EVOLUTION – Biology 4250
                                                                                                                        Dr. Adams
Review Sheet Number 2 – Test 2

In most cases, you will be responsible for examples of the concepts that are presented in the
text book, as well as the other examples that I present in the classroom.

Chapter 6:  Mendelian Genetics in Populations -- Selection and Mutation
    You should know how to use a Punnett Square to determine potential offspring from a
cross between two parents with specified genotypes.

The Hardy-Weinberg Equilibrium Principle:  A review
    Populations with no selection (or mutation, genetic drift, etc.) should not evolve, meaning no
change in allelic frequencies from generation to generation.  If such populations existed, then
frequencies of both the alleles in the population as well as genotypes of individuals should remain
    We will talk about the Numerical example involving "mice" on pages 181 to 188, or some similar
case where we can actually COUNT the alleles in a fictitious population, then apply this to the more
general case of the H-W Equilibrium principle.
    The General H-W equation (for a trait with two alleles, one dominant/one recessive):
            p = frequency of dominant allele (freq. A)            Hopefully it is clear that in this case
            q = frequency of recessive allele (freq. a)                 p + q = 1 (which represents 100%)
        Additionally, inserting frequencies into the Punnett Square (see pages 184 & 187),
            p2 = frequency of homozygous dominant individuals in the population (freq. AA)
            (pq + pq) = 2pq = frequency of heterozygous individuals in the population (freq. Aa)
            q2 = frequency of homozygous recessive individuals in the population (freq. aa)
                            And again, hopefully it is clear that in this population,
                                p2 + 2pq + q2 = 1 (100% of the individuals in the population)
    YOU WILL be expected to reproduce the above equations and to be able to use them to figure
        out frequencies in (artificial) populations (a homework assignment is coming up next week).

Are populations in H-W equilibrium, and, if not, what use is the H-W principle?
    Clearly, the H-W equilibrium principle has several assumptions:
        1.  No selection; individuals contribute equally to future generations regardless of phenotype.
        2.  No mutation.
        3.  No immigration (followed by mating) or emigration (pop in isolation) -- no GENE FLOW.
        4.  No chance events that allowed some individuals to mate a lot more, or chance events that
                killed individuals of a certain genotype.
        5.  Mating is random; no mate choice based on mate characteristics.

    The point?  Clearly, NONE of the above assumptions are likely to be true in any natural population.
So what use is the H-W equation?  The H-W equilibrium is, in essence, the NULL hypothesis for
evolution, because if the allelic frequencies are NOT in equilibrium then it means  . . .  evolution.
And, since the assumptions are clearly not met, then what can we say?  EVOLUTION is occurring!

Selection and its effects
-- testing assumption #1 of the H-W principle
        As we already know, different phenotypes have different fitness, based on how well adapted
individuals are to the current environmental conditions -- remember, fitness has two components:
1) survival to reproductive age and 2) reproduction once reaching reproductive age.  The point is that
there are selection pressures (weather conditions, food/water/shelter/mate availability, etc.)
on individuals in the population, and different phenotypes may fair better or worse.  It can be complex,
however, since an individual better at getting shelter will not necessarily be better at getting mates.
        On pages 192 - 193, selection is added to the mouse example (from earlier) such that one of the
alleles is now somewhat detrimental in the heterozygous individuals, and even moreso in the
homozygous individuals.  End result?  Decrease in the frequency of that allele by the next generation.
The experimental example of altering AdhS/AdhF allele frequencies over 50 generations in Drosophila
(fruit flies) by providing ethanol in the diet of some strains but not in others, is a simple
and elegant example of selection.  Other examples (like HIV resistance in humans due to the CCR5-Δ32
) are also discussed.
        You should be able to CALCULATE allele frequencies from one generation to next when given
simple selection for/against percentages for different genotypes.

