As the Glenn Morton Phylum-Level Evolution page points out, one group of sponges has indeed evolved a completely different body plan from the rest of its group, or from any other living thing.
Vacelet and Boury-Esnault (1995, p. 335) relate:We don't actually assign a phylum to every unique body plan if we can see clear relationships by other means.
Our results raise fundamental questions about the validity of characteristics used to distinguish the phyla of lower invertebrates. A sponge is defined as a sedentary, filter-feeding metazoan which utilizes a single layer of flagellated cells (choanocytes) to pump a unidirectional water current through its body. Except for being sedentary, the cave Asbestopluma and presumably all Cladorhizidae lack these basic sponge attributes. In an extreme environment where active filter-feeding has a low yield, cladorhizids have developed a mode of life roughly similar to that of foraminiferans or cnidarians. Their feeding mechanism relies on passive capture of living prey and on transfer of nutrients into the body through intense cell migrations, the analogue of cytoplasmic streaming in foraminiferan pseudopodia. