Notes from Altenberg, part II

{What follows is the second of three installments (one per day) of raw notes from the Altenberg (Vienna) workshop entitled "Towards an extended evolutionary synthesis," which I have organized together with Gerd Muller. These are unedited notes only, and I am solely responsible for their content (I may have misunderstood or not adequately represented the individual speakers' thoughts). An article on the workshop appeared this week in Science magazine, and MIT Press will publish the full proceedings next year.}

Altenberg workshop, Day 2

David Jablonski (University of Chicago): Pattern and process in macroevolution

Evolutionary novelties appear in non-random fashion during evolution. Macroevolution is not a simple extrapolation from microevolutionary dynamics, stasis and species selection are pervasive.

A clear cluster of novelties appeared of course at the Cambrian explosion, can be tracked for instance by the appearance of new marine orders or genera. Innovations appear to be inversely diversity-dependent (meaning that they appear in a relatively brief period of time, preceding taxonomic differentiation). Time dependence of novelties may depend on intrinsic factors (decreased evolvability?) or extrinsic ones (incumbency?).

Key developmental pathways were already in place before the explosion, and early on disparity (morphological diversity) outpaces species richness (number of species).

Mass extinctions generate complex dynamics, and surviving one such event is no guarantee of recovering diversity for a particular group of organisms (they may dwindle to extinction later, or be pushed to marginal ecological niches).

Interestingly, while marine orders start onshore, low-level (i.e., within orders) novelties do not: they appear as commonly on- and off-shore (Triassic to mid-Jurassic data). So major innovations are habitat dependent, low-level diversity is habitat independent.

Post-paleozoic marine benthic orders tend to appear first in the tropics. Lower level novelties also tend to start in the tropics.

Over paleontological time, directional change is rare, most change is compatible with either a random walk or with stasis (based on 251 traits in 53 lineages), with a large proportion of random walks probably actually representing stasis (because of fluctuations in geographical range of species).

Species selection is a well established possibility, theoretically and with some empirical support (even Coyne agrees!). Geographical ranges are a clade-level trait with high heritability (N=95 genera from the Cretaceous). This is important for macroevolution, though it turns out that genus- (not species-) level geographical ranges predict survivorship after a mass extinction.

Massimo Pigliucci (Stony Brook University): Phenotypic plasticity as a causal factor in evolution

The concept of phenotypic plasticity has undergone a remarkable trajectory since its inception at the beginning of the 20th century. Introduced at the same time as the realization of the difference between genotype and phenotype, it has been in the background of evolutionary theory and empirical research for many decades. It underwent a renaissance beginning in the mid-1980s, and papers on plasticity now regularly appear in evolutionary journals.

Still, there is much confusion among biologists about what plasticity is, and more importantly about the role it plays in our understanding of organic evolution. Plasticity includes a bewildering array of phenomena, for instance polyphenisms (discontinuous, environmentally-dependent, alternative developmental phenotypes).

Plasticity is a primitive (not necessarily adaptive) condition, which can then be selected (change in reaction norms) or eliminated (leading to environmental homeostasis). The mechanistic basis of plasticity include not only specific regulatory genes (so-called “plasticity genes”), but hormones, and epigenetic effects.

Evolutionary phenomena that involve plasticity, such as for instance genetic assimilation, may have important consequences for apparently disparate phenomena, such as invasive biology and speciation processes.

Discussion of the “two-legged goat” effect emphasized by West-Eberhard and the relationship between phenotypic and genetic accommodation. The latter two terms update ideas that have been explored here and there throughout the 20th century, particularly the work of Baldwin, Schmalhausen and Waddington. West-Eberhard’s scenario for a four-step evolutionary process where developmental plasticity gets things started and “genes follow” (i.e., they fix changes induced by the environment on the pre-existing plasticity).

Plasticity may also lead us to reinterpret (and make new sense) of “classic” evolutionary phenomena, like mosaic evolution (which may occur quickly as a result of correlated plastic responses among traits) and pre-adaptation (which may actually result from pre-existing plasticity exposed to correlated, not entirely novel, environments).

Eors Szathmary (Collegium Budapest): Evolution by natural selection in the brain (with notes on the origin of life)

On the origin of life: We still don’t understand how life originated, but we do know much better what we don’t know. Present day bacteria are obviously too complex to provide a good model, so one needs to ask what basic elements had to be in place.

Ganti (1974) proposed that for life to originate one had to have metabolism, template copying and membrane growth, each of them being autocatalytic systems. The new field of system chemistry may be now poised to address these issues. Interestingly, this line of research has potentially important practical applications, as it relates to nanotechnology and the spontaneous assembly of nano-bots.

Embedded into this problem is the question of evolvability itself. Distinction between weak and strong self-referential replicators. Any RNA coupled with an appropriate enzyme is a weak replicator, but strong replication requires a (replicable) replication machinery. It is the latter that makes evolvability possible.

Now, about natural selection and the brain: Changeux and Edelman proposed selectionist but not evolutionary systems of learning. “An organism cannot learn more than is initially present in its pre-representations.” This means selection of pre-representations, but no account for variability.

Crick (1989) strongly criticized Edelman’s idea of “neural Darwinism,” dismissing any analogy between natural selection and brain development. However, Adam introduced the idea of a “synaptic mutation,” which coupled with the possibility of neural circuits being copied opens up the possibility of articulating a multilevel selection theory for neural processes.

Gerd Muller (KLI): Epigenetic innovation theory

The core issues of evo-devo cut both ways: on the one hand we want to know how development evolve; but on the other hand we are interested in how the characteristics of developmental systems influence the further course of evolution. The latter deals with constraints, evolvability, innovations, modularity and body plans.

