15. See Simpson (1944).
16. See MacFadden (2005) as well as Simpson (1944) and Bell (2012).
17. Although scientists usually refer to them as Ammonoids, after the subclass Ammonoidea of cephalopods to which they belong, I will refer to them by their better-known common name of ammonites.
18. Nautilus was one but not the only discoverer of the principle of jet propulsion, which is also used by other water-living animals, such as jellyfish.
19. More precisely, the second quantity is the diameter of the umbilicus, the distance between the central axis and the inner wall of the outermost whorl, but the two are closely related. See Raup (1967) and Chapter 4 of McGhee (2007).
20. Center and right ammonite drawings after Figure 6 of Saunders et al. (2004).
21. See Chamberlain (1976, 1981).
22. See page 73 of McGhee (2007).
23. See page 360 of Chamberlain (1976), as well as Chamberlain (1981).
24. See page 75 of McGhee (2007), as well as Chamberlain (1981), McGhee (2007), and Saunders et al. (2004).
25. See Saunders et al. (2004) and McGhee (2007). Factors other than swimming efficiency, including buoyancy control and equilibrium control, may also matter, and the Ammonoid fitness landscape may thus be more complex than their analysis lets on. See McGhee (2007) and Chamberlain (1981).
26. See Hay-Roe and Nation (2007).
27. See Benson (1972).
28. See Brown (1981).
29. See Brown (1981), as well as Brower (1994, 2013).
30. See Haffer (1969), and also Knapp and Mallet (2003).
31. If we could “replay the tape” of life’s evolutionary history, a thought experiment famously conducted by the late paleontologist Stephen Jay Gould, we would probably observe butterflies whose warning colors were just as effective as but completely different from those alive today. Their adaptive landscape would have different peaks.
32. See Majerus (1998), especially Chapter 6.
33. One of the reasons why the peppered moth captured the imagination of early population geneticists is that in its populations, genetic dominance—a phenomenon first described by Mendel—varies. In some populations, the light form is dominant over the black form—crosses of white and black moths yield mostly or only white moths—whereas in other populations the black form is dominant over the white form. The causes of this phenomenon of dominance modification led to a dispute between Fisher and Wright that lasted years. See Provine (1986).
34. Morgan’s white allele is of special historical importance because it was the first allele with sex-linked inheritance—a different inheritance pattern in males and females—that pointed to the existence of sex chromosomes.
35. See Wright (1932).
Chapter 2: The Molecular Revolution
1. See Watson and Crick (1953).
2. More precisely, many hormones are short chains of amino acids also called peptides.
3. As I will discuss later in this chapter, through a process called alternative splicing cells can make more than one protein from a gene, such that the number of proteins in the human body is much greater than the number of genes.
4. And that is a gross underestimate of the size of the adaptive landscape on which Drosophila evolves, because the genomes of different fruit flies may differ in not just one but thousands or millions of different nucleotides. In addition, many genes are much longer than one thousand nucleotides, and many mutations can occur outside genes in the nonprotein coding regions of the genome.
5. See Mackenzie et al. (1999).
6. See Kauffman and Levin (1987).
7. 7 billion humans × 100 years × 356 days per year × 8.6x105 seconds per day equals 2.2 × 1020. A landscape of four hundred gene loci with two alleles would have 2400 = 2.6×10120 genotypes, or 1.2×10100 as many.
8. See equation (2) in Kauffman and Levin (1987) with N = 15,000.
9. See equation (4) in Kauffman and Levin (1987).
10. The number is given by the binary logarithm of 15,000; see page 23 of Kauffman and Levin (1987).
11. The notion of a protein space explored by evolution goes back at least to the biologist John Maynard-Smith. See Maynard-Smith (1970).
12. See Weinreich et al. (2006). The “conventional” beta-lactamase I am referring to is a member of a much larger family of proteins called TEM beta-lactamases, which share a similar amino acid sequence. It is the prototypical member of the family, also called the reference sequence. Four of the five mutations in question alter the amino acid sequence of the encoded protein, and the fifth is a regulatory mutation that occurs in the nonprotein coding part of the gene.
13. Even though only four amino acid changes are involved (and one DNA change in a regulatory region of the gene), the numbers still hold. TEM-1 is 263 amino acids long, and there are already 1.94×108 different ways of choosing just four amino acids for replacement. Each of them can be replaced by nineteen different other amino acids, so that for a mere four amino acid replacements, there are 2.5×1013 possible strings. That number has to be multiplied by another factor of three if one takes the fifth, regulatory change in DNA into account.
14. Under the condition of Weinreich’s experiments, which used very strong selection for better variants, horizontal paths that do not affect fitness were also prohibited, even though they can play an important role in nature, as we shall see in Chapter 4.
15. Szendro et al. (2013) provide an overview of these experiments. I note that the genetic changes in such experiments need not all be single-letter changes in a protein. Some of them may affect the regulation of a gene rather than alter the letter sequence of an encoded amino acid string. Also, some changes are deletions or insertions of short pieces of a genome. However, the same principle applies here as to studies focusing exclusively on amino acid sequence variation. Different mutant alleles can arise in different orders, and only some of the resulting paths toward a fitness peak will be accessible.
16. Although it may not be the most important mechanism for such tuning. See Miranda-Rottmann et al. (2010). For this and other examples see also Graveley (2001).
17. Most splicing occurs in organisms like us—eukaryotes—and comparatively little occurs in bacteria.
18. See Hayden and Wagner (2012).
19. See Jimenez et al. (2013). This molecule is chemically very similar to the much more familiar ATP, containing a guanine in place of ATP’s adenine.
20. See Badis et al. (2009), as well as Mukherjee et al. (2004) and Weirauch et al. (2014). The actual design of such a microarray is more complicated than described here, for example because some DNA words can contain gaps—regions that are not recognized by a regulator.
21. See Weirauch et al. (2014).
22. See Aguilar-Rodriguez et al. (2017).
Chapter 3: On the Importance of Going Through Hell
1. See Hawass et al. (2010).
2. See Alvarez et al. (2009).
3. Ibid.
4. I am referring to recessive diseases here, the most common diseases caused by mutations in a single gene, where both alleles of a gene in a diploid organism are required to have suffered a disease-causing mutation. Dominant diseases, where only one mutation suffices, would manifest themselves both in inbred and outbred individuals. It is also important to distinguish mutations in somatic tissues that do not contribute to reproduction from mutations in the germ line—the cells that get passed on to the next generation. Only the latter directly affect inheritance and future generations.
5. This also means that the best way of maintaining the lineage is to outbreed blue-eyed cats with non-blue-eyed cats, which would yield 50 percent offspring with blue eyes.
6. See Pusey and Packer (1987).
7. See Pusey and Wolf (1996).
8. For evidence supporting the Westermarck effect, see Shepher (1971), as well as Lieberman et al. (2003).
9. See Chapter 15 of Futuyma (2009), as well as Charlesworth and Willis (2009). It bears mentioning that some orga
nisms, especially plants that have survived very small population bottlenecks through the ability to self-fertilize, do not necessarily suffer from inbreeding depression, because many of their recessive deleterious alleles may have been purged already from their populations.
10. This is an important point. Inbreeding depression may be only one among many reasons why inbreeding is avoided in nature. In humans, for example, incest taboos can help forge family alliances. For relevant literature, see Charlesworth and Willis (2009), Pusey and Wolf (1996), and Szulkin et al. (2013).
11. The relevant mathematics comes from population genetics and, more specifically, from coalescent theory, which describes how many generations one must go back in time before a given number of individuals share a common ancestor. The details depend on whether one studies haploid or diploid organisms and whether one studies relatedness among all genes or just one gene in a genome, but the principle that this time depends linearly on the number of individuals in the population remains unchanged. See Hartl and Clark (2007).
12. See Coltman et al. (1999).
13. See Keller (1998) and Keller et al. (1994).
14. Wright also already proposed, in the form of his well-known shifting-balance theory, a process by which populations can descend from a peak through a valley to another (higher) peak of a complex fitness landscape. Some of the theory’s details are complex and controversial, but one central element is simple and widely accepted: genetic drift can help populations traverse adaptive valleys. See Chapter 9 of Provine (1986).
15. I simplify the genetics and evolution of eye color here for the sake of having a concrete example to illustrate the concept of genetic drift. Although high school biology classes sometimes discuss eye color as an example of a trait influenced by a single gene, it is actually affected by multiple genes, some of them more important than others. An especially important gene is OCA2, which is involved in the synthesis of brown pigment in the iris, and a single mutation that changes this gene’s expression is sufficient to turn brown eyes into blue. See Eiberg et al. (2008). It is not clear that eye color is a neutral trait—i.e., not affected by natural selection—because blue eyes have spread rapidly since their origin some ten thousand years ago. Also, eye color affects the incidence of diseases such as macular degeneration and uveal melanoma. See Sun et al. (2014). If I nonetheless use eye-color alleles to explain the phenomenon of genetic drift (rather than more unambiguously neutral alleles) it is because few neutral alleles affect a human trait as plainly visible as eye color.
16. The reason is that a random letter change in a long DNA text is much more likely to create a new variant than to revert a mutation that has occurred previously.
17. This holds as long as selection does not affect the sampling. For example, a specific class of “selfish” genes can promote their own propagation to the next generation at the expense of the organism’s genetic health in a phenomenon called meiotic drive. See page 290 of Futuyma (2009).
18. Even though eye color is technically a polygenic trait—influenced by multiple genes—brown eyes are usually dominant over blue eyes, meaning that only one of the alleles in an individual’s genome would have to be “brown” for the iris to be brown, whereas both alleles would have to be “blue” for blue eyes. Even so, because 50 percent of populations would get fixed for the blue allele, in those populations all individuals would have blue eyes.
19. I am using here two basic insights from population genetics. The first is that a gene evolving only under the influence of genetic drift, with a frequency of p in one generation, will have a random allele frequency in the next generation whose mean is p and whose variance is of the order of p(1-p)/N, where N is the population size. If one were to choose the standard deviation as a measure of dispersion, allele frequencies would fluctuate by an amount that is inversely proportional not to N but to the square root of N. The second insight comes from coalescent theory, which shows that the amount of time one has to go back in time to find the common ancestor of two (or all) alleles in the population is of the order N. I note that all these expressions are for haploid organisms. For diploids, N needs to be replaced by 2N, which does not, however, affect the order-of-magnitude argument of the main text. See Hartl and Clark (2007).
20. Alleles causing dominant disease, where only one of two copies needs to be mutated for the disease to occur, cannot spread as easily through genetic drift, because natural selection would prevent their spreading. The reason recessive alleles can spread through drift is that they only manifest their negative effects when in two copies, which is very unlikely to happen unless they already have a large frequency. More generally, most genetic diseases are complex diseases caused by mutations in multiple genes, where any one mutation may contribute very little to disease risk and can thus spread far through a population by drift alone. Also, many naturally occurring mutations have fitness effects that are deleterious, but very weakly so, such that natural selection may delay but does not prevent their spreading by genetic drift. See Hartl and Clark (2007).
21. More precisely, they do not shuffle genomes every generation like we do. However, they engage in other forms of sexual reproduction that affect only part of their genomes and that do not occur every generation, such as bacterial conjugation. See Griffiths et al. (2004).
