This is the answered version of the Marker Assisted Selection practical. Every ❓ question is followed by a worked model answer (green box). All the interactive steps still work, so you can keep running the GWAS and the breeding programme while you read, and the self-test quiz at the end is unchanged.
Plant breeders want to combine desirable traits (for example disease resistance, drought tolerance, or high yield) into a single crop variety.
Traditionally this is done by crossing plants and simply looking at which offspring look the best. This is called phenotypic selection. It is slow: some traits only show up late in a plant's life, and some are expensive or destructive to measure (you would have to expose a plant to salt stress just to see if it survives).
Genetic markers offer a shortcut. A marker is a specific, measurable position in a plant's DNA where individuals can differ from one another. Here we simplify this to just two possible values, 0 and 1, for every marker:
0 = this plant does not carry a particular DNA variant at that position1 = this plant does carry that DNA variant
If we know that a marker is reliably linked to a trait we care about, we can screen very young plants (often from a small leaf sample, days after germination) instead of waiting months. This is the core idea behind marker assisted selection. In this practical we will:
Use a genome-wide association study (GWAS) to find a genetic marker associated with salt resistance.
Use that marker to cross a high-yielding but salt-sensitive crop with a salt-resistant one, making it more salt resistant while losing minimal yield.
1 Explore a genetically diverse population
We want to breed plants that are salt resistant, but also have a high yield. To study what makes plants salt resistant, we first need a population that is genetically diverse: the plants differ from one another at many marker positions, and also differ in the trait we care about.
This is essential: if every plant were genetically identical, we could never work out which marker is responsible for a trait, because there would be nothing to compare.
Scroll sideways to see all the markers.
❓ Question
What do the rows and columns in the table above represent?
✅ Answer
Rows and columns. Each row is one individual plant in the population. Each of the columns labelled M0 to M49 is one marker, that is one fixed position in the DNA, and the 0 or 1 in the cell says which of the two variants that particular plant carries at that particular position. The last two columns are different in kind: they are not DNA at all, they are the two phenotypes we measured on the plant, its salt resistance and its yield. That mix is the whole point of the table. Everything that follows comes from asking whether any of the 0/1 columns lines up with one of the two measured columns.
2 Find markers associated with the trait (a GWAS)
To find out which marker(s) are responsible for salt resistance, we run a Genome-Wide Association Study (GWAS). A GWAS statistically tests, separately for each marker, whether plants with a 1 at that marker tend to have a different phenotype than plants with a 0.
The result of each test is a p-value. A small p-value means it is unlikely that the difference we observe happened purely by chance, so it is more likely to reflect a real, biological effect. Because p-values can become extremely small, we usually plot -log10(p-value) instead: this turns tiny p-values into large, easy-to-compare bars. The higher the bar, the stronger the evidence that a marker is associated with the trait. This plot is called a Manhattan plot, because the tall bars resemble a city skyline.
👆 Click a bar to inspect that marker below.
Relation between Marker ? and Salt Resistance
❓ Questions
What does this plot show, and why does the population have to be genetically diverse for it to work?
Which marker do you think is most promising? Click its bar. Do the two groups (0 vs 1) separate in the scatter plot below?
Before you drag the slider: what do you expect will happen to the Manhattan plot as you add more and more plants? Predict first, then drag the slider up to 100 and check.
Show a hint
Statistical tests become more reliable (more powerful) the more data points you give them, so the evidence for a real marker gets stronger as the population grows.
✅ Answer
What the plot shows, and why diversity matters. Each bar is one marker, and its height is -log10(p) from a separate statistical test asking whether plants carrying a 1 at that marker have a different salt resistance from plants carrying a 0. A tall bar means that difference is unlikely to be a fluke. The population has to be genetically diverse because the test is a comparison: if every plant carried a 1 at a marker, there would be no 0 group to compare against, the test would have nothing to work with, and the bar would be flat. No variation means no information, however many plants you measure.