Are phenotypic novelties a distinct class of phenomena? Two distinct levels: the origin of basic body plans relatively early in evolution, vs the evolution of new structures/functions on existing body plans. Clearly the latter is a more tractable problem, both developmentally and evolutionarily.

Focus on novelties that are not homologous to pre-existing structures, either in the focal organism or in its ancestors. Not all phenotypic innovations are “key” innovations (those open up the exploitation of new ecological niches, but they can in fact derive from pre-existing homologous structures).

Epigenetic innovation theory: innovations arise from threshold responses of developmental systems, because of the non-linearity of epigenetic processes. The typical example is limb development, proceeding by the addition of skeletal elements but the reduction of the number of digits. There is now sufficient knowledge at the molecular level to build models of the entire developmental process that account for the observed patterns of variation/evolution.

A generic patterning system allows the addition of new elements as an indirect result of general changes in size and shape of the organism, particularly through threshold effects. This can be addressed experimentally, by artificially making embryos smaller. The resulting animal is not a simple scaled down version of the original, but one that loses a number of skeletal elements in a specific sequence.

Similar experiments can be done to test the developmental plasticity to mechanical stimulation of cartilage and other skeletal elements.

The non-linearity of developmental systems transforms gradual into non-gradual variation. Innovations arise from threshold effects acting during development. This “inherency” at the mechanistic level may be the major conceptual contribution of evo-devo.

Stuart Newman (New York Medical College): Dynamical patterning modules

Proposal: metazoan form originated and rapidly diversified by the action of dynamical patterning modules (DPM). The latter are complexes of proteins that mediate cell behavior or cell-cell interaction by mobilizing physical forces or effects.

Example of gastrulation. There are five main types, they involve engulfement-like tissue rearrangement, which can arise from differential adhesion of cell subpopulations. Gastrulation may also involve buckling-like movements of tissue sheets (e.g. in Drosophila).

Another crucial physical-chemical process for development is diffusion, which originates gradients, and therefore differentiation along the axis of the embryo. Several molecules now known to be involved in this, well studied in Drosophila.

Segmentation by biochemical oscillation may underlie somitogenesis, the formation of paired segmental blocks of tissue along the vertebrate embryo midline. Chemical oscillation is a well understood phenomenon, as in the Briggs-Rauscher reaction.

All of this can be formalized in mathematical models, which in turn allow for simulations of cellular and developmental processes. All these processes are “generic,” and they could be mediated by many alternative gene products. Hence, the corresponding dynamical patterning modules could have been invented over and over during metazoan evolution. Turns out that the genes are more highly conserved than the morphological structures which they underlie.

Basic DPMs are present in metazoan and related clades, including cnidaria, arthropods and choanoflagellates, though they are often doing different things in those contexts.

Marc Kirschner (Harvard University): Facilitated variation

Three sub-theories in evolution: genetic variation, selection, genesis of phenotypic variation. The latter is the current weak point.

The Modern Synthesis established selection as central and genetic variation as the source of phenotypic variation. The question is what kind of evolution is explained solely by selection and genetic variation.

Phenotypic variation, for the MS, is assumed to be non-limiting, small in extent of change, copious in amount, and isotropic. Genetic variation becomes both the measure and cause of phenotypic variation.

The molecular revolution revealed an unexpected amount of conservation at the genetic level. Consider that the human genome is only 1.5 times that of fruit flies, and only 6 times that of E. coli, which raises the question of where all the additional phenotypic variation comes from.

Innovation came in waves, followed by regulatory change: the origin of life, then metabolism & DNA, then the eukaryotic cell, then multicellularity, then complex body plans, then the origin of limbs. In between waves, evolution involves variation, modification and integration of the components in place at that time.

So, variation is not isotropic. The theory of facilitated variation is meant to explain the variation component of evolvability. It is based on the existence of exploratory processes, which generate many variable states (in the cell or entire organism), then selecting among them for function. Examples include vertebrate adaptive immunity, microtubules, ant foraging, and several others. For instance, spatial polarization of microtubules is achieved by variation and selection. Exploratory processes reduce the requirement for simultaneity in the evolution of extreme novelty.

Weak linkage is another useful phenomenon in evolution. For instance, weak linkage in the input-output circuits of neurons, where an electrical system is the linking element between input and output, which means that the same signal can work in a variety of different systems, as long as the substrate allows the (physically) weak interaction to take place.

Gunter Wagner (Yale University): Modularity, evolvability, and the evolution of genetic architecture

Do we need an Extended Evolutionary Synthesis? The old synthesis has real conceptual limitations and self-imposed limitations. The first include the role of development in evolution, and the dominance of selective explanations. The second include a limited focus on a certain set of questions (adaptation, speciation) and ignoring the variational properties of evolvability.

Why do so many population geneticists get uncomfortable about talk of the evolution of evolvability? As proposed by Pigliucci (2007), evolvability can be short term (genetic variation, heritability), medium term (production of high fitness phenotypes by mutation), and long term (ability to produce phenotypic novelties).

Evolvability is context dependent: e.g., mammals are evolvable for body size but not for number of limbs. Many factors influence evolvability: pleiotropic, epistasis, gene number, etc.

Evolvability addresses a serious challenge for evolutionary theory, the ability to explain the evolution of complex structures (Darwin’s organs of extreme perfection, Fisher’s geometric model). The evolution of evolvability can also play a role in the evolution of genome organization and developmental evolution.

There are several misconceptions about evolvability: that it requires group selection, or that direct selection for evolvability is not possible in any system that has recombination. On the contrary, direct selection for evolvability is possible without either of those restrictive conditions (as shown in Wagner 1981 and Wagner and Burger 1984).