The paradox of accelerated evolution. Column in ComputerreOnline #46

←
Dmytro Shabanov
→

← Dmytro Shabanov → Challenges of Career Guidance The Paradox of Accelerated Evolution When Does Selection Become Ineffective?

Column in KompyutereOnline #45 Column in KompyutereOnline #46 Column in KompyutereOnline #47

In the previous installment I argued that in moving toward the III synthesis, evolutionary biology must overcome the limitations of the II synthesis—the Synthetic Theory of Evolution (STE), which was formulated about seventy years ago. It is time to discuss the shortcomings of STE. For our purposes it will be sufficient to use the definition of STE given by Alexander Pavlovich Rasnitsyn. According to this theory, the evolutionary process is represented as the dynamics of allele frequencies in populations, controlled by selection (through viability assessment and, consequently, differential reproduction of phenotypes) and stochastic processes. Not entirely clear? I will explain by carefully presenting the classic STE viewpoints (even those with which I do not fully agree). It is easier to do this while commenting on the figure. [IMG_1] Fig. 1. Darwinian evolution according to STE. Phenotype is selected, genotype is reproduced The five rectangles in the figure represent three generations of very simple organisms: bacteria, for example. Is the terminology clear? The complex of hereditary endowments of an organism is called the genotype, and the set of its expressed traits is the phenotype. The genotype is the set of genes; genes can exist in different states—alleles. Within the phenotype one can distinguish individual traits, which may be in various trait states. During an organism’s development its phenotype changes, and these changes depend on the genotype. The organism interacts with the environment, building itself at its expense. If the phenotype is adequate to the possibilities offered by the environment, the organism develops successfully and leaves offspring; inadequate (maladapted) phenotypes are eliminated or prevented from reproducing. Do you see the feedback loops in the figure? The genotype determines the phenotype, the environment evaluates the phenotype and allows carriers of optimal genotypes to reproduce. Errors in the genotype arise during reproduction. In the figure selection eliminated the change in gene B and retained the change in gene C. Thus we observed a basic evolutionary event: a shift in allele frequencies in a population. Recall that new information is always the consequence of remembering the outcomes of random choice. Darwinian evolution is the inevitable consequence of reproduction, variation, inheritance, and selection. The tetrapod in Fig. 1 is present—so evolution will proceed. Of course, individual success and failure are not always regular; selection acts by altering probabilities, which is why Rasnitsyn’s definition mentions stochastic processes. The overall picture is clear; the problem lies in the details. For selection of phenotypes to effectively change genotypes, phenotypes and genotypes must be unambiguously linked. This is not always the case. Unfortunately, a universally accepted classification of traits (and their possible states) depending on how their development is regulated has not yet been established. The traits shown in Fig. 1 are regulated in the simplest way: each depends on a specific gene. Thus, gene B may determine the structure of an enzyme that allows the bacterium to metabolize a certain substance. When the gene is normal (allele B0), everything is fine. In the second generation, mutation B1 disrupted the enzyme, impaired nutrient uptake, and led to the organism’s death. Mutation C1 is more interesting: it did not reduce but increased viability. Perhaps it strengthened the bonds between neighboring amino‑acid residues in a protein, making the protein more resistant to elevated temperatures. Are deleterious and beneficial mutations equally probable? The more complex and refined an organism, the less likely a random change will be advantageous. Some authors even consider the probability of a beneficial change astronomically low. I think this is not the case, and convincing experiments (dedicated precisely to bacterial adaptation to higher temperature) show that beneficial mutations are rare but observable. But does every trait depend on a single gene? No. For example, in the just‑mentioned experiment on the evolution of thermotolerance in bacteria, the effect of many mutations could not be evaluated separately without considering the allelic state of other genes. The same mutation, together with certain alleles of other genes, is advantageous; combined with different alleles, it is harmful. This means that the fate‑determining state of a trait depends on several genes. Do gene‑gene interactions reduce the efficiency of selection? Yes, especially for sexually reproducing species. I will give an example for such organisms, each of which carries two identical or different alleles of each gene and transmits only one of them to offspring. Imagine a lucky individual acquiring a combination of two rare (perhaps recently arisen) alleles, each of which is deleterious on its own: A1 and B1. In the population the common alleles are A0 and B0. The combination A1 + B1 is beneficial, whereas A1 + B0 and A0 + B1 are harmful (both relative to the normal A0 + B0). I will not explain the logic of solving genetic problems in detail; instead I will outline the offspring group ratios from crossing the fortunate carrier of the advantageous double‑mutation combination with the most probable partner. Consider the cases where both alleles are dominant (affecting the phenotype even when present in a single copy) and where they are recessive (expressed only in the phenotype of an organism that has two copies). If alleles A1 and B1 are dominant, only one quarter of the offspring will have an improved phenotype (shown in bold), and two quarters will have a deteriorated phenotype (underlined):

A1A0B1B0 × A0A0B0B0 → A1A0B1B0 : A1A0B0B0 : A0A0B1B0 : A0A0B0B0.

A1A0B1B0 × A0A0B0B0 → A1A0B1B0 : A1A0B0B0 : A0A0B1B0 : A0A0B0B0.

If the considered alleles are recessive, all offspring will have a normal phenotype:

A1A1B1B1 × A0A0B0B0 → A1A0B1B0. What will appear in the next generation? A1A0B1B0 × A1A0B1B0 → A1A1B1B1 : 2 A0A1B1B1 : 2 A1A1B1B0 : 4 A1A0B1B0 : 2 A1A0B0B0 : 2 A0A0B1B0 : A0A0B1B1 : A1A1B0B0 : A0A0B0B0.

The beneficial trait will appear in only one out of sixteen offspring, while six out of sixteen will have a deteriorated phenotype! In these (and other, unexamined) cases of such allele interactions, the alleles will more often be found separately in the population—where selection works against them—than together in a selection‑favoured state. Situations where selection maintains a complementary pair of alleles will be rare compared with cases where it eliminates the alleles individually. Gene interactions reduce the efficiency of selection! And the more complex the organisms, the more complex their traits, the less directed phenotype selection will lead to directed allele selection. In truly complex organisms, several additional regulatory mechanisms intervene between genotype and phenotype, complicating allele sorting by selection. One of them, in STE terminology, is the expansion of the reaction norm. Organisms with identical genotypes can differ because they develop under different conditions (or simply by chance). Imagine three human triplet siblings with identical genotypes. One is raised by a tribe of primitive hunters, another attends a sports boarding school, the third goes to a specialized math school. Despite identical inheritance, they will grow very differently—physically and behaviorally. Unlike undirected mutations, their differences increase the fitness of their carriers. In selection (for example, sexual selection—winning the heart of a potentially fertile beauty) these three twins will have different chances, because selection (including sexual selection) evaluates phenotypes! A second mechanism involves epigenetic regulation of gene activity, especially characteristic of mammals. Depending on the organism’s state, it transmits to offspring genomes with different chemical “marks” on genes (methylation or other modifications of the genetic text). These marks affect gene activity and thus traits. Usually epigenetic marks provide the developmental variant (within the reaction norm) that best matches the current environmental conditions. When conditions change, the marks can be erased or rearranged. How do these changes affect selection efficiency? I will not draw a multi‑generation scheme; I will only show the logic of trait formation in a single individual. [IMG_2] Fig. 2. Only some of the mechanisms that complicate allele sorting by selection, typical for highly developed species, are shown here Beyond the basic feedback loop (genotype influences phenotype; by selecting the best phenotypes, the environment selects the best genotypes) new regulatory levels have appeared. Observe: the development of trait 1 is still governed simply—it depends on a single allele. In contrast, traits 2, 3, and 4 depend on all the shown genes and on the environment, to varying degrees. Interactions between gene products and environmental influences create a highly intricate network of interwoven causes and effects. Gene activity itself depends on the outcome of this interaction: depending on it, genes undergo epigenetic editing—reversible hereditary modification. These changes reduce the efficiency of allele sorting by selection. The link between phenotypes and genotypes becomes highly ambiguous (except for “simple” traits like trait 1 in Fig. 2). The evolutionary mechanism that worked efficiently in Fig. 1 will stall in Fig. 2. Evolution is a race to produce new adaptations quickly. Return to the first figure: in the second generation shown there, the normal genotype won, and in the third generation the carrier of the recent norm already lost because more adapted organisms appeared. How could organisms arise whose selection efficiency (and efficiency of generating new adaptations) had decreased—they should inevitably lose to those evolving faster? The point is that in complex organisms selection efficiency does not decline but increases. Again I quote Rasnitsyn: …age (in millions of years) of the half‑life fauna (in which half the species are extant, half extinct) for large mammals (proboscideans and ungulates) – 0.2, for small mammals – 0.5, for birds and fish – 0.7, for insects – 3–7, for molluscs – 3.5–5, for diatom algae – 15. The half‑extinction time (analogous to half‑life, the time for half of the original number of species to disappear) in the same units is smallest for elephants – 0.18, and on average for mammals – 0.54, for bony fish – 3.5, for graptolites – 1.3, for echinoderms – 4.2, for bivalves – 7, for planktonic foraminifera – 5, for benthic – 18–24, for diatom algae – 5.5, for dinoflagellates – 9. At higher taxonomic levels the differences are similar, often more pronounced. For example, the half‑life age for mammalian genera is 4 Myr, for birds – 10, for reptiles – 20, for fish – 30–50, for insects – 40, for molluscs – 60, for foraminifera – 230. In addition to the reasons outlined above, there is a whole set of illustrative factors that should make the evolution of large animals very slow (Rasnitsyn discusses them in detail). For instance, each individual proboscidean produces only a few offspring, and generation length is catastrophically long. While a single elephant calf is growing, dozens (if not hundreds) of insect, diatom, or foraminifer generations—each with countless individuals—will have turned over. Proboscideans simply lack sufficient material for selection; they must evolve extremely slowly… [IMG_3] Fig. 3. Over several tens of millions of years of proboscidean history, more than forty genera and over three hundred species have arisen! In terms of evolutionary speed (measured both by the rate of new species formation and by the rate of trait change) proboscideans are among the record‑holders in the biosphere. From the STE perspective this phenomenon is inexplicable. What is needed is an idea of “evolution of evolution”—the improvement of the mechanisms that generate adaptations as life itself develops. I will give just one example. The family Hominidae sharply increased its rate of generating adaptations when it refined cultural inheritance mechanisms and added them to genetic inheritance. Likely, this improvement of the evolutionary mechanism was not the first in our history. Elephants, like other mammals, evolve using both STE‑described mechanisms and other means. Additional feedback loops in the implementation of their genetic information do not reduce but increase the efficiency of their evolution. How—remains not fully known. In any case, our knowledge still has room to evolve! ← Dmytro Shabanov →

←
Dmytro Shabanov
→

Challenges of Career Guidance The Paradox of Accelerated Evolution When Does Selection Become Ineffective?

Column in KompyutereOnline #45 Column in KompyutereOnline #46 Column in KompyutereOnline #47