The promising marker. It is Marker 12, the single bar that climbs far above all the others and above the significance line. Click it and the scatter plot below separates cleanly into two horizontal bands: the plants with a 0 sit around 30 percent salt resistance, the plants with a 1 sit around 80. The two groups barely overlap, which is exactly what a marker with a large effect looks like. Try clicking a few other bars for contrast; there the two groups sit on top of each other and you cannot tell them apart.
Growing the population. At 3 plants the plot is noise: several bars look tall, and Marker 12 is not obviously special. As you drag towards 100, the bar at Marker 12 climbs higher and higher while every other bar stays low and flat. Nothing about the biology changed, only the amount of evidence. With few plants a coincidence is easy, so no marker can be distinguished from chance; with many plants a real difference accumulates evidence and the coincidences do not. This is why a GWAS needs hundreds or thousands of individuals, and why a peak found in a small population should be treated with suspicion.
3 Introgress the marker into an elite variety
Now that we have pinpointed a marker for salt resistance, we want to move that specific piece of DNA into a plant that farmers already grow: one with a high yield, but which lacks the marker. This process is called introgression: transferring a gene (or a small region of a chromosome) from one variety into another through repeated crossing and selection.
We use two starting populations:
a resilient population: genetically diverse and carrying the salt-resistance marker. Think of it as a wild relative or old landrace: hardy, but usually a poor yielder.
an agricultural population: today's elite, high-yielding variety that farmers actually plant, but which lacks the salt-resistance marker.
Notice the trade-off: one plant is salt resistant but low-yielding, the other is high-yielding but salt sensitive. That trade-off is exactly what we are about to break, by combining the best of both.
4 Marker assisted backcrossing
Now it is your turn to run the breeding programme. Backcrossing means: take the population you are currently holding, cross it with one other population, then keep only some of the offspring based on a marker. The offspring you keep become the population you hold for the next round. Repeat for a few rounds.
You start out holding the resilient donor. Each round, you decide which population to cross it with, and which marker (and allele) to keep the offspring on. Use what you found in steps 2 and 3. After each round, watch the scheme and the graph to see how salt resistance and yield change.
You are holding: 🌾 Resilient donor
Salt resistance Yield elite target yield
❓ Questions
Which population should you cross back to each round, and which marker and allele should you keep? Why those choices?
What is the advantage of selecting each population based on a marker, rather than on how the plant looks?
How can you tell from the graph whether the breeding was successful?
Try a few rounds where you choose "don't filter (keep all)". What happens to salt resistance, and why?
✅ Answer
What to cross with, and what to select on. Cross back to the elite (agricultural) variety every round, and keep the offspring with Marker 12 = 1. The reasoning has two halves that pull in opposite directions. Each cross to the elite parent replaces roughly half of the remaining donor genome with elite genome, so after a few rounds the plants are almost entirely elite and the yield climbs back towards the elite level. But that same random halving would also throw away the salt-resistance allele about half the time, which is why you filter: selecting on Marker 12 = 1 protects the one piece of donor DNA you actually wanted. Crossing back to the donor, or to the population itself, makes no progress towards elite yield, and selecting on any other marker does nothing to protect resistance.
Why select on a marker rather than on the plant. You can read a marker from a leaf sample of a seedling, days after germination, instead of growing the plant to maturity, and you never have to expose it to salt to find out whether it is resistant. That is faster, cheaper and non-destructive. It is also more reliable: a measured phenotype is the sum of the genetics and the environment, so a resistant plant in a bad spot can look sensitive and mislead you, whereas the marker reads the genetics directly. In this simulation you can see the environment as the noise term that scatters the points in the step 2 scatter plot.
Reading success off the graph. Success is both lines ending up where you want them, not one. Salt resistance should start high and stay high, near 80, round after round. Yield should start low, because the donor is a poor yielder, and climb towards the dashed elite target line as the elite genome is recovered. If salt resistance collapses you lost the allele; if yield flattens out well below the dashed line you are not backcrossing to the elite parent. Watch the round cards too: they tell you how many of the 20 offspring passed the filter each round.
