Was Darwin Wrong?

[Update: I understand that the cover of New Scientist says "Darwin was wrong" and as Bjorn has just told me, this is the original article. I'll read that one as soon as I can.]


A new article in the Guardian (hat tip Bjorn at Pleion) has the headline :

Evolution: Charles Darwin was wrong about the tree of life


I think this headline, and the spin of the article in general is a rather extreme over-simplification, and more importantly, it is subject to misinterpretation by anti-evolutionists.

The main points I want to make are:

1. It is an oversimplification to say that Darwin was "wrong" on this point. It is not a clear cut case of right or wrong. Instead, the facts as we understand them today are more complex than what Darwin envisioned, or could have envisioned (given he didn't know about DNA).

2. The primary, most general implication for the history of life is not changed. Darwin's tree of life posits common ancestry of all life. This is the central scientific fact that anti-evolutionists rebel most against (because they don't want to admit we are all related to slime-molds, etc). In fact, the new observations about biology continue to reinforce Darwin's history-changing insight that all life shares a common ancestry. Yes, we share a common ancestor with a chimp and a fungus, get over it.


Below is what I wrote about our changed view of the Tree of Life in a paper with Michael Rose, published here in the Open Access journal Biology Direct.

Complications for "The Tree of Life"

Nineteenth and 20^th Century biologists generally conceived of a "Tree of Life" – a mostly bifurcating graph connecting species in an order that reflects their common ancestry. At least three processes complicate such a view of a tree of life, horizontal transfer, symbiogenesis, and differential lineage sorting of genes. Each of these processes are at odds with fundamental assumptions of the Modern Synthesis [7,8] and a Tree of Life for the new biology is necessarily more complex than a graph joining species.

In the middle part of the 20^th Century, it was often supposed that organisms and their cells are sleekly functional (Fig. 2A). Given such assumptions, passing genes from one species to another would not be favorable if those genes were finely tuned for the necessary functions of the species from which they originate. Even the movement of genes within a single genome was not accepted by the biological mainstream at that time, despite McClintock's early discovery of accessory elements in maize [9]. Nevertheless, molecular characterization of transposable elements in the late 1970s finally undermined the view of the genome as a static, well-organized library of genetic information [reviewed in [10]]. With the advent of genome sequence data, researchers studying the molecular phylogenetics of bacteria realized how common prokaryotic horizontal transfer is [11,12].

Similarly, modernist preconceptions led some to discount the importance of endosymbioses in the origins of new life forms, like eukaryotes. Broad theories of endosymbiotic origins for species had been suggested in the late 19^th and early 20^th Centuries [7], but were ignored save for a few well-established cases like lichens. By the 1980s, the evidence for symbiogenesis in major cell biological events was voluminous [13,14].

Even systematics has had to abandon many strictures that were part of the Modern Synthesis. If species are the durable unit of biology, and if natural selection quickly molds genes to current utility, then most genes should diverge at the time of speciation events, given views like Mayr's. Here again, analyses of newly abundant sequence data in the late 20^th Century showed that rather than a highly congruent coalescence of genes at the times of speciation events, the coalescence times of alleles among species are highly variable. As such, species trees and gene trees often cannot be equated [15,16].

These phenomena complicate the tree of life. Rather than a graph connecting species, the tree of life itself is hierarchical: A universal tree of species is largely a human-imposed ideal because the components of any particular species have evolutionary histories that are not congruent with each other. This incongruence has a clear and well documented mechanistic basis in horizontal transfer, symbiogenesis and differential lineage sorting (not to mention gene duplication explained above). These processes together undermine the existence of a tree of life defined only at the level of species, pointing instead to branching histories that often differ among levels of organization and scales of analysis.

Figure 2. Old and new views of the evolution of prokaryote genomes. (A) 20^th Century biologists sometimes assumed a close congruence between gene history and species history. Horizontal gene transfer was assumed to be uncommon, as the process of genes entering a new genome is counter to the idea of a sleek and well adapted genome. (B) After analyzing the genomes of many prokaryotes, biologists recognized that horizontal gene transfer may be a common event. Furthermore, prokaryote species trees may be viewed as a patchwork of gene trees with varying levels of congruence. A similarly hierarchical view of eukaryote evolution has been articulated by Maddison [15], except that differential coalescent times – usually not horizontal transfer – is the primary mechanism used to explain incongruence of gene and species trees.

And in the response to reviews:

2. Rose and Oakley bring up the newly apparent prevalence of horizontal gene transfer as one of the major blows to the 20^th century perspective in biology. This is, certainly, true, but I think the discussion in the paper stops short of really driving the nail down. The real issue is that, when fully conceptualized, extensive HGT undermines the very notion of the Tree of Life (the TOL paradigm) which, certainly, is a big part of the Modern Synthesis (as well as the classical, Darwinian foundation of biology). Simply put, although trees are crucial in depicting certain phases and aspects of life's history, there is no TOL as such, i.e, evolution of life cannot be presented as a tree, so Darwin's famous simile fails as an overarching generalization. The demise of the TOL paradigm is covered in several recent papers [91,92]. Again, this is related to the problem of "eukaryotic chauvinism": the tree pattern might hold for the evolution of the major divisions of eukaryotes (although not necessarily for all eukaryotes taken together) but, certainly, not for prokaryotes, let alone the entire history of life.

Authors response – Here we disagree with Dr. Koonin. HGT does not necessarily undermine Darwin's "Tree of Life" completely, even though in post-Modernist biology this Tree of Life is much more complex. Today's Tree of Life, as Dr. Koonin points out, is different from what Darwin envisaged, in that it is multi-dimensional – branching histories characterize in a complex way multiple levels of organization, not just the species level. Further, as discussed in the article, HGT is not the only blow to a two-dimensional tree; paralogy, endosymbiosis and lineage sorting also contribute to a new, highly multi-dimensional view of evolutionary history.

