The concept of natural selection has been a disaster for philosophy, but it didn't have to be. A Darwinian ontology, in which the only significant forces are the gradual mutation of an otherwise stable substrate by which properties are carried from one generation to the next, and the self-evident fact that these properties will only be preserved if the current generation successfully reproduces before dying, has severely undermined not only timeless inquiries into the very nature of knowledge and subjectivity, but has severely shrouded many practical questions in a haze of poorly tailored concepts and assumptions. The issue isn't that natural selection doesn't happen or that genetics aren't of profound scientific importance, but that these things are by no means the whole of how evolution operates or even what evolution is.
If anything, however, a fuller understanding of evolution shows just how far-reaching a concept it is while showing the myriad places where neo-Darwinian pretenses hold no water. Before talking about the missing piece, I'd like to list the utterly baseless assumptions one has to make in order to say that this or that property of an organism simply comes from natural selection:
1. That there is always a simple and reliable correspondence between genotype and phenotype
If natural selection selects for something physically in the world, a phenotype, then the only way for the "fit" traits to reliably propagate while the "unfit" traits do not, is for those traits to be reliably heritable. Since the means of inheritance is DNA, this means that these phenotypes must be reliably "remembered" in genetic material.
Given the way a gene works on a physical level, this is already suspect. Genes do not on their own generate organic matter, they exist within a substrate of organic matter that itself grows and develops in accordance with (but not necessarily completely determined by) its genes. Genes are code, and what material developments they engender is therefore always dependent on how their corporeal substrate chooses to read them. That could easily enough change depending not just on the nature of the substrate, but also the context in which a gene exists: just like the word "spare" means different things in the sentences "I have a spare tire" and "spare me the nonsense", a gene that ostensibly gives an elephant its trunk could not be implanted into the genome of a banana to create a banana with an elephant's trunk. There is, of course, in some sense of the word, a degree of stability in the information that DNA holds, not least because of its largely read-only nature, but this stability is of a nebulous nature given the ways in which context changes the meaning of any gene.
2. That an organism is fully optimized for its environment
And supposing that natural selection is the explanation for any and all traits, this means that there must always be some identifiable pressure by which such a trait is shaped. In other words, it must be the case that there was something about the environment that made it untenable for smurfs to be green, maybe because being green made them invisible and therefore made it difficult for smurf women to spot them during mating rituals amongst the foliage, or because the color blue is better for absorbing vitamin D and that made blue smurfs outlive green smurfs and slowly muscle in on their gene pool.
This is absurd for many reasons. First, if something were perfectly optimized for its environment in this manner, it would create two big problems off the bat: (1) that any slight change to their circumstances would severely damage their chances of survival in the same way that an "optimal" amount of leverage in your financial portfolio becomes a serious liability if you suddenly encounter an unexpected difficulty in acquiring the cash to cover it, and (2) no big or interesting leaps in evolution can happen if everything is always optimized, because any slight deviation off course would be immediately corrected by natural selection; unless, of course, genes don't strictly correspond to phenotypes, in which case you'd have no mechanism to ensure optimization in the first place.
3. That selective processes always converge on some specific set of properties
This is more or less a variation on the previous point, but worth going into in a bit more detail. For "natural selection" to be the reason that a certain trait exists, there has to be a path that unequivocally leads to that point. Organisms may reproduce or die for many reasons, many of which, at least from a naive standpoint, have no rhyme and reason to them. For there to be some actual selection pressure on a specific trait would require that one can show that this trait is consistently necessary for survival/propogation, or more precisely, that one can demonstrate a limit such that regardless of the timeline, the process ends in a neighborhood in which this trait necessarily exists amongst all members of the group. In other words, one must demonstrate a kind of ergodicity.
4. That any identifiable property of an organism is ontologically disjoint from its environment
All of these nonlinearities are major problems, but are a mere shadow of a much more damning issue: that you cannot simply define a trait in isolation. If someone has a specific taste for strawberry ice cream, how can you define that without recourse to a food which has only existed for a relatively short period of time? Many would go on to reduce this to some craving for excess energy in the form of carbohydrates, but that doesn't account for liking this one flavor better than another. If this seems too pedantic, then I'll raise the stakes with something more direct: the ability to grasp branches.
At this point, I fully expect that the response is that even if trees didn't exist, hands still could. Maybe so, since hands have other functions and besides which I pointed out that evolution requires some degree of redundancy. All that being said, it is not the part, taken out of context, that allows an organism to thrive and reproduce, but the functionality it provides. In fact, part of why it's unrigorous to assume any given trait exists or dominates based on natural selection is that if an organism is selected for anything, it's selected for function, and the same function can be fulfilled by potentially many different means. But to define any such function requires reference to the affordances offered by the environment, with any traits that one can talk about "in a vacuum" being some kernel that obtains across these functions.
Nor could one even define a hand in any meaningful way without such functionality in general: it's not the skin or the bones that make hands what they are, and for that matter not even their nerves, but the way in which all of these things are connected to the organism's body/mind, which in turn is something in large part defined by the way in which it traverses its environment. Without this qualification, you can't define anything essential about a hand, but only mere material accidents. Any such trait an organism has is therefore identifiable by its place within an encompassing syntax, each organism a prototype that may yet expand such a syntax by insofar that things initially accidental become definable according to some formal relationship with the environment.
Nor is it simply the case that a trait exists at the intersection of different ways an organism can utilize pre-existing affordances; an organism can itself create new affordances by utilizing existing ones in unprecedented ways. Not only do trees offer an entirely different kind of affordance to beavers than pretty much any animal prior to them, the very dams those beavers build are themselves things that other organisms could exploit, and by extension, evolve around. Taken together, this reciprocal creation and utilization of affordances is known to biologist as co-evolution. (Co-)evolution therefore is not merely some simple dynamic playing out on a static semantic space defined ahead of time, but a fundamentally contingent unfolding of an ever-changing field of relationships.
It follows that "meaning" is never a-priori a simple correspondence by which signifiers denote underlying noumena, but a dynamic gestalt by which new relationships and potentials are engendered by unforseen enactments
An affordance is a material potential for action, it does not strictly define what can be done but it does provide constraints. There's no simple limit to the amount of things I can write or draw or scribble on a piece of paper or to the many different ways I could go about doing it, each such stroke of the pen an enactment of its own. In the context of a larger system, I can think of even more things I could do with it: perhaps I could use it to order some item from a store. Of course, at the same time, this context the uniqueness of each and every enactment is no longer significant: it's now a question of some function that any number of sufficiently "similar" strokes of the pen will fulfill.
Even then, however, that does not make such a uniqueness totally irrelevant: perhaps I need to inform somebody of some bad news, in which case as long as I say the words they'll have the information delievered, but the way I say it could have a different effect. Perhaps it's not about information: perhaps the real "function" of what I'm doing is to help them through it, in which case a difference in how I say it is now relevant to function, but even then the point remains that every difference between enactments (and no two enactments are ever the same: "repetition is not generality") has the potential to alter the world moving forward even if it doesn't do so with regards to some syntax.
The unique signature of any such enactment, each one of which is a utilization of some affordance, is the stuff of which interpretation is made: not some ghost in the machine floating between one's ears, but a specific way of traveling through a medium that alters or supplements its meaning by changing the webs of relationships that define it in various respects. Even if one's own reading of a book never escapes the play of soup and sparks between their ears, it still changes their sensorimotor relationship to some part of the world or themselves. These material grooves by which such relationships exist like water flowing through them constitute the text, that through which an actor travels as a means of interfacing with the real world (of which the text itself is also a part); the text itself written through, you guessed it, the co-evolutionary spawning of affordances.
Deleuze said that philosophy was in the business of creating concepts, where art created affects. Given the co-evolutionary material genesis of all "meaning", however, it would seem that such a simple difference is not tenable, and that whether one is talking about art, philosophy, or even something as formal as mathematics, it's ultimately aesthetics all the way down.