A concept that has always puzzled me is: how do new species arise? Of course, you're thinking, mutations accumulate, until you produce a brand new organism. But if that's the case, why don't we see human-monkey hybrids, or rhinohorses, or . . . you get the picture. What puzzles me is where do the intermediate steps between two closely related species go? There are so many mutations that make one species diverge from another, yet you don't see one organism for each new mutations that arises. You see the endpoints. What happens in the middle? At the DNA level you see all shades of gray, but at the species level all you see is black and white (horse or non-horse). Why aren't those mutations expressed in new organisms as things progress from rhinoceros to horses, for example?
I don't know if there's only one answer to this question, but what really helped me clarify the issue were the concepts of epigenetic landscape and canalization.
The expression "epigenetic landscape" was coined by British developmental biologist Conrad Hal Waddington (1905-1975). He compared gene regulation during development to marbles rolling downhill towards a wall. As cells differentiate, the different cell fates are represented by the lowest points in a landscape made of several pits, some lower than others. Waddington also coined the expression "canalization" to describe the ability of organisms to produce the same phenotype against genetic and environmental variations.
Let's understand this better. I often talk about mutations, genetic variation, and how mutations can cause diseases or increase the risk of disease. At the same time, there are mutations that have no effect whatsoever. The math geeks out there will appreciate this metaphor: genotypic variation is not in a 1-1 correspondence with phenotypic variation. In fact, there are phenotypic traits that do not change despite genetic and/or environmental changes.
"There seems to be a strong robustness of some phenotypes against genetic and non-genetic change or perturbation. More generally, the amount and quality of phenotypic variation can differ dramatically within and among populations. Some traits are highly invariant within species while simultaneously being highly variable among closely related species; other characters seem to be highly conserved among species or clades ."On the one hand, canalization provides a buffer to phenotypes that are optimized in terms of fitness against genetic and environmental variation, making them more robust and stable. On the other hand, it allows for non-expressed genetic variation to accumulate: since it's not expressed, there's no selection, hence it persists in the population. So, even when a species "looks" homogeneous, it doesn't mean it's actually genetically homogeneous. There are underlying genetic differences that do not affect the phenotype. To me this represents a "hidden reservoir" of genetic variability. So long as the phenotype is fit, it is convenient to keep it that way. But once a selective sweep takes place, epigenetic changes take place and are likely to act differently on different genomes, thus providing a phenotypically homogeneous population with the potential for genetic change.
"The cryptic pool of genetic variation accumulated under canalization can be phenotypically expressed again if genetic or environmental change uncovers the silent genetic variation, thereby increasing phenotypic variation in the population. Decanalizing conditions can be due to environmental perturbations that change environment-dependent gene expression or allele substitutions that render canalizing mechanisms nonfunctional."So, you see what's happening: when a certain phenotype is stable and fit, it doesn't mean it's not changing genetically. It is, but there's a buffering mechanism that prevents the genetic changes to be expressed. It's like a marble trying to get out of a pit: it rolls around the lowest point, and small perturbations will make it roll higher or lower along the walls of the pit. In order to make it out of the pit, it needs a big perturbation, one large enough to provide energy to make the jump. That's when the cryptic variation suddenly becomes expressed and forms a new species.
I'm sure biologists are raising their brows at me, as there are probably better and more technical ways to say this, but I wanted to express it in simple words.
Now to a more deeper look, for those of you who are interested in a bit more technicalities: what are the molecular mechanisms of canalization? In other words, what prevents genetic and environmental changes to affect the phenotype of an organisms?
"The proximate (molecular) mechanisms causing canalization, as well as the nature of the perturbations, can be manifold. For example, the canalizing mechanisms can potentially be located at any level of the biological hierarchy, from gene expression, RNA stability, protein structure and folding, intermediate metabolism and physiology to morphology, behavior, and life-history traits."Furthermore, stability in one trait may depend on the variability of other traits, and the buffering may indeed involve complex epistatic pathways. In simpler words: a change in one locus may involve other genes whose expression needs to change in order to maintain the phenotype despite the change. This makes canalization particularly difficult to measure because not only it's not always obvious at what level the regulatory mechanisms are taking place, but also how many genes or loci it actually involves.
 Flatt T (2005). The evolutionary genetics of canalization. The Quarterly review of biology, 80 (3), 287-316 PMID: 16250465