okay so welcome to my lecture on selection on phenotypes okay so from the uh previous lectures you should now have a pretty good idea about the genetics of uh continuously distributed traits so the aim of this lecture now is to introduce you to some of the basic methods that we use in order to study patterns of selection on those traits on those continuously distributed traits and also so you understand the multivariant nature of selection so today's lecture is going to be broken up into three segments the first uh will cover basic patterns of selection are a quick recap of um some of the things we've already dealt with but then we get on to um the sorts of selection the forms of selection that we can see next we'll get on to uh talk about both direct and indirect and uh correlated selection and here you're going to start appreciating how selection can Target more than one trait at the same time or for instance how traits that are linked with Fitness may not actually be the true targets of selection and finally we're going to um expand our um basic um sort of Univar analysis that we do to take account of these um complexities if you like about how selection acts on multiple characters so I'll be introducing introducing to a concept called the multivar Breeders equation which is essentially the multivariate form of the unit VAR breeders equation which I've already introduced in my previous lectures okay so let's just consider now first um patterns of selection but before we do that a very quick recap so um you'll recall that there are basic elements required for phenotypic evolution to occur in the first place okay so obviously um here we can see in the top graph um I've given an example in this case from flowers um that we require there to be phenotypic variation for the trait that we're interested in and of course um there has to be um some additive genetic variance in those traits too which we get to which we've already talked about now critically of course in order for selection to happen on those traits uh there needs to be a relationship between variance in the trait of interest and an organism's Fitness so in other words there has to be some selective Advantage uh that's associated with having a certain trait um rather than another or a certain value of a trait I should say and finally of course as I've already said there has to be additive genetic variance underlying the trait of Interest so there needs to be heritable genetic variance underlying the phenotypic variance that we see in nature now we've already dealt with the heritability part of that in the previous lecture so I don't really want to cover that here what I want to cover are these two basic um uh elements of um selection so the fact that we need fintiv variance in the first place and obviously a relationship with Fitness and I now want to recall uh the concept of the unic Breeders equation which I've already bought in okay so you're recall that the univ gers equation says that the response to selection is a function both of the heritability of the trait in question and uh selection acting on that trait now we've already as I said dealt with heritability this lecture deals with the S part of that equation which is the selection differential and if you remember this is the difference between the selected progeny the track the the trait value of the selected progeny and the mean of the population that it came from okay so we need a simple way first of all of visualizing how selection occurs on phenotypes so let's just consider a simple biar plot here between affinic trait um again we could say for instance this could be petal shape in radishes and relative Fitness okay so this then would generate what we're going to um term throughout this lecture something called a fitness function so first of all let's just describe what these different axes actually describe now um on the y- axis there we've got a little term that looks like a w this is the symbol we use for relative Fitness and relative Fitness is a very difficult thing to measure if you're an evolutionary biologist what is Fitness but let's just think about this as something like lifetime production of Offspring okay or it could be something linked to survival um something that determines whether or not you are successful as an organism so relative Fitness is called relative Fitness literally because it's the Su the success or the fitness value of an individual over the population mean okay the average fitness for the population now the phenotype traditionally or usually in studies of phenotypic selection we standardize to a value that will have a uh mean of zero and a standard deviation of one and we call that a zed score okay and we simply calculate this quite easily from the individual phenotypic value we take that value from the population mean value and divide that by the population standard Dev ation that will then neatly turn any Fitness trait into something that we can pair across other traits because it's on a standardized scale okay we can then start to look at the fitness function okay now the fitness function is simply the relationship between the phenotypic trait of interest and our measure of relative Fitness okay and if we draw a line through that relationship in this case it's obviously a linear relationship between um trait variance and fitness um we have uh a a straight line which describes the fitness function which is essentially the strength of selection that acts on that trait so let's just explore what we mean by the fitness function so the fitness function is quite simply it describes the strength and the form of selection acting on the phenotype so the strength is the slope of that line that I've just shown you and the form is the shape of that line so there are three basic forms that we're going to consider in this first segment of the lecture that is directional sexual sele oh sorry directional selection stabilizing selection and disruptive selection so let's just consider each of those in turn so directional selection is quite simply a linear relationship either positive or negative uh between the trait of Interest which is the um the value of Zed uh and relative Fitness now we can describe that relationship with a basic linear regression equation which you should be familiar with with so this is basically saying that the relationship is uh a is The Intercept so where that line passes through the X through the y- axis I should say and bz is the slope um for um the change in the phenotype with Fitness okay and that describes for us the strength of selection so the slope between um phenotype and relative Fitness so let's just imagine that this is the underlying distribution of the trait um that we're interested in for the population that we're studying so we have a normal distribution here pictured on the right hand graph okay so this is just a frequency histogram or frequency distribution for the phenotype across our population now if we were to impose directional positive directional selection on uh on the on the trait described here we of course would move the mean Zed the value of altic the average value of ourant towards the right of that graph which I've pictured here okay so we're moving mean Zed um to more positive values because we have positive directional selection on that trait so of course we move the mean and this of course would be if you like our response to selection which would be described by our Univar breeders equation but notice also that we've reduced the variance in that trait and that's because um selection favors individuals at the extreme end of the distribution so this is termed truncation selection and this should erode variance in the trait let's now consider the second form of selection that can happen on the fite this is stabilizing selection this is a form of nonlinear selection so I'll just picture here a sort of classic case of stabilizing selection um and because we have nonlinear selection we have to use a different form of regression so we have a regression that's very similar to the first one in the sense that we're interested in the linear component too but we introduce a new term which I've called here y Z squ this is a quadratic term um and under pure stabilizing selection of course we would expect B the linear portion of that relationship to be zero and Y would be negative so you have an Ever decreasing slope which produces a um a form of selection that looks like this okay so now let's consider the underlying or the effect of selection or stabilizing selection on a given trait so if this is the the underlying distribution of um phenotypes in our population on the right hand side uh just like the previous one but now let's explore what happens to that when we impose stabilizing selection on it question here is how will stabilizing selection influence the mean and the variance of trait Zed well if you think about it there shouldn't be any change in the mean right so the mean should stay the same after generation of selection that's because selection is favoring individuals with intermediate phenotypic values for Zed okay so the highest Fitness if you look at the graph on the left is at that intermediate value of the trait so of course selection is going to continue to favor those individuals but what stabilizing selection does is of course it acts against the outliers so again we see a reduction in the phenotypic variance of the trait over successive generations of selection I want to give you a a tangible example of stabilizing selection in our own species in this case birth weight in humans okay so if we were to plot and this comes from a classic paper um published in the ' 50s this shows that um more we can see here the normal curve the gray curve on the figure shown here shows the distribution of birth weights for a population of humans and as you can see you have a normal distribution uh where the average birth weight is around 7B okay so it's in old old units around just 3 kilog or so now the green graph shows how selection acts on birth white and as you'll see you have um the sort of classic u-shape function here stabilizing I'm sorry um yeah stabilizing selection occurring now we can introduce that this is stabilizing selection because this green line is showing us the mortality of offspring of of infants okay and what you can see is that um the mortality of Offspring is lowest at intermediate birth size so if you like that's flipping that Rel that um distribution that I've shown you uh before that pattern of stabilizing selection I should say that I've shown before in its head by showing the mortality rather than the fitness so as you can see you have lowest mortality at um the intermediate birth sizes now um the it's quite obvious from this graph that state the um mortality curve here is coincides the lowest motility mortality I should say coincides with the um average birth weight for the population so that is to say that birth weight birth weight in humans has stabilized around the optimum value um and of course that's what we see in the population as our average birth weight the third form of um selection I'd like to uh go over today is disruptive selection okay disruptive selection um if you like is the opposite of stabilizing selection it's also a form of nonlinear selection but the key characteristic of disruptive selection is uh there's an intermediate minimum value for Fitness unlike stabilizing selection where it's maximized at those intermediate values and fitness is higher at the phenotypic extremes so either if this was size for instance of those leaves this would suggest from the graph from the left that Fitness is maximized when you either have very small leaves or very large leaves and that you have the lowest value of Fitness for those intermediate values so now think about with our underlying distribution here I've shown in the graph on on the right for our population the frequency of Zed in our population that trait Zed how will disruptive selection influence the mean and variance of this tra okay so it should be fairly obvious to you that what's going to happen of course is just like stabilizing selection we see no difference in the average um phenotype in our population uh this is because um selection is equally favoring those large and small individuals or for instance any extremes of the trait we're interest in so overall the average of course stays the same but importantly because we're now favoring outliers uh we see an increase in the variance in the trait under question so disruptive selection will tend to add variance to traits that nevertheless might be under very strong selection of course this is um enormous implications uh for uh why we see such high variants for instance in traits that are very closely aligned to Fitness and obviously um the potential for speciation where uh groups can diverge so far that they actually become different species I want to give one very um simple example of disruptive selection here from coo salmon now coo salmon um uh like many salmon uh exhibit what we term alternative mating tactics okay uh now in the case of um uh one such group uh we call those precocious par now there's a figure here that I've just shown shows something called a Jack at the top so that you've got two alternative life um history um if you like directions that fish or males can take they can either mature into these very small Jacks they're called precocious they don't go out to the ocean in order to grow and then come back they simply mature very very soon after um after um birth and what they are effectively are large bags of testies okay they they've specialized in a mating tactic that's called sneaky mating and they don't try to attract females now the alternative phenotype that a male can adopt is to become a large hooknose um uh male Now hook knows its term because they've got this very prominent hook which you can see from the picture on the right of your slide these are the fish that go out to sea they're an adris fish they go out to sea to feed they take a long time to grow so of course they pay a cost to get there but then they come back to the rivers to compete for matings and of course uh they have very high success in terms of competition over indiv idual that mature a smaller size so basically three things to consider here small and large males both gain access to females but they do so in different ways the Jack specialize in sneaking so they will steal if you like matings that are already happening by putting sperm spawning sperm onto um eggs that females are laying for these hooknose males whereas the hooknose males will specialize in fighting now it turns out that from work that gross has done uh and published in nature is that both tactics enjoy equal average fitness so selection favors both forms so of course you would expect to see disruptive selection on body size in this species importantly any male that chooses an intermediate strategy will do do poorly in either um tactics so they won't be good sneaks because they'll be seen coming they'll be too large to be effective at sneaking but they'll also be too small to compete effectively for matings now there normous evolutionary implications of disruptive selection I've just touched on the possibility of speciation okay um let's just consider these so first of all uh disruptive selection as I said will generate phenotypic and genetic variance which is really important because traits closely aligned with Fitness are thought to exhibit reduced variance because of course we expect selection sometimes to erode that variance well disruptive selection doesn't do that it can also result as I've already alluded to into adap active differentiation and and eventually speciation I want to give just one example of that here this is the Lord um Lord how Island and uh Below in this Photograph Montage you can see the Lord um how palms and here uh they've had a history of um what we term ecological character displacement whereby um these trees which had originated from the same species two species have basically adopted different strategies in terms of their soil preferences and therefore the timing of their flowering and eventually this has actually driven these populations apart to the point where they can no longer um they can no longer uh breed together if you like to use that term um because the timing of flowering is so different between these alternative morphs and this is eventually led to speciation which of course then opens up um uh um this sort of speciation that we see on Islands so that's the end of this segment now uh we're now going to move on to to um consider situations where selection can Target traits other than um the for instance the focal trait that we're interested in or indeed can Target multiple traits at the same time