could be
answered by writing out the seventy different orders for four girls and four boys, using the rule
of products to calculate the probability of each specific order, and using the rule of sums to add
these separate probabilities together. Employing the binomial (or multinomial) equation offers a
more efficient approach.

The Binomial Probability Distribution: Whether obtaining heads or tails when tossing a coin
or girls or boys when having children, the outcomes will be binomially distributed. The
binomial distribution applies to any situation with a set number of independent trials in which,
for each attempt, there are two possible outcomes that are mutually exclusive and exhaustive
(i.e. there are two and only two distinct outcomes possible) and the probability of obtaining a
particular outcome does not vary from one attempt to the next. The following are examples of
binomial distributions where N equals the number of trials, p equals the probability of success
(arbitrarily defined), and q equals the probability of failure (q = 1 – p):

N
1 1p1q0 1p0q1
2 1p q
2 0 2p q
1 1 1p0q2
3 1p3q0 3p2q1 3p1q2 1p0q3
4 1p4q0 4p3q1 6p2q2 4p1q3 1p0q4
5 1p q
5 0 5p q
4 1 10p q
3 2 10p q
2 3 5p q
1 4 1p0q5

For each term in the above binomial distributions, the exponents indicate the number of
successes and failures and the coefficient represents the number of different orders in which that
number of successes and failures can occur. (Note: Remember that any number raised to the
zero power is equal to one.) For example, the second term in the binomial distribution for N = 5
gives the probability (5p4q1) of obtaining four successes and one failure out of five trials where
p4q1 gives the probability of four successes (say heads or H) and one failure (tails or T) in a
particular order (e.g. H,T,H,H,H) using the rule of products and 5 is the number of different
orders in which four successes and one failure can occur (The failure can be on the first, second,
third, fourth or fifth toss.). Since each specific order with four successes and one failure has the
same probability of occurring, multiplying the probability of a specific order by the number of
different orders possible is simply an application of the rule of sums. The coefficients of the
binomial distribution can be obtained from Pascal’s triangle where each coefficient is the sum of
the two adjacent numbers in the next row up (e.g. 6 = 3 + 3). Generating complete binomial
distributions is both unnecessary and tedious when, as is often the case, the concern is about the
probability of one particular outcome (e.g. the family of four girls and four boys mentioned in
the discussion of the rule of sums

The probability of a particular outcome is most efficiently calculated by employing the binomial
equation:
N !
pXq(N−X)
(X !)(N−X)!

(Note: The factorial (!) of a number is that number multiplied by

every smaller integer and 0! is equal to one.)

where pXq(N-X) equals the probability of getting X successes and (N-X) failures in some specific
order (rule of products), N!/((X!) (N-X)!) equals the number of different orders in which X
successes and (N-X) failures can occur, and the product of these two terms equals the probability
of getting X successes and (N-X) failures when order is not specified (rule of sums). The
probability of a family of eight consisting of four girls and four boys can now be calculated as
follows:

8
!
)4=7
)4(1/2
(1/2 )8=7
0(1/2 0/2
5
6
!(8−4
4 )
!

If there are three, rather than two possible outcomes for each trial, the binomial equation can
easily be modified to cover this trinomial situation as follows:

N! X Y Z
p qr
X!Y!Z!

Similar modifications can be made to treat situations with more than three outcomes possible for
each trial. Note that in all of these equations, the sum of the exponents must always equal N and
the sum of the probabilities of the different outcomes (e.g. p+q+r) must always equal 1.0.

Conditional Probabilities: Thus far, all calculations have been to determine the probability of
obtaining a particular outcome out of all possible outcomes. Occasionally, it is important to be
able to calculate the probability of a particular outcome given a condition that excludes one or
more possible outcomes. Returning to the case of all possible outcomes from tossing two
quarters –

Right hand Left hand Probability

H H 1/2 x 1/2 = 1/4
 H T 1/2 x 1/2 = 1/4
T H 1/2 x 1/2 = 1/4
T T 1/2 x 1/2 = 1/4

- given the condition that at one least quarter comes up heads, what is the probability that the
other quarter is also heads [Pr(2 heads | at least one head)]? Since only 3/4ths of all possible
outcomes meet this condition and 1/4th of all possible outcomes both meet the condition and have
the other quarter heads, 1/3rd of the outcomes with at least one head have the other quarter also
heads. The general approach to calculating conditional probabilities is as follows:

pr
o
ba
bi
l
i
tyo
f
de
s
ir
ed
ou
t
c
ome 1/4 1
c
o
nd
i
t
io
na
lp
r
o
ba
bi
l
i
ty = = =
pr
ob
a
bi
l
i
tyo
f
co
nd
i
ti
o
n 3/4 3

(Note: The desired outcome must meet the condition (e.g. if the condition is at least one head, the
desired outcome cannot be two tails).) It is important to remember that both the probability
distributions and conditional probabilities discussed above are based on the rule of products and
the rule of sums. A thorough understanding of probability is required for the easy mastery of
both subject matter and problem solving in the areas of transmission and population genetics.

Sample Problems:

1) When rolling a die, what is the probability of getting a number divisible by three?

2) If you flip a quarter three times, what is the probability of getting heads, tails, heads in
that order.

3) In a family of five children, what is the probability of having the following:

a. The oldest child a girl?
b. The youngest child a girl?
c. The birth order girl, boy, girl, boy, girl?
d. Three girls and two boys with birth order not specified?
e. At least one girl?

4) A psychologist is interested in the effects of sibling gender on the behavior of young girls
and collects data on a large number of families of size three where at least one of the children is a
girl. In what proportion of these families should all three of the children be girls?

5) A deck of cards contains 52 cards with 13 cards in each of four suits – clubs, diamonds,
hearts, and spades. If a heart is defined as success, and all other outcomes as failure, how would
you answer the following questions:
a. In drawing four cards at random from the deck (note: after drawing each card, the
card is replaced in the deck before the next draw), what is the probability of drawing success,
failure, failure, success in that order?
b. In how many different orders can two successes and two failures occur?
c. What is the probability of obtaining two successes and two failures if you do not
care about order?
d. How would your answers to the above change if you removed all the diamonds
from the deck before you drew your cards?

Mendelian Genetics

Gregor Mendel founded the science of genetics in 1865 with the publication of Experiments in
Plant-Hybridization. To this day, Mendel’s work remains a beautiful example of experimental
design – he carefully constructed his initial experiments, collected a large amount of data,
formulated models to explain his results, designed additional crosses to test his models, and
clearly stated his conclusions. Some of the reasons why Mendel succeeded in elucidating the
basis of inheritance where others before him had failed are the following:
a) He made an excellent choice of experimental organism. The garden pea (Pisum
sativum) has a reasonably short life cycle and can be grown in large numbers in relatively small
spaces, typically self-pollinates but is easily crossed, and existed in many different varieties.
Mendel made certain that the plants in his original crosses bred true for the character states being
studied (i.e. a plant with yellow peas produced only plants with yellow peas when self-
pollinated).
b) He chose discontinuous characters for study that existed in two alternate states and
this allowed him to always unambiguously classify each individual in subsequent generations as
being of one parental type or the other. Such unambiguous classification is not possible when
treating continuous characters such as height in humans where individuals fall along a continuum
rather than into discrete classes. Height was a discontinuous character in Mendel’s peas since his
plants were either six feet tall or one foot tall.
c) Mendel was careful to separate his results by generation and to record not merely
what types occurred in each generation but also their frequencies.
Mendel’s two laws are based on the results of his monohybrid and dihybrid crosses.

Monohybrid Crosses: In his monohybrid crosses, Mendel crossed two true-breeding lines that
differed in a single characteristic. Mendel obtained the following results when he crossed lines
differing in the character pea color:

Parental Generation (P) Yellow X Green

First Filial Generation (F1) Yellow (self)
Second Filial Generation (F2) 6022 Yellow:2001 Green

Mendel called the character state that appeared in the F1 dominant and the character state that
disappeared in the F1 and reappeared in the F2 recessive. He obtained the same results for each
of the seven characteristics he studied – only the dominant character state appeared in the F1 and
both character states appeared in the F2 in the ratio of 3 dominant: 1 recessive. With these
observations in hand, Mendel followed the standard practice of developing the simplest model
that would explain his results. He concluded that he was dealing with particulate determinants
for the characteristics since the green F2 plants were identical in appearance to the green
parentals even though yellow F1 plants produced these F2 green plants.

The first decision to be made in constructing the model was how many determinants for each
character were present in each individual. The key to answering this question lies in the F1
individuals that must contain a determinant for yellow since they are yellow and must contain a
determinant for green because they produce green progeny. If, as Mendel did, we symbolize the
dominant determinant with A and the recessive determinant with a and assume that the two
determinants in each individual segregate in gamete formation (i.e. gametes contain only a single
determinant for each character), we have a model that will explain all of the monohybrid cross
results as follows:

Parental Generation (P) Yellow X Green

AA aa
First Filial Generation (F1) Yellow (self)
Aa
Second Filial Generation (F2) 6022 Yellow:2001 Green
A- aa

Since the F1 yellow plants will produce two types of gametes (A and a) in equal frequencies, the
rule of products and the rule of sums predict that the random union of these gametes will produce
an F2 generation containing 3/4 dominant to 1/4 recessive individuals. This application of the
basic rules of probability is typically presented in the form of a Punnett square –

Male Gametes
(1/2) A (1/2) a

Female (1/2) A (1/4)AA (1/4)Aa

Gametes
(1/2) a (1/4)Aa (1/4)aa

Some terminology:
Alleles – alternate states of a gene (e.g. A and a are alternate states of the “A gene”)
Homozygotes – individuals carrying two identical copies of a gene (e.g. AA and aa )
Heterozygotes – individuals carrying two alleles (e.g. Aa)
Phenotype – the appearance of an individual (e.g. yellow and green)
Genotype – the genetic composition of an individual (e.g. AA, Aa, and aa)

Mendel did two additional crosses to test his model. He testcrossed (a cross to a recessive
homozygote) the F1 plants to the parental green line (Aa x aa) to produce the predicted 1/2
yellow: 1/2 green result and he selfed the F2 yellow plants to demonstrate that 1/3rd (AA) bred
true and 2/3rds (Aa) produced both yellow and green progeny (Note: that the 1/3rd and 2/3rds are
conditional probabilities- i.e. given that an F2 plant is dominant, the probability that it will breed
true is (1/4)/(3/4) = 1/3). Once the results from these two additional crosses confirmed his
model, Mendel formulated the Law of Segregation – individuals contain two determinants for
each characteristic and these determinants segregate from one another in gamete formation.

