Creationists are fond of the "it can't happen" argument: they like to point to things like the complexity of the eye or intricate cell lineages and invent bogus rules like "irreducible complexity" so they can claim evolution is impossible. In particular, it's easy for them to take any single organism in isolation and go oooh, aaah over its elaborate detail, and then segue into the argument from personal incredulity.

Two things, one natural and one artificial, help them do this. Organisms are incredibly complicated, there is no denying it. This should be no solace to the anti-evolutionists, though, because one thing natural processes are very good at is building up complexity. The other situation that has helped them is our current reliance on model systems.

We use a few model systems extensively to study development—Drosophila, C. elegans, Danio come to mind—and they give us an unfortunately rigid view of how developmental processes occur. The model systems that are favored for laboratory work are those that have rapid, streamlined development with a great deal of consistency to the pattern—variability is avoided, and we tend to look for reproducible rules. We get a false impression of the rigidity and inflexibility of developmental systems.

How to correct that? We use the model systems as a starting-off point, and look at related organisms. As we start to accumulate information about diverse species, the variability in the patterns of development becomes more prominent, and we see that the evolutionary pathways aren't difficult to see at all. The worm vulva is a great example of how phylogenetic studies of development can inform our understanding of evolution.

I described some of the details of vulva development in C. elegans before, so I'll just give the two-penny summary here. The key players are the anchor cell, a specific cell in the gonad, and an array of generic epidermal cells called p3.p, p4.p, p5.p, p6.p, p7.p, and p8.p. The formation of a vulva is triggered by an inductive signal from the anchor cell to the nearest epidermal cell, which is usually p6.p. This cell then activates the genes responsible for forming the vulval opening itself, and adopt the 1° fate. Then, p6.p produces a lateral signal which recruits the adjacent cells, p5.p and p7.p, to follow the 2° fate, becoming support cells that make the muscle attachments and other details needed for the vulva to function. The remaining cells become 'mere' epidermis in response to other signals in the environment.

Schematic summary of signaling interactions during vulva formation in C. elegans. An inductive EGF-like signal originates from the AC (arrows). P6.p signals its neighbors to adopt a 2° fate via 'lateral signaling' (dotted arrows). Negative signaling (black bars) prevents inappropriate vulva differentiation.

It all sounds very programmatic and pre-determined, doesn't it? These cells and signals are all critical and needed for proper formation of a fairly important organ; how could they have ever arisen without careful planning?

Let's look at a few pieces of the story. First, the anchor cell: this is the linchpin of the whole assembly, so where does it come from? In C. elegans, there's actually a fair amount of flexibility in which cell becomes the anchor cell—there are two equivalent cells, named Z1.ppp and Z4.aaa, that negotiate with one another to determine which will become the anchor cell (AC), and which will adopt another role as the ventral uterine precursor (VU). Either one can do the job, it's random which one becomes the AC (50% of the AC cells in C. elegans are derived from Z1.ppp, 50% from Z4.aaa), and the signalling is symmetric. Picture two people equally qualified to do a job, who settle which one gets it with a game of rock-paper-scissors. This is the situation portrayed in the top left box of the diagram below (which has very tiny print…click on it to see the larger, readable version).

In nematodes of the genus Acrobeloides, though, there is a slightly different situation. There isn't a 50:50 chance anymore; there is a predisposition for Z4.aaa to get the job. There are still negotiations between the two cells, and if Z4.aaa is deleted, Z1.ppp is still fully capable of taking over and functioning as an anchor cell, but there is a bias for Z4.aaa. In this case, picture two people vying for the same job, but one happens to be the boss's nephew, with an inside track on the position.

In yet another species, Cephalobus cubaensis, we see a hardening of the situation. Z4.aaa becomes the anchor cell 100% of the time. If an experimenter deletes Z4.aaa, Z1.ppp can step in and take over, but this doesn't occur under normal circumstances. There is only one signal here, with Z4.aaa instructing Z1.ppp to become the VU. The analogy here is two competent people vying for a job, but one is the boss's son and has an absolute lock on the job, barring getting run over by a bus.

Finally, the last condition in Panagrolaimus is a rigid, lineage-dependent specification of cell fates. Z4.aaa always becomes AC, and Z1.ppp always becomes VU, with no communication between the two. Deleting either one makes no difference, and neither can take over the other's job.

