Wednesday, November 16, 2016

Drafting Sisters or Public Goods - how tree-like is evolution?

I am drafting a paper exploring how tree like evolution is. A prime focus will be on "cell type trees" and "organ trees" - phylogenetic trees of those entities using gene expression data.

I want to set the stage for that discussion in the context of similar questions for gene trees and species trees.

Here is a draft of 8 paragraphs starting in that direction. Note, these are not yet referenced fully, and are a hastily written draft. Any feedback, comments, omissions, disagreements, etc are most welcome.....

Sisters or Public Goods? How tree-like is the evolution of genes, modules, cell types, organs, or species?

The metaphor of a tree of life occupies a central place in our understanding of evolution, but how often do features evolve strictly by bifurcation? Is there an alternative metaphor to the tree of life? At the level of species, horizontal transfer, hybridization, and incomplete lineage sorting often interrupt strict bifurcation (or “treeness” (Cavalli-Sforza and Piazza 1975)), causing incongruence between the history of genes and the species that contain them. Therefore, the history of all species cannot accurately be visualized as a single bifurcating tree. Instead, that history is a network. What about other levels of biological organization? Protein domains, genes, modules, cell types, and organs may also evolve by furcation (Oakley et al. 2007), and if so, their history could be visualized as phylogenetic trees. However, at each of these levels, processes analogous to horizontal transfer, including domain shuffling and co-option, also interrupt strict treeness. Therefore, those histories are also often networks rather than trees. An alternative to strict tree-thinking may be a public goods metaphor (McInerney et al. 2011), borrowed from economics, where biological entities are ‘non-excludable’. In the context of species phylogeny, ‘non-excludable’ means that the the parts of species (e.g. genetic material) can be transferred horizontally from species to distantly related species. What, if any, are the implications for our understanding of evolution if we adopt a public goods metaphor instead of the tree of life?

Thinking about evolutionary history at multiple levels
If evolutionary history is treelike, or if we can determine subsets of life’s history that are treelike, we can use the statistical machinery of phylogenetics, developed over decades, to analyze the rising deluge of RNA-seq data and address questions of homology, convergent evolution, cell-type evolution, and more. If evolution is usually not treelike, we may need a fundamental shift in how we analyze comparative data sets. Before exploring whether evolution is treelike, I use this section to explain some background, introducing how we might think about evolutionary history at levels of organization that include protein domains, cell-types, organs, and morphological characters.
All of life shares common descent and biologists use tree thinking (Plachetzki and Oakley 2007) to organize patterns of common descent. Patterns of common descent result from mechanisms that split lineages. I’ve previously termed lineage splitting at any level “furcation” (Oakley et al. 2007). The mechanisms that split species and genes are speciation and gene duplication, respectively. In addition, mechanisms may split lineages at other levels. Parts of genes, like protein domains furcate through “exon shuffling”. Developmental fields furcate by field splitting (Friedrich 2006; Oakley et al. 2007; Buschbeck and Friedrich 2008; Oakley and Rivera 2008) and cell types furcate to become sister cells, perhaps often by subdividing functions of the ancestral cell (Arendt 2008; Arendt et al. 2009). If new biological entities arise strictly by furcation, then a bifurcating tree, the tree of life metaphor, is the result. Biologists are used to thinking about bifurcating gene trees and species trees. In addition, some scientists have begun to explore the idea of phylogenetic trees of protein networks (Plachetzki and Oakley 2007), morphological features, tissues, and organs (Geeta 2003; Oakley et al. 2007; Buschbeck and Friedrich 2008), and cell types (Arendt 2008).
The tree of life metaphor makes a strict assumption that during lineage splitting, all components are inherited vertically, from direct ancestor to descendant, and are never passed horizontally, from evolutionary cousin to cousin. We know this assumption is often violated through mechanisms that vary by level of organization. The components of genes (protein domains) are often exchanged between distantly related genes (Haggerty et al. 2013) by duplicating domains independently of full genes. Therefore, a gene tree may not be strictly bifurcating, forming a network. Multiple mechanisms can cause incongruence of species tree and gene trees, such that genes are sometimes exchanged horizontally. Horizontal transfer occurs routinely through various copying mechanisms in prokaryotes, but is also important in animals (refs). Hybridization and introgression are being recognized as a common mechanism (Hahn and Nakhleh 2016; Pease et al. 2016) that we might call horizontal transfer, even though close relatives are involved. In cell type and tissue phylogenies, the components also may not be inherited from direct ancestors, as in the sister-cell model, but rather could be co-opted from distant cell-type or tissue-cousins. When the assumption of vertical descent holds, there is great potential to use existing phylogenetic methods to understand evolution. However, given that strict bifurcation is commonly violated, we may want to explore other models and metaphors for macroevolution. One of those metaphors is a public goods model.

Economic classification of goods
In economics, “goods” may be classified into a 2x2 matrix, forming four categories. In practice, the categories are usually not discrete alternatives, but instead form axes with continuous variation. One axis asks how “excludable” and the other axis asks how “rival” is a particular good. Non-excludable goods can be ubiquitously accessed, whereas excludable goods have restricted access. For example, the air we breath has low excludability because we cannot prevent anyone from breathing the atmosphere. However, a theme park is quite excludable; only those with a ticket may enter. On the other axis, highly rivalrous goods are exhaustible, but non-rival goods are unlimited. A parking space is rivalrous because if I park there, no one else can. In contrast, a public radio signal is non-rivalrous. My listening to 92.9 KJEE does not prevent someone else from also listening. “Public goods” are defined as both nonrivalrous and nonexcludable. The extreme corners of these two axes form four named categories and there are many everyday examples for each category (Table 1).

Table 1. The classification of goods

Excludable (high)
Excludable (low)
Rivalrous (high)
Private Goods
Car, apple, parking spot
Common Pool Goods
Fish stock, public park bench
Rivalrous (low)
Club Goods
Private golf course, satellite TV
Public Goods
Air, public radio signal,

Public Goods in Macroevolution
Recent papers applied the concept of public goods to macroevolutionary topics like the tree of life and novelties. McInerney et al (2011) considered genetic material (“genes” for short) to be a public good and considered biological species to be the consumers in this economic metaphor. McInerney et al (2011) contrast their public good hypothesis of genes with the traditional idea of a universally bifurcating ‘tree of life’, with vertical transmission from ancestor to descendant species. With only vertical transmission, genes are highly excludable between species because they can only be present in a genome if inherited from a direct ancestor. Genes also have low rivalrousness because no matter how many descendents evolve, they all can have those genes in their genome. Non-rivalrous, excludable goods of the tree of life model are “club goods”. Instead, McInerney et al (2011) argue that genes should be considered “public goods” because horizontal transfer is very common. Horizontal transfer makes genes much less excludable because they could be transferred from any clade to any other clade.
Erwin (2015) applied the concept of public goods to major innovations during evolution. He extends public goods thinking in macroevolution beyond genes to include environmental and ecological goods. In particular, he suggests that the origins of particular public goods were associated with major transitions in evolution. A prime example is the production of oxygen, first as a waste product of photosynthetic cyanobacteria. Once oxygen accumulated in the Great Oxygenation Event, it became both non-excludable and non-rivalrous, a public good that was exploited by other organisms.
The macroevolutionary applications above differ from public goods applications to microevolution, especially by ignoring costs. Frank (2010) defined a public good as “An individually costly act that benefits all group members”. The conflict between cost and benefit is central to microevolutionary topics like altruism or cooperation, kin and group selection (Hamilton 1975), parasite virulence (Frank 1996), ‘tragedy of the commons’ scenarios, and cases where microbes produce public goods like nitrogen (West et al. 2002) or iron-scavenging molecules (Kümmerli et al. 2009). One link of these microevolutionary topics to macroevolution posits that cooperation between units led to changes in the level of selection that precipitated major transitions in evolution (Smith and Szathmary 1997).  Unlike the microevolutionary applications, where a cost is incurred by individuals producing the good, the macroevolutionary applications discussed above are not concerned with costs to the producers. The production of oxygen is presumably cost-free waste for the photosynthetic organisms that produce it and only a benefit to some organisms that do not produce it, like animals. Oxygen production does not then set the stage for conflict between producers and benefactors that are so central to the microevolutionary topics listed above. Similarly, if genes are public goods for all species through horizontal transfer, the costs to producers again are not apparent. There would seem to be no direct cost to a species if one its genes are copied into a distant relative. Perhaps because of this explicit absence of costs (and any quantification of benefits), very little research in macroevolution uses the mathematical framework of public goods that is so prevalent in microevolution. Instead, the macroevolutionary research simply points out that some critical elements of evolution are nonrivalrous and nonexcludable.

Measuring Excludability with Treeness

Wray G.A., Hahn M.W., Abouheif E., Balhoff J.P., Pizer M., Rockman M.V., Romano L.A. 2003. The evolution of transcriptional regulation in eukaryotes. Mol. Biol. Evol. 20:1377–1419.

No comments: