Has anyone proven with math that natural selection is non-random?

I've recently been having a rather lengthy debate in the MSN "Nature VS God" forums about whether or not I.D. has conducted any "real" scientific research. During the course of this debate he stated that no one has constructed any significant statistical models about any particular, claimed evolutionary pathways using mutation and natural selection. Of course, I immediately turned this back and asked if no one has done that, then how can anyone claim it works, much less call it a "fact"?

More importantly, it got me to thinking about natural selection and how proponents of Neo-Darwinism claim it is a "non-chance" fitness filter; has anyone ever challenged them to produce a proof that demonstrates that natural selection is, in fact, "non-chance" or "non-random" - that it is a "fitness filter" at all?

For all I can tell, "natural selection" seems to me to be totally random if one figured in all the potential internal and external factors that "decide" if a genetic change survives; how can they claim it to be non-random, if they've never mathematically modeled it? How do they know if "fitness" is even a meaningful commodity in evolution, if there is no algorithm that can predict what would be a "more fit" genetic expression in any given scenario?

For all they know, sequences of genetic success aren't based on natural selection at all, but rather on some kind of common teleological process that is deliberately searching for and filling an evolutionary niche; this could be an inherent commodity of all life and coded for in all DNA. I mean, if RNAcan "accidentally" figure out how to self-replicate, and we know humans can plan and design outcomes, why is it such a stretch to think that DNA itself doesn't have such a teleological mechanism?

Couldn't it also be that DNA "deliberately" mutates itself (or, at least, that's how it would appear to us) to accomplish evolutionary goals?

Allen MacNeill (and other biologists) actually tries to break natural selection into several different “neat” categories to fit the evidence that is consistently found in the fossil record (very sudden appearance, then rapid diversity, then slow decline in diversity over periods of time.) Let me explain.

All naturally occurring populations exhibit what Fisher called continuous variation. That is, a range of variation in various traits that, when plotted in Cartesian coordinates, approximates a normal distribution (bell-shaped curve). In a trait that exhibits continuous variation, most of the individuals exhibit the trait at a value reasonably close to the mean, with a relatively small number of individuals exhibiting relatively large or relatively small values for that trait.

An example is height in humans, which is often illustrated in introductory biology texts with a group of students arranged along a football sideline in order of height. There are a few very short and a few very tall people, but the vast majority form a bulge in the middle of the curve. As noted by Eric that major problem is with the "engines of variation".

Given a population that exhibits continuous variation for a trait, it is claimed that there are three different patterns of natural selection that can result:

Stabilizing selection, in which individuals from both extreme “tails” of the normal distribution are not preserved over time (i.e. they do not have as many offspring that survive to reproduction), compared with those in the middle bulge of the curve. Under such conditions, the mean value for the trait does not change over time (hence the term “stabilizing selection”). In our example of height, stabilizing selection would be the result if individuals of average height had the most surviving and reproducing offspring (assuming that height is heritable from parents to offspring, of course).

Directional selection, in which individuals from one (but not the other) extreme “tail” of the normal distribution are not preserved over time (i.e. they do not have as many offspring that survive to reproduction), compared with those in the middle bulge of the curve. Under such conditions, the mean value for the trait changes over time, shifting toward the tail of the curve that includes the surviving individuals (hence the term “directional selection”). An example would be a population on an island all becoming pygmies over time.

Diversifying selection, in which individuals from the middle of the range (but not either extreme “tail”) of the normal distribution are not preserved over time (i.e. they do not have as many offspring that survive to reproduction), compared with those at the extreme tails of the curve. Under such conditions, the mean value splits and becomes bimodal, with two new mean values increasing in frequency, and the old mean value disappearing. This process would eventually (depending on the intensity of selection) produce two phenotypically different populations where one had previously existed (hence the term “diversifying selection”).

Thus, with these different versions of natural selection Darwinism becomes unfalsifiable. For they can explain both rapid diversification of a fossil type and then can explain lack of diversity of fossil type thereafter. He can explain anything he wants in the fossil record whenever he wants his theory to fit the evidence. i.e. his “new” theory has greater weight than the evidence has to falsify it.

Now many ID proponents don’t dispute the notion of stabilizing selection. They dispute the notion that there’s any kind of selection other than stabilizing selection. This does not mean that it does not happen per se (think finch beaks) but that it's not special enough that should not warrant a separate categorization. Or at least that directional selection is exceedingly rare and can only operate under limited conditions/environments and thus for a very short amount of time (or at least it better be short lived...directional selection tends to decimate a population as was seen with the finches). Personally I'm fine with him making these distinctions since they've only been shown to be capable of trivial changes.

Now as I've pointed out before the major issue is that natural selection is essentially a funnel, and it must be balanced in order to produce results. I've already belabored this point in the blog post GA Chess. It must also be balanced long enough that the trait becomes fixated. The problem with the finch example is that once the environment changes back to normal the finch population also reverts back to being a mixed population. As in, the changes purportedly funneled by directional selection don't stick (they are not fixated). Some Darwinists like to say that in order for such changes to fixate that the environment must be permanently altered as well. Well...in the finches case it's apparent by their dwindling numbers that this would likely cause extinction of that population within that environment. Even if they did survive and the trait did fixate within the population it's unknown whether the finches would permanently lose the ability to produce beaks of different sizes if the environment changed once again far off into the future.

A Darwinist put it this way: "sufficient conditions for long-term improvement [and fixation, I might add] to be likely are quite complicated." Tell me about it... Here's an example with flying squirrels, which have numerous balanced morphological changes in order to properly glide.

Dawkins speculated that falling from trees provided the environmental funnel. How many squirrels died jumping out of trees before some of them found out that they were lucky enough to have mutant extra skin along with modifications to the spine and ligaments in order to allow them to glide? How many squirrels have to fall to their deaths for such a change to become fixated in the population? Do we have any data at all on deaths caused by falls or is it all speculation? The automatic tendon locking mechanisms of such creatures should keep most of the corpses of natural deaths up in the trees I would imagine. What environment would provide this selective pressure? Unfortunately for such speculations, ordinary squirrels have been observed to fall from great heights with little or no injury. So are we now forced to hypothesize a limited set of environments which may include trees that would regularly cause death by falling?

The reason I ask all this is because evolutionary biology claims to have all this predictive power, so answering these questions should be easy. If this particular hypothesis (death by falling providing the environmental pressure) does not match reality what scenario is plausible? After all, there needs to some sort of plausible scenario since these traits are shared in divergent species and are supposed to be the result of convergent evolution.

The recent article HOW TO MAKE A FLYING SQUIRREL: GLAUCOMYS ANATOMY IN PHYLOGENETIC PERSPECTIVE (2007) makes the suggestion that since leaping distance scales with size that a smaller species would benefit more from gliding. So perhaps the selective pressure would be a smaller species competing with a larger species? They also briefly mention that evolving from a ground-based ancestor would be unlikely, presumably because of the low positive selective pressure for gliding. But again, we're back to the problem of needing regular directional selection in order to fixate these changes in the population. Also, in order for these changes to be beneficial in the first place they have to be balanced (look up that article to see just how balanced). And if they're not balanced they're unlikely to provide much benefit (it's neutral) and thus will be lost.

Having said all that, in general I don’t see an issue with unguided Darwinian mechanisms being capable of making these particular changes considering their relative simplicity and apparent modularity (then again, it may be front-loading). I just think it disconcerting that the focus of that recent article–which should represent the latest findings on this subject–seemed to be on making comparisons between samples. Darwinian mechanisms as the source of evolution were generally assumed to function, without any evidence of this being the case. The problems related to natural selection were never addressed. This is ironic since the article is entitled “HOW To Make a Flying Squirrel”.

Now Darwinists always start with the assumption of simplicity giving rise to higher complexity. Some ID proponents present this alternate scenario: What if ALL of the original squirrels could glide? After all, it's far easier to suffer a deleterious mutation, and the survival benefit from this particular feature is negligible in most circumstances. The same could be said of the bat, where some species have echolocation and others do not. What if the original bat had echolocation and then over time some divergent lines lost it? Now this scenario is compatible with YEC/OEC and front-loading hypotheses where the change program terminates at the final form and then deletions occur.

Another issue is that often times Darwinists are dealing with mathematical models. It is claimed that fitness should not be measured by actual success (or actually, lifetime reproductive success, LRS). Instead, fitness should be the mathematical expectation of LRS in the environment. So it's possible that Darwinism may "work" in the mathematical models but the models do not match reality.

Needless to say, I'm not sure if all this makes the issue ever more confusing.

The old example of sexual selection conceived by Darwin apparently has gotten shot down:

The feather train on male peacocks is among the most striking and beautiful physical attributes in nature, but it fails to excite, much less interest, females, according to new research.
...
Across the board, the researchers were unable to link the elaborateness of a peacock's train with his mating success. In fact, Takahashi and her team found little train variance among males in the population they studied. They also couldn't detect any link between a particular male's fitness and his train.

While the report concludes that the elaborateness of the train does not influence mating success, it also appears to say that there is some selective advantage for having trains that can generate more rustling during the shivering display. The bigger the train the more rustling (ie size still matters). Evolution is not uncomfortable with any finding (Dawkins makes precisely this poing in his discussions on 'Zahavi's handicap'. It is precisely because 'anything goes' that evolution lacks the element of falsifiability:

(Excerpt From My Book The Circular Temple)

Of course there are those that are severely critical of the claim that Darwinian theory is not falsifiable. The philosopher John Wilkins for example claims that Darwinian theory can and does exclude certain possible outcomes. According to Wilkins, natural selection, “rules out the existence of inefficient organisms when more efficient organisms exist". However Wilkins forgets that even here we do not have an a priori definition of what an efficient or an inefficient organism actually is before observing them in nature. Perhaps not surprisingly, even when there is a clear disadvantage for the possession of a particular adaptation, Darwinists have been able to come up with the final word- an advantage for having such a disadvantage. Amotz Zahavi’s hadicap theory, for example, tells us that for adaptations that seem to be burdensome and dangerous, such as the elaborate feathers of a male peacock, the very demonstration to the opposite sex that the male can carry such a cumbersome adaptation is in itself a display of physical prowess. The handicap is all too evident- such a display is energetically draining as well as outright dangerous (imagine the ease by which a predator can home in on the exuberant peacock). Richard Dawkins has suggested a more general application of Zahavi’s handicap principle, with natural selection favouring skepticism in females to such a degree that males have to display their prowess simply for the sake of demonstrating their prowess

“The premise of Zahavi’s idea is that natural selection will favor skepticism among females (or among recipients of advertising messages generally). The only way for a male to authenticate his boast of strength…is to prove that it is true by shouldering a truly costly handicap- a handicap that only a genuinely strong male could bear. It may be called the principle of authentication”

Like most other people, I can admit that salesmen have to be more cunning to convince me of the authenticity of the product that they are trying to sell me and only those that pass the ‘authenticity threshold’ will manage to sell me anything. However, we see in Zahavi’s handicap principle, the potential for misuse since, like in all the illustrations above, it lacks the element of falsifiability. If the outcome were any different- for example, if only peacocks with stubby feathers survived- we could just as easily conclude that natural selection, or sexual selection in this case, had favored stubby feathered varieties. After all, what a handicap it is for a peacock to carry it wondrous feathered display of beauty.

ARN picked up on the peacock story:

http://www.arn.org/blogs/index.php/literature/2008/04/01/sexual_selection_falsified_in_the_case_o

Fisher’s Fundamental Theorem of Natural Selection: the death sentence for Darwinism

Sal did oversimplify the topic too much and Bob immediately clarified, but otherwise it "appears" to be a good overview. I'm not sure if Sal is overstating his case. After all, we DO have real cases of natural selection at work (finches, blind cavefish, malaria, etc.).

Anyway, the major issue is like Eric stated: natural selection is not operating with regard to SOME traits. As in, it's not operating uniformly but with only limited cases.

The reason I think this is an issue is since natural selection usually relies on environmental factors. While some factors are generalized, some factors must be very specific in order for the funneling effect to work. What if, like with these peacock feathers, the factors are very rare or don't even exist? That means that in order for Darwinism to work not only does CSI have to emerge it must be paired with a rare event that offers selective pressure.

Let me bring up the flying squirrel again. Since I last wrote about it at OE I had a long conversation with a friend who played as devil's advocate. HOW TO MAKE A FLYING SQUIRREL asserted that JUST small gliding squirrels competing vs large squirrels would enough, but that doesn't make sense...what factor would cause the large squirrels to be dying in large amounts? As in, what factors would be enough for directional selection to work?

1. My friend posited a forest where a population of squirrels are isolated. This forest has VERY tall trees with very little branches, where a tumble would likely mean death (assuming there's no branches to grab onto on the way down).

2. We ignored the difficulties of the emergence of variation...we just assumed that a squirrel with fully functional gliding abilities appeared. We also assumed they had somehow become 5-10% of the population, which is very generous. BTW, I don't know what informational content separates a glider from a regular squirrel so I can't say whether this is CSI, so it "might" be within the reach of Darwinian processes.

3. Squirrels are not stupid. They don't fling themselves to death on a regular basis. So what provides the environmental pressure? My friend posited a fire. The only squirrels who could jump from tree-to-tree fast enough to escape would be the gliding variants.

4. Well, wouldn't the regular squirrels far enough away from the source of the fire escape as well? My friend posited multiple lightning strikes to start fires all over. But that added difficulty for the gliding squirrels as well. On top of that squirrels living near the edges of the forest would escape. In such a chaotic environment it's impossible to assign values, but for this story (which is all it is) we'll say the population is now half regular and half gliding. As in, 90-95% of the population was eliminated.

5. The final issue is that these factors are not consistent. Once the fire is gone so goes the positive selective pressure. What's to keep a deleterious mutation from eliminating this trait? In fact, with the forest now supposedly decimated the gliding squirrels might have a disadvantage. Skins flaps are not suited well for clambering through bramble and bush.

6. Is even this convoluted scenario enough for the trait to fixate?