Patterns of Selection
    Selection on recessive/dominant alleles -- the Tribolium example, with a recessive lethal allele;
           Huntington's chorea, a dominant allele in humans
                  Which is stronger?  Selection in pops with recessive or dominant lethals?
    Selection on homozygotes/heterozygotes -- eg., sickle cell heterozygotes in malarial regions
            Overdominance -- selective advantage for heterozygotes.
            Homozygote advantage (underdominance) -- eg., Drosophila compound chromosomes
    Frequency dependent selection (most often this involves selection for rare morphs) examples: 
            1.  Yellow and purple elderflower orchids (see text pgs. 213 - 214)
            2.  male mosquito "song" pitch (females mate more frequently with males whose wing beat
                   frequency is different producing a different pitch in their "buzz"
            3.  snail shell pattern and "search images" formed by bird predators; search images formed
                    much more easily for the more common patterns, even if uncommon patterns might stand
                    out a bit more
        The end result of frequency-dependent selection is to maintain variation in populations, as can
            overdominance, as indicated above.

Mutation and its effects -- testing assumption #2 of the H-W principle:
        We've already talked about how mutations occur, and to an extent the effect mutation can have
on allele frequencies.  Clearly, one effect is establishment of completely new alleles. More often, how-
ever, is mutation from one allele to another, resulting in a change in allele frequencies. Remember, how-
ever, that mutation rates will NOT be a potent force in short term changes in frequencies of alleles
in populations.  Mutation rates are typically QUITE small, most are somatic (not passed on) and
most are neutral or detrimental (at most eliminating one individual at a time from the population, if that).
        However, mutations can occur that convert a dominant allele to a recessive and vice versa, which,
as long as both alleles are selectively equal, can alter allelic frequencies with "no harm done".  From the
example in the book (pgs. 217-218), you will note that there is still little effect, but "little effect" is NOT
the same as "no effect".  Still, mutation will never be a major player in the short term, but do not forget
that in the long term, mutation is EVERYTHING, supplying the new genetic material for real change.

Mutation and Selection
        So, as we have already discussed, mutation, which by itself would alter frequencies at a snail's
pace, COMBINED with selection, becomes a potent, indeed THE potent, evolutionary force.  This
is perhaps most clearly seen in populations of organisms that cannot recombine genetic material, i.e.,
can only reproduce asexually.  With mutation, followed by selection, even asexual strains can be altered
through time.  Without mutation, only natural clones would be produced and evolution would grind to
a halt
.  Even rapidly reproducing sexual organisms can also exhibit rather quick change this way, how-
ever.  See the Drosophila example on page 219.
        As mentioned above, however, in most cases, there will be a mutation-selection balance.
Since many mutations are deleterious, selection tends to eliminate the mutations, not reinforce them.
        Occurrence of Cystic fibrosis genes as an example of mutation-selection balance -- as it turns out,
heterozygotes for CF alleles tend to be significantly more resistant to typhoid fever (see page 223-224).

Chapter 7: Mendelian Genetics in Populations: Migration, Genetic Drift and Nonrandom
                            Mating. The H-W assumptions 3 through 5
   Note: the chapter starts out with a discussion of the Florida Panther -- declining health and populations,
etc. – make sure you read the introduction, as we will come back to a discussion of the Florida Panther
towards the end of the chapter.

        From an evolutionary standpoint, migration is the movement of alleles between populations. This
means immigration/emigration of individuals followed by mating by these individuals – in other words,
gene flow
from population to population. 

     Example of Empirical Evidence on influence of migration on allelic frequencies:
         Water Snakes on mainland Ontario/Ohio and the islands in between in Lake Erie.
              The Water Snake (Nerodia sipedon) varies in pattern from very plain tan (unbanded) to
strongly banded with darker brown. On the mainland, all populations seem to be completely strongly
banded forms; on the islands are lightly banded to unbanded tan forms. Since the snakes have a unique
basking "platform" of limestone along the shores of the islands, it would seem that the unbanded form,
especially in the young, would be much better protected from predation. Indeed, mark-recapture
of snakes from juvenile to adult stages directly indicates greater survival rates in the unbanded
forms than any of the banded forms. So, how come there are ANY banded individuals on the islands?
Answer: continued migration from mainland, with subsequent mating (gene flow). A related question:
How come those on the mainland are virtually all strongly banded?? Answer . . . ???
                    NOTE:  Migration is working in opposition to selection on the islands.
            In general, gene flow (migration) tends to homogenize populations, making the populations
more similar to each other (which can offset to an extent the different selective pressures the populations
are experiencing). So, gene flow reduces differences between populations, but can (though doesn’t
necessarily) increase variation within populations by sharing more alleles.