Crossing without a filter. Salt resistance drifts down, usually in steps, and often collapses to around 30 within a few rounds. The reason is chance rather than selection: each offspring inherits each marker from one parent or the other at random, so an offspring of a resistant plant crossed with the elite has only a 50 percent chance of receiving the resistance allele. With no filter, the fraction of carriers halves on average every round, and once it reaches zero it cannot come back, because nothing in the remaining population carries the allele any more. This is the single most important thing the simulation shows: the cross alone does not preserve a trait, the selection does.
5 Marker assisted selection for another trait: yield
So far we selected for salt resistance. Now let's run exactly the same GWAS approach for yield and see what we find. As before, drag the slider to grow the population, then investigate the Manhattan plot yourself.
Manhattan plot for Yield, instead of Salt Resistance. 👆 Click a bar to inspect that marker below.
Relation between Marker ? and Yield
❓ Questions
Grow the population to a large size. How does this Manhattan plot compare to the one you made for salt resistance in step 2?
Click the tallest bars one by one. Does any single marker split the plants into a clear low-yield and high-yield group, the way Marker 12 did for salt resistance?
Does yield look like it is controlled by a single marker, or by many? What is your evidence from the plot?
What would that mean for marker assisted selection of yield? (In the next practical you will meet a method built for exactly this situation: genomic selection.)
Extra time? Go back to the simulator in step 4 and try to breed for maximum yield instead of for salt resistance.
✅ Answer
Comparing the two plots. They look completely different, and the difference does not go away as the population grows. The salt plot develops one dominant spike at Marker 12 that keeps climbing. The yield plot instead develops a whole cluster of modest bars, none of which dominates, and many of them stay near or below the significance line even at 100 plants. Adding plants sharpens the picture but never produces a single obvious winner.
Clicking the tallest bars. No, none of them does. Whichever bar you pick, the two clouds of points overlap heavily: plants with a 0 and plants with a 1 both spread over almost the whole range of yields, and the two group averages (the white lines) sit only a little apart. Compare that with Marker 12 in step 2, where the two groups formed two separate bands with a clear gap between them. The yield markers are not fake, each one really does shift the average slightly, but on its own each explains only a small part of why one plant yields more than another. That small shift is exactly why these bars stay modest instead of towering like Marker 12.
One marker or many. Many. The evidence is in the shape of the plot. No single bar towers over the rest the way Marker 12 does. That pattern is what a polygenic trait looks like, one where many genes each contribute a small amount to the phenotype, and it is the normal situation for complex traits like yield, height or flowering time.
What that means for marker assisted selection. It largely breaks the method. Selecting on any one of those markers gains you only a sliver of the trait, and stacking them one at a time would take an impractical number of crossing rounds, with each round risking the loss of what you gained before. Worse, a marker with a small effect is hard to distinguish from a false positive, so you may spend rounds selecting on something that does nothing. The method needs a single marker carrying most of the effect, and yield does not have one. What you need instead is a way to use all the markers at once, adding up their small contributions into one number per plant, which is exactly what genomic selection does.
Extra: breeding for maximum yield. The first surprise is that the dashed elite line is a ceiling, not a target you can beat. Yield in this simulation is built from Markers 0 to 9, which each add to it, and Markers 20 to 22, which each subtract from it. The elite parent already carries a 1 at Markers 0 to 8 and a 0 at all three of the yield-lowering markers, so it is nearly the best plant these two parents can produce. It is missing only Marker 9, and the donor does not carry that one either. A cross can only shuffle the alleles the parents already have, it cannot invent one, so no amount of breeding here will take you above the elite level.
Getting there is easy: cross to the elite variety every round and the yield line climbs towards the dashed line on its own, even with no filter at all, because every round replaces more donor genome with elite genome. If you filter on a yield marker instead (keeping Marker 20 = 0 to shed one of the donor's yield-lowering alleles, say, or one of Markers 0 to 8 = 1) yield climbs a little faster, but watch the green line while you do it: salt resistance drifts away and collapses, because you spent your one filter on something else. That is the real lesson of the exercise. Single-marker selection lets you protect exactly one thing per round, so with a trait built from a dozen markers you are always trading one gain against another. Selecting on Marker 12 = 1 and letting the backcrosses bring the yield back is the best compromise available to you here, and doing better than that needs a method that can weigh all the markers at once.