Allopatric and synpatric speciation are the pits

Suvrat Kher in a comment to the ongoing discussion of pits here, asked:

"Are you applying this equally to both sympatric and allopatric modes of speciation? My thinking is that the difference between your speciation model and random drift model will be more apparent for the sympatric situation where natural selection may cause a divergence within a population while drift may not.

In case of allopatry both natural selection and/ or drift may cause divergence."

Yes, you are right. The first example shown would be allopatric sympatric speciation, as the new population forms from within the parent population.

Sympatric Allopatric speciation would be different, as it includes the physically separating of a sub-population from the parent population. In my model, this separation would also occur in morphospace instantly, as the allele frequency of the parent population has changed.

While the allele frequency of a population is an average over the whole population, this frequency would not occur at every point within the population. At any given point some allele would be more frequent, others less frequent.

If a sub-population becomes isolated, it's allele frequency will not change much, but the allele frequency of the parent population will. This is because some allele have now been relatively concentrated in the parent population (those alleles occurring less in the sub population), compared with others that have been relatively diluted in the parent population (those occurring more in the sub-population).

Thus, in this model, isolation would be represented as a shift in the position of the parent population in morphospace, with the sub-population orbiting close by.

I'll try and get some diagrams up shortly.

New pits - now with diagrams!

I've been discussing pits here, here, and here, if you are interested, but so far without diagrams. There's a reason for that. I'm rubbish at diagrams. OK, I don’t have a decent graphics program, but that wouldn’t matter anyway. I'm still rubbish. That’s why I prefer my fossils in 2D! However, here is an attempt at a graphic representation of what I’m talking about.

The first diagram represents speciation via a concentration of alleles, adaptation, whatever.

A) This represents the current species as a morphospace topography. The shape doesn’t really matter, nor does the orientation, as in morphospace all three dimensions are equal, and there is no “up” or “down”. The shape is just to invoke some idea of 3D. If you want to get fancy, there may be gaps on close-up, but for now we are just giving an impression of the extent of the topography that defines this species. Basically the topography represents the extent of allele distribution that represents this species.

B) A beneficial allele appears at a location and its presence in the populations begins to increase preferentially. The circle represents the increasing occurrence of the allele.

C) Frame of view rotated to show what’s happening underneath. The allele increases in concentration at this location faster than it occurs elsewhere in the population, and so the topography warps in response, forming a pit. As the allele concentration increases, the local allele distribution departs from that of the surrounding ‘normal’ allele population and so the bottom of the pit continues to move away from the ‘normal’ topography. (This is pretty much the same as the Mt. Improbable version, but reversed.)

D) Speciation! Well incipient speciation anyway. The local allele distribution departs from the ‘normal' to the extent that further interaction with the original species ceases. The new population is now isolated and begins to change independently of the original species.

E) Speciation. The new daughter species is established. The new species clusters together with the old species in morphospace, since they share far more alleles that the new species shares with any other species out there. The topography of the new species is similar to the old species, which also to invokes a relationship. There is no connection between them, so no allele swapping, so a new species. The topography of the old species also changes slightly due to the change in allele frequency.

OK now lets see what happens what the daughter species starts the same speciation process.

A) A beneficial allele occurs and starts to become preferentially concentrated at one location in the population. The pit forms and continues to grow as the concentration of the beneficial allele continues to increase. This time the pit isn’t ‘downward’ because there is no ‘down’. The pit forms at 90 degrees to whatever the surface orientation is.

B) Incipient speciation as before, with the new population separated from the main population.

C) The new species is established, and again clusters close to the parent species as they share more alleles in common with each other than they share with any other species.

This group could also be classed as a Genus as we have a direct ancestor-descendant relationship. Also we can wind the tape back and show that relationship as the daughter species run back into original species.

This model is also internally consistent, because the topography can represent the origin of species, genera, families, even phyla. The same process, the same imagery. It's like a Mandelbrot set, with the same imagery showing species, families, or phyla, depending on the magnification.

For example, the first diagram could be used to represent the evolution of birds from dinosaurs (albeit very crudely - it'll need some work), with the original topography representing dinosaurs and the daughter "species" representing birds. In this instance, the next image in the sequence "F" would show the dinosaur topography shrinking, and eventually disappearing altogether as the as the dinosaurs moved towards, and finally became, extinct.

As a bonus, here's how genetic drift would look like.

A) The morphospace topography of the population.

B) A new allele occurs and starts to spread in the population.

C) Instead of being preferentially concentrated in one area as a beneficial allele would, the allele slowly increases by spreading out into the surrounding population. In other words it spreads out 'faster' than it concentrates in one location.

D) The new allele is fixed in the population, causing the topography to change somewhat as the new allele frequency is slightly different than the original one.

Johnny, I'll try and get around to your questions in the next couple of days.

Even more pits

OK this is the latest in an ongoing conversation I'm having with Johnny on the issue of pits, the first two installments can be found on his blog Johnny provided some comments on my previous pit post and I have brought them up to the front to answer them. Johnny's comments are blockquoted.

Since we agree that Mt. Improbable is a good analogy for its intended purpose, I’ll move on.

Um, hang on. I say that Mt Improbable is ok as far as it goes. The problem is that it doesn’t go very far! The analogy doesn’t cover all evolutionary processes, only adaptation. There is no climbing involved, and evolution is directionless, it doesn’t proceed upwards. So, apart from that, yeah, it’s OK.

Viewing the two blog post here, as well as the two comments you were kind enough to provide to the “rebuttals” at my site, I think that, broadly speaking, there are a couple of assumptions or implications being made that are fundamentally false – false, based on my understanding.

1. You describe Natural Selection as acquiring beneficial alleles through a “random process.” This is incorrect; NS is a non-random cumulative process.

The generation of alleles is random, the concentration of them may or may not be. The concentration of beneficial alleles is non random, but “non-beneficial” alleles can also be fixed.

2. You imply that genetic drift contributes to fitness when, in fact, drift is a random process that is entirely disconnected from fitness. (Big point)

Actually I didn’t mean to imply that GD contributes to fitness. It may do, if the fixed alleles subsequently provide the starting materials for beneficial alleles, but at the time, no, there is no increase in fitness.

3. Through descriptions of lateral jumps between peaks, distances between pits and by suggesting that new “groups” can arise by “breaking through new fitness landscapes,” you’re implying that a Genus, Class or Family can erupt from a Species. This is false. Species rise from “speciation” at the species level. There is no jumping between points – only steps - a continuous gradation.

I’m a bit confused here. Genera, Classes and Families, while artificial constructs, are formed initially from one species – the common ancestor of the group. The original population undergoes a concentration of alleles into a pit which sags If that concentration of alleles becomes intense enough, ‘bottom’ of the pit separates from the original population’s topography, producing a new species (if there is no link then there can be no allele transfer = genetic isolation and so a new species.) We now have two species, represented by the original topography, and the new topography, species 2, which is orbiting close by. Say the original species undergoes another speciation event to produce species 3. Now we have three species cluster closely together in morphospace, as they share more alleles in common with each other than they do with any other species.

This grouping of the common ancestral species and its descendant species would comprise a genus, but with the caveat that the boundaries of which species would be classed as in the genus or outside is an arbitrary one. Speciation still occurs over time, with no major jumps.

4. You imply that “Generalist” and “Specialist” are definitive forms by saying that one has survival value over the other. These are relative terms. You may view/classify a particular insectivorous mammal as a specialist because its diet is limited to only insects – I could counter by saying that it’s a generalist because it eats a wide variety of insects as oppose to limiting its consumption to termites. Without knowing the type and extent of a theoretical catastrophe – there is absolutely no way to predict what niches will remain intact or the rate to which any event survivors will rebound.

There is no doubt that specialists can be very successful, and you are right that there is no telling which niches will survive. But specialists with adaptations to a specific environment may well be at a disadvantage should the environment shift and close out the niches to which it has become adapted. Under those circumstances the species is carrying a load of alleles that are no longer useful. If those alleles produce adaptations that have a high production and maintenance cost, then they are disadvantaged compared with other species which, while not dominant in the previous environment, do not carry the extra load and so may be in a better position to survive.

On an individual species basis significant adaptation may well provide dominance in a certain niche, but that species is more likely to disappear if the environment changes compared with other species that are more general and hence can survive in a number of environments. Generalists generally do not dominate any niche, but exist in several. Specialists dominate in one niche.

5. You create a false dichotomy between peaks and pits; both are landscapes and as the term “landscape” suggests, both represent a varied ever changing state complete with valleys, ridges, prairies, bumps, holes – they are one in the same –metaphors for change.

Yes, but as I mentioned above the Mt Improbable analogy doesn’t cover all evolutionary processes, only adaptation. There is no climbing involved, and evolution is directionless, it doesn’t proceed upwards.

6. You imply that organisms adapt or imbed themselves into a static environment, when in fact static environments don’t exist. Ecology is a balancing act, therefore so is adaptation.

The environment is not static, and adaptation works to balance this, but become too specialised, and have the environment shift too much, and the species is left high and dry.

7. When comparing short-term success with long-term survival you seem to suggest that evolution is a fundamentally different process following some catastrophe or major ecological shift. This isn’t so; the same processes driving speciation, drift and natural selection today are the very same processes that will drive evolution following an asteroid impact. If adaptation and specialization are what is successful in the here-and-now, they are also going to be successful strategies following a catastrophe.

Actually, evolutionary processes are the same, it’s just that after a major ecological shift it is more like to be generalists that repopulate first and become the source of new species (yes, by adaptation).

8. Describing both Natural Selection and Drift in terms of allele frequency is fine; however keep in mind that NS works with phenotypes as well as genotypes; this builds the link to fitness. Drift does not have this link.

Yes Natural Selection works well, there’s no argument there. But NS isn’t the only process at work, and the Mt Improbable analogy only covers adaptation. A description that covers all evolutionary processes would be better.

9. The idea that a current level of adaptation exhibited by an organism somehow reflects its future potential or its available genetic plasticity is erroneous.

OK, what if we have two organisms. A with small feet adapted to dry ground and B with larger feet. A is dominant in the environment, but B is OK on dry ground, but can also access marshy ground. Now, suppose the environment changes to all marshy ground. Can we say nothing about the future potential of these two organisms?

However, I do admit, the visual image of a “sag” being created by the weight of an area with increased allele concentration is pretty catchy…

Yeah, one does tend to gravitate towards it . . .

More Improbable Pits

The reasoning behind all this is not to attack the Mt Improbable analogy – which is to show that you do not need large-scale jumps to produce adapted forms, but that they are produced through gradualistic processes.

And a good job it does. What I am looking at here, is whether we can formulate an analogy that can be expanded to encompass more of the evolution process, and hence a series of internally consistent descriptions. The Mt Improbable analogy was not formulated to encompass the depth and range of evolutionary processes, and it would be wrong to criticise it on that score.

Close scrutiny of the Mt. Improbable analogy shows that there may be better ways to represent adaptation that can be expanded to other representations of evolution in a consistent way.

Riding Mt Improbable
Adaptive or fitness peaks are a reflection of the current state of allele frequencies. In other words they reflect which allele frequencies are delivering advantages in a particular environment (advantage = greater reproductive success = a higher concentration of alleles). They do not represent the best possible frequency/fitness solution to a particular environment – such a solution would be impossible, or at least constantly changing - influenced as it would be by the environment, the starting point of allele frequencies, and new alleles produced by mutation (and also impossible to achieve as adaptive success would be 'better than everyone else', not the best possible - which is similar to the old joke of two hikers being chased by a bear. Hiker 1 stops to put on a pair of running shoes. Hiker 2 says, "They wont help you outrun the bear”. Hiker 1 says, "I don't need to, all I need to do is outrun you".)

Peaks represent actual frequency/fitness values measured from actual populations/species. Since this is the case, a population/species sits on top of a peak, because the peak is defined by it. In other words the peak represents a particular groups of alleles that produce an advantage and hence are reproductively favoured, and so are concentrated at that point. The greater the concentration here, as opposed to elsewhere in the population, the greater the peak.

Therefore a population always rests on the peak. As the alleles continue to provide an advantage, the concentration of those alleles increases and the peak increases in ‘height’ (or more accurately distance from the fitness plain). Imagine a lava lamp. Once warmed up, a central peak starts to form. As the heat increases, the peak ‘grows’ upward. This is what is happening in the fitness landscape. The population/species sits on the peak and the peak grows underneath it upwards away from the landscape, as the allele concentration increases.

This means that there is no climbing involved. We should be taking about Riding Mt Improbable, not climbing it!

More than adaptation
The next issue we have is how to describe speciation. With the Mt. Improbable analogy this would result in peaks sprouting from peaks like horns, as populations split. The analogy therefore starts to groan under the strain of trying to be consistent. Again I wish to emphasis that the analogy was not meant to do this, and so this is not a criticism, I am pointing out that maybe there is a more consistent analogy we can use.

The other problem is that the peaks are still attached to the fitness landscape, which implies that there is still a route that populations can take to follow in the footsteps (slime trail?) of the new species. In truth, this would not occur, as once the new species has been formed, all connection with the ancestral population are severed. So we would end up with isolated ‘peaks’ suspended above fitness plains, again stretching the Mt. Improbable analogy too far.

"My God, it's full of pits"
So can we establish an analogy that can be expanded to fit into a more encompassing explanation for evolution generally? I think we can.

Firstly we need to explain the fitness landscape. A possible better analogy is one where populations occupy discrete areas, or topographies, of morphospace. Morphospace itself can be considered as an essentially limitless three-dimensional space within which morphospace topographies describe populations.

These topographies can be any shape as there is no “up” or “down”. Populations are described by the current spread of alleles, and so only areas that correspond to current allele frequencies have a topography. No populations, no topography, just empty space. Populations can expand into empty space as allele frequencies shift, or contract to leaving empty space – if they contract far enough, they disappear = extinction. But there is no set landscape that populations occur in. Populations define the extent of the topography.

OK, a population defines a topography that can be any shape. As the alleles in the population shift, the topography shifts. In this analogy, alleles that confer an advantage become concentrated within a sector of the population. This causes the topography to sag. As the advantage continues and resulted in reproductive success, the allele concentration increases, increasing the depth of the pit gradually (though not necessarily constantly). It should be noted that, since the topography could be any shape and orientation, the direction of the pit could be horizontal, vertical or anything in between, as the sagging will be at 90 degrees from the surface topography (which could be at any angle). This does away with the implication (unintended as it is) from other analogies that evolution is directional and upwards.

The pit therefore describes the state of the population in terms of allele frequency, with the bottom of the pit representing the highest concentration of the advantageous alleles and the sides representing decreasing concentrations of the allele. So if people ask how did the population get to be in a pit so deep in one go, to answer, hand them a shovel.

Speciation occurs when the pit separates from the rest of the topography. This does two things. It forms a satellite topography that represents an independent population – a new species – which is free to form it’s own topography, and start the formation of new allele concentrations and eventually new species.

It also causes the topography of the old population to retract away from the new topography as the allele frequencies realign back towards the old population frequency since there has been a significant removal of alleles with the new species.

A word here on genetic drift. The Mt Improbable analogy doesn't cover drift. But in my anaolgy, drift would be represented by a broad shallow pit as the allele becomes incorporated into the population faster that it is at the point of initiation so the pit expands outward through the topography rather than into a pit. Drift is then a ripple in the topography that, once fixed, leaves the topography slightly lower/higher that it was prior to the fixation event.

Separate topographies = no allele transfer = species. It is possible that the old population has retained enough of the advantageous alleles that the separate populations grow back together. But that would need to happen very quickly, before the frequencies become to divergent (as in populations that are geographically isolated but can still share alleles if brought together).

So a cluster of interconnected topographies, and some closely aligned but separate topographies, denote a species. This is the highest magnification. Ratchet the magnification back a notch, and clusters of closely positioned topographies become Genera. Ratchet back again and clusters of genera topographies become Families, etc. The closeness of topographies between separate species, genera , etc can make the decision as to which topographies go where, difficult - as in real life.

Ratchet the magnification back far enough and we can see the all of current life on Earth represented.

But, it would show relationships as they are today, with groups occupying discrete areas of morphospace, separated from each other. It cannot show the connectedness of life because we are viewing it as it is today, after 3.5 billion years of evolution. To show the connections, the evolutionary relationships, we need the 4th dimension – time.

What we can do is run a time sequence backwards, that would show morphological topographies coalescing, firstly species recombining backwards into founder species, then genera, etc, until we see the major groups contracting back together, metazoan topographies coalescing back into single celled topographies, eukaryote topographies coalescing back into prokaryote topographies etc.

Running the sequence forwards we would see the reverse. For example, it could show the origin of the dinosaur topographies from other reptile topographies. Such topographies would increase in number and morphospace coverage, and then start to shrink back over time, but one group of topographies produces a flurry of new topographies that expands and continues to develop and produce new offshoots, even as the main dinosaur topographies reduce in number and finally disappear. This new group of topographies would be the birds.

The pit analogy would then, connect with a more internally consistent consistent group of analogenic (is there such a word?) descriptions of evolution.

It’s pit’s all the way down!

Climbing Pit Improbable

In the ongoing Adaptationist v. Pluralist debate, both sides agree on a surprising amount. Both sides agree that there is more to evolution than adaptation by natural selection. However, Adaptationist would argue that adaptation by natural selection is the most important, or even the overwhelming, evolutionary process, and that evolution can be described as climbing Mt. Improbable – with adaptation to environment similar to climbing a fitness landscape peak towards optimal fitness (but please note, never, ever, reaching the top!)

I disagree. I think genetic drift accounts for most of the evolution that occurs, and natural selection, while very important – especially in creating diversity – accounts for a smaller percentage. However, both have worked together to produce the diversity of life on Earth.

I also have a problem with the Mt Improbable analogy . . . well, actually I have two problems.

1) It perpetuates the idea that evolution is an upward striving process, and that derived or adapted groups are higher, than the less derived or less adapted and, as a consequence, fitter, advanced . . .better. (the old Tree of Life analogy problem.)

OK, maybe it is applicable to a fitness landscape, but there is no reason that the landscape has to have the peaks pointing upwards, . . . is there? Surely it's the distance between where you are on the peak and the schmucks on the fitness plane that is important, not the direction of that distance?

Plus, fitness landscapes, are not permanent, or even solid. They change with the environment. A population/species, or whatever, may be quite "high" (see how hard it is to use neutral language)on a fitness peak one minute, and find itself down on the plane, or even in a fitness trough, with hardly any change in allele frequency, but a significant change in environment. In other words the fitness landscape moved underneath it.

2) The real problem with the Mt Improbable analogy though, is that it gives the impression that as hard as it is to ‘climb up’ (and it is), the analogy suggests that it is relatively easier to ‘climb down’ - and it isn’t because its actually harder. OK that might be pushing the analogy a bit far – but that’s the point, it doesn’t hold up to detailed scrutiny.

The real problem is that adaptation, in the broader picture, is an evolutionary cul-de-sac.

Adaptation means that certain alleles are being selected for because they confer an advantage in a particular environment. If the environment changes, then the alleles that conferred an advantage may no longer do so. Worse, the very process of selecting for certain alleles may well have stopped other alleles getting fixed through drift – alleles which might be beneficial in the new environment. Even worse, the alleles that were originally selected for may be costly to produce and maintain where they confer no advantage, and thus be deleterious.

But the really bad thing is that, as hard as it is to gain the alleles that provided an advantage, it is even harder to loose them, as this would require specific mutations to affect those particular alleles (rather than the random process that produced them). You could reduce them to a vestigial level, provided you survived long enough to do so. Difficult though, if you are struggling to survive in a new environment where the competition does not have the adaptive dead weight (unless you have some other advantage.)

The more adapted a group becomes, the more imbedded it is in a particular environment, and the more sensitive it is to changes to that environment.

Eventually all strategies lead to extinction, but during environmental change, it’s the generalists that survive, not the specialists. Adaptation generally leads to extinction. Highly adapted groups/species and ecosystems delicately balanced on a web of interconnected adaptations, will crash once environments change.

Adaptation is not climbing up Mt Improbable, it’s climbing down Pit Improbable! The pits are hard to find, but once in, it’s easier to go down than it is to back out, and if you adapt too far, you are trapped in a cul-de-sac with no way out when the environment changes. The generalists that flirt with the rim of the pit, or on the fitness plane have a better chance of surviving to become the stem stock for new adaptations.

It may well be that some species or groups of species in a pit break through to new fitness landscapes and produce new groups (e.g. birds and mammals from reptiles) because fitness landscapes are not flat, but curved.

But for most populations/species, adaptation is a pit of no return.

Photo credits
Mountain image from copyright-free-photos.org.uk
Pit image from larrydsmith.com


UPDATE
There are two discussions here and here, and a new blog post here.