Genetic Drift
            This concept involves any change in allelic frequencies due to chance events; typically these
changes are much more evident when population size is small (as you will see). These chance events
be anything from "sampling error" in selection of gametes to potentially catastrophic events that cause
death of individuals at random. This, in essence, is "blind luck" functioning as a mechanism of evolution.

        Mathematical model of drift:
            In a hypothetical and very small population of ten mice with a starting freq. of A = 0.6 and freq.
of a = 0.4, selecting gametes at random from the population to produce a new population of ten mice
will result in a filial population in equilibrium with the parental (p = 0.6, q = 0.4) only about 18% of the
time (see Fig. 7.9, pg. 243). Although this is hypothetical, it does show that changes can result solely
as a function of chance events.
            Why is population size so important? The in-class "falling rock" example.

        The Founder Effect -- sampling "error" as a mechanism of evolution
            Where (and/or when) are populations naturally small?
            The most likely occurrence of populations that are small is when new populations are being
. The founders are a small subset of the parental population, and, by chance, the frequencies
of alleles in the founders can therefore be different from the averages in the parental population. 
Sometimes, founders could even be single individuals, such as gravid female arthropods.
            Polynesian Field Crickets (Teleogryllus oceanicus) in Australia and Pacific islands (p. 244).
            Humans can show founder effects as well. Achromatopsia is seen in 1 out of every 20 (of
3000) Pingelapese people, whereas in the world population the frequency is less than 1 in 20,000.
This is a founding effect due to a typhoon that hit the Pingelap Atoll in 1775, leaving 20 survivors,
one of which was heterozygous for the condition.

            With nothing else influencing allelic frequencies (which, of course, rarely if ever happens), drift
tends to decrease heterozygosity, and fix alleles in the population, though this is not inevitable,
especially in large populations and with other factors influencing allelic frequencies. Still, alleles could
conceivably be fixed by drift, even in somewhat larger populations (SEE pg. 247). This would result
in a loss of variation.

            So, genetic drift can be an important evolutionary event because:
                1. EVERY population experiences drift, which means EVERY population follows its
                        own unique evolutionary path.
                2. Given enough time (without other significant influences, a MAJOR assumption),
                        drift can produce substantial change, even in fairly large populations.
                3. Small populations may be strongly effected by drift in fairly short time periods.
                4. Genetic drift tends to reduce variation within populations, though increase
                        differences between populations. This, of course, would be offset by migration
                        (and moderated by selection effects).       

        Experimental Evidence for fixation of alleles:
            Brown eye alleles and Drosophila melanogaster (see page 250)
                Started with allelic frequencies at 0.5.
            After 19 generations of 16 flies (eight males, eight females), out of 107 lines, 30 had lost the
brown eye allele completely, 28 others had it fixed at a frequency of 1 (though the overall frequency
for all lines of the brown and "normal" eye alleles for all lines remained close to 0.5) -- as expected,
this reduced heterozygosity and approximately half of the lines are fixed after 19 generations.
        Genetic Bottlenecks: Examples of Empirical evidence for genetic drift
                The Ozarks Collared Lizards – seven distinct fixed genotypes among the populations
                        (see Fig. 7.19, pg. 253)
                Among four species of plants (Fig. 7.20, pg. 254), smaller populations almost
                        invariably had lower heterozygosity and polymorphism.
                Separate breeds of dogs

            How quickly can NEW alleles "take over" (become fixed)? How fast does evolutionary change
by drift proceed? New alleles are, of course, produced by mutation. Those that are disadvantageous
may immediately be eliminated (though may reappear through mutation). Some, however, may persist 
at very low levels. Neutral mutations, which, of course, include silent (synonymous) mutations, have
drift as a MAJOR influence in their evolution, (see Fig. 7.21, pg. 255) and a new neutral allele may
be substituted for another by drift over the course of time. Of course, a mutated allele with a selective
may more rapidly substitute for another.

Genetic Drift and Molecular Evolution -- some salient points
        Silent substitutions (as defined immediately above) are far more common than replacement
    substitutions for a wide variety of proteins (Fig. 7.25 pg. 262), with some significant variation.
    Genes responsible for the most vital cellular functions have the lowest replacement rate (not sur-
    prising).  By the way, why am I talking about mutations in this section on genetic drift?

Positive selection -- in some genes for some organisms, there appears to be selection FOR higher muta-
        tion rates, meaning that mutation must produce more positive effects for these genes than "usual". 
        In these cases, replacement mutations outnumber synonymous (silent) mutations (pgs. 265-266).
        Examples include the ARS (antigen recognition site) of MHC class proteins involved in specific
        immunity.  For others, see "Which loci are under strong positive selection?" on page 268.

Selection on "Silent" mutations
    If silent mutations were truly silent, then we should see EQUAL distribution of various codons that
        code for single amino acids.  We do NOT see this.  We see a significant codon bias for particular
        codons over others, especially in highly expressed (more frequently transcribed) genes (See Figs.
        7.30 and 7.31, pg 269).  How is such selection possible?  Why would it happen?
    The leading hypothesis is selection on translational efficiency -- some tRNA's more common than
        others.  This may indeed explain why "silent" mutations do not accumulate as rapidly as mutations
        in pseudogenes.   

Fixation of non-selected alleles by "hitchhiking", or selective sweep -- chromosome number 4
    in certain Drosophila species (melanogaster and simulans, page 270-271).  No recombination
    (crossover) takes place along its entire length, so the entire chromosome is inherited as a single
    linked set.  Strong selection for one allele on chromosome #4 can "sweep" other alleles around
    it to fixation.  Indeed, researchers (Berry, et al, 1991) found virtually no polymorphism in a 1.1
    kilobase section of chromosome in modest sized samples of these species. 
        On the flip side, negative selection can reduce frequency of closely linked alleles as well.

Nonrandom Mating
        Nonrandom mating, which virtually always indicates some mate selection, probably occurs in
virtually all populations of living organisms. It should be pointed out that nonrandom mating does not
necessarily drive evolution (change allelic frequencies). Nonrandom mating can occur different ways,
including inbreeding and mate choice (sexual selection). We will concentrate on inbreeding here.

            In chapter 7, the authors talk some about inbreeding and its effects (the founder effect).
Inbreeding occurs because the breeders involved are a small subset of the overall population.
        An Empirical Example – Inbreeding and Inbreeding depression
            As suggested above under the genetic drift section, inbreeding will reduce heterozygosity
(increase homozygosity).  This, in turn, leads us to the concept of inbreeding depression.  Sea Otters
indeed show lower than predicted numbers of heterozygotes in natural populations, if the populations
were mating at random.  There are LOTS of examples we can point to, including lots of self-fertilizing
plants (Pink Lady Slipper orchids, the Cheetahs mentioned above, egg hatch in many birds (see Fig.
7.39, pg. 282), and even humans (see Fig. 7.37, pg. 281). Needless to say, the vast majority of
organisms have evolved mechanisms to avoid inbreeding.  Mate choice (with the ability to recognize
close relatives), dispersal (migration) drives, and self-incompatibility (in plants) are all important
mechanisms for avoiding inbreeding. Still, in small populations, inbreeding may be unavoidable, and this
may present a formidable challenge when trying to save rare and endangered species which, needless
to say, may be represented by one or a few small populations.
            So, what does all of this have to do with Florida Panther? Hopefully, by now, you’ve figured it
out! What DOES all this have to do with Florida Panther?  Reduced population size and isolation have
led to a genetic load of somewhat deleterious mutations,  a "mutational meltdown" if you will -- inbreed-
ing depression for sure. The solution?  Import pumas.  In 1995, eight Texas pumas were introduced into
Florida, and heterozygosity and numbers have improved.  It is significantly less likely to go extinct now.

