In 2012 a team of Russian scientists managed to germinate a 30.000 year old seed of an indigenous Siberian flower, called the Silene stenophylla. The seed blossomed into a plant, whose flowers produced fertile seeds.
The Russian team found the seed in a 38 meters deep permafrost layer in Northeast Siberia. An artic squirrel, burrowed the seed there 30.000 years ago.
Moderne Silene Stenophylla, foto: shams bahari (Wikipedia)
The scientists also compared the prehistoric Silene stenophylla with the modern Silene stenophylla, that still exists in the Siberian flora. Even though the plants of the old and modern Silene stenophylla had the same morphology, the petals of the flower of the prehistoric Silene stenophylla had a different form compared to the petals of the modern Silene stenophylla. The different shape of the petals of the old and modern flower revealed that the old seed contained other gene types than the seeds of the modern Silene stenophylla.
Not only did the 30000-year-old seed grow into a plant and blossomed, it also produced viable seeds. Therefore, the seed was able to bring old gene types back into the population of modern Silene stenophyllas. This storage of seeds with old gene types that can revive later on, is an example of a natural seed bank.
Natural seed banks occur more often in nature than you would expect. There are not only many plant species that build a seed bank in the ground, but there are also bacteria populations that are able to build a seed bank. Probably you can imagine that seeds who germinate only after a few years can help a plant species to survive, for example in case of drought. But the consequences of a seed bank are bigger than that.
For example, micro biologists found that the resistance of bacteria to antibiotics is closely related to seed banks. Also, worldwide there are artificial seed banks build to help our crops survive the climate change. We will show how this seemingly invisible phenomenon changes the evolution of species and why its consequences could be so big.
Mimicking a natural seed bank to survive climate change
A natural seed bank arises from seeds of plants that fall into the soil, but do not germinate in the next spring. This happens for example if a storm covers the seeds with thick layer of sand. Or when, like in the case of the Silene stenophylla, a squirrel burrows the seeds, but does not dig them up later on. If some years later the layer of sand disappears and the weather is good, the seed can still germinate.
The seeds in the soil that do not germinate directly form together a repository of genetic material. While the evolution of the plants continues aboveground, the seeds in the soil store older genetic information. When after some years such an “old” seed germinates and grows into a plant, this plant will have another older gene type than the plants that grew from seeds that dropped in the soil last autumn. When a new disease threatens the plants, this genetic diversity can help the plant population to survive the disease, because the more different genetic types of the plant there are, the larger the probability that there is one that is not susceptible to the disease.
Plant species with a seed bank do not only survive diseases more easily than species without seed bank, they are also better fitted to deal with changing climates and new predators. Again, a larger diversity in gene types enhances the probability that there is a gene type that can deal with higher temperatures, more humidity or that is toxic to the new predator.
Therefore, to protect our crop to climate change and new predators, people all around the world collect seeds in artificial seed banks. One of the biggest seed banks is the Svalbard Global Seed Vault in Spitsbergen.
Hundreds of thousands of seeds are nowadays stored in this naturally low temperature, nuclear-proof seed bank that is excavated in a mountain. Again, the idea is that with the seeds there we are able to breed new crops, or simply revive old crops that are able to cope with new environmental conditions.
Svalbard Global Seed Vault, foto: Bjoertvedt (Wikipedia)
Bacteria that build seed banks
More surprisingly than the seed bank of some plants species is that also bacteria populations show a seed bank phenomenon. When there is not enough food for a bacteria population, a part of the population can decide to become dormant. A bacterium that becomes dormant forms a so-called endospore. An endospore does not need any food or water and does not reproduce itself. If the level of nutrients in the population is again sufficient, the endospore will wake up and start reproducing itself. In this way the endospores form a seed bank.
Some of the endospores formed by bacteria can survive extreme conditions. Apart from the fact that they do not need food, they can be immune to ultra violet light, high temperatures, extreme freezing and even some chemical disinfectants. Endospores can become thousands of years old and still produce a bacterium. The oldest viable endospore was found in the stomach of a 250 million years old fossil of a bee.
In the same way as we saw for plants, the seed bank in bacteria populations increases genetic diversity. However, in this is case the increase of genetic diversity is not necessarily positive. During an antibiotic treatment the bad bacteria can transform themselves into endospores. These endospores are able to survive the antibiotic treatment. When treatment is over, the endospores can transform back to a viable bacterium. In this way the endospores play an important role in the formation of antibiotic resistant bacteria.
How big is the effect of seed banks?
Here mathematics enters the story. The seed bank seems to have large consequences, but how can these be measured? To do this the concept of genetic drift is used. Let’s see how genetic drift work in a concrete example, a meadow with poppies.
A meadow with poppies
Suppose one day you inherit from a great uncle a tiny, wooden, well maintained cottage on a small hill. When during the summer you visit your new property for the first time, the meadow surrounding the cottage is spoiled with poppies, purple and red. It is this view of the mixture of red and blue poppies in the high green grass what you appreciate the most of your new little house.
After a day of reading in your great uncles rocking chair in front of the cottage, while enjoying the sounds of buzzing bees, you suddenly ask yourself: “Is it possible that one type of these poppies in my meadow will die out and my cottage will be surrounded by only purple poppies or only red ones?”
So, the question is: How does genetic diversity evolve in your meadow with poppies?
Each of the poppies in your meadow produces many seeds. A large part of these seeds will not germinate. Maybe they are flooded by rain, blown on a paved road or a rabbit has eaten them and pooped them out in less favorable soil. In other words, the seeds that germinate are the lucky ones. Just by chance they end up in a place where the conditions are favorable to become a poppy themselves.
In this way chance plays an important role in the genetic evolution. In biology this chance factor is called genetic drift.
Hence to predict how the genetic composition of the population of poppies in your meadow will look like next year, some probability theory is needed.
To keep the numbers tractable, suppose for the moment that there are just eight poppies in the meadow, five purple ones and three red ones. Additionally, each of the eight poppies produces the same amount of seeds and each of the seeds has an equal probability to be the lucky one and germinate.
Suppose you leave your cottage and you come back one year later and you see again the same amount of poppies in your meadow, in our example eight. The poppies of last year have died out, their seeds fell on the ground and some of them germinated. How will your meadow look like one year later? It could be full of red, full of purple or a mix of the two. How probable are these scenarios?
Let’s do some computations. You had eight poppies in your meadow. A fraction of five out of the eight of the produced seeds will have the purple gene, since they descend from a purple poppy. A fraction three out of eight of the seeds will have the red gene, by descending from a red poppy. Therefore, each seed that will germinate and flower has a probability of 0.625 (5/8) to become a purple poppy and 0.375 (3/8) eight to become a red poppy. In the figure above you see one of all the possible outcomes of our meadow in the second year. Below you can click to see what the probability is of three purple and five red poppies in the next generation.
The probability to have in year two exactly five red poppies and three purple ones equals:
The first term indicates in how many ways you can order 5 red poppies and 3 purple poppies, that is
Now we know something about the odds of the composition of our meadow in the next year, we can try to predict what will happen to the diversity in the meadow. We will call the meadow diverse if there are both red and purple poppies. If we use the above computations, we see that with probability 0.375 the first germinating seed has a red gene. Again, with probability 0.375 the second germinating seed has a red gene. Hence with probability 0.141 (0.375^2) the first and second germinating seed have both a red gene. Continuing like this we see that with probability 0.0004 (0.375^8) all eight germinating seeds have a red gene. Since a seed has a purple gene with probability 0.625, we see that with probability 0.023 (0.625^8) all the eight germinating seeds will become purple poppies. Note that diversity is lost if there are only red or only purple poppies left. Therefore this will happen with probability 0.0234 (0.023+0.0004).
The Wright-Fisher model
The computational example with eight poppies is actually a special case of the Wright-Fisher model. The Wright-Fisher model gives us the opportunity to predict the genetic composition of the meadow in the next summer. The beauty of the Wright-Fisher model is that it does not only apply to poppies, you can use it to make predictions about the future evolution of many types of plants or bacteria.
What is this Wright-Fisher model exactly?
In the Wright-Fisher model we have a fixed number of individuals that form together a population. In our example these individuals were the eight poppies. Each individual in the Wright-Fisher model has one of two gene types called P and R. So, in the example we could have called the purple gene type P and the red gene type R. When there is a new generation, which happens with the poppies every year, the individuals in our population reproduce themselves. Since we have a fixed number of individuals, the number of individuals in the new generation is the same as the number of in individuals in the old generation.
The chance that an individual in the next generation is of type P equals then
Number of individuals of type P in the current population
In the figure below a possible scenario is shown for the next three generation of our population on the left and the corresponding scenario in the Wright-Fisher model.
The figure above shows a possible scenario for the evolution of our meadow. The letter t indicates the year. On the right we have described the corresponding outcome in the Wright-Fisher model. The black lines denote the parents of the poppies. So the two purple poppies on the left in year 1 both descend from the most left purple poppy in year 0.
In each population where individuals have one of two gene types and they have only one parent who determines this gene type, you can use the Wright-Fisher model. Also, you can adapt the number of individuals in your population. You can study meadows with 20 poppies or even 200 poppies. Using this Wright-Fisher model and a lot of probability theory, mathematicians are able to say something about the expected diversity in a population. By doing these computations they can give you an estimate about when they think genetic diversity will die out. And for your meadow with 8 poppies they expect that it will take more than 4 generations before all the poppies are of the same color. With probability 0.625 there will be only purple poppies left by that time and with probability 0.375 only red poppies will be left.
When you would have started with 12 red and 8 purple poppies the Wright Fisher model predicts that after approximately 11 generations diversity will die out. The probability of only red ones left is then 0.6 (12/20), while the probability of only purple poppies left is 0.4 (8/20). In case of a meadow of 60 red poppies and 140 purple poppies, we expect it to take roughly 85 years before one type extincts and now the red poppies survive with probability 0.3 (60/200), while the purple ones survive with probability 0.7 (140/200). So, we see that which gene survives does only depend on the number of red and blue poppies you start with, while the number of generations you have to wait before extinction happens grows if the number of poppies grows.
The secret of the poppy
Year after year you visit your tiny wooden cottage on the hill and each year there are still red and purple poppies welcoming you upon arrival. Even after 8 years, you can still enjoy the mixture of red and purple poppies in the meadow, much longer then the mathematicians told you. So, are you just extremely lucky?
Or maybe not….
The poppy is one of the plant species with a seed bank. Therefore, to study the diversity in the poppy population, we should adapt our model.
The Wright-Fisher model with seedbank
The Wright-Fisher model with seed bank was introduced by Jochen Blath, Adrian Gonzalez Casanova, Noemi Kurt and Maite Wilke-Berenguer in 2016. This model is more involved than the model without seed bank, but it still allows us to do some calculations.
After 10 years you drive to your cottage and see, indeed only purple poppies are left. Maybe you are a little bit disappointed, but you knew it was once going to happen. But then, a year later you return to your property and are welcomed by a lot of purple poppies, but also one red poppy. And one more year later there is even a big part of your poppies red.
Maybe you can already guess what has happened. To make it clear you can look at the figure below. In a certain generation you see no genetic diversity anymore on the meadow. In a population without seed bank this would have been the end of the story. But here genetic diversity was hidden in the seed bank and both type of poppies come back in the population. This is exactly how a seed bank increases the genetic diversity.
The seed bank, an astonishing phenomenon
The poppy example has shown us how a seed bank can prolong the diversity in a population. Instead of poppies, we could have had depicted the Siberian flower Sylene Stenophylla, or even a colony of bacteria. Even though the model is a very simplified version of the reality, it illustrates how the seed bank can increase genetic diversity.
Researchers investigating the Wright-Fisher model with seed bank even found that a seed bank can prevent extinction of a species when the seeds can stay viable in a seed bank for very long time. A beautiful example of this is the Siberian flower Silene stenophylla. It shows us how valuable the storage of genetic material can be. Although a seed bank effect in bacteria populations is not always favorable, the seed bank phenomenon in plants is a useful tool to maintain genetic diversity. Therefore, to defend our crops against the coming climate changes, let us fill the Svalbard Global Seed Vault and all the other seed banks in the world.
We owe the featured image to Zbynek Burival from Unsplash.
Would you like to stay up to date whenever a new post appears on the Network Pages? Then subscribe to our mailing list, follow us on Twitter or on LinkedIn.
In February 2014, a big bank in the Netherlands suffered from an internet banking malfunction which led many costumers to accidentally perform duplicate bank transfers.
Applications of machine learning models are everywhere, with many online platforms and major science fields using tools relying on machine learning. Take, for example, image recognition and computer vision. But did you know that the results of supposedly perfect and accurate machine learning models can be deceived by slight perturbations in the data?
Between 1973 and 1986 multiple rapes and murders were committed in the state of California. Years later the idea was raised that these crimes might be connected. But traditional DNA analysis from the samples found at the crime scenes, could not identify the culprit.
One of the main building blocks of modern AI-tools are artificial neural networks, abstract models inspired by the structure and functions of biological neural networks which enable machines to "learn". In this article, I will discuss some thoughts on this topic.
