About Rationally Speaking
Rationally Speaking is a blog maintained by Prof. Massimo Pigliucci, a philosopher at the City University of New York. The blog reflects the Enlightenment figure Marquis de Condorcet's idea of what a public intellectual (yes, we know, that's such a bad word) ought to be: someone who devotes himself to "the tracking down of prejudices in the hiding places where priests, the schools, the government, and all long-established institutions had gathered and protected them." You're welcome. Please notice that the contents of this blog can be reprinted under the standard Creative Commons license.
Saturday, July 12, 2008
Notes from Altenberg, part I
Altenberg workshop, Day 1
John Beatty (University of British Columbia): Chance, history, and natural selection
“Contingency” used in two different ways, not distinguished by Gould. a) Replaying life’s tape, unpredictability; b) Not randomness, but dependency on previous sequence of events. Both kinds of contingency limit the “importance” of natural selection.
Darwin did address chance variation (beginning with the book on orchids, the first after the Origin), acknowledging different types of chance variation. He focused on the appearance of alternative, apparently equivalent, “solutions” to the same biological problem, like pollination syndromes in orchids. Even so, the emphasis is on natural selection, subsuming chance variation.
Lyell wrote to Darwin that selection could preserve or destroy, but not create. Lyell believed that creative variation comes from intelligent design. Asa Gray agreed with this view. Darwin rejected the suggestion on two grounds: it was theologically unsatisfactory, and it undermines the importance of natural selection, to the point of making it superfluous.
Creative role of selection re-emphasized by architects of Modern Synthesis, like Dobzhansky. This was a reaction to the persistence of the idea of directed variation at the turn of the 20th century. Because this was largely successful, there was less need to consider contingency in the literature, except for random drift/sampling error.
More recently, the interest has shifted to the second sense of “contingency.” Lenski and Trevisano studied long-term evolution in bacteria, finding significant variation in the fitness of their lines, which they attributed to the sequence of mutations that had characterized each line’s evolution (i.e., contingency in sense (b) above). Weinreich et al. (2006) concluded that Darwinian evolution can follow only very few mutational paths to fitter proteins (i.e., again, order of appearance of mutations matter).
Sergey Gavrilets (University of Tennessee): High-dimensional fitness landscapes
Modern Synthesis made possible by the mathematical formalization of Fisher, Wright and Haldane. Most of that work was on adaptation, but a significant portion dealt with the question of speciation (reinforcement, sexual selection, shifting balance theory).
Wright introduced the idea of fitness landscapes in 1932, it has become a standard tool for formal mathematical modeling of evolutionary processes. Distinct landscapes for individual gene combinations or population averages. Dimensionality can be very high, particularly for population-type landscapes.
Wright thought of rugged landscapes, the result of widespread epistasis/pleiotropy. Fisher thought of smooth, single-peak landscapes, subject to “mass selection.” Kimura, later on, introduced the idea of flat landscapes in the context of the neutral theory of molecular evolution. Later landscapes were thought of in terms of quantitative traits, and people considered non-static landscapes.
Dimensionality of realistic landscapes must be on the order of several thousands, if not millions. This implies a redundancy of the genotype-fitness map, where many genotypes are roughly equivalent in fitness. Once can derive conditions under which n-dimensional landscapes behave as quasi-neutral (in the sense of having large clusters of genotypes with similar fitness).
Evolution on “holey” multi-dimensional landscapes: Microevolution can be thought of as climbing from a “hole”; Macroevolution becomes movement along the holey landscape; Speciation is seen as movement around a “hole” (the latter a generalization of the Dobzhansky-Muller model for two loci with two alleles).
Speciation mechanisms can be classified as geographic (sympatric, parapatric, allopatric); by evolutionary forces (mutation, drift, natural selection, sexual selection); by commonality of mathematical description (spontaneous clusterization, isolation as by-product of adaptation, reinforcement).
We now have (thanks to mathematical formalism) a better understanding of conditions leading to sympatric speciation, about which many biologists are still skeptical. Not many clear empirical cases available, but part of the problem is that sympatric speciation is empirically difficult to distinguish from other modes.
We are now beginning to develop a mathematical formalism of adaptive radiations.
Current work also on human speciation, characterized by evolution of brain size and of social coalition. “Social brain” hypothesis links the two, has been explored mathematically, with model supporting its implied mechanism and allowing articulation of empirical predictions.
David Sloan Wilson (Binghamton University): Multilevel selection, major transitions, and human evolution
Problem with evolving traits for social life is that they generate a conflict between individual relative fitness within the group and group fitness relative to other groups. Group selection can in part overcome the problem if the between groups selective pressure is sufficiently strong.
The naive group selectionism of the ‘60s was rejected, to be eventually replaced by current multi-level selection theory. Interestingly, the group selection controversy does not coincide with the development of the MS (in fact, some exponents of the MS were in some respect naive group selectionists).
Wilson et al. in the ‘60s articulated the logic of group selection, affirming it as a theoretical possibility. They then rejected group selection on empirical/plausibility grounds.
Revival of multiselection theory follows a “Russian doll” logic: genes within individuals; individuals within groups; groups within metapopulations). Indeed, kin selection must be seen as a special case of multilevel selection theory, and hence of “group” selection.
Development of pluralism, where all levels of selection can be described in terms of effects at gene level (gene level not necessarily as causally more important, but can always be used for “bookkeeping”).
Major transitions as a concept was not on the radar screen from Darwin to the MS, until Margulis proposed symbiotic theory of cell formation (but still not within the context of multilevel selection theory). Maynard-Smith, Szathmary and others then generalized the idea. Major transitions are rare events with momentous consequences: origin of life, first cells, first eukaryotes, multicellularity, eusociality (in insects, shrimp, mole rats etc.), and language in humans. During transitions, within-group selection is suppressed but never eliminated.
Human success as the result of selection for teamwork; we have become the primate equivalent of a single body or a beehive (in contrast with Machiavellian brain hypothesis, framed as within-group competition). Implications for evolution of moral sense (suppression of deviants who undermine teamwork).
Concept of “Darwin machines,” open-ended processes: behavioral and cultural change are like the immune system.
Evolutionary psychology in narrow vs broad form: Narrow is the version of Cosmides, Tooby et al, based on the idea of massive modularity, with open-ended, domain-generic learning processes deemed impossible. Broader version is now developing, where people are applying an evolutionary perspective to all aspects of human condition, but without the restrictive assumptions on narrow evo-psych.
Greg Wray (Duke University): Gene regulatory networks and natural selection
Several periods of integration/synthesis before and after the “official” Modern Synthesis (paleontology, molecular biology, developmental biology, genomics). Part of the idea of a synthesis is that different disciplines/theories have a limited explanatory scope (e.g., population genetics is not meant as a theory of phenotypic evolution).
Current working model of genes is outdated, incomplete and restrictive. It ignores how genes function and what their relationship(s) to phenotypic traits are. For instance, noncoding sequences have a diverse range of regulatory roles.
It is now possible to deploy statistical methods to test for selection on non-coding sequences, distinguishing negative selection (functional constraint), drift, and positive selection (adaptation). Applied to comparative analysis of regulatory sequences in primate genomes. Found more positive selection on 5’ flanking than coding regions.
Genomics of the evolution of the human brain: development, physiology, cognition. You can’t just make a bigger brain (lethal), there have to be changes in physiological support systems. Also, bigger brains do not necessarily entail language, tool use, abstraction, and so on. During human evolution there has been no enrichment of coding sequences expressed in the brain; on the other hand, there has been enrichment in 5’ flanking (regulatory) regions. Much more evolution of coding sequences happened in the immune system, by comparison.
Standard approaches to study mutation are limiting: uniform genetic backgrounds underestimate gene-gene interactions; non-stressful conditions limit detection of gene-environment interactions; studies at single life history stages underestimate trade-offs.
Human genome differs from chimp “only” by 1-2%, but that still means something on the order of 100 million mutations, a huge search space for comparative analyses. Hence focus on individual case studies: e.g., lactose hydrolase evolution. It is normal for adult mammals to be lactose intolerant, and it is unique to some humans not to be. The ability to digest milk as adults evolved recently, in association with pastoralism during the past 50,000 years. This capacity evolved several times independently, by changes in the promoter region.