Chapter 8: Evolution at Multiple Loci: Linkage and Sex

        If we start from the model of selection presented in Chapter 6, the model predicted very well the
course of evolution in flour beetles (Tribolium) over 12 generations (page 203) – a powerful model
under the circumstances. However, it must be pointed out that the conditions under which the flour
beetles were grown were very controlled, and the evolution investigated involved a recessive lethal
allele. As the authors correctly point out, for other alleles under complex environmental conditions
the price of mathematical modeling is oversimplification.
        In this chapter, we WILL learn how to apply a model to circumstances looking at two (or more)
alleles at the same time. This may seem hopelessly abstract, though there are two payoffs to this
approach: 1) this approach can be used to reconstruct history of populations and genes (we will finally
delve a bit into the CCR5-∆32 allele that provides some HIV resistance, where it came from, and why
it is at the moment virtually only found in Europe), and 2) this provides insight into why organisms may
use sexual reproduction (as opposed to asexual).

Evolution at two (or more) loci: Linkage Equilibrium/Disequilibrium
        Investigating two different genes at the same time means that technically those two genes could
be anywhere on the chromosomes. In this section, we will talk for the moment about genes that are
Linked genes, of course, are passed together to gametes, giving those gametes (and the
chromosome) their respective haplotype.
        We will go over the numerical example discussed on page 293. Understand that "g" = the
frequency of whatever follows "g"; "D" = coefficient of linkage disequilibrium; and "r" = the recombina-
tion rate (the crossover frequency between genes). A crucial point about this numerical example is
that the two populations on the page have equal individual allele frequencies.

        Linkage Equilibrium
– when alleles of two different genes appear to be inherited independent of
each other; the freq. of any haplotype can be determined by multiplying the frequency of the individual
alleles; in other words, freq. of AB = freq. of A X freq. of B, and so on. With this, the coefficient of
linkage disequilibrium
(D = gABgab – gAbgaB) is zero.  (Hopefully, it will also be clear that genes on
chromosomes will be in "linkage" equilibrium, as they are, of course, NOT linked!)
        Linkage Disequilibrium
– a nonrandom association between alleles at different loci.  This is due
to the genes being physically linked. Disequilibrium can be generated in three different ways:
1) selection on multilocus genotypes, 2) genetic drift, and 3) population admixture. D will not equal
if linkage disquilibrium exists, and for reasons that will become apparent, D can range only between
0.25 and -0.25.

Linkage disequilibrium – possible causes (besides being closely linked).
        Continuing with the numerical example from page 293, if we add selection against any individual
having two or more recessive alleles, we end up with the results as shown at the top of page 298.
As you can easily see, this population will now be in linkage disequilibrium. In this example, there is
multilocus selection – selection acting on BOTH genes.
        In a finite population, a mutation followed by selection can lead to linkage disequilibrium. It is the
mutation (a chance event) happening once (infrequently) that led to the disequilibrium (see Fig. 8.5 and
text on pg. 299).
    Population admixture
        If you have two populations with different frequencies of haplotypes, if they are then mixed this
will establish new frequencies of haplotypes that can easily be considered in disequilibrium.

Reduction/Elimination of Linkage Disequilibrium
        Genetic recombination (crossing over) is THE event that reduces/eliminates D – indicates why
sexual reproduction
(meiosis) is an important part of this discussion.
        A crucial aspect of this is that the more closely (physically) linked the genes are ("r" close to
), the more difficult to remove the disequilibrium (see Fig. 8.7). The reduction of linkage disequilib-
rium has been demonstrated in the lab (see Fig. 8.8 for a Drosophila example).

Why does this concept matter?
        . . . Because if genes are linked, selection for an allele at one locus will INFLUENCE frequencies
of (most/all) alleles that are linked to it. What this means is that if A is linked to B, and there is selection
on A, it can change the frequency of B. So, someone studying JUST the frequency of B/b could get the
mistaken impression that there was selection against B, when in actuality selection against A is reducing
the frequency of B. This is the concept of "hitchhiking" mentioned above.
        So, one would EXPECT linkage disequilibrium when there is a strong selective advantage or dis-
advantage for certain alleles. See the VERY intriguing study done with human chromosome #5, the
ergothioneine transporter gene and frequency of Chron's disease.  The apparent link between the trans-
porter and disease, however, is due to linkage on the chromosome and not an actual causation.
In general, observed disequilibrium in genes studied is quite low, suggesting that even for those that are
physically linked, crossing over is frequent enough to bring "D" close to zero. For populations that are
significantly inbred, even genes on separate chromosomes can appear to be in linkage disequilibrium.
But even occasional outbreeding appears to significantly reduce linkage disequilibrium.

