More on Mendel

We blogged about Gregor Mendel's insights in a recent post, and we mentioned various questions about what he did and what he thought he did and wanted to do.  There is no question, however, that his work was used to produce one of the most powerful research methods in the history of science.  Combined with the methodology that resulted from Darwin's ideas (that is, the understanding that life has diverged from common ancestry, and the traits that have evolved had to be viable in their respective environments), biology made great progress in understanding development and evolution.

Mendel

The  so-called 'Modern Synthesis' of the 1930s and 40s united patterns of inheritance with understanding of how inherited variation accumulated over time in diverging species, making a consistent underlying framework for understanding much of the nature of life.  The result today can be called the 'Darwinian' method (we blogged about this earlier in a series of posts starting here).

However, Mendel's work also led to the often unstated expectation of clear-cut 'phenogenetic' connections--that is, between genes and traits.  More specifically, his work led to the idea that genes 'cause' traits and that natural selection only works on genetic effects.  That's because non-inherited effects, which certainly do exist, disappear with the individuals bearing them,while genetic effects are 'written' in genetic posterity until modified by further genetic change.  This is over-simplified, but still a powerful and valid way to view life.

But the truth is that phenogenetic connections are not all that precise.  When traits are controlled by many genes, each varying in a population, then each individual with the 'same' trait may have it for different genetic reasons.  The expectation of strong phenogenetic connections is the problem limiting GWAS and other genome-based attempts to explain, much less to predict,  disease or other traits. Overall, the problem can be seen in the common confusing of inheritance of genes, which does happen, with the inheritance of traits which doesn't (except for those present in a fertilized seed or egg).  Traits arise from genes, at least in part, and may resemble those in ancestors, but it is inaccurate if not misleading to think of traits as inherited directly.

Seen another way, when traits are the result of many different contributing genes that can each vary, as is so common, then even the same trait in an offspring and its parent (e.g., how tall they are), are affected by genes but are not due to the same genotype.  There are many traits in which variation really is largely due to single genes, but these are the luscious hooks that have addicted us to Mendelian principles when they don't really apply.

Furthermore, from an evolutionary viewpoint, genetic (reproductive) success is a relative concept among individuals in a population, and hence does not lie in genes alone, but in the environment in which they find themselves, and that is always changeable.  Even the future genetic environment is not entirely predictable, nor are the non-genetic aspects of environment (like our lifestyles).

Also, Mendel's rules are about the transmission of elements--genetic variants--from parent to offspring when the parents are dipoid, that is, when there is sexual reproduction and each of two parents carries two copies of each gene.  Bacteria and many other species (which means most species!) are haploid (only one copy of their genome) and don't transmit in a Mendelian way which is to choose between two copies.  This is what got Mendel into trouble when he began to experiment with hawkweed.

This means that the Modern synthesis was about diploid evolution, not evolution per se.  The evolutionary principles don't require Mendelian 'laws'.  This shows, in perhaps another way, that what Mendel did was provide a method for understanding variable transmission, that was used to discover what genes are and where they are in cells.

These issues make practical differences in science.  Today, GWAS and direct genome sequence comparisons used to look for causal genetic variation (say, of diabetes or autism) compare individuals sampled not from families as was necessary before we could really identify specific gene sequences, but sampled by their trait (cases vs controls, for example), with the genes searched directly to find variation preferentially associated with some specific trait value.  We can see this in the fact that new mutations arise all the time, that may have causal effect but were not transmitted from the person's parents.

In a sense, the properly lauded Modern Synthesis arose by the rejection of Mendelian causation, in the sense that clear-cut trait-inheritance was not required either for biological traits or for their evolution, if genetic variants, no matter how small or even statistically undetectable their effects, was transmitted in Mendelian fashion.  Rejecting the idea that traits were largely inherited as distinct effects made things consistent with gradual evolution of complex traits, and that made the synthesis possible.

But mixing Mendelian genetic transmission principles with genetic causal principles has led, we think, to expecting too  much of specific genes.

Mendelian inheritance and evolution. Part III

We've discussed some problems with the idea of Mendelian inheritance as a misleading legacy from the kinds of work that Mendel had to do, that has persisted deeply into modern thinking in ways that we think are misleading at best, or even plain wrong for several reasons.

First, because genetic units, not traits are what is inherited.  A fertilized egg doesn't have legs or a brain nor pollenated pea seed have color, wrinkles, or plant-height.  What is inherited are genetic variants that affect these traits in the adult organism.

