Based on exercise materials provided by Brian Kinghorn with Genup program.
Introduction
This exercise uses a computer simulation to both demonstrate and test simple selection theory. To aid in conception, the example relates to
selection for fleece weight in 'Borderdale' sheep. For the present purposes it is not necessary to distinguish between the two sexes, and
generations are considered to be discrete - i.e. selected animals are bred only once. These factors simplify the situation sufficiently for the
simulation results to be easily interpreted. On startup, the fleece weights of 200 sheep in each of five generations are shown by histogram.
You will see responses to selection per generation as changes in the overall mean of the population. The animals which are selected and
used as parents are shown in yellow, and the average performance of these individuals is also marked.
As the user, you will determine the values of the three main components of selection: Proportion selected p, heritability h2 and phenotypic
standard deviation P. Recall that Response=ih2P, and that selection intensity i can be got from tables, given p.
Module SEL
1.
Try running the program with proportion selected p =0.1, h2=0.25, and P =0.5Kg, and year effect standard deviation kept at zero.
Note that the distribution of phenotypic values differs in detail between generations. This is what you would expect in real life due to
random chance effects, and is generated in the computer the random number generator. What type of distribution is approximated
within a generation?
How many generations of selection contribute to the difference between the means of the first and fifth generations? Note that despite
notable response, there is still considerable overlap of yearling weights between these generations. There are no deviations in
generation environmental effects (such as climatic effects affecting whole generations, not to be confused with the environmental
deviation each individual within a generation has), as shown in the bottom right-hand graph. Thus in this case, genetic and observed
(phenotypic) trends are equal.
Make a note of the total response from generation one to five, by reading from the scale at the bottom of the screen, or by clicking on
the 'Show generation means' button. Do another run using exactly the same parameters. You should get a slightly different result (but
possibly the same by chance, if so try another run) and the distributions will differ in detail from the previous run. How does this
relate to real life? Make a note of your total response from generation 1 to generation 5.
2.
Using your last output from question 1, find the approximate value of the selection differential in the first generation. How close is
the result to what you would predict? Approximately what proportion of this superiority of parents over their contemporaries was
transmitted to the next generation? What have you just estimated? Find the sum of the selection differentials for those generations in
which selection contributes to response in generation 5. How close is this value to what you would predict? Divide the total response
by this value - what have you just estimated?
3.
Calculate the response you would have predicted for question 1 (R=i h2P per generation of selection). You can compare this to the
response you achieved in SEL - or maybe to the average of a number of replicate runs.
4.
Rerun as in 1 but with p=0.25. What does this do to the selection differentials? Can you detect any effect on realised heritability?
Would you expect any effect? Calculate a predicted response and compare this with your realised response at generation 5. Note the
steadiness in genetic trend - this will be referred to in the next question.
5.
Rerun as in 1 but with p=0.01. What does this do to the selection differentials? Calculate a predicted response and compare this with
your realised response at generation 5. You may see a more inconsistent genetic trend (if not this will be because of "good luck",
rerun the program until you see this effect). Can you explain what might cause it? (Hint: never put all your eggs in one basket). What
negative effects of such intensive selection may appear?
6.
Rerun as in question 1 but with h2=0. Explain the result. Repeat this using h2=1, and explain this result.
7.
Rerun as in 1 but with P =.1Kg. Compare your selection differentials and realised response to those from questions 1 and 2, and
explain the results. How does this realised response compare to what you would predict?
8.
Increase the standard deviation of generation environmental effects from 0 to 0.5Kg. Try a run using the other parameters as in 1. As
a farmer running such a selection program, how would you react to the results achieved as the program progresses? Try more than
one run. Generation means can be read after clicking the 'Show generation means' button.
Note the steadiness in genetic trends, despite large fluctuations in observed trends. Genetic trend can be found by subtracting the
environmental trend from the observed trend. Can you suggest how the farmer can identify his genetic trend in real life?
Proportion
selected
Selection
intensity
Proportion
selected
Selection
intensity
Proportion
selected
Selection
intensity
0,001
0,002
0,003
0,004
0,005
0,006
0,007
0,008
0,009
3,364
3,167
3,047
2,959
2,889
2,831
2,781
2,737
2,698
0,01
0,02
0,03
0,04
0,05
0,06
0,07
0,08
0,09
2,663
2,419
2,266
2,153
2,061
1,984
1,917
1,858
1,804
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,755
1,400
1,159
0,966
0,798
0,644
0,497
0,350
0,195
Module Age
Chose the option “show only whole numbers of animals within age classes”
9.
10.
11.
12.
13.
Set up parameters of module Age according to the example: selection on fleece weight in sheep. The flock consists of 1000 dams,
their age structure is as follows: 250-two years, 200- three years, 180 – foru years, 150- five years, 120-six years, 100-7 years and 20
rams: 12 –two years, 8-three years. First progeny is born at the age of 2 years. Mating ratio is 1:50. Weaning rate is equal to 80%.
Assuming mass selection, heritability of 0.3 and phenotypic standard deviation of 0.4 calculate what is the expected reaction to
selection per year. Which survival rates have you chosen?
The length of using animals as parents affects response to selection. Manipulating only this parameter find settings which maximize
response to selection. Why is it not possible to shorten use of dams to less than 2-3 years? Explain, why the maximal selection
response is not achieved with extreme lengths of herdlife.
Compare settings from point 9 with ones maximizing response to selection. What reasons can prevent the breeder from organizing
herd structure to maximize response to selection?
Change the weaning rate, how does this parameter affect genetic gain? Which part of the formula of selection response is changedwhy?
Manipulate sire and dam survival- what is the effect of these parameters? Which conclusions can a breeder draw from points 12 and
13?
Module Loops
Module Loops helps you to visualise the various paths and loops in a pedigree diagram, and how they contribute to an individual's
inbreeding coefficient.
Using notepad you can input your own pedigree. The file should have the following structure
The first number in each row is the animal number. This should be sequential. The next two numbers are the sire and dam of each
animal. A value of 0 denotes parent unknown. All animals in the pedigree must have their own unique row.
Use the following pedigree. Calculate inbreeding coefficient of animal H and verify your result with paths from Module loops.
n
1 1
Fx     1  Fa 
2 i  2
E
C
A
B
D
I
J
F
G
H