This emerging understanding of the trees of life is pluralistic, encompassing the branching history of biological units at all different levels of organization [81]. The evolutionary histories of units at different levels (gene domains, genes, species, etc) are not always congruent with each other, yet there are still branching histories that characterize each of these levels. Branching history is a pattern that results from well known mechanisms including exon shuffling, gene duplication, genome duplication (polyploidy), co-option, speciation, and vicariance of multiple species. HGT is one example of a mechanism that causes branching histories at different levels of organization to be incongruent. It clearly points out the failed assumption that the history of components is congruent with the history of the higher level unit to which it belongs. Nevertheless, this assumption can be used as a valuable null model to understand macroevolutionary patterns and processes [93]. As we discussed in this article, the species was usually seen as the durable unit driving branching at all levels, but the existence of multiple evolutionary levels and mechanisms violates this assumption.

Processes to split biological units pervade all levels of the biological hierarchy. Protein domains duplicate within genomes and may be "horizontally transferred" from one gene to another. Genes may form units of synteny or operons, but individual genes may also be copied from one part of the genome to another or from one genome to another, independently of the rest of a synteny unit or operon. Whole chromosomes and whole genomes may also duplicate by various mechanisms. All these processes create the new tree of life. But that tree is a postmodern tree, rich in complexity. Components coalesce to form units with a congruent path for a time, only to be broken up. There is no reason to provide anti-intellectual, anti-evolutionists with quotes like "The Darwinian paradigm is dead", because this complexity only enhances Darwin's most profound insight – the universal common ancestry of life.

PT eye evolution flare up

I was notified that there was a little flare up at Panda's Thumb about our recent article on eye evolution, entitled:

Opening the "Black Box": The Genetic and Biochemical Basis of Eye Evolution by Todd H. Oakley and M. Sabrina Pankey (PDF) published in Evolution: Education and Outreach.

Warning: Long Post. If you only read one thing, read this:

One might argue that we are just pushing back the origins, changing the question of "phototransduction" origin to the question of "transduction" origins. In a way this is true, but it is also a fundamental insight about how evolution works. New features are not breathed into organisms by some unknown force, they evolve by duplication/divergence or recombination of existing features. Trace a feature like phototransduction back far enough in evolutionary time, component by component, and it grades into something else altogether.

Here is the story:
A week or two ago, I was contacted by an undergraduate from a university in the UK (I see no reason to reveal her name). She had questions about the article, writing:

I've just been reading your 2008 paper 'Opening the Black Box: the genetic and biochemical basis of eye evolution' and found it really interesting! I'm just writing an essay on eye evolution atm and am trying to get to the crux of the issue, and find out how the phototransduction cascade itself evolved. After explaining that opsin probably arose by a mutation in a serpentine gene/protein, you mention in your paper that:

"In yeast...these receptors [GPCR's - the serpentine proteins] are sensitive to pheremones, and they even direct a signal through proteins homologous to non-opsin phototransduction proteins."

What I'm wondering is, is it this whole yeast pathway that has been modified for the metazoan phototransduction cascade? Or is it only the opsin which has been derived from it? (With the subsequent molecules involved in the phototransduction cascade being co-opted from other proteins not involved in the yeast signalling pathway).

Based on this email, it seemed the student had a pretty good grasp of the issues, and I replied:

Thanks for you questions.  I found out after writing the paper that the yeast pheromone proteins are not the "rhodopsin type" GPCR, so they are distantly related at best to opsins.  So they should not be considered anything like direct ancestors of opsin.

As for other components of the yeast pheromone cascade, these are different than phototransduction.  Yeast pheromones activate a MAP kinase cascade. So, I think all that is similar is the GPCR and G-protein.

So, it really is an open question as to what the ancestral function was of some of the genes of phototransduction, although some of these genes do function in other sensory transduction pathways...

Unfortunately, she misinterpreted this email, understanding it to mean (written over at PT, using the pseudonym "Green"):

"Yeah I knew that co-option thing would be coming. Turns out there was a mistake in that Oakley and Pankey paper.

The yeast intracellular signalling cascade turns out not to be homologous in any way to the metazoan signalling cascade. It just got published before the authors realised."

First of all, this is an incorrect interpretation of what I wrote. In fact, I wrote "So, I think all that is similar is the GPCR and G-protein." This is far from "not homologous in any way", as claimed by the student. The G-protein is undeniably homologous, and the yeast pheromone receptor is a 7-transmembrane protein, at least conformationally like opsin (whether or not the opsin and pheromone receptor sequences are homologous is a trickier issue).

But the larger issue is I think an issue of "linear thinking", which I address quite often on this blog. The student seems to think that if we cannot identify in yeast (taken as a linear ancestor of animals) a cascade identical to phototransduction except for opsin, then the origin of phototransduction requires numerous simultaneous mutations. This is not the case. First of all, yeast is a more distant relative of animals with phototransduction than is sponges. I just mentioned the yeast pheromone photoreceptor in the paper as a well studied example of a pathway outside of animals with partial homology (some components homologous, some not) to photoreception. There are closer "relatives" of phototransduction in sponges (poorly studied functionally, but the genes are known) and in other animals (better studied).

Another issue is a difficulty that people have with thinking about partial homology - that some components can be homologous and some not, depending on the time scale of the comparison (see my post The Red Herring of Eye Evolution).

Partial homology is a pattern that indicates a mechanism of co-option in the evolution of features. Co-option is the combination of existing things in a new way (analogy: dijonaisse = dijon mustard plus mayonaisse). All of the components of phototransduction pre-date animals, except opsin. And if we consider opsin to be a GPCR, which it is, then all of the components of phototransduction pre-date animals. This may be considered a pattern of co-option, or exaptation. Signaling pathways were already present before phototransduction. Some of the phototransduction components function together as far back as the yeast + human common ancestor (GPCR + G-protein). Other components of phototransdcution function together in non-phototransduction cascades of other animals. This indicates that phototransduction did not assemble all at once, but built incrementally upon an existing scaffold.