Mendelism was wrong because most variation, the variation that enabled the modern synthesis to unite discrete Mendelian genetic inheritance with gradual Darwinian evolution, was generally minor relative to Mendel's purposes.  Clear discrete states and dominance are not the general ground state of biology.  In a population individual genes have many more than 2 alleles (variant sequences) and the effects of their paired combinations in individuals (plants and animals) is associated with more than two trait states (yellow or green).  Even at single loci, even when some alleles are in the traditional sense dominant relative to others, the dominance is usually not complete or invariate.

Mendelism was wrong in that the key to uniting discrete genetic inheritance with gradual variation in traits and their evolution, was that many different genotypes--combinations of alleles--yield the same trait.  When many genes co-contribute, as is widely the case, the trait can be called complex or 'polygenic'.   This is because the individual effects are generally small relative to the variation in the trait.  Small basically 'additive' effects predominate, rather than strongly determinative ones.  We had to give up on notions of Mendelian inheritance were then (and still widely are today) widespread.  But that this was the key to the modern synthesis was not widely perceived in these terms.

One could object that oh, yes what we say applies to multigenic traits, but not to single-gene traits.  Aren't there hundreds of these on the books, in human disease and in other species as well?  We'd respond that in real life most clear dominance or single-locus traits are much less dichotomous or simple that in general perception or in the textbooks, and  adaptation and complex function are demonstrably cooperative and multigenic rather than singe-factor competitive.  Complex organisms couldn't really function or evolve if they were just a bag of individual dominant, Mendelianly inherited traits.  And not only are most traits multi-genic, but genes, no matter how our definitions what a gene is change, are multi-allelic in natural populations.

Once we realize this, we have to accept that in general, statistical dominance, which we explained in an early post in these series, is a population rather than biologically inherent property of individuals.  It is the exception, and only a partial exception that is classically Mendelian: partial, because often even in these cases we overlook variation because it may not be great relative to our pragmatic purposes (such as diagnosing the presence of a disease).  But pragmatic considerations like this often turn out to be wrong and science is about understanding nature.

The problems mainly follow from falsely confounding inheritance of genes and with inheritance of traits, and from using the extreme of a distribution--the few nearly-dominant examples--to characterize the whole distribution of the effects of individual genetic variants. 

At the time of the synthesis, not enough was known about genes (or, at least, seriously and widely enough accepted) to recognize the implications of polygenic inheritance.  But now there is no excuse for clinging to theoretical concepts that are misleading and, at best, inaccurate ways to understand life. Doing so has led us down the paths that we imagine will promise simple answers of immediate clinical or commercial value. Each approximate 'hit' entices us to go deeper into the woods of our dreams.

Mendelian inheritance and evolution. Part II

The modern evolutionary synthesis reconciled what Mendel had showed about the discrete nature of inheritance (or so it was thought), and what Darwin had insisted on was the major gradual, more or less continuous, nature of the variation in traits and their evolution.

A lot of things contributed to this reconciliation.  Many experiments, largely stimulated by the 1900 rediscovery of Mendel's 1866 work, began to find mutations that were not grotesque, but had small effects, and were perfectly viable for the organism.  Big bad mutations did still arise and had been obvious and easy to see, but now the majority of mutations seemed to be of the lesser type.

Quantitative traits like stature varied gradually rather than by discrete jumps (such as between yellow and green peas).  Darwin's distant cousin Francis Galton had shown formally what had been obvious, that even for quantitative traits relatives resembled each other--indeed, they did so more than they might for traits with strong dominance (yellow peas did not resemble their green-pea ancestors in that trait).  A variety of investigators, most notably the statistical geneticist RA Fisher (in 1918) recognized how the two types of inheritance could be brought under the same umbrella.  If complex or quantitative traits were caused by the effects of many different genes, each making a small contribution, then the resemblance among relatives, the quantitative variation, and Mendelian inheritance were all consistent.

If that was so, then Darwinian gradual evolution was possible even with Mendelian inheritance as the basic fact of life.  That's what the modern synthesis showed.  It seemed wholly legitimate, indeed perhaps strikingly insightful, and it became the core theory of biology (and in many eyes, we think somewhat erroneously in ways similar to the adherence to Mendelism as the ground-state of inheritance, it still is).  The formal theory of evolution or, to many, of life was the mathematical theory called population genetics.  Well, Isaac Newton said something to the effect that if it can't be expressed mathematically it can't be a law of Nature, so we needed something real, not soft for biology!

It is strange for a central theory in a serious science to say nothing whatever about actual traits of the entities--organisms--that it purports to explain, but that's what population genetics does.  It says only that variation is comprised of discrete Mendelian (that is, discretely transmitted) units of inheritance ('genes' has been the word for them), and that by chance and mainly by the systematic force of natural selection, changes in the frequency of genetic variants were responsible for variation among organisms and among species.  Surprisingly, this theory became accepted even before there was any real theory of what a 'gene' is!  That means it was such a blanket view, so mathematically rigorous, that it could be universally applied to entities that hadn't even been discovered.  Reconciling this theory with actual traits has only come in recent decades under rubrics like EvoDevo (the evolution of development).  Our book, Mermaid's Tale, deals with these issues extensively, and we try to show that there has been far too much stress in biology that rested on the Competition-among-clearcut-units-is-almost-everything worldview of the modern synthesis.

The dogma became that Mendelism and Darwinism are compatible.

But in a real sense, we think, this is fundamentally wrong.  The modern synthesis essentially was possible only because Mendelian inheritance was itself wrong!

Mendelian Inheritance and evolution. Part I

We recently discussed problems with the idea of Mendelian inheritance, and the way that assuming that was a mode of trait inheritance has been so misleading and, we think for that reason incorrect.  (Here's the first of the five posts in that series.)  We'd now like to consider some of the implications of our point of view, when it comes to evolution.  We think this in fact reinforces that point of view, as we'll try to show.

Some traits do seem clearly to 'segregate' (be distributed) among family members in a present or absent mode, associated with a single allele (variant) at a gene, and if one allele is more likely to prevail that is called 'dominant'.

We pointed out that Mendel chose traits that 'worked'--that provided clear-cut material that allowed him to understand the nature of 'inheritance.'  That word is actually somewhat off, because Mendel was interested in the nature of hybridization (inheritance in a mixed cross between strains), and there was the confusion that still widely reigns, between inheritance of traits and inheritance of genes.  By pragmatically equating the two, because that wasn't too far off in his chosen model situation, Mendel got his result.  Even he knew that most traits didn't work that way (and even some traits that should, in some species he later studied, didn't).

For many decades after Mendel, it was thought that such traits had two important properties:  they were stable, and they made clear-cut qualitative trait differences.  If either wasn't true, then Mendel's rules wouldn't work, hybridization wouldn't work as expected, and how inheritance worked would remain obscure.

And, when Darwin's and Wallace's ideas came along, it was clear that evolution wouldn't work, either!  The most one might expect would be that one of the alleles would replace the other over time, by natural selection.   But this was not satisfactory, because evolution clearly involved smaller, gradual changes more than big dramatic ones.  And evolution, even by Darwin's mistaken genetic ideas, would continually generate new variation for selection to screen.  That was a big problem, because the observed new alleles, traits in offspring that were not present in their ancestors, seemed mainly to be grotesquely harmful, not material that selection could work with to adapt to changing circumstances.

Darwin was aware of much of this, too, in fact.  He could not see how immutable traits, like the dominant/recessive ones, could evolve and also he felt that evolution simply had to be gradual and more or less continuous. Among his reasons were his erroneous largely Lamarckian notion of the physical basis of inheritance, and the fear that without gradualism traits or species might arise suddenly....requiring explanations for their origin.

As a result, for decades leading biologists, realizing that Darwin's gradual inheritance theory (that he called 'pangenesis') was fundamentally wrong, were not convinced that natural selection could account for adaptive evolution or, in particular, that this had anything to do with the known mechanisms of inheritance.

The solution that was widely accepted and prevails (not always accurately) to today, was worked out in the 1930s and later, based on some theory and data from earlier in the 1900s.  This solution was known by names like the 'modern evolutionary synthesis.'   It reconciled Mendelian inheritance with gradual evolution in ways that were consistent with variation within and between species, and with the fossil record.

There are many things to dispute about the modern synthesis, and we'll point them out (and have mentioned them before here on MT).  But one thing that is important, but so far as we know hardly if at all recognized, was that the modern synthesis was possible only if Mendel was wrong, in ways we tried to outline in our recent previous series.

We'll elaborate on these thoughts in the next installment.....so stay tuned!