Posts Tagged ‘evo-devo’
Two evolutionary pathways meet phylogenetics
In a previous post and elsewhere (e.g. here Biological Theory 2(1) 10-22 gated and here Spontaneous Generations ungated ) I suggested extending Van Valen’s definition of evolution to incorporate development and ecology and to acknowledge the fact that there are two distinct pathways to evolution by natural selection. New genes can reconstruct old environments or new environments can reinduce the expression of old genes or both, but even if both, one or the other is likely to lead and the other to follow initially. In an article this month (Trends in Ecology and Evolution V26#3 here), Schwander and Leimar call the first “genes as leaders” and the second “genes as followers”. P. Z. Myers once exquisitely analysed an experimental case on Pharyngula here (a case originally published by H. F. Nijhout in Science).
It is important that the possibility of “genes as followers” not be misunderstood as sneaking Lamarckian explanations for adaptation in by the back, or should I say by a side door! Environmental influences on phenotypes beyond the historically experienced range, just as novel genes expressing a norm of reaction beyond that historically expressed, are more likely to be maladaptive than adaptive – which is not to say that either cannot have a selective effect – albeit one usually negative, at least at first.
Now to the new development. Schwander and Leimar make the novel (but in retrospect obvious) proposal that which direction change has taken place in any particular case can be detected using phylogenetic methods. Recall that Darwin said that his theory of descent with modification was composed of two great principles – “the unity of types” (i.e. history) and the “conditions of existence” (i.e. natural selection). Since the work of Harvey and Pagel in the late 1980s and early 1990s, it has become common in evolutionary ecology, and even in some cases in anthropology, to control for the former in testing particular hypotheses about the latter. However such methods have never before to my knowledge been applied to this question.
Schwander and Leimar apply “ancestral state reconstruction methods” to the question of how commonly switches (and losses) have taken place between genetic polymorphisms and polyphenisms. They admit to many limitations of their study. They consider only discrete and not continuously varying characteristics. They use selected examples, i.e. the evidence is anecdotal. Nevertheless, by mapping genetic polymorphisms and polyphenisms (as well as losses) for the same phenotypic characteristics in the same larger group such as winged and winglessness in carabid beetle species and right and left handedness in Heteranthera plant species for example onto their phylogenies, they have begun to pave the way towards answering this fascinating question. Bravo!
From the series of examples considered, they conclude that there is “no clear tendency for genes to be followers or leaders overall”. However it is important to understand that an historical change to, does not necessarily mean a change by means of . An evolutionary change from a genetically polymorphic species to a polyphenic one could nevertheless have been initiated by a genetic mutation or recombination followed by selection for adaptive phenotypic plasticity under conditions of uncertainty but with reliable cues. However, given that virtually the identical alternative phenotype was present as a genetic alternative in the ancestral polymorphic species, it is more likely that the alternative phenotype was latent in the formation of the new species and came to be induced there at least initially by an altered environment. In other words, it was likely a change initiated by environmentally inductive means as well as a change to environmentally inductive control. Still, as the authors eventually make clear, “the frequency and direction of transitions between them depends not only on how often either system emerges but also on how often one system is more beneficial than the other. Polyphenism should be favoured when a phenotype-determining environmental cue accurately predicts the selective condition for which the corresponding morph is suited, whereas genetic polymorphism should be favoured when such environmental cues are lacking and there is local frequency dependence or spatial variation in conditions, combined with limited gene flow.”
Leigh Van Valen’s definition of evolution extended
Bill Wimsatt mentioned to me lately in an e-mail that Leigh Van Valen had died in the fall which I hadn’t known. By coincidence, in sorting through some old papers the next day with the aim of disposing of some of them, I came across a copy of Van Valen’s paper on “Three paradigms of evolution” from his journal (Evolutionary Theory 9:1-17, July 1989). I knew immediately why I had held onto this paper for so long – because it contained a mention of his previously published definition that “evolution is the control of development by ecology” which had struck me forcefully at the time and which I had always remembered.
Much later when Van Valen’s dissatisfaction with the definition of evolution as “change in gene frequencies in a population” had become much more widespread and in thinking about incorporating ecology and development in a definition (as well as taking into account standing or cryptic genetic variation, niche construction, and epigenetic inheritance), I came up with this extension of his definition:
“Microevolution by natural selection is any change in the inductive control of development (whether morphological, physiological or behavioural) by ecology and/or in the construction of the latter by the former which alters the relative frequencies of (genetic or other) hereditary elements in a population beyond those expected of randomly chosen variants.”
Before I published it in a couple of articles, Leigh was kind enough to tell me by e-mail that “I agree generally although not in detail.” I wish now I had quizzed him more and thereby perhaps have learned something else! The following is the rationale I eventually provided for the definition ( in Darwinian Sociocultural Evolution, 2010, 166-8 minus references).
“7.2 Two Pathways to Natural Selection – Preadaptations and Niche Construction
Natural selection is widely thought of as a sieve, filter or sorting device. For example, Dawkins writes “Each generation is a filter, a sieve: good genes tend to fall though the sieve into the next generation; bad genes tend to end up in bodies that die young or without reproducing” (1995:3). The origin of this metaphor is unknown to me but it is obviously intimately related to the genetical theory of evolution which defines evolution as “a change in gene frequencies in a population” (see any text in population genetics). It has been widely observed that this definition includes genetics and evolution but omits development and ecology. Once these latter are also brought into the picture, it is obvious that the “sieve” metaphor is so simplified that it is positively misleading (what follows in this section is adapted from Blute 2007, 2008a).
Natural selection never acts solely “backwards” as a sieve, filter or sorting device. Instead it always acts sooner or later inductively in a “forward” direction, altering the development of individuals. There are two fundamentally distinguishable pathways. First, an ecological change can induce some individual(s) of a pre-existing hereditary background to develop differently whether morphologically, physiologically or behaviorally relative to others. This is possible because phenotypes are plastic (Pigliucci 2001). This in turn can change relative fitnesses, and hence ultimately the frequencies of genetic or other hereditary elements in a population. If a new food source becomes available and is made good use of by some which are heritably different than others, the former are not just chosen; but changed. They may be induced by their altered nutritional status to grow bigger, live longer or produce more offspring for example. Similarly, a new antagonist like a parasite or predator does not bloodlessly choose; it too changes. In this pathway, the sequence is ‘eco-devo-evo-geno’. An ecological change induces a developmental change, which causes an evolutionary change (by changing the relative fitnesses of organisms – it is organisms not genes which survive and/or produce offspring), which causes a genetic change (a change in the frequencies of genetic or other hereditary elements in the population). This pathway (minus the ecological and developmental content) was traditionally called a “preadaptation” in evolutionary theory and thought to be relatively uncommon although more recently under rubrics such as “exaptation” and “co-optation” its probable commonness has been emphasized more.
The second pathway, instead of beginning with an ecological change against a pre-existing hereditary background, begins with a hereditary change such as a new genetic mutation or recombination against a pre-existing ecological background. A hereditary change leads some individual(s) to develop morphologically, physiologically or behaviourally in such a way relative to others that they perceive, define or construct a pre-existing feature of the ecological environment differently, changing it, thus changing themselves, thus changing relative fitnesses, and hence ultimately the frequencies of genetic or other hereditary elements. If a new hereditary element becomes available enabling its carriers to consume and make good use of a previously unutilized resource, the ecological environment is changed. That change in turn again does not simply sieve, filter, or sort but inductively alters the affected organism(s) improving their nutritional status, resulting ultimately in a change in gene frequencies. The sequence here then is ‘geno-eco-devo-evo-geno’ i.e. a genetic or other hereditary change is then followed by the same eco-devo-evo-geno change sequence as previously. This second pathway implies that niche construction, (Blute 1995, Odling-Smee et. al. 1996, 2003 but in substance see also Hansell 1984, 2000) is not a once-in-awhile phenomenon. Instead, it is the pathway through which all evolutionary change initiated genetically is achieved.