A note on Punnett squares: In order to be adept at solving complex problems in genetics, you
must know the monohybrid Punnett square so well that you can visualize it without writing it
down. You should be able to visualize the 3:1 phenotypic ratio, the 1:2:1 genotypic ratio, the 1:2
true-breeding dominants to non-true-breeding dominants ratio, all without putting pen to paper.
In addition, you should be able to visualize the 1:1 monohybrid testcross ration. We are a visual
species, and the Punnett square can help you master monohybrid crosses. In most cases, you
should not, in spite of the fact that this is often done in textbooks, apply the Punnett square
approach to more complicated crosses.

Dihybrid Crosses: Mendel next performed dihybrid crosses between pure breeding lines that
differed with respect to two characters in order to determine whether determinants for different
characters were inherited independently of one another. For example, he crossed round yellow
peas times wrinkled green peas to produce the following: (where A = yellow, a = green, B =
round, b = wrinkled)

Parental Generation (P) Yellow Round X Green Wrinkled

AABB aabb
First Filial Generation (F1) Yellow Round (self)
AaBb
Second Filial Generation (F2) YellowRound A-B- 315
YellowWrinkled A-bb 101
Green Round aaB- 108
Green Wrinkled aabb 32

Note that there is a 3:1 ratio of dominant to recessive for both color (416:140) and texture
(423:133) and that for any particular character state 3/4 will be dominant for the other
characteristic and 1/4 recessive (e.g. of the yellow peas, 315 are round and 101 are wrinkled).
This suggests that the determinants for the different characters are indeed inherited
independently, and, if it be true that texture is independent of color, the probability of any
particular outcome can be determined using the rule of products as follows:

Yellow Round A-B- (3/4)(3/4) = 9/16 315

Yellow Wrinkled A-bb (3/4)(1/4) = 3/16 101
Green Round aaB- (1/4)(3/4) = 3/16 108
Green Wrinkled aabb (1/4)(1/4) = 1/16 32

Mendel obtained this 9:3:3:1 phenotypic ratio for all of his dihybrid crosses. Mendel tested the
model of independence with dihybrid testcrosses (AaBb x aabb) and obtained the expected
1:1:1:1 ratios. Based on these observations, Mendel formulated the Law of Independent
Assortment – determinants for different characters assort independently of one another in
gamete formation. You should be able to visualize the rule of products approach to dihybrid
crosses just as clearly as you see the monohybrid Punnett square. Once you have mastered the
rule of products approach to the dihybrid cross, you are ready to easily generate expected ratios
for more complicated crosses. Once again, you should not try to master dihybrid and more
complicated crosses using the Punnett square.

Mendel chose Aa, Bb, etc as allelic symbols for the different characters he studied and these
symbols are an appropriate, often preferred, choice when constructing genetics models to solve
any problem in transmission genetics. However, practicing geneticists usually choose symbols
that somehow reflect the character being modeled. For example pea color is now symbolized by
Y for the dominant yellow and y for the recessive green. It would have been equally acceptable
to have chosen G for yellow and g for green. What is not acceptable is to choose Y for yellow
and g for green as this would indicate non-allelic determinants (i.e. determinants at different
genes). Standards of genetic nomenclature have been established for various organisms. For
example, in Drosophila genetics, a gene is always named based on the mutant phenotype and, if
that mutant phenotype is recessive, the first letter of the symbol for the gene will be lower case.
The determinants for the normal phenotypic state (wild type) are indicated with a superscript
plus. Here are some examples that illustrate genetic notation in Drosophila:
- The white-eye mutation (w) is recessive to the brick-red wild type (w+).
- The carnation-eye mutation (car) is recessive to the brick-red wild type (car+).
- The curly-wing mutation (Cy) is dominant to the straight wild type (Cy+).
When you construct genetic models, chose a unique symbol for each gene and modify those
symbols in some clearly distinguishable manner to represent different alleles. There are
exceptions to this practice (one frequently encountered example is the choice in many texts of M
and N as the allelic symbols for the MN blood group gene), but such choices are generally
inappropriate (LM and LN are the formal symbols for the MN blood group alleles).

Hypothesis Testing – The Chi-Square Test:

When flipping a coin, there is a one-half probability of obtaining heads and a one-half
probability of obtaining tails. However, when a coin is flipped ten times, there is less than a one
in eight chance of obtaining exactly five heads and five tails. Most outcomes will deviate from
five heads and five tails due to random sampling error. This random deviation from expected
will also occur in the results obtained from genetic crosses. The chi-square test can be used to
calculate the probability of sampling error producing a deviation as large as or larger than the
deviation observed. If the probability of sampling error producing the deviation is small, it is
concluded that the deviation from expected is not due to chance, but rather due to some problem
with the model – the model must be rejected. In genetic crosses (where there are a limited
number of possible models), when the probability of sampling error producing the deviation is
not small, the model can be accepted. Chi-square is calculated as follows:

(O−E)2
χ2 =∑
E

where O and E equal the observed and expected numbers of individuals in each phenotypic class.
Analysis of the results of the following cross in fruit flies (Drosophila melanogaster) will
illustrate the use of the chi-square test: (Scarlet and ebony are mutant phenotypes while wild type
is the normal phenotype for the character in question.)

Parental generation (P) ebony body Females X scarlet eye Males

First Filial Generation (F1) wild type (F1 X F1)

Second Filial Generation (F2) wild type 950

scarlet 330
ebony 250
scarlet ebony _ 70
1600

Goodness-of-fit chi-square calculation:

wild scarlet ebony scarlet ebony total
observed 950 330 250 70 1600
expected* 900 300 300 100 1600

*In a goodness-of-fit chi-square test, the expected numbers are obtained by multiplying the
probability of that phenotype (according to the model tested) by the total (e.g. 9/16 X 1600 =
900). Here goodness-of-fit to the 9:3:3:1 ratio expected in a dihybrid cross is being tested.
In all chi-square tests, the total for the expected must equal the total for the observed.

(9
5
0− 90)2 (3
0 3
0− 30 0−
0)2 (2
5 30)2 (7
0 0−10)2
0
χ2
= + + + =2
6.9
7
9
0
0 3
0
0 3
0
0 10
0

Table of chi-square values:

Probability(p)*:
df** 0.95 0.90 0.70 0.50 0.30 0.20 0.10 0.05 0.01 0.001
1 .004 .016 0.15 0.46 1.07 1.64 2.71 3.84 6.64 10.83
2 0.10 0.21 0.71 1.39 2.41 3.22 4.61 5.99 9.21 13.82
3 0.35 0.58 1.42 2.37 3.67 4.64 6.25 7.82 11.35 16.27

*The values listed are the probabilities of obtaining a chi-square value that large or larger by
chance.
** df stands for degrees of freedom and is equal to the number of independently variable
classes in the data set. Given that the total for the expected is fixed by the total observed, df
is typically equal to the number of phenotypic classes minus one (e.g. In the chi-square test
above, once three of the expected numbers have been calculated the fourth is fixed because
the four numbers must sum to the total. There are three independently variable classes and,
therefore, three degrees of freedom.)
Since the chi-square value calculated for goodness-of-fit to a 9:3:3:1 model is greater than that
listed under p = .001, there is less than one chance in a thousand of obtaining deviations as large
or larger than those observed due to sampling error. Generally, if the chi-square value obtained
is larger than that listed under p = 0.05, the deviations are considered significant and the model
must be rejected. This practice means that a correct model will be rejected one time in twenty
but is designed to balance two types of errors – rejecting a correct model and accepting an
incorrect model.

The 9:3:3:1 model tested is based on three underlying assumptions – a 3:1 ratio of wild to scarlet,
a 3:1 ratio of wild to ebony and independence of scarlet and ebony. If we had been able to
accept the 9:3:3:1 model, our analysis would be complete. Since we had to reject the model, it
would be interesting to find out which of the three underlying assumptions hold and which do
not. To do this, the above chi-square is partitioned into two goodness-of-fit chi-squares and one
independence chi-square as follows:

Goodness-of-fit chi square calculations:

wild scarlet total
observed 950+250=1200 330+70=400 1600
expected 1200 400 1600

Obviously, the chi square value for these data will be zero as there is a perfect fit
to the 3:1 model for scarlet.

wild ebony total

Observed 950+330=1280 250+70=320 1600
Expected 1200 400 1600

(1
2
80 −1
20 0 −4
0 )2 (3
2 0)2
0
χ2 = + =2
1.3
3
1
20
0 4
0
0

With one degree of freedom, this chi-square has a probability of p < 0.001.
The results deviate significantly from those predicted by the 3:1 model in that
there are too few ebony flies relative to wild type. (It turns out that the ebony
mutation makes the flies darker and reduces their viability.)

Independence chi square calculations: (expected values are in parentheses)

wild scarlet total

wild 950 (960)* 330 (320) 1280

ebony 250 (240) 70 (80) 320

 total 1200 400 1600

*If being scarlet is independent of being ebony, the expected number of wild flies
can be calculated using the rule of products as follows: multiply the probability of
being in row one (1280/1600) times the probability of being in column one
(1200/1600) to give the probability of being in both row one and column one,
then multiply times the total (1600) to get an expected number (960). The other
expected values can be obtained by subtraction from the row and column totals.
Clearly, there is only one degree of freedom in this chi-square test. In an
independence chi-square test, the df=(#rows-1)(#columns-1).

0−
(9
5 0)2 (3
9
6 3
0− 0)2 (2
3
2 5
0− 0)2 (7
2
4 0−80)2
χ2
= + + + =2
.0
8
93
1.5 31
0.5 26
8.5 8
9.5

With one degree of freedom, this chi-square has a probability of 0.20>p>0.10. The
results do not deviate significantly from independence.

Sample Problems:

1) Determine the expected F2 phenotypic ratio in a trihybrid cross.

2) In a cross between AaBbccDdEE X AaBbCcddEe (where upper case letters are

dominant alleles),
a) How many different phenotypes will be present in the next generation?
b) How many different genotypes will be present in the next generation?
c) Determine how many different types of gametes each parent can produce.
d) What proportion of the progeny from this cross will breed true when selfed?

3) The following chart shows the results of matings between individual jimsonweed
plants that had either purple or white flowers and spiny or smooth pods, where
purple (P) is dominant to white (p) and spiny (S) is dominant to smooth (s).
Indicate the genotypes of the parents for each of the crosses. (attribution unknown)

parental phenotypes progeny phenotypes

purple white purple white
spiny spiny smooth smooth

a. purple spiny x purple spiny 94 32 28 11

b. purple spiny x purple smooth 40 0 38 0

c. purple spiny x white spiny 34 30 0 0

d. purple spiny x white spiny 89 92 31 27

e. purple smooth x purple smooth 0 0 36 11

4) Use the chi square test to test goodness-of-fit to the appropriate model for
cross (d) above.

5) Albinism in humans is a recessive trait. In a study on a large number of two-child

families with normal parents but at least one albino child, a total of 68 normal and
72 albino children were observed. How do you explain the fact that albinos were
more common than normal children? Do the numbers deviate significantly from
expected? (attribution unknown)

Extensions of Mendelian Genetics

Dominance Relationships: In Mendel’s crosses, one character state was always completely
dominant over the other. That is not always the case. Two other frequently encountered
dominance relationships are partial (or incomplete) dominance and codominance. The following
cross involving flower color in four o’clock plants illustrates partial dominance:

Pure breeding parentals: Red x White

F1: Pink (F1 x F1)

F2: 1 Red:2 Pink:1White

Partial dominance refers to that situation when the phenotype of the F1 falls somewhere between
the phenotypes of the two parents. When this is the case, the F2 phenotypic and genotypic ratios
are identical. In contrast, with codominant alleles, the heterozygote shows the phenotypes of
both homozygotes at the same time. Codominance is often found when studying genetic
variation at the molecular level. The following discussion of blood groups will illustrate both
codominance and the fact that, while any diploid individual can carry at most two alleles,
multiple alleles can occur in a population.

Blood Groups: Blood groups are determined by antigens on the surface of red blood cells
(RBCs). Antigens are substances that are capable of eliciting an immune response. Part of that
immune response may be the production of antibodies that can recognize and bind to the specific
antigen that elicited their production. The ABO blood group classification is based on presence
of particular RBC antigens, but also involves naturally occurring antibodies. Four ABO
phenotypes are present in most populations: Type A, Type B, Type AB, and Type O. The
presence of the RBC antigens is controlled by three alleles at the I gene (IA, IB, and i) with
codominant alleles IA and IB both being dominant to the recessive i allele. The ABO system is
summarized in the following table:

Blood Type A B AB O
 RBC Antigen A B A and B neither
Serum Antibody anti-B anti-A neither anti-A and anti-B
Genotype I IA or IAi
A I IB or IBi
B IA IB ii

The presence of the naturally occurring antibodies makes the ABO blood group particularly
important in blood transfusions. For example, if Type A blood is transfused into a Type B
individual, the anti-A in the serum of the recipient will cause the A antigen carrying RBCs from
the donor to clump together, blocking capillaries, and lyse, releasing high concentrations of
hemoglobin. To avoid such transfusion reactions blood types are matched prior to transfusion.
Since the RBCs of Type O individuals do not carry either the A or the B antigens, these
individuals are referred to as universal donors while Type AB individuals, whose serum lacks the
naturally occurring antibodies, are universal recipients. Another frequently encounter blood
group in genetic studies in the MN blood group with three blood types (Type M, Type MN, and
Type N) controlled by two codominant alleles.

Another medically significant blood type is the Rh blood group. This system was first studied
by Landsteiner who injected Rhesus monkey (hence Rh) RBCs into rabbits and then tested
human blood with the rabbit anti-Rhesus serum. The RBCs from about 85% of the people he
tested clumped when mixed with the anti-Rhesus serum indicating that the RBCs from these
people carried an antigen that was similar to an antigen on the Rhesus monkey RBCs. He called
these individuals Rh+ and those whose RBCs did not carry this antigen rh-. The Rh+ phenotype is
dominant to the rh- phenotype. This system is of medical significance when an rh- woman is
carrying an Rh+ fetus. It is common late in pregnancy for some fetal RBCs to leak into the
maternal circulation. If Rh+ fetal RBCs leak into the circulation of an rh- woman, her immune
system will recognize the Rh antigen as foreign and mount an immune response to the fetal cells.
In a first pregnancy when her immune system encounters the Rh+ RBCs for the first time, a
primary immune response ensues with gradual production of a low level of antibodies. This
primary immune response often does little or no damage to the fetus, but leaves the mother with
immunological memory cells. The next time she encounters the Rh antigen, these memory cells
will lead to a secondary immune response with rapid, massive production of antibodies. These
antibodies will flood across the placenta into the fetal circulation usually resulting in death of the
fetus. Physicians circumvent this maternal/fetal incompatibility problem by passively
immunizing the expectant rh- mother with low levels of anti-Rh antibodies so that the fetal Rh+
RBCs are removed from the mother’s circulation before her immune system can mount a
primary response. If this procedure is followed with every pregnancy, the rh- woman will never
mount a secondary response to the Rh antigens of her fetus.

Lethal Alleles: A mutation in chickens called creeper results in chickens with legs so short that
these birds appear to “creep” across the ground. When a creeper chicken is crossed times a
normal chicken, the F1 generation will contain a 1:1 ratio of creeper to normal chickens. When
two F1 creepers are crossed the F2 progeny will consist of two-thirds creepers and one-third
normal chickens. The F2 normal chickens indicate that creeper (Cr) is dominant to normal (cr)
and the 1:1 F1 ratio (monohybrid testcross ratio) indicates that the parental creeper was
heterozygous (Cr//cr). The cross between F1 creepers (Cr//cr) did not result in the expected 3:1
ratio of dominant:recessive because the creeper allele, while dominant in its effect on leg length,
is also a recessive lethal (i.e. creeper homozygotes (Cr//Cr) die). Creeper is an example of a
pleiotropic gene, a gene that affects more than one aspect of phenotype. The 2:1 ratio is a
conditional probability with the condition being that the chicken is alive. The above crosses are
diagramed below.

Parental Creeper x Normal

(Cr//cr) (cr//cr)

F1 1 Creeper : 1 Normal
(Cr//cr) (cr//cr)
F1 Creeper x F1 Creeper
(Cr//cr) (Cr//cr)

F2 2 Creeper : 1 Normal
(Cr//cr) (cr//cr) (Cr//Cr die)

Sex Determination and Sex Linkage: Sex is a phenotype that is determined in different ways
in different sexually reproducing organisms. In some species, sex is determined by environment,
but here the focus is on those instances in which sex is determined chromosomally. In
honeybees, sex is determined by a haploid:diploid mechanism where unfertilized eggs develop
into haploid males and fertilized eggs into diploid females. In grasshoppers, males and females
also differ in chromosome number, but the difference is in a single sex chromosome (the X
chromosome). Female grasshoppers are XX and produce gametes that all have the same haploid
chromosome compliment (females are homogametic) while the XO males have only one sex
chromosome and produce two types of gametes with respect to chromosome compliment (males
are heterogametic). Obviously, female grasshoppers, with pairs of autosomes and two X
chromosomes, have an even number of chromosomes while males, with the same number of
pairs of autosomes but only one X chromosome, have an odd number of chromosomes. Sex is
also determined by sex chromosomes in the animals most frequently studied in genetics (mice,
fruit flies, humans) but, in these species, males and females contain the same number of
chromosomes. In all these cases, females are homogametic (XX) while males are heterogametic
(XY) with the two sex chromosomes in males differing in their morphology and genetic content.
In some animals (e.g. birds and butterflies), the females are the heterogametic sex (ZW) and the
males are homogametic (ZZ). While the chromosomal basis of sex appears similar in mice, fruit
flies, and humans, this is not the case. In Drosophila, sex is determined by the ratio of X
chromosomes to haploid autosome sets with a normal female having one X chromosome for each
set of autosomes and normal males having one X chromosome for two sets of autosomes. The Y
chromosome, while necessary for fertility, is not involved in sex determination, and
chromosomally abnormal XO Drosophila are phenotypically male. In mammals, it is the
presence or absence of the Y chromosome that determines sex and chromosomally abnormal XO
humans are phenotypically female (Turner’s syndrome).

In 1910, Thomas Hunt Morgan, the father of Drosophila genetics, published a paper on a white-
eyed mutant male that he had found in his bottles of wild-type red-eyed fruit flies. When he
crossed this white-eyed mutant male times its red-eyed sisters, he obtained the following results:
White-eyed Male x Red-eyed Female

F1 1,237 Red-eyed Males and Females

3 White-eyed Males

These F1 results suggested that the wild-type red was dominant to the mutant white. With
respect to the three white-eyed F1 males, Morgan stated “The occurrence of these three white-
eyed males (F1) (due evidently to further sporting) will, in the present communication, be
ignored.” Morgan then crossed the F1 red-eyed flies to obtain the following:

F1 F1 Red-eyed Females x F1 Red-eyed Males

F2 2,459 Red-eyed Females

1,011 Red-eyed Males
782 White-eyed Males

Since eye color was clearly associated with sex, Morgan explained his results by placing the
white-eye gene on the X chromosome. White thus became the first sex-linked gene (sex-linked
means X-linked) and the first gene to be placed on a specific chromosome. Sex linkage explains
the results as follows (where w is the symbol for the recessive white and w+ is the symbol for the
dominant red):

White-eyed Male x Red-eyed Female

w/Y w+//w+

F1 1,237 Red-eyed Males and Females w+/Y or w+//w

3 White-eyed Males

F2 2,459 Red-eyed Females w+//w+ or w+//w

1,011 Red-eyed Males w+/Y
782 White-eyed Males w/Y

Since genes on the X chromosome are typically not found on the Y chromosome, males will
have only one copy of each sex-linked gene. Males therefore cannot be classified as either
homozygotes or heterozygotes but, rather, are referred to as hemizygotes. The pattern of
transmission that identifies genes as sex-linked is sons only from their mothers and fathers only
to their daughters. Y-linked genes are transmitted only father to son and are referred to as
holandric.

Since sex-linked genes typically have nothing to do with sex, the one gene found in males must
produce as much gene product as the two genes found in females. In Drosophila this is
accomplished by having the one sex-linked gene in males produce gene product at twice the rate
of each copy of the gene in females. In mammals, this dosage compensation is achieved by the
random inactivation of one X chromosome (which is now referred to as a Barr body) in each cell
of a female early in development. Once this random inactivation occurs, each cell will produce a
clone of cells with the same X chromosome inactivated. Females are thus mosaics of clones of
cells expressing either the maternally or the paternally derived X chromosome. This X
chromosome inactivation explains the coat-color pattern seen in calico (or tortoise-shell) cats that
are heterozygous for a sex-linked coat-color gene as well as why these cats are, almost without
exception, female.

Before moving on from sex linkage, a final comment should be made about Morgan’s paper. He
had an exciting contribution to make and decided to publish the paper even though he could not
explain all of his results, i.e. the F1 white-eyed males. In publishing his paper he did what
honesty required – report these three flies as unexplained- and then he gave the problem of
explaining their existence to one of his students, Calvin Bridges. Bridges went on to
demonstrate that these phenotypically abnormal F1 flies were sterile XO males that had received
their X chromosome from their fathers and hence had white eyes.

Sex-Influenced and Sex-Limited Characters: Sex-influenced characters are typically under

the control of autosomal genes, but the expression of these genes is dependent upon the sex of
the individual. Two examples of sex-influenced genes in humans are pattern baldness and short
index finger (index finger shorter than ring finger). In males, bald is dominant to non-bald and
short index finger is dominant to long index finger. These dominance relationships are reversed
in females. Unlike sex-linked or sex-influenced characters, sex-limited characters are expressed
in only one sex. Sex-limited characters are typically under the control of autosomal genes that
determine primary or secondary sex characteristics. Milk yield in cattle is one example. Both
sexes carry the genes for milk yield, but cows express these genes and bulls do not.

Gene Interaction: Comb shape refers to the red structure on the heads of chickens, and, by
considering the following series of crosses, it is possible to construct a genetic model for
variation in the shape of this character (note: the parental lines are all pure breeding and rose,
pea, single, and walnut refer to different comb shapes).

Parentals Rose x Single Pea x Single Rose x Pea

F1 Rose (F1 x F1) Pea (F1 x F1) Walnut

F2 3Rose:1 Single 3 Pea:1 Single

The first two crosses are clearly monohybrid crosses with rose dominant to single and pea
dominant to single. The third cross was done to determine the relationship between rose and pea.
The F1 walnuts might be due to partial dominance of rose and pea and, if that were the case,
crossing the F1 walnuts would produce a 1 rose:2 walnut:1 pea phenotypic ration in the F2.
However, crossing the F1 walnuts produces the following result:
 Parentals Rose x Pea

F1 Walnut (F1 x F1)

F2 9 Walnut:3 Rose:3 Pea:1 Single

The 9:3:3:1 F2 ratio clearly indicates that two genes must be interacting to control comb shape in
these crosses. By letting R and r represent the alleles for rose and single at one gene and P and p
the alleles for pea and single at a second gene, we can construct a genetic model to explain all of
the above results.

Parentals Rose x Single Pea x Single Rose x Pea

(RRpp)x(rrpp) (rrPP)x(rrpp) (RRpp)x(rrPP)

F1 Rose (F1 x F1) Pea (F1 x F1) Walnut (F1 x F1)

(Rrpp)x(Rrpp) (rrPp)x(rrPp) (RrPp)x(RrPp)

F2 3 Rose (R-pp) 3 Pea (rrP-) 9 Walnut (R-P-)

1 Single (rrpp) 1 Single (rrpp) 3 Rose (R-pp)
3 Pea (rrP-)
1 Single (rrpp)

Many gene interactions are more complicated than that involved in determining comb shape in
chickens. Two more examples will illustrate this point. In the first, pure breeding lines of white
Leghorn, white Wyandotte, and colored chickens to produce the following results:

Parentals White Leghorn x Color White Wyandotte x Color

F1 White (F1 x F1) Color (F1 x F1)

F2 3 White:1 Color 3 Color:1 White

Clearly the genetic basis for white is not the same in the two white breeds of chickens: the white
in the Leghorn breed is dominant to color while the white in the Wyandotte breed is recessive to
color. When the two white breeds are crossed the following results are obtained:

Parentals White Leghorn x White Wyandotte

F1 White (F1 x F1)

F2 13 White:3 Color
This must be a dihybrid cross since 13:3 has a base of 16 and is simply a modification of the
basic 9:3:3:1 dihybrid ratio where 3/16 of the progeny develop color while the other 9/16 + 3/16
+ 1/16 do not. This can be explained by having the white Leghorn chickens carry a dominant
allele (I) that inhibits color formation while the white Wyandotte chickens lack this dominant
inhibitor of color. The Wyandotte chickens are white because they are homozygous for a
recessive c allele that indicates an inability to produce color. Pure breeding chickens with color
have the genotype iiCC. The I- genotype at the inhibitor locus is epistatic to (hides the effects
of) the C gene. Note that epistasis describes an interaction between genes that is similar to the
relationship of dominance between alleles. However, the term dominance (in all its forms)
should only be used to describe allelic relationships and never interactions between genes. The
above gene interaction is, however, appropriately described as dominant epistasis because it is
the dominant genotype at the epistatic gene that hides the effect of the second gene. Here is the
model for the dihybrid cross:

Parentals White Leghorn x White Wyandotte

(IICC) (iicc)

F1 White (F1 x F1)

(IiCc) x (IiCc)

F2 9 White (I-C-)
3 White (I-cc)
3 Color (iiC-)
1 White (iicc)

The second example involves crossing two independently derived white-flowered lines of sweet
peas.

Parentals White #1 x White #2

(AAbb) x (aaBB)

F1 Purple (F1 x F1)

(AaBb) x (AaBb)

F2 9 Purple (9A-B-)
7 White (3A-bb + 3aaB- + 1aabb)

This is an example of complementation where dominant alleles at the two genes compliment
one another to produce color. A homozygous recessive genotype at either locus results in white.
If you truly understand the basic dihybrid ratio, you should be able to explain any F2results
involving gene interaction (e.g. 9:6:1 or 9:4:3 or 15:1). One challenge you will face is deciding
when you are dealing with results that involve gene interaction. If you think you understand
what is happening in a cross and then something surprising happens (e.g. single showing up in
the F2 progeny of a rose times pea cross or color showing up in the F2 of a cross between two
white breeds of chickens or crossing two pure breeding white-flowered plants and getting an F1
that has purple flowers), you should consider the possibility that you may be dealing with gene
interaction.
Sample Problems:

1) A Siberian iris with purple flowers and long stems was crossed with an iris with white
flowers and short stems. The F1 plants were crossed and the following F2 plants were obtained:

purple flowers, long stems 3/16

purple flowers, short stems 1/16
blue flowers, long stems 6/16
blue flowers, short stems 2/16
white flowers, long stems 3/16
white flowers, short stems 1/16

Explain these results. (attribution unknown)

2) The following five mothers, (a) through (e), with the given phenotypes, each
produced one child whose phenotype is described. The phenotypes of the five
fathers are also given. For each child, select the appropriate father.

Maternal phenotype Child’s phenotype Male phenotypes

a) A M Rh+ O M Rh+ (1) A MN rh-
b) B N rh- O N rh- (2) B MN Rh+
c) O M rh- A MN Rh+ (3) O N rh-
d) A N Rh+ AB MN Rh+ (4) O M rh-
e) AB MN rh- AB M rh- (5) A MN Rh+
(from Genetics by Monroe Strickberger, 1969, The Macmillan Company)

3) A woman of type A blood and normal color vision (color blindness is a sex-
linked recessive trait) produced five children as follows: (a) male, A blood
type, colorblind; (b) male, O blood type, colorblind; (c) female A blood type,
colorblind; (d) female, B blood type, normal color vision; (e) female, A blood
type, normal color vision. Of the two men who may have mated with this
woman, # 1 had AB blood type and was colorblind, and #2 had A blood type
with normal color vision. Determine, if possible, which of these men fathered
each child. (From Genetics by Monroe Strickberger, 1969, the Macmillan Company)

4) Retinitus pigmentosa, a form of blindness in humans, may be caused either by a

recessive autosomal gene, a, or a dominant autosomal gene, R. Thus only A-rr
individuals are normal. An afflicted man whose parents are both normal
marries a woman of genotype AaRr. What proportion of the children are
expected to suffer from this affliction if R and A are inherited independently? (from
Genetics:Questions and Problems by John Kuspira and G.W. Walker, 1973, McGraw
-Hill, Inc.)

5) A mouse breeder is sent an unusual male mouse that totally lacks body hair –
nude. The nude mouse is crossed with pure breeding black females to produce
an F1 generation consisting of 21 nude mice and 20 black mice. The F1 nude
mice are then crossed to produce the following F2: 96 nude, 27 black, 9 brown,
12 cream. Construct a genetic model to explain these results. (attribution unknown)