What these represent are evolutionary changes towards increasingly stringent specification of cell fates, from a flexible, regulatory species like C. elegans to a more rigid, hardwired pattern that we see in Panagrolaimus. It's relatively easy for evolution to ratchet a regulative, symmetrically-interacting, self-organizing set of cells into a rigid mosaic of pre-determined cell identities. If we saw only Panagrolaimus, we might have an even more biased opinion of the invariance of nematode cell lineages.

The second column of the figure above is illustrating another point: even within one species, we see this whole range of degrees of specification in different cell types. By looking at multiple kinds of interactions between cells in different organs, we can find examples of each step in this evolutionary pattern.

Similarly, there is variation in the modes of interactions between the vulval cells in different species. C. elegans builds a vulva using a mechanism combining A and B from the diagram below, but others use versions of A through E.

Patterning mechanisms of the vulva precursor cells. The vulva develops from precursor cells (Pn.p cells, oval) aligned in the ventral epidermis. The fates of these precursors form a centered pattern with inner vulval fates (black), outer vulval fates (grey) and nonvulval fates (white). The panels show five modes of patterning of such a field of cells. A: Graded. The gonad ( for instance, the anchor cell, square) sends a graded signal: a high dose of signal induces the inner vulval fate, a low dose induces the outer vulval fate. B: Sequential. The anchor cell sends a localized signal inducing the inner vulval fate in one precursor, which in turn induces its neighbors to adopt the outer vulval fate. C: Two-step induction. The gonad/anchor cell sends a first signal specifying the outer vulval fat in three cells, a second signal inducing the inner vulval fate in the central one (or its daughters). D: Self-organization. The vulva precursor cells signal each other, so that the central one receive more signal. E: Prepattern. A pre-established positional system (for instance by differential expression of HOM-C genes) specifies the fate among these cells. All or most of these mechanisms are used more or less redundantly to ensure the reproducible pattern of vulval fates in C. elegans, but A and B are predominant. The presence and the weight of these different mechanisms vary among nematodes.

It's not just the signaling that varies, but also the origin of the cells that respond to those signals. The diagram below illustrates the lineage pedigrees of the vulval cells; in (b) we can see the invariant pattern of cell divisions that occurs in each of the Pn.p cells in C. elegans in response to their cell fate commitments, and also the subtly different divisions that occur in Pristionchus and Oscheius. In (c), there are even mutants in Oscheius that modify the lineage, yet still result in functional vulvas.

Schematic summary of vulval fate patterning and vulval lineages in Oscheius wild-type and mutant animals. (a)Vulval cell fate specification in Oscheius. P(48).p form the vulva equivalence group and P(57).p form the wild-type vulva with a 2°1°2° pattern. (b)Comparison of vulval cell lineages between Caenorhabditis elegans, Pristionchus pacificus and Oscheius sp. CEW1. L, longitudinal division; N, no division of the Pn.p granddaughters; S, cells fusing with the hypodermal synctium; T, transverse division; U, no division, but cells lack the typical characteristics of N cells as described in C. elegans. (c)Oscheius mutant animals as described by Dichtel et al. Vulval precursor cells undergo hypo- or hyper-proliferation without changing terminal vulval cell fates. Cells forming vulval tissue are underlined.

I can predict one argument from the creationists: despite these variations from species to species and within a species, these are all still just nematode worms. However, these are also fundamental changes in developmental processes that are far greater in magnitude than anything known or proposed between humans and chimpanzees. If one is going to be blasé about these differences within 'just worms', one is going to have to accept the differences between mammals as relatively trivial, as well.

What we have in this particular system is a set of confirmed properties that make evolution an inarguable consequence of natural, biological processes.

• the existence of morphological differences between species
• natural mechanisms that can generate comparable morphological differences
• compensatory and regulatory mechanisms to tolerate variation within species
• the demonstration of common molecular mechanisms underlying those differences
• experimental perturbations that mimic species differences and produce viable individuals

There are no magical barriers to evolution. This is what modern biology is telling us: the study of model systems has provided us tools to probe the underlying mechanisms of development change with growing facility, and allowing us to explore phylogeny with greater and greater depth…where we are finding fluid transformations between different forms.