22. In the interest of simplicity, I have taken several liberties with the landscape concept here. First, as opposed to Chapter 1, where the units of study—individuals on the landscape—are for the most part genotypes, here the unit of study is an entire population. Viewed as an object on the landscape—one can think of this object as also representing the center of mass of a group of individuals—natural selection drives this object uphill. But as a result of genetic drift, the location of this object fluctuates because the population’s allele frequencies fluctuate. So, strictly speaking, it is not the landscape itself but the object that is being shaken. A better but more technical analogy is that of a particle under Brownian motion with a diffusion coefficient proportional to the inverse of population size.
23. Technically speaking, I am referring to a general observation from population genetics that the selection coefficient of an allele must be smaller than approximately the inverse of the population size (which is proportional to the generation-to-generation variation in neutral allele frequency) in order for drift to be able to overcome the pull of selection. See Hartl and Clark (2007). The exact number depends on whether organisms are haploid or diploid (where population size N has to be replaced by 2N), on which aspect of fitness one considers, and on the units in which this aspect of fitness is measured. Also, I note, for the cognoscenti, that whenever I am discussing population size, I am considering what population geneticists call the effective size of a population, which is the relevant quantity for genetic drift, but may be smaller than a population’s census size. See also Lynch (2007).
24. See Eyre-Walker and Keightley (2007) and Chapter 5 of Freeman and Herron (2007).
25. That’s because the (effective) population sizes of bacteria usually exceed 108 individuals. See, for example, Chapter 4 of Lynch (2007).
26. See Sun et al. (2014).
27. See Table 1 on page 219 of Whittaker and Fernandez-Palacios (2007).
28. See Sulloway (1982).
29. See pages 228–229 and Table 9.3 of Whittaker and Fernandez-Palacios (2007).
30. The oldest islands on Hawaii and Galápagos are Kauai and Española. See Geist et al. (2014), as well as page 220 of Whittaker and Fernandez-Palacios (2007), for relevant dating information. It is important to be aware that in volcanic archipelagos, islands can arise and become submerged again, such that an archipelago may be older than its oldest island visible today. Molecular clock dating can be used to estimate the age of the oldest organismal lineage on an archipelago, which shows, for example, that on Hawaii few lineages are older than ten million years, a short amount of evolutionary time.
31. The examples in this paragraph and others can be found in C
hapter 9 of Whittaker and Fernandez-Palacios (2007).
32. Island life in particular shows recurrent patterns of change that include gigantism in plants and some animals, as well as reduced dispersal; for example, through flightlessness. See Grant (1998).
33. See Montgomery (1983).
34. Most scientists believe that population bottlenecks on islands play an important role in adaptive radiations. See Chapter 7 of Whittaker and Fernandez-Palacios (2007). However, they are not the only factor. Just as important is the reduction of competition that is experienced by species that occupy previously empty ecological niches when they colonize an island. More intense competition often means more stringent selection, and reduced competition on islands means that the intensity of selection is reduced, another case in point that curtailing selection—by whatever means—can facilitate innovation.
35. See page 453 of Lynch (2006), as well as Carbone and Gittleman (2002).
36. Again, when I refer to population size here and throughout, I am referring to what population geneticists call the effective population size, which is typically much smaller than the census population size and reflects the fraction of individuals or genes that contribute to the next generation’s gene pool. It is influenced by several factors, including the mode of reproduction and variation in census population size over time. See Hartl and Clark (2007).
37. An additional factor whose effect is difficult to predict is that the kind of selective pressures we are subject to are changing. On the one hand, we suffer to a greater extent from diseases related to our modern lifestyle, such as type II diabetes. On the other hand, medicine has made huge progress in helping us compensate for some defects caused by our genetic heritage.
38. In another difference from eukaryotic organisms like us, several of E.coli’s regulatory proteins are themselves part of the RNA polymerase that transcribes genes. In the interest of brevity, I am omitting a few other roles of non–protein coding DNA, such as antiviral defense or assistance in the initiation of DNA replication, because gene regulation is so prominent among them.
Life Finds a Way Page 26