One might argue that we are just pushing back the origins, changing the question of "phototransduction" origin to the question "transduction" origins. In a way this is true, but it is also a fundamental insight about how evolution works. New features are not breathed into organisms by some unknown force, they evolve by duplication/divergence or recombination of existing features. Trace a feature like phototransduction back far enough in evolutionary time, and it grades into something else, component by component.




There were other comments, too. Again at PT, she also commented:

Yeah I read Oakley and Gregory’s articles on eye evolution a couple of weeks ago. Unfortuantely neither address the crux of the issue: namely the origin of the biochemical phototransduction cascade.

To be fair, Oakley’s article (the ‘Black Box’ one) at least tries to give some biochemical details. But it only scratches the surface by suggesting a potential origin of the opsin protein. Unfortunately the origin of a new opsin protein is not equivalent to the origin of an entire phototransduction cascade.

So it seems the Darwinian account still falls quite far short of any satisfactory biochemical explanation. Descriptions of morphological change, comparisons of genes, crystallins, etc. all skirt the issue if it cannot be shown how the phototransduction cascade itself arose

The difficulty with this comment is that the origin of opsin defines the origin of phototransduction. The other components of the cascade were already there, they all predate opsin, as described above.

This is also a "God in the Gaps" argument, or maybe, a "God under the surface" argument, stating that describing the origin of the keystone molecule of phototransduction (opsin) "only scratches the surface".

Also, I don't understand what the difference is between "comparisons of genes" and "biochemical explanation". What would a biochemical explanation be for the evolutionary origins of things that doesn't involve "comparisons of genes". The genes of the phototransduction pathway have biochemical interactions with each other, many mediated by interactions between specific amino acids.

How Complexity Evolves

[Slightly altered excerpt from an invited review on the evolution of (nervous system) complexity]

The evolution of complexity is an enduring and fundamental topic in biology. Recent research is allowing new insights into the origins of complexity. Namely, scientists now have access to the components of complex structures, and their evolutionary histories. "The eye" is no longer an anonymous collection of components - Darwin may have known an eye is composed of lens retina and nerve - but he did not know what lenses or retinas were composed of, nor did he know in much detail how they functioned. We now know many of the protein components of structures like eyes and how they function, and we know that these components have evolutionary histories. Since biological entities from genes to ecosystems arise from existing entities, general patterns emerge as to where those entities came from. They either duplicate/diverge; split/diverge; or fuse in new combinations.


What is biological complexity?
Numerous definitions have been suggested for biological complexity, from mathematical and information theoretic formulas [2,3] to intuitive definitions, such as the numbers of parts or components that can be counted in a system [4,5]. In practice, a researcher’s choice of definition for complexity often boils down to his ability to measure it. In investigator-defined computer simulations [2] or circumscribed biological systems [3], information theoretic definitions using explicit mathematical formulas are tractable. However, in extensive biological systems, especially those that encompass multiple levels of biological organization with varied properties, the relationship between structure and function (e.g. genotype and phenotype) is often largely unknown, and may itself be complicated and variable. These unknowns complicate the formulation of explicit, yet general, mathematical definitions of complexity for extensive biological systems. Instead, some have argued that counting the numbers of parts of a system is an appropriate measure of complexity [4]. Although often focused on structural units, such as cell types, genes, or species, and not on the functions of those units, McShea [5] has also argued that counting structural parts should often be a good estimate of functional complexity.
Herein, we follow previous authors [4,5] and define more complex systems as those that have more parts. “Part” is used as a term that spans levels of biological organization [4]. Species are parts of ecological communities. Organs (like brains or ganglia) are parts of species. Cell types (like neuronal types) are parts of organs and of species. Proteins are parts of cells and of networks, and domains and amino acids are parts of proteins. Many other biological units are also parts [6]. For the current discussion, we are concerned less with defining or counting parts, and concerned more with how new parts originate during evolution. According to the definition of complexity that we employ, new parts (that do not come at the expense of existing parts) increase biological complexity. Therefore, those mechanisms that cause the evolution of new parts are of particular interest because those are the mechanisms that cause increases in biology complexity.

"those mechanisms that cause the evolution of new parts are of particular interest because those are the mechanisms that cause increases in biology complexity."

General patterns of increased complexity
By definition, complex systems have many parts, and the histories of those parts are varied. As such, we cannot expect a simple, one dimensional answer to the question of how complexity evolved. Nevertheless, at all levels of biological organization, conceptually similar patterns (Figure 1) have resulted from a varied array of mechanisms that historically increased complexity.



Figure 1 - Generalized patterns of increased biological complexity. The top of the figure represents an ancestral state, the bottom a descendant state. Shapes represent “parts”, a generic term for a biological feature at any level of organization. Parts can be species, genes, protein domains, pathways, brain regions, and many other biological units [see 6]. A. Copying and divergence B. Fission and divergence C. Copying and fusion. Without copying or fission, complexity does not increase because divergence (or fusion) maintains the same number of parts. Without differential divergence or fusion of parts, complexity only increases marginally.



First, parts may exhibit a pattern consistent with differential divergence of copied elements (Fig. 1a). A prime example of a specific mechanism leading to this pattern is gene duplication plus divergence. Duplicated genes are initially identical, and they gradually diverge over time, increasing genomic complexity. Second, parts may exhibit a pattern consistent with differential divergence of split elements (Fig. 1b). Here, a prime example is speciation, where populations of individual organisms, originally all of the same species, split into multiple populations that diverge to the point of becoming separate species. Splitting (fission) can also occur in an asymmetric fashion (Fig 1c), generating two uncoupled parts that together would sum to one ancestral part. Third, parts may exhibit a pattern consistent with fusion of copied parts (Fig. 1d). For example, copied protein domains often join together to generate a new gene. Another example of fusion is expression of genes in new combinations, a process termed co-option [reviewed in 7]. A primary goal of this paper is to review cases in nervous system evolution that provide specific and more detailed mechanisms that account for these patterns at different levels of biological organization.


The patterns in figure 1 are produce by a number of different mechanisms, including gene duplication, alternative splicing, retrotransposition, co-option, etc.

[The specific mechanisms leading to the general patterns is the topic of the rest of the paper from which this excerpt is taken].



References Cited
1. Striedter GF: Principles of Brain Evolution. Sunderland, MA: Sinauer; 2005.

2. Adami C, Ofria C, Collier TC: Evolution of biological complexity. Proceedings of the National Academy of Sciences of the United States of America 2000, 97:4463-4468.
3. Adamowicz SJ, Purvis A, Wills MA: Increasing morphological complexity in multiple parallel lineages of the Crustacea. Proceedings of the National Academy of Sciences of the United States of America 2008, 105:4786-4791.
4. Bonner JT: The Evolution of Complexity. Princeton: Princeton University Press; 1988.
5. McShea DW: Functional complexity in organisms: Parts as proxies. Biology & Philosophy 2000, 15:641-668.
6. McShea DW, Venit EP: What is a part? In The Character Concept in Evolutionary Biology. Edited by Wagner GP: Academic Press; 2001:259-284.
7. True JR, Carroll SB: Gene co-option in physiological and morphological evolution. Annual Review of Cell and Developmental Biology 2002, 18:53-80.

Evolutionary Novelty: Hair

Mammals have hair but no other animals do. As such, hair is a clear evolutionary novelty, present in one group but absent in all others. In my macroevolution course (EEMB 102), I use hair as a clear character that can be used in phylogenetics. Hair groups all mammals to the exclusion of other organisms. In systematics jargon, hair is therefore a “synapomorphy”, grouping mammals together.

We can map the trait of hair on a family tree of animals. From this perspective, we can infer that the ancestor of all mammals very likely had hair, but that the ancestor of sauropods (birds, reptiles, and mammals) lacked hair. Therefore, hair originated prior to the common ancestor of all mammals.



Figure 1 – Hair originated before mammals, but after the common ancestor of birds, reptiles and mammals. Grey ellipse (hard to see except as a broken branch, I'd fix it but I'm too lazy) is the origin of hair keratin protein.



So where did hair come from - how did this evolutionary novelty evolve? A new paper by Eckhart et al in PNAS [link] provides evidence that the building blocks of hair pre-date the origin of hair itself. Namely, they found alpha-keratin (“hair keratin”) proteins are encoded in the genomes of chickens and the green anole lizard. In the green anole they studied, ‘hair keratin’ proteins were used in claws.


Just ten years ago, results like this clarifying the molecular components of trait evolution were rare, but they have become common now that genome sequences are available for many species. Before we had some idea of gene function, and before genome sequencing, scientists could only examine one level of biological organization – the trait (hair in this case). And that could only get science so far. In the case of hair, it mainly got science as far as Figure 1, which leads to the inference that hair evolved a bit before the common ancestor of living mammals. But “hair” is not one thing. It is a complex of building blocks, including structural genes (like keratin) and developmental processes. Today, scientists can decompose a trait, like hair, into its components and study the evolutionary history of each part separately, tracing the parts through various genomes.

What do we expect for the evolution of hair’s components? Figure 1 suggests that “hair” and all its components arise at the same time, near the origin of mammals. The origin of “hair” on figure 1 can be considered a first-pass hypothesis for the origins of ALL the components of hair. If hair itself originated near the origin of mammals, a logical idea is that the components originated then too.

Today, we can test this first-pass hypothesis because we know some of the molecular components of hair. A particularly important part is “hair keratin”. Mutate this protein and the hair built from that mutant protein is fragile and brittle. The expectation based on figure 1 is that hair keratin proteins originated with hair itself. But the discovery of these genes in an anole indicates an earlier origin for this component. In other words, components of hair originated before hair itself. In this case the protein “hardened” by mutations to cysteine amino acids that may have functioned to molecularly harden the proteins. Since these changes were later useful in the structure of hair, they may be considered exaptations, features that originated for functions other than current utility: keratins may have hardened before that feature became useful for hair formation. [Note for scientific accuracy – the biochemistry of the anole protein has not been studied, so while it is cysteine rich, we don’t know yet if the anole protein is ‘hardened’].

This work also illustrates that in evolution, new things do not appear from nowhere [see my post Coming to Grips]. In evolution, new things come from the duplication/differential modification and recombination of existing parts. Morphologists know this, as one dominant idea about the origin of hair is that hair evolved by modification of scales. Hair keratin is not expressed in anole scales, so the scale hypothesis is not supported by the new PNAS paper. Also unfortunate for the scale hypothesis is the fact that the fossil record retains no transitional forms between scale and hair. Even though morphological relatives of hair are ambiguous, the molecular relatives in this case are clear. Hardened keratin comes as two types, which share an evolutionary relationship, and hardened keratins may share an evolutionary relationship with soft keratins, proteins that are present in numerous tetrapods, and therefore have a more ancient origin than the hard variety. In sum, keratin has an ancient heritage, and through gene duplications and differential modification, two related groups of these proteins have specialized as hair keratins. Fascinatingly, some of the hair keratin modifications pre-dated hair itself.







If you are interested phylogenetic analyses of trait evolution, and the evolutionary history of trait components, this is a common theme of research in my lab.

We’ve found:

Synaptic components are present in sponges and therefore may predate synapses. [paper] [blog]

Phototransduction components were first assembled for vision in the eumetazoan ancestor (cnidaria + bilateria), yet some components pre-date animals [blog] [blog] [paper] [paper]

See also: Red Herring Blog

Phylogeny, evolution, biodiversity and ecology

We are in the midst of massive upheaval in the world’s ecosystems, driven by species invasions and the sixth mass extinction in Earth’s history. How will these changes in biodiversity affect the functions of ecological communities? Will the functions of ecological systems that humans rely on for survival, such as production of oxygen, be impacted by all this upheaval?

Answering these questions requires that biologists have a good metric for biodiversity. New research by Marc Cadotte, Brad Cardinale and I, and published this week in Proceedings of the National Academy of Sciences, indicates that one particular metric of biodiversity - evolutionary diversity - is a particularly strong predictor of the biomass produced in plant communities: The more biodiversity present (measured as evolutionary diversity), the more productive the community. In fact, for the datasets we examined, evolutionary diversity was a better predictor of productivity than raw species number or number of functional groups in the community. This suggests that the most evolutionarily diverse communities may function best, and that the most evolutionarily distinct species might be the best targets of conservation efforts aimed at maximizing ecosystem productivity.

Measuring Biodiversity
One common theme in current ecological research is to ask questions about how changes in biodiversity impact or influence ecological systems or communities. This has obvious importance when we know that many species are going extinct, and that many species are being shipped around the world by human transportation. Often in ecological research, a measure of biodiversity is placed on the X-axis, and some predicted response is placed on the Y-axis, to test if there is a strong relationship. For example, one might predict that less diverse and simpler communities are more susceptible to invasive species compared to more diverse and complex communities. One might also predict that more ecological diversity leads to a healthier ecosystem, as measured by higher production. Namely, a higher diversity of organisms could make more efficient or more complete use of available resources, ultimately leading to a healthier, better functioning ecosystem.

These types of ecological studies usually use a measure of biodiversity referred to as “species richness” as the X-axis variable. “Species richness” is simply a count of the number of species. However, simply counting species makes the assumption that, say, two very closely related grass species contribute the same amount to the diversity measure as two much more distantly related species, such as a grass and a magnolia. In contrast to this implicit assumption of “species richness”, phylogeneticists often think of biodiversity in terms of evolutionary relationships, assuming that differences between species (one way to conceive of diversity) accumulate over the time since they last shared a common ancestor. To a phylogeneticist then, species that share a very recent common ancestor – like the two similar grasses mentioned above – should be nearly identical and therefore represent less total diversity compared to the much more distantly related grass and magnolia species.

We wondered if evolutionary diversity really does matter for predicting how much biomass a community produces, one measure of the health of ecological communities. Decades of experiments have already established that species number (“species richness”) is in fact correlated with productivity – the more species of plants growing together, the more biomass is produced. We extended these studies, weighing different species by how closely related they are evolutionarily. Could we better predict biomass production by also accounting for evolutionary (phylogenetic) diversity? Based on our analyses, the answer was a clear “yes”. Incorporating evolutionary distances into our biodiversity metric resulted in better predictive power of the productiveness of experimental plant communities. The metric including evolutionary history was better than “species richness” and better than the number of functional plant groups, two commonly used metrics of biodiversity.

Our data set was a collection of 40 different previously published experimental studies, conducted around the world using a total of 177 different species of flowering plants. Researchers planted experimental ecological communities, using many different combinations of plant species, and using different numbers of species. Then they let the communities grow, and measured the biomass produced by the different combinations. We added an analysis of the phylogenetic relationships of the plants using publicly available genetic data from four different genes commonly used in other studies of plant phylogeny.

Phylogenetics and Nihilism

Do all ecologists now need to become phylogeneticists? This question is similar to one asked of comparative biologists in the mid 1980’s.

In 1985, Joe Felsenstein wrote a landmark paper introducing the method of phylogenetic independent contrasts, which is now standard in comparative biology. The core message is that we cannot treat species as independent entities because they share a nested set of common ancestors. In other words, species are similar because of descent, not only because of adaptations, and traits might be correlated across species because of shared evolutionary history. At that time, comparative biologists were told they must consider phylogeny when testing for correlations among traits. Felsenstein addressed the question, “What if we do not take phylogeny into consideration [in comparative biology]?” His answer:

“Some reviewers of this paper felt that the message was “rather nihilistic,” and suggested that it would be much improved if I could present a simple and robust method that obviated the need to have an accurate knowledge of the phylogeny. I entirely sympathize, but do not have a method that solves the problem…. Comparative biologists may understandably feel frustrated upon being told that they need to know the phylogenies of their groups in detail, when this is not something that they had much interest in knowing. Nevertheless, efforts to cope with the effects of the phylogeny will have to be made. Phylogenies are fundamental to comparative biology; there is no doing it without taking them into account.”

-Felsenstein (1985)

Although other systems and other questions might differ from our study in how diversity relates to ecological processes, it seems to me that counting species is far too simplistic of a metric of biodiversity. If adding phylogenetic information was valuable in one case, it seems worthy of strong consideration any time a metric of diversity is below the X-axis in a graph. To paraphrase Joe, ecologists may understandably feel frustrated upon being told that they need to know the phylogenies of their groups in detail, when this is not something that they had much interest in knowing. Nevertheless, the evolutionary history of their focal communities or systems will often have a lot to tell them. Species are not independent entities, and biodiversity cannot be measured as if they were.

M. W. Cadotte, B. J. Cardinale, T. H. Oakley (2008). Evolutionary history and the effect of biodiversity on plant productivity Proceedings of the National Academy of Sciences, 105 (44), 17012-17017 DOI: 10.1073/pnas.0805962105

Quick impressions of “Evolve: Eyes” on the History Channel

I recorded the TV show Evolve:Eyes, which aired on the History Channel last night, and I watched it today. Last night during the original airing, I was in LA watching the Dodgers with my family. I think my son was quite taken with professional baseball. This was not his first game (I had to make sure that for both kids, their first game was in Milwaukee, where mine was), but I don’t think my son remembers seeing the Brewers in Miller Park. So when we walked through the gate of Dodger Stadium, I saw the awe in his eyes, as he took in the stadium’s brightness, the impossibly green grass, and as we heard 40,000 fans cheer together. There is something to be said for appearance and spectacle, which can make even things like playing baseball games seem important.

This was the primary strength that I found in the History Channel show Evolve:Eyes; it looked really slick; it had its share of spectacle. There were fantastic graphics. Three dimensional eye figures exploding into component lens, cornea, and retina. The linear progression of eye evolution was re-told graphically, a few photoreceptors morphing into a deepening pit. Then a lens appeared, until the human eye had “evolved” right before my own eyes.

There were also dramatic film sequences of real scientists. There was a re-enactment of my friend (we were postdocs together in Chicago) Nate Dominy of UCSC doing his field research observing primates (I’m guessing they filmed Nate in Santa Cruz with binoculars and cut in footage of primates). There was footage of my friend Chris Kirk (haven’t seen him for a while, but we were grad students at Duke at the same time) pulling eye balls out of jars, complete with lab coat (most biologists I know don’t wear lab coats, unless they’re being filmed, or maybe if their pulling eyeballs out of a smelly formalin/ethanol solution). And there was footage of my colleague Alex Goodell collecting Polyorchus (a cnidarian) and shining LED lights on them to watch their behavior: green light, then blue, illuminating the tank and animal like neon lights in the dark. Scott Edwards, alone in an arid landscape, artistically filmed full-body and off-center, commented on how many times eyes evolved, my proverbial red herring, deciding to re-tell the take of a fellow Harvard-man, the late Ernst Mayr. Spectacle. Beautiful.

But all was not positive from my perspective. Namely, the show fell into the two most common traps that plague the understanding of evolution by the public. 1) Evolution is not adaptive story telling and 2) Evolution has not progressed in a direct line from simple to complex; evolution most certainly does not culminate with the human species. In nature, humans are no more special than bacteria. We are just another twig on the great tree of life (that was for Eisen).

Very little of this show was actually about evolution, and most of the “evolution” that was mentioned did not explain the actual science behind it (there were of course some excpetions). Instead, there was a lot of adaptive storytelling. Eagles evolved amazing vision to allow them to be better hunters. I don’t doubt this is true, but this show very often simply asserted and explained current function, with little mention of how evolution of those traits proceeded. Jellyfish try to escape UV/blue because it is damaging. Again, this may be true, but it was simply asserted in the show. Evolution was simply assumed, with little mention of how we actually have established that evolution works.


But the most egregious to me was how the narrative and graphics re-enforced the non-existent linear progression from jellyfish to humans. In fact, the graphics were a line, shown between stories, proceeding from 600 million years ago to present day primates. In this hypothetical line, first came the “jellyfish” (they actually showed a hydromedusan). The camera simulated the “primitive” jelly’s blurry vision, far inferior to our own, no mention of the cubomedusan’s highly complex eye. Next came the trilobites, whose calcite compound eyes are among the oldest fossil eyes. (To the show’s credit, they did mention that trilobites are not direct ancestors of living arthropods, a slight nod to “tree thinking”. Also, the Cambrian and trilobite animations were pretty cool. More spectacle.)


Then came the vertebrates. Not much mention of fish, instead we jump to dinosaurs. More flash here, tremendous spectacle, shining lasers on models of T. rex, the icon of charismatic species. Then the story moves up the ladder to early mammals, “primitive mammals” eeking out an existence, with only “black and white” vision, living in the shadow of dinosaurs. Then the great KT event knocks out rex and mammals can finally flourish. This next bit really got me. They compare human (and some other primates’) vision to that of early mammals, saying how great we are to be able to see red. What a shallow point of comparison, just to inflate the human ego, and maintain the march of progress up to humans! Why not compare to our eyes to stomatopods (mantis shrimps), who see circular polarized light, and have upwards of 12-chromatic compared to our tri-chromatic vision. Or even compare us to adult maggots (i.e. flies), who see polarized and UV light that we cannot detect. In my book, the evolutionary ladder of eyes proceeds up to mantis shrimp, not humans.



But perhaps I am being a bit too critical, expecting too much nuance from a show aimed to keep viewers from switching to, say, ESPN to see how the Dodgers are doing (they won by the way). Spectacle can be something to celebrate!

After the Dodger’s game last night, I could tell my son was inspired by spectacle, not by nuance. In fact, I doubt he got any nuances of the game. He was oblivious to strategic pitching changes; clueless about a rare call that gave the Dodgers an extra run in a tight game (the Giant's left fielder juggled the ball out of play, something I’ve never seen, after many years of watching baseball, e.g. I’ve seen in person a triple play, and Ricky Henderson’s record breaking stolen base). No, it wasn’t nuance that compelled my son to ask when he gets to start soccer, and if I can video record his games so we can watch them on TV. It was spectacle, fan screaming, light flashing, ice cream in a Dodger helmet spectacle.

So who am I to complain if somewhere, some kid is taken by the spectacle of eye evolution, as so vividly portrayed last night? Taken by the idea of shining lasers on T. rex, or neon LED’s on pulsating jellyfish in dark rooms? Maybe an African-American kid saw Scott Edwards on TV last night, and dreamed a dream of studying evolution. Maybe the child of an anti-evolutionist now has some questions.

There is a place for spectacle. If spectacle can inspire, nuance can come later.

Choose your own descent

I was interested to see that over at The Tree of Life Jonathan Eisen discussed a few different metaphors for common descent, which include:

The tree of life
The coral of life
The watershed of life
Blood vessels of life
Lungs of life
Shrub of life


I think metaphors are useful for teaching the idea of common descent and tree thinking, and so I try to think of new metaphors from time to time, since I teach a course on Macroevolution and also talk a lot about phylogenies in an Invertebrate Zoology course.

