Documenting allometry and patterns of size and shape changes in evolution are helpful as descriptive approaches to the evolution of development. But these phenomena are themselves the result of developmental mechanisms at the molecular and cellular level. We can often say without reservation that there has been a change in development during evolution, but how that change in development was achieved is yet another question. We improve the description somewhat by saying that changes in development result from changes in the: 1) spatial organization of cells, 2) timing of gene action and tissue differentiation and 3) geometry of tissues and organs. But how are these changes mediated?

Consider the comparison between humans and chimps: the adult morphology is obviously distinct, but at the genetic level we are extremely close to chimps: >99% similar at the genetic level. This is less genetic divergence that seen between sibling species (can't tell them apart) of Drosophila and some mammals These observations indicate that morphological evolution has proceeded faster than molecular evolution suggesting that regulatory evolution has proceeded faster than DNA sequence evolution. Where are the important mutations (short arrows)? in the coding sequences of genes or in the regulatory sequences upstream from them? May depend on how the product of a gene interacts with other genes (longer arrows).

Thus perhaps the key to understanding the evolution of development is the study the evolution of the genetic regulatory mechanisms that control development. Now the question becomes: what do we know about genetic regulation of development?

A fair amount is known in Drosophila. The exciting point here is that in recent years there have been increasing numbers of papers describing the existence of gradients across the egg or early embryo in the concentration of specific proteins encoded by a handful of loci. These proteins can be thought of as morphogens ("form creators"), molecules that, for years, were postulated to exist by embryologists. With a gradient across the embryo of such a morphogen, there is the possibility the other proteins that might interact with such a morphogen can obtain position information from the gradient such that high concentration means "anterior" (or "limb end" in vertebrate limb bud) and low concentration means "posterior" (or "limb base").

The significant point in all this is that Drosophila geneticists have been able to identify specific developmental mutations (mutations in the genes that code for morphogens, or genes that code for molecules that interact with morphogens) that disrupt specific events in development. One such example is the bicoid gene: when this gene is mutated, its normal gradient is disrupted and the embryo has two tails (bi-caudal). The point is that there are specific genes that determine the major body axes and one can envision that evolution of major new developmental programs might proceed by naturally occurring mutations in these genes that would move/alter the gradient, or, equally as significant move/alter the cellular localization of the receptor of a morphogen.

On a more theoretical level, morphogens have been hypothesized to operate in a threshold-like manner in more localized examples of pattern formation such as the generation of additional bristles in Drosophila or specific patterns of striping in mammals (see figs. below). Specific molecules causing "prepattern" such as one sees in zebras have yet to be identified, in contrast to the major advances made in Drosophila, but mating zebras is a major undertaking.

There is solid support for such ideas in Drosophila development. The RNA encoded by the bicoid gene is localized in the anterior portion of the embryo. The protein translated from this localized RNA is distributed as a gradient from anterior to posterior across the embryo. The bicoid protein affects the distribution of the RNA of another gene, hunchback. This RNA (hunchback) is not distributed as a gradient but in a discrete way: present in the anterior, absent in the posterior. Thus there appears to be positional information in the concentration of bicoid which is read by hunchback as a threshold. One could imagine that a mutation that affected the localization of one morphogen could alter the localization of important thresholds of different morphogens, which would in turn lead to the development of new morphologies.

The genes controlling the early events in the development of Drosophila can be classified into three broad categories: Gap genes are a set of genes that act to define broad regions of the early embryo; these can regulate the expression and action of Pair rule genes which further define the broad regions into more numerous segments; the pair rule genes can affect the expression and action of Segment polarity genes which will determine the fate of certain structure within each segment. As with the gradients of morphogens described above, one can envision mutations that alter the interactions between these broad classes of genes controlling the developmental fate of parts of the organism which, if established in the population, could lead to the evolution of new morphological "plans" (.g., a new Bauplan).

There is good evidence for such a supposition in another very important set of genes: the homeotic genes. Certain mutations in these genes result in homeotic mutations where one body part is transformed into the structure of another body part. The best examples are the Antennapedia complex and the Bithorax complex which are large regions of the chromosome containing several genes each. The positions of the genes on the chromosome have a remarkable correlation with the segment of the body in which they are active! (see figures below). The genes contain a region of DNA that codes for a highly conserved stretch of amino acids, known as the homeobox that generally are involved in the determination of body segments (but bicoid has a homeobox and it is more involved with anterior-posterior determination). While mutations that move a leg to the position of an antenna (in the Antennapedia complex) or transforms the balancing organs (halteres) into a pair of wings (in the Bithorax complex) is of dubious fitness value to the organism, it does show that modifications of the general body plan be achieved by mutations in one or a few genes, i.e., there is genetic evidence that Hopeful (hopeless?) Monsters could be produced.

These phenomena are compelling in light of the belief that arthropods (insects, crustaceans, etc.) evolved from annelids (segmented worms; see figure below). One can envision that sequential modification of body segments, through mutations such as those described above, might allow for the evolution of insects from a worm-like ancestor. Suggestive of this is the observation that when the Antennapedia complex and the Bithorax complex are mutated the larval stage of the fruit fly is transformed into a larva with many thoracic segments rather than the wild-type pattern of differentiation into maxillary, labial and abdominal segments (see fig.below). This "throw back" to the ancestral form (i.e., the middle segments of worms are relatively undifferentiated) is called an atavism.

In thinking about all possible morphologies one might be able to get with bizarre mutants in flies, and looking out at the incredible diversity of form in the natural world we might best think of this problem in terms of the question: why this and not that? Are there forbidden morphologies that development cannot produce?. There is some nice evidence that the different forms seen between species may be the result of the "playing out" of discretely different developmental programs. When the developing limb bud of one salamander is treated with an inhibitor of mitosis, the number and pattern of digits developing resembles that of another species (section reading). This suggests that there are developmental constraints, i.e., that development is constrained to proceed in a certain way. If different developmental programs are carried by different lineages of organisms as they diverge from one another, these developmental constraints become phylogenetic constraints: there is no chance that horses will sprout wings because the lineage of horses (and ungulates in general) are constrained to develop and use their forelimbs in very different ways than bats, lets say.

A conceptual model for this notion of constrains is to think of there being canalization of developmental programs. Waddington's model of development as a ball rolling down a landscape suggests that the program is canalized to follow a particular trough. Mutations and/or environmental fluctuations (next lecture) might knock the ball around in the trough, and if these perturbations were strong enough might throw the developmental program over into a new canalized ontogeny. In this model the location of the troughs suggests that events that perturb development early on are more likely to result in major changes in the developmental plan. Perhaps developmental programs become more "canalized" as they proceed through development. This is of particular significance in light of the network of genes described above that establish body plan during early embryogenesis.