We all know the outlines of Darwin's theory: heritable phenotypic variations in a population; competition for resources; survival of the fittest, constituting "natural selection."
One wag in Darwin's time quipped that "the theory explains the apple tree by cutting away its limbs." That misstates the deeper issue: Where do the "fitter" forms come from that are then selected? That is, from whence "the arrival of the fittest," or at least the "fitter."
Here are three partial answers.
1) Bacteria become resistant to virtually any antibiotic within a few dozen years. But the biosphere has never seen these newly synthesized molecules in many cases. There are trillions and more of possible molecules. How come resistance as an adaptation arises so readily. Why is it so "sufficiently likely?"
A partial answer can be seen in the human immune system. We make on the order of 100 million differently shaped antibody molecules. In order to be effective in protecting us from foreign "antigens," these antibodies must recognize almost any incoming "non-self." How is this possible?
Because, on the scale of molecular binding site recognition, say a few tens of angstroms in length, height and width and several other features such as polarity, van-der-Waal forces, and so on, there are far fewer effectively different molecular shapes than there are kinds of molecules.
Data would suggest that there are only about a million, not 100 million, effectively different molecular shapes. So 100 million for the size of molecular "shape space" seems a safe bet. That's probably why bacteria, making hundreds of millions of mutant proteins, can find resistance easily. The fittest arrive.
2) How can evolution evolve the control network among genes by which they activate and inhibit one another, the "genetic regulatory network," in such a way that cell types are preserved most of the time in evolution, but sometimes a new cell type is added?
To address this we need a theory of the relation between genetic regulatory networks and cell types. In such networks, the genes modulate one another's activities at the same time. The total system follows a pathway, or "trajectory" in its "state space" consisting of all the possible patterns of gene activities, and typically settles down to something like one of a number of alternative steady state patterns of gene activity.
Each such pattern is called an "attractor" mathematically. Many pathways, i.e., trajectories, flow into each attractor, as many streams flow into a lake. Here the lake is the analogue of the attractor. The set of pathways, or "trajectories" that flow into a given attractor are called its "basin of attraction." We know cell types are stable over division cycles, and distinct from one another. Now consider the hypothesis that cell types correspond to attractors.
This hypothesis, now partially proven experimentally, explains why different cell types in you, liver and kidney, are distinctly different. More, because each attractor tends to be stable to small perturbations, cell types exhibit homeostasis and memory over cell divisions. Finally, if cell types are attractors, noise or signals that induce transitions between an attractor and another attractor are just what we need for cell differentiation from the fertilized egg, the zygote, to the adult down branching pathways of differentiation.
Fine, but suppose that mutations to the regulatory network altering its connections and gene regulating rules, altered all or most attractor cell types? This would typically be lethal, so adaptation could not occur. But remarkable recent results help us. Such networks behave in three ways: ordered, critical, and chaotic. It turns out that mutations to chaotic networks do alter all or most attractors. So if cells were chaotic, mutations would be lethal. But for ordered, and even better for critical networks, most mutations leave cell type attractors intact, and sometimes add new attractor cell types.
In short, a vast class of ordered and critical gene regulatory networks allow the arrival of the fittest, i.e., adaptation to occur. Such networks are robust but evolvable.
3) The third strand is new and of unknown importance so far. In past posts I've discussed the fact that a dividing cell, e.g., a Paramecium, must achieve a closure in a complex set of tasks, including membrane formation, ATP production, construction of chemosmotic pumps and so on. More this "task closure" passes via the biotic and abiotic environment of that cell. Now consider this: as proto-organisms and organisms became more complex, their parts had ever more causal consequences.
But for an evolving cell, all that is required is that some one or several causal consequences find some use in the cell and niche that enhances fitness. Then that evolving cell is fitter and selected by natural selection if there is heritable variation for the features in question. In short, the more complex an organism and its parts, with more causal consequences per part, the more readily it should be able to adapt. We have a third, unexplored, strand in the "arrival of the fittest" and a new reason to expect that evolution will tend to increase the complexity of organisms, as has happened for whatever reason.
Then why is Darwin's "bank of life" tangled? Because, in part, the "arrival of the fittest" is sufficiently likely. Adaptation is able to happen only because of this. Selection winnows, as the wag said, but the abundant possibilities of the ways of life yields the arrival of the fittest.