One that I think is really powerful, is found in Richard Dawkins' book Ancestor's Tale. The idea there is to reverse time in common descent, and imagine humans walking on a path to join our ancestors. First chimps join the band, later other apes, then other primates, other mammals, etc. This analogy is good for understanding the history of a particular entity, like the human species. However, since it has a specific starting point, it tends to re-enforce the way people conceive of evolution as a linear series of events. This is true if we have one specific endpoint in mind (like humans), but evolution does not have one specific endpoint.


So, this morning, this was going around in my head, and a new metaphor sprung into my brain: A choose your own adventure book. These books have a section of the story, and then the reader must choose between two (or more) different paths, each with a different outcome to the story.


Congratulations, your character has just achieved bilateral symmetry! If you would like to proceed down the path of protostomy, go to page 150. If you would like to proceed down the path of deuterostomy, go to page 376. It might be fun to write an actual evolution book this way!

Coming to grips with common descent

Talk about a bump. Since PZ mentioned this blog, visits are through the roof. I'm not sure I really ever intended to blog for the masses; I was just using a blog as a motivator to write more often, and to collect ideas for a book that I am very slowly writing. But I feel a bit obliged to have something of substance, since so many intelligent people are spending time looking at these pages.

Since I am swamped finishing up our quarter, and preparing for a family vacation to Europe, I'll pull something off my hard drive, which is pretty good. Here is a little set up: Only the most irrational and fundamental creationists object to natural selection, or "microevolution". The real barrier to the acceptance of evolution is recognizing global common descent ("macroevolution"), and especially the inference that humans are not specially created, but a twig in the tree of life. At the same time, natural selection is often directly equated with Darwinism (see this post for example) and evolution itself. I'm very slowly working on a book that explores the implications of common descent (pattern) and the processes that produce it (speciation, gene duplication, and "duplication" of biological units in between genes and species, like gene networks and traits). This post is an excerpt from a draft of one chapter. I've edited it down a bit, but it's perhaps still a little long and technical for a blog post - but pharyngulites are a smart bunch, so I have faith. For what it's worth:



Coming to grips with common descent

One of the most profound insights in the history of human thought is Darwin’s proposal that all species are united by common descent. The idea transformed the view of humankind’s place in the natural world from a self-congratulatory perch atop the Scalae Naturae to an arbitrary outpost alongside other apes as cousins to every palm, gastrotrich, and canary (note the goal here was to rhyme names of organisms with "Tom Dick and Harry", the colloquialism meaning "everyone"; not sure if anyone ever gets that though). We are still coming to grips with the idea that each and every living thing is related through a common evolutionary heritage. Despite being accepted for about a century and a half, I argue that a truly intuitive grasp of common descent remains difficult. Nevertheless, coming to grips with common descent is of fundamental importance for achieving the central goal of evolutionary biology: to understand the origins of biological diversity. Most notably, I argue that a conspicuously missing topic in evolutionary theory is the explicit comparison of hierarchical patterns and processes of lineage splitting and common descent.

[For brevity, I deleted a part describing Mayr's "horizontal processes" - like speciation; and "vertical processes", like natural selection. These are equivalent to cladogenesis and anagenesis. This horizontal is not to be confused with horizontal gene transfer. I first point out that traits, like teeth vertebrae and eyes, can duplicate within a species by mutation. Therefore, these traits can form phylogenetic trees like gene families. This is the subject of the book, so it's only briefly introduced here.]

1. Common descent is non-intuitive

The most formidable obstacle to a fully intuitive grasp of common descent may be the sheer magnitude of differences that life outwardly expresses. An oak tree and a dung beetle are cousins!? Not to mention our more obscure relatives, the oxymonads, slime molds, and E. coli’s of the world! How can it be that such enormous differences in appearance, life style, size, and behavior have evolved? The answer is not difficult to understand logically. For example, to Darwin the immensity of biotic diversity could in large part be explained by time itself. Geological time is so vast as to be practically incomprehensible. Who then could logically argue with time’s ability to allow for the origin of an equally incomprehensible biological variety – even if starting from a single point? Many of the tell tale similarities expected of cousins would logically erode away after so many millions of generations. But logic quite often does not guide intuition. Such is the case with common descent. Even after learning crystal clear indicators, we continue to be surprised by common descent.

Perhaps the clearest signal of globally shared evolutionary heritage is the genetic code. On October 12, 1962, the headlines of the New York Times read “The Genetic Code is Held Universal: All Living Things Are Said to Use Same Chemicals for Heredity Transfers”[1]. More than a full century after Darwin’s “Origin”, common descent was again in the public eye. Who could help but marvel at the thought of a universal genetic code, in place nearly since the dawn of life? A code that remained constant through the origins of cells, of gastrulation, of brains and of societies. The few exceptions now known to the universal code have not diminished the significance. The conserved hereditary machinery provides a concrete thread linking every organism in the tree of life.

One might expect that the triumphant generality of the genetic code would have precluded any more surprises relating to common descent. After discovering a “universal code”, perhaps we finally could come to grips with the universal relatedness of living things. But more surprises were in store. The advent of molecular developmental genetics would again cause biologists to re-examine their expectations regarding common descent. This time the papers read, “From worms to cows, one class of genes spells out the blueprint”. “Class of genes” of course refers to the Hox genes, the conserved transcription factors involved in body patterning. Discoveries of other genes with highly conserved roles were to follow. Of the gene Pax-6, the newspapers read, “science outdoes fiction” when a human gene was shown to step in seamlessly for a defective fly gene or to produce eyes in unnatural places.

The lesson in this tripartite sketch of the history of biology is that at three levels of organization – species, genes, and developmental processes – biologists have been surprised to discover common descent. To me, this indicates a rather fundamental lack of intuition for the process. We seem to have to relearn the lessons of common descent at every level of biological organization. In each case, we start with the assumption – or perhaps with the notion (something less formal than an assumption) – that biological entities somehow arise de novo in a massively parallel manner. In Darwin’s time very few biologists considered common descent of species (exceptions include Buffon and Erasmus Darwin), and no one else neared the sweeping generalizations for the relatedness of most or all of life that Charles Darwin suggested. Instead, Darwin’s contemporaries and predecessors believed that each species had a separate origin. Some biologists broke the strictly typological thinking common of the time – to Lamark species transformed and to Cuvier they sometimes went extinct – but no one else had formed anything near a Darwinian concept of a global tree of life. Perhaps it is the strong intellectual roots in creationism, or perhaps it is the conceptual difficulty in linking together the entire enormity of biodiversity, but even long after Darwin’s ideas of common descent have been accepted for species, a biologist’s intuition often falls back on the notion of parallel origins when considering other levels of biological organization.

Current support among biologists for common descent of genes (see also this post) is strong– but this was not always so. The 1960’s and early 1970’s was the time when biologists were learning that common descent holds for genes and proteins as well as species. One telling quote, used recently by both Gould {2002} and Carroll {1999}, can be found in an influential book by Ernst Mayr, Animal Species and Evolution.

"Much that has been learned about gene physiology
makes it evident that the search for homologous
genes is quite futile except in very close relatives. If
there is only one efficient solution for a certain
functional demand, very different gene complexes
will come up with the same solution, no matter how
different the pathway by which it is achieved.”

Ernst Mayr (1963)

I do not think that this was a particularly radical viewpoint of the time. For example, I was struck when reading Walter Fitch’s citation classic defining the terms orthology (descent by speciation) and paralogy (descent by duplication), because almost the entire paper is concerned with a topic rather different from orthology vs. paralogy. Instead, Fitch lays out statistical methods for discriminating homologous from analogous proteins. In other words, a real concern of the time was that proteins might have evolved similar sequences in parallel. This is of course a worthy consideration and is a hypothesis that is tested thousands of times per day using similarity searches of genetic databases. Nevertheless, I find comparison with biologists of Darwin’s time extremely interesting. In both cases, massively parallel and de novo evolution was taken almost as the null expectation, no matter how unparsimonious. To Darwin’s contemporaries, species were created separately and similarly. To Fitch’s biochemist and especially to Neo-Darwinist contemporaries of the 1960’s (e.g. Mayr above), the very real possibility remained that genes were “created” separately and similarly in different species. I don’t mean to imply religious-type creationist leanings in the biologists of the 1960’s. On the contrary, their “creator” was strong (Darwinian) natural selection for a particular function. They imagined natural selection to be so powerful and dominant as to allow multiple re-inventions of genes, thereby erasing most of the historical component of genes. No matter the creative force behind the purported independent origins of species or genes, the point here is that deeply rooted common descent was a hard earned realization, not an immediately obvious intuition of biologists studying the problem.

In many ways, it was history repeating in the 1980’s and 1990’s when biologists began discovering deep homologies in body patterning genes like Hox genes. The prevailing view and intuition of the time, especially of morphologists and systematists, was that many organismal structures like segments, eyes, limbs, and hearts, evolved essentially de novo, multiple times independently in various lineages. Like the pre-Darwinian concept of species or the concept of genes in the days preceding the advent of molecular biology, the dogma was that organismal structures and the developmental processes leading to them often arose separately in different lineages – especially when comparing different phyla, which are considered to have different “body plans”. But the discovery of conserved developmental genetic processes for patterning the bodies of taxonomically different organisms forced biologists again to consider common descent at new levels of biological organization.

Eye evolution provides a canonical example. A comprehensive, and well-cited paper on photoreceptors by Salvini-Plawen and Mayr from 1977 surveyed the morphological types of eyes in all animals, concluding that photoreceptors must have evolved essentially de novo 40-65 times independently. But the 1990’s saw a chain of discoveries that showed many genes involved in eye development are homologous between phyla. That visual pigment genes (opsins) are conserved was known for some time. Instead, the watershed discovery in eye evolution was that similar mutant (disease) eye phenotypes in flies, mice and humans are caused by mutations in homologous genes {Quiring, 1994}. These genes are named Pax-6 in vertebrates and eyeless in flies. Multiple other phyla were subsequently shown to utilize homologous Pax-6 proteins during in eye development. Furthermore, multiple other genes besides Pax-6 and opsin were subsequently shown to have a conserved role in eye development across phyla. Similar stories can be told for limbs, hearts, and segments. Naturally, many biologists began to question the generality of the assumption that organs and especially developmental processes evolved multiple times independently. Just as biologists recognized deep common descent of species in the 1860’s and genes in the 1960’s, they began to recognize a real possibility for deep homology at levels of biological organization in between genes and species. These conserved developmental processes and the genes that code for them have been termed “the genetic toolkit for development”.

There are many possible reasons why common descent is hard to intuit, deep-seated intuitions about ladders of progress, and an overemphasis on natural selection seem to be especially important. Yet once we start thinking about genes, traits, and species in terms of common descent, we can be greatly enlightened about how evolution has produced the enormous complexity that we see every day.

---

[1] The Genetic Code is Held Universal. Robert K. Plumb New York Times Octocter 12, 1962.

"Lineage"

It is interesting that evolutionists quite often use the term "lineage" to refer to a clade. I feel I've seen this in many contexts. Gene families may be called a lineage. A clade of species may also be called a lineage.

What is linear about these groups of biological entities? I think almost nothing. A clade of genes or species contains a common ancestor. A line can be drawn from that ancestor to any of the terminals. However, the decision of which line to draw is arbitrary, unless a single gene or species is all that is of interest in the clade.

Clades are not lineages, they are clades. Clades are branching diagrams without any inherent linearity. I think referring to clades as lineages provides yet another example of how linear thinking is embedded in the way people see the world. "Tree thinking" is a skill central to understanding biology. But to achieve this skill requires people to overcome innumerable difficulties. Some, such as referring to clades as lineages, seem subtle, but nevertheless probably have an effect on our ability to see biology as a collection of branching processes.