A Practical Application -- the GBA-84GG allele and the CCR5-∆32 allele
        Both of these are loss-of-function mutations, the first exclusive to Ashkenazi jews, the second 
in Europeans. Remember that when a mutation first occurs that automatically puts it in D with
surrounding alleles.  So as D decays, you can follow this and make predictions about the origin of the
mutation and the time left to equilibrium (see Fig. 8.17).  Also, this allows you to predict that alleles
that are in SIGNIFICANT D with surrounding alleles are YOUNG (new).
        The CCR5-∆32 allele is a loss-of-function mutation at the CCR5 locus, such that HIV cannot
enter target cells. Individuals homozygous for the ∆32 allele are protected from sexually transmitted
HIV strains. So . . .
        Where did the ∆32 allele come from? Why is it only in European populations (at the
        An analysis of chromosome #3 shows that the CCR5-∆32 allele is found almost
exclusively together with the marker GAAT and marker AFMB (two non-coding regions with
no effect on fitness very close to the CCR5 allele) in humans with the HIV immunity. So, this
indicates that a mutation resulting in the ∆32 variant occurred just once, on the chromosome with
the two markers; the end result is that ∆32/GAAT/AFMB is inherited as a unit.
The linkage disequilibrium is breaking down a bit, as crossing over has resulted in other haplo-
types. Estimates of recombination rate (crossing over frequency) and mutation rate originally put
the estimate of the origin of the ∆32 allele at around 700 ya (Stephens, at al, 1998). However,
since that time, it has been shown that the original chromosomal map was a bit flawed, and
the presence of ∆32 in bones of 2900 year old humans in a Lichtenstein cave, and further marker
crossover studies suggest the origin of the mutation was a approx. 5000 years ago. 
        So, is the allele under selection, or could drift have increased the frequency of the allele to
its current level (somewhere between 10 and 20 percent) in European populations?  The bones
discovered above suggested a frequency of about 12% at that time, which means the changes
since 2900 years ago COULD be due to drift.  The %age has not changed much.  Selection
COULD be involved, but it is not clear that it HAS been involved. Selective forces are not
known (though see immediately below), and if the mutation has occurred elsewhere, it has NOT
persisted (not been selected for).  There are a couple possibilities as to what selective pressures
could have been (other epidemics): the bubonic plague ("black death") that struck Europe during
the 14th century; and smallpox.

Chapter 9:  Evolution at Multiple Loci: Quantitative Genetics.
    Selection on Quantitative Traits – Quantitative Genetics
Traits showing continuous variation are called quantitative traits – such traits typically
involve additive affects of many genes, as well as some environmental influence. Two quick
examples in us: height and skin color (in humans). These traits do not exhibit an either/or
phenotype (either you have it or you don’t, which is what you see with traits controlled by a single
gene with two alleles). Quantitative traits tend to show normal (or near normal) distributions (with
the associated bell-shaped curve).
        For these traits it is appropriate to ask: What fraction of the variation in height is due to variation
in genes, and what fraction to differences in the environment? In other words, we are looking for the
(broad sense) heritability of these quantitative traits.

h2  =  Heritability  =  VG   =      VG__       P  = phenotypic, G = genetic, E = environmental
                                 VP       VG + VE

Furthermore, h2   =   Heritability   = VA         VA______    A = additive, D = dominance
                                                        VP     VA + VD + VE

The second equation above represents what is called the narrow sense heritability, and is that which
is due ONLY to the effects of additive genes (NOT typical dominant/recessive variation).

The concept of variability being both environmental and genetic is actually quite easily testable: we
will do a height plot similar to what is shown on page 330. Figuring out HOW the genetic and environ-
mental interact, and how MUCH is genetic, is more difficult.
        Remember, offspring can resemble parents due to similarities of environments as well, so to truly
"figure out" the heritability, you need make sure that similar environmental influences are excluded from
the analysis – this is none too easy, and certainly not viable for human studies!! However, check out
the Song Sparrow example on page 346; also note the human studies of mono- vs. dizygotic twins.

