In the last few posts in this series, we introduced the concept that individual characteristics (such as individual gene sequences) may not always match the phylogeny, or species tree for a group of related organisms. Incomplete lineage sorting is one way for this to occur, but another is for similarities to arise through independent events. Such features would have the superficial appearance of being inherited from a common ancestor, but in fact would be examples of homoplasies (singular = homoplasy): features shared between species that were not inherited from a common ancestor.
Birds of a feather
One classic example of a homoplasy is powered flight in birds and (some) mammals (i.e. bats). The species tree for birds, bats and non-flying mammals (for example, mice) places all mammals together as more closely related to each other than any is to birds. In order to explain the shared feature of powered flight for bats and birds, then, one needs to model it as a homoplasy – as independent events arising on two separate lineages:
The alternate explanation – that powered flight is homologous between bats and birds (and thus present in their last common ancestor) – would require that all mammals except for bats have lost this ability (to say nothing of the reams of DNA sequence data that support the above species tree). Beyond this evidence, there is also good reason from comparative anatomy to think that powered flight arose independently in bats and birds. Birds use feathers attached along the length of their forelimbs to provide lift. In contrast, bats use a membrane to form their wings, and this membrane is attached between their digits as well as to their body:
Both solutions work well, but when we break down the larger trait of “powered flight” into its component parts, we see that though the trait as a whole is convergent, the underlying components are not. This observation further supports the conclusion that powered flight in birds and mammals arose separately.
Homoplasy vs. homology
We can illustrate an example of how a simple DNA sequence homoplasy arises using a phylogeny. Suppose three species have the following sequence for a portion of the same gene:
Based on these data alone, the simplest (most parsimonious) phylogeny would be as follows:
Based on these data, the ancestral sequence would be inferred to be “TCATCC”, and the branch of the phylogeny leading to Species A would have one mutation to explain the observed sequence difference. In the absence of other evidence, this phylogeny would be the best fit for the data.
This tidy picture, however, could be upset by more data – data that demonstrates that the simple species tree we have drawn above is in fact incorrect. If so, then we need to fit the above sequences into a different species tree – meaning that we will need to explain the pattern using more than one mutation event. Let’s work through a hypothetical example to show the process.
Let’s suppose that sequence data for several hundred additional genes are compared for these three species, as well as for a number of other related species not shown in our species tree. Let’s also suppose that these data strongly support a different species tree than the one we just generated – in the vast, vast majority of cases, the data supports a tree with Species A and B as closest relatives, with Species C as more distantly related. This would “force” us to redraw the species tree as follows, placing our original short sequences into a different pattern along with their species:
Let’s also suppose that the DNA sequence data for this particular gene sequence from the additional species not shown on the species tree indicate that the ancestral sequence actually had a “T” in the second position rather than a “C”:
Now we have to account for all three species in our species tree having a non-ancestral sequence at the second position, as well as try to make sense of the mutation events that led to the pattern we see here. Note that we are still constrained to make the most parsimonious explanation for the whole of the data, but for this particular gene, we are forced to invoke multiple mutation events to fit the pattern to the species tree. We make this choice, however, because it would be even more unlikely for multiple mutation events to have shaped the pattern of the hundreds of other gene sequences in a coordinated fashion – and those other sequences support this version of the species tree.
If you take the time to try “solving” the gene tree by adding mutation events to the species tree you’ll soon realize that at least three mutation events are needed to produce the observed pattern. There are also solutions that use more than three mutation events, but they are less likely explanations. One of the possible solutions is shown below:
In the branch of the phylogeny leading to Species A and Species B, a (T to G) mutation occurs prior to the A / B divergence (represented by the red bar). A second mutation then occurs on the lineage leading to Species B that changes the G at the same position to a C (represented with a blue bar). Independently, the lineage leading to Species C also has a mutation at this position, changing the ancestral T to a C (also represented with a blue bar). The end result is that two of the sequences (in Species B and C) have become identical – but neither inherited the “C” at the second position from their common ancestor. In other words, they have arrived at the same “destination” from different starting points, or “converged” on a common sequence. Not surprisingly, this phenomenon is known as convergent evolution. For these two species, the “C” at the second position is not homologous (a similarity inherited from a common ancestor), but rather a homoplasy – a similarity that resulted from independent events on two lineages.
