Quick, in ten words or less, how does evolution work? The old phrase "survival of the fittest" quite likely sprung to mind, and you'd (mostly) be on the right track. But that phrase, coined by Herbert Spencer, is a gross oversimplification of Darwin's theory of natural selection.
The key is how we define "fitness." In the evolutionary sense, fitness refers to an organism's ability to not just survive, but reproduce. And if an organism can successfully produce offspring—in essence, copies of itself—it will propagate faster than one that can't.
Seems obvious, right? Well, again, this is another oversimplification. If one creature is better at reproducing itself than another creature, the first one is going to eventually win out. But how does a creature become "better" in the first place, and how does this drive evolution?
That brings us to this video by Bjørn Østman and Randy Olson of Michigan State, which features fitness landscapes, a method of visualizing evolution that was first proposed by Sewall Wright in a 1932 paper. Personally, I love them because they show how evolution is like an ever-morphing puzzle.
A peak in the chart represents a highly-fit genotype, or the set of genes that makes an organism what it is; you might also think of it as the "ideal" genotype for a specific ecological niche. The dots are individuals; the closer together they are, the more similar their genotypes are.
As Østman and Olson explain in the video, over time, organisms move towards higher ground on the chart—becoming more fit—because as they get closer to that "peak," they're able to reproduce more, and some of their offspring will produce more, and so on.
A fitness landscape from Sewall Wright's 1932 paper.
It helps show how evolution isn't an active process on the part of organisms—despite what you may have heard, chimps didn't just decide one day that they wanted to become humans. Instead, it's a matter of probability: when an opportunity opens up, those organisms best suited to take advantage of it will produce more of themselves than competitors, and the most successful of those offspring will produce even more, and so on as the population climbs up the fitness mountain.
Now, if everything in the world was completely static, we'd eventually see organisms evolve towards a single solution for their specific environment. You can imagine this if you take a mental snapshot of the world right now; a polar bear is pretty highly suited for its environment, food sources, competition, and so on. But the world isn't static at all. The environment, geography, animal behavior, biodiversity, and competition—both intra- and interspecies—are all continually in flux, meaning the "peaks" on the fitness landscape of the world are always shifting.
And as they shift, so do the successful genotypes that are able to climb up them. In essence, that's how evolution works on a very macro scale: Because the rules of success for a specific ecological niche are always changing, the ideal solution also must change.
One interesting case is when fitness peaks change rapidly—because the population doesn't have time to shift fully towards one or the other, organisms that are suited to both might arise, or the population might drift towards one peak and not the other. That's often when you'll get new species.
In another case, where fitness is density-dependent—having too large a population makes the whole population less successful—you see populations becoming more specialized quickly, like that of Darwin's famous finches, which evolved beaks to deal with very specific food types.
When you look at a fitness landscape on a broad scale, like Wright's original diagram above, you can start to see that there are plenty of fitness peaks in the world, and that those genotypes closest to the peak will most likely continue to evolve in that direction as they continue to get more and more fit. And when populations split and shift to this ever-changing landscape of fitness peaks, the end result is the ever-changing mosaic of species that have evolved on Earth.