Survival and Reproductive Success – the components of fitness
        Note selection differential (S; difference between means of two populations for some additive
character/overall mean of the two populations), selection gradient (slope of line representing fitness in
relation to some additive character), relative fitness components of discussion (on pages 348 – 350).

        In the end, you can simplify the evolutionary response to selection with the following equation:

                                                R = h2S

An example from nature: Alpine Skypilots and Bumblebees (Galen, 1996)
        Skypilots from above treeline (tundra) are 12% larger than those at treeline. Previously, Galen
had documented larger skypilots attracted more bumblebees, and those that attracted more bumble-
bees had more seedset. So she asked two questions:

         1. Is selection on flower size by bumblebees responsible for larger tundra flowers?
        2. If so, how long does it take to generate a 12% difference in size?

        First, need to estimate heritability: a scatterplot of offspring flower size to maternal flower size
shows a heritability of around 1, but with significant scatter, suggesting she could only safely conclude
that 20% (.2) of phenotypic variation was due to additive genetic variation (and so the rest of the
variation is due to . . . ?). Second, need to estimate strength of selection differential imposed by bum-
blebee pollinators: she found a selection differential S = .74 mm/15 mm for these flowers (meaning
flowers pollinated by bumbles were on average .74 mm bigger), which results in a S @ 5%, in turn
meaning the plants that win have flowers that are 5% larger than the average of the entire population.
So, again using the conservative 20% estimate from above, the response R = .2 x .05 = .01 (or R =
1 x .05 = .05 if using the high end estimate for heritability). So, that means bumblebees should
promote a 1 (up to 5%) change in average flower size in a population of skypilots moved up to the
tundra in a generation. The explanation for the size difference is that the skypilots at treeline are
pollinated by a variety of pollinators, while those above are only pollinated by bumblebees. If bum-
blebees are excluded, skypilots below treeline set seed, but those above do not.

Modes of selection and the maintenance of Genetic Variation
: Finally getting to the "meat"
        Directional – examples (several)
        Stabilizing – gall example from book is a good one (pg. 358, Fig. 9.28)
        Disruptive – beak sizes in various birds (including Darwin’s finches); mimetic forms

        Understand that all three are STILL selection, meaning that low fitness individuals are eliminated,
and overall mean population fitness increases.

        In general, it has been typically assumed that WITHIN populations, directional and stabilizing
selection are rather common, and disruptive rather rare. However, if that is the case then genetic
variation (at least in some traits) should be significantly reduced/eliminated completely over time. So,
what helps maintain the variation? We’ve answered this partly before, but we’ll add more detail here.
        1. Most populations are not in evolutionary equilibrium with their environment in terms of
directional/stabilizing equilibrium. There is a slow, steady supply of new mutations. Besides that,
different traits may be experiencing differential selection, and linked traits may maintain some varia-
tion with differential selection, at least until recombination unites alleles that are both favorable under 
the current conditions.
        2. In most pops., there is a balance between deleterious mutations and selection. We’ve dis-
cussed previously that, although selection removes deleterious alleles, most will remain at low
frequency (in heterozygotes and through continued mutation). However, since with additive effects
of quantitative traits, any deleterious mutation at any one locus may have a very small influence on
fitness, so that there may be significant variation maintained simply because of the many genes
involved in the trait.
        3. Disruptive selection, or other patterns (frequency-dependent selection) may be more
common than generally recognized.

Important take home messages
        1. If a trait has high heritability in two different populations and those populations have different
means in the additive traits, this does NOT tell us anything about the CAUSE of the differences in the
traits – the different environments can still cause significant differences, even with a high heritability
(the IQ argument fallacy).
        It is, of course, difficult to show this with humans, but examples from other organisms have been
tested. For instance, see the Clover Aphids versus Alfalfa Aphids common garden experiment on pg.
363.  Alfalfa aphids have average higher fecundity than Clover aphids.  BUT when grown in two com-
mon gardens, one with Alfalfa and one with Clover, each type was more fecund in its environment of
origin.  This was unanticipated -- each population is superior in its own environment of origin, and
indicates some significant genetic component to the fecundity.
        2. Heritability tells us nothing about the role of genes in determining traits that ALL members of a
population share.