As a consequence of the need to incorporate ecology and development along with evolution and genetics and because of the existence of these two distinguishable “inductive” and “constructive” pathways, Blute (2007, 2008a) suggested the traditional definition of evolution by natural selection as a change in gene frequencies in a population be replaced by building on Van Valen’s (1973) observation that “evolution is the control of development by ecology”. This yields the following definition:
Microevolution by natural selection is any change in the inductive control of development (whether morphological, physiological or behavioural) by ecology and/or in the construction of the latter by the former which alters the relative frequencies of (genetic or other) hereditary elements in a population beyond those expected of randomly chosen variants.”
The evo-devo hour glass goes molecular but what explains it?
In the nineteenth century the great German embryologist Karl Ernst von Baer proposed laws of embryological development – that general characteristics and structural relations develop before special ones; the form of an embryo does not converge on that of others but diverges from them; and that the embryo of a animal never resembles the adult of another animal but only its embryo. In modern language, embryological development is a process of differentiation in the senses that parts of an embryo become differentiated from each other and that members of different but related groups become more different from each other. Although Von Baer was not an evolutionist, Darwin thought that the early similarity of members of different groups was the best evidence for his theory of common descent. As he wrote in the Origin, “Community of embryonic structure reveals community of descent.” (Von Baer and the reception of his laws are discussed by Scott Gilbert here.)
While evolutionary changes in development themselves are presumably attributable to selection and/or drift, this pattern of change in development – less change earlier, more later – is most readily explained by constraints. A genetic mutation or recombination which affected the earliest stage of development would have more side effects down the road than would one which first acted later, and many of these would likely be maladaptive. Analogous phenomena are found in other selection processes. For example, a rat learning a maze with a series of choice points reinforced at the end eliminates mistakes in a backwards direction i.e. change takes place more readily in the later than in the earlier stages of the entire sequence. The philosopher of biology, William C. Wimsatt, dubbed the constraint principle in the evolution of development “generative entrenchment”.
In the 1990’s however, it turned out that the empirical generalization ‘more evolutionary change has taken place later than earlier in development’ does not quite hold – rather the difference between members of different but related groups resembles an hourglass (discussed for example by Raff in The Shape of Life). Groups differ more in the very earliest phase of development, converge to become more similar (the hour glass narrows to what is called the phylotypic stage in animals), and then diverge again quite a bit for the bulk of the rest of development. Kalinka et. al. writing in Nature (gated) have recently confirmed this hourglass pattern for gene transcription involved in key developmental processes by comparing six species of Drosophila.
I argued (in Darwinian Sociocultural Evolution pp. 146-8) that the constraints logic implies that the in-between phylotypic stage of minimal change/differences between groups must represent the actual historical origin of the taxa involved. This interpretation has now been borne out by Domazet-Lošo et. al. writing in the same issue of Nature. Using “phylostratigraphic” methods they had previously pioneered, they found from the transcriptome (all RNA molecules present), that the genes expressed in Zebra fish in the equivalent of the animal phylotypic stage are indeed older than those from other, including the earliest phase. They also confirmed this for some other groups using data from the literature.
But another question remains, again if the constraints logic is valid, how has so much evolutionary change in the earliest phase of development been possible? I argued that it must be because what appears to be the earliest phase of development is in actuality a later phase. Such would be the case if it were largely a maternal effect i.e. a later phase of the mother’s development. Intriguingly Domazet-Lošo et. al. did find significant differences between the age of genes expressed in the late juvenile and adult phases of males and females in Zebra fish with more newer in females than in males. The authors suggest this may be related to recent sexual selection (although it should be noted that Zebra fish are not notably sexually dimorphic and there is little evidence that female choice is more significant than male-male competition among them – if anything, the reverse may be the case). However, the sex difference is not really a test of the maternal effect hypothesis anyway because on my understanding (and I am not a molecular biologist) while maternally inherited RNAs would show up in the transcriptome, what would not show up would be if they were maternally inherited, or transcribed from maternally imprinted genes, or expressed as a consequence of maternally inherited protein transcription factors for example.
In short, I eagerly await whether the maternal effects hypothesis of ‘too much’ change early in development can and will be tested. If confirmed, Von Baer’s law would in a sense be restored.
Blute Blog 