Michael Purugganan (New York University): Epistasis, selection, and the evolutionary synthesis in the age of genomics
Studying transposable elements as an instance of “genomic ecosystems.” Retrotransposons not only jump from one genomic location to another, but can increase their copy numbers. To this day we do not know what determines transposons copy numbers in a genome. Moreover, different organisms have different percentages of retrotransposons and DNA transposons (which jump, but increase in copy number less easily).
An important evolutionary connection is that at least some introns originate from transposons, a phenomenon that may lead to a variety of splicing patterns, and hence a variety of gene products. Retrotransposons can also originate flanking regions of normal regions, evolving into promoters.
The study of epistasis is moving away from the purely statistical approach because of the increased availability of genomic tools. Example of flowering time in the model system Arabidopsis thaliana, where a large amount of information is available about the (rather extensive) genetic network underlying the trait of interest. For instance, FRIGIDA gene exists as active or null in natural populations; the active form follows a clear latitudinal cline, while the null form shows no correlation with latitude. FRI in turns affects Flowering Locus C, which is a nodal point in the flowering time network.
The study of natural selection is also changing in the genomic era, now that it is possible to scan entire genomes for candidate instances of positive selection (see Wray’s talk).
Eva Jablonka (Tel Aviv University): The epigenetic turn: the challenge of soft inheritance
If one wishes to challenge the MS, one needs to articulate exactly what the theory says and what it left out. Assumptions include: heredity by transmission through the germ line; heredity from recombination and mutation; heritable variation has small effects; unit of selection is the gene (added in the 1970s); phenotypic innovations are a result of cumulative gene mutations; targets of selection are individuals; evolution is a matter of descent with modification from a common ancestor. All of this culminated with Dawkins’ selfish gene metaphor.
Challenges to the MS, therefore, are posed by non-DNA heritable variation and other forms of “soft” inheritance. Also, high rates of variation appear to be triggered by conditions of stress. A small kit of genes has played major evolutionary roles. The focus shifts to the properties of networks of genes. Phenotypic plasticity is a primitive property. Group selection is common. The pattern of evolutionary divergence includes web-like regions, not only trees.
The “epigenetic turn” which began in the 1990s has three components: the Waddingtonian approach (genes as followers in evolution); the structuralist approach (generative plasticity); the heredity-oriented approach (soft epigenetic inheritance).
Epigenetic inheritance is a component of epigenetics, which is a broader term referring to a variety of supra-genetic phenomena during development, not all heritable. Broad and narrow conception of epigenetic inheritance: the broad sense includes any kind of heritable variation that does not require changes in the genes, including for instance cultural inheritance, various types of developmental inheritance, and even niche construction; the narrow sense focuses on cell-level phenomena. The latter includes mitotic inheritance, and is not just across generations (cell lineages).
Examples include cortical inheritance in Paramecium, stable morph types in Candida albens, prions, epimutations in Linaria’s flowers, and many others in bacteria, protists, plants, animals.
Adaptation can occur through the selection of heritable epiallelles, without genetic change. It is suggested that epigenetic inheritance played a role during major evolutionary transitions. We need to study the origins of mechanisms of transmission at various levels.
John Odling-Smee (Oxford University): Niche construction and niche inheritance
Example: ants creating gardens in forests, killing all plants except their own host by releasing formic acid. The standard version of evo-theory needs to be updated with the concept of niche construction/ecological inheritance. The latter depends on organisms that are “ecologically related,” but not (necessarily) genetically so.
Simple two-loci population genetic models of niche construction show some surprising phenomena, like the fact that one locus does not have to be present in the same population as the second locus whose interaction leads to niche construction.
What each organism inherits is a complex of genetic and environmental resources. So ecological inheritance introduces a second “channel” of vertical inheritance between generations, with the difference that this second channel connects multiple, possibly genetically unrelated, organisms.
Concept of semantic information: anything that reduces uncertainty about selective environments relative to the fitness of organisms. Through channel one organisms inherit not only semantic information (genetic, epigenetic, maternal effects), but also energy and matter. Channel two organisms inherit energy and matter (ecologically modified selective environments), but also semantic information (e.g., through learning, symbolic inheritance).
Adaptive niche regulation can be achieved through a developing, phenotypically plastic, niche-constructing organism. The classic example of “ecosystem engineering” is the beaver’s dam: the wetlands created by the dam can persist for centuries, and obviously modify the environment for a whole local ecosystem.