Homoplasies can be as simple as single DNA monomer changes (as in this example), or as complex as the independent reorganization of multiple systems with numerous genes and body parts to converge on a solution (as in the case for powered flight in birds and bats). In both cases, however, we can determine that they arose as independent events on separate lineages because these features do not fit onto the species tree as unique events.
The power of convergence
Since homoplasies act as markers that flag repeated evolutionary events, looking for homoplasies in species trees is a useful way to test hypotheses about the reproducibility of evolution, or how often species converge on similar solutions. As it turns out, evolution is remarkably repeatable for many general traits. There are numerous examples of repeated, independent innovations over evolutionary history, some of which we will examine in more detail in upcoming posts:
- Streamlined body shape: the streamlined body form of aquatic life such as fish, ichthyosaurs, whales, seals and diving birds (e.g. penguins) are all independent, convergent adaptations to an aquatic lifestyle.
- Powered flight: in addition to birds and bats, powered flight also evolved independently in insects and pterodactyls.
- Echolocation: some mammals, such as bats and whales, have independently developed systems that allow them to locate food through detecting how sound that they generate echoes off structures and prey in their environment.
- Camera eyes: the repeated evolution of camera eyes (i.e. eyes that use a lens) is one of the most striking examples of convergent evolution. Camera eyes have independently evolved in cnidarians (certain jellyfish), cephalopods (such as squid and octopus) and vertebrates (birds, mammals).
One thing to note is that these widespread examples of convergence are all shaped by the physical environment of the organisms in question – the perception of light (eyes), the ability to fly through air (wings), or move efficiently through water (streamlined body). The fixed presence of these environmental features would be expected to shape the adaptation of many species.
Previously, we introduced the concept of a homoplasy – a similarity in form in two lineages that arises due to independent events. In the example we looked at last time, birds and bats independently obtained powered flight through convergent evolution – with bats arriving at membrane-based wings, and birds at feather-based wings. Since the last common ancestral population for bats and birds was a species that did not have powered flight, this is a good example of a homoplasy – one that arose through convergent evolution.
Underneath this convergent event, however, there is a deeper connection. Bats and birds are both tetrapods – organisms with backbones and four limbs. The tetrapod body plan was already a feature of their last common ancestral population, and has been maintained in both lineages. As such, when considered strictly as a forelimb, bat wings and bird wings are homologous structures. In birds and bats, forelimbs have been shaped through natural selection for flight in different ways, but the starting point for both was a homologous structure. In other words, underneath the convergent event of powered flight in bats and birds is a deeper homology – the limb upon which both lineages independently constructed a wing. To represent this on a phylogeny, we would place the tetrapod body plan prior to the divergence of all tetrapods, and powered flight as two events on the appropriate lineages:
This pattern – convergent events with deeper homologies lurking beneath them – is one that is seen time and again in evolution. In fact, these deeper homologies improve the odds that convergent events will occur, since they provide a common basis that separate lineages can use for independent innovation. For bats and birds, adaptations leading to flight were possible because both lineages had forelimbs that could be modified, over time, from one function to another. While this example is at the anatomical level, these sorts of “predispositions” and the convergent events that arise from them can be observed at the molecular level as well.
The eyes have it
As we mentioned in the previous post in this series, camera eyes are one of the most striking examples of convergent evolution, having appeared independently in several lineages (the most common examples of which are vertebrates, cephalopods such as octopus and squid, and certain jellyfish). Camera eyes have a light-sensitive cell layer (the retina) as well as a lens that focuses light on the retina. Explaining the distribution of camera eyes among these three groups requires us to invoke three convergent events on their phylogeny (“cnidarians” are the group in which jellyfish are found):
At first glance, it seems wildly improbable that three distantly-related lineages would independently converge on such a remarkable structure as a camera eye. As it turns out, however, a key homology between all three groups greatly improved those odds – the molecules that act as light sensors.
At its most basic form, sensing of the external environment requires that the environment induce a change within cells. Accordingly, sensing light requires a light-induced change of some kind. The key molecules that perform this function in all three of the above groups are proteins called opsins and their chemical partners (a group of compounds called retinals). Each opsin protein has a retinal attached to it, and together the opsin/retinal pair acts as a light sensor. Retinals change their shape when they interact with light (i.e. absorb a photon, represented by the gamma in the diagram below). This shape change in turn alters the shape of the opsin protein attached to the retinal:
The opsin/retinal system of detecting light is a very widespread system – in fact, all animals that can detect light use these molecules as the physical basis for doing so, whether they have camera eyes or other eye types (such as compound eyes, or merely patches of light-sensitive cells). This is strong evidence that the opsin / retinal system predates the divergence of the three groups we are considering:
With this knowledge in hand, we can see that the development of camera eyes in these lineages is not as improbable as we might have thought at first. In all three cases, these lineages built a camera eye around a preexisting molecular system for detecting light. The camera eyes themselves might be convergent, but they are based on a deeper underlying homology that improved the odds that they would appear through successive modifications of an ancestral system. And as we saw for bird and bat wings, there are differences between the camera eyes in these lineages that support the hypothesis that they are the result of convergent events (the most well-known example of which is that the vertebrate and cephalopod eyes have their nerve “wiring” in opposite orientations).
Hearing is believing
A second example of “molecular predisposition” leading to convergence can be seen in the molecular machinery underlying a different form of sensory perception – the ultrasonic hearing required for echolocation in bats and toothed whales. Both groups use highly tuned echolocation for navigation and seeking prey in an environment where visual perception is limited or lacking altogether. The evidence that the development of echolocation in these two very divergent groups of mammals is due to convergent evolution is strong – no other mammals more closely related to either group has such an ability.
The cellular / molecular basis for detecting sound in mammals is a set of cells in the ear that extend hair-like projections (called cilia) that vibrate in response to different wavelengths of sound. Cilia also change their length and vibratory properties in response to different auditory stimuli. The vibrations are used to change the flow of electrical charge in these cells, eventually leading to nervous system signals that the brain perceives as sound. All mammals use a protein called prestin as part of the auditory system. Prestin is a “motor protein” that can change cell shape by moving internal structures around – and mammals use it for modifying cilia in response to sound.
The cilia/prestin system is known to predate all mammals, so it is not surprising that toothed whales and bats use this system for hearing. What is interesting, however, is that in these groups the prestin protein has been independently shaped through natural selection to be tuned to high frequency (ultrasonic) sound more useful for echolocation. In fact, in a phylogeny restricted to prestin sequences, bat prestins and toothed whale prestins appear to be the most closely related to each other – a finding wildly at odds with the species tree for bats and whales. Further examination, however, shows that these striking similarities are the result of convergent evolution, not a more recent shared ancestry. In both cases, the prestin protein was available to become attuned to ultrasonic wavelengths, and similar (though not identical) mutation events in both lineages were selected for along the way – an additional example of a “deep homology” favoring independent convergent events.
Summing up: evolution as a non-random process
One common misconception I encounter about evolution is that it is predominantly a random process – one that is mainly influenced by chance events. While we have already shown that evolution has a strongly non-random component (natural selection), this discussion of convergent evolution further demonstrates that evolution is repeatable in certain important ways. When natural selection affects distantly-related groups in a similar fashion, we often observe similar outcomes. These similar outcomes are in many cases favored by prior history (homology) and arrived at through similar, but not identical paths (demonstrating that contingency and chance are present as well). Evolution is thus a balance of contingent events (mutations and other chance events) and emphatically non-contingent events (selection, convergent evolution).
In the next post in this series, we’ll return to bat echolocation to explore how evolution of one species can be greatly shaped by another species in close relationship with it – a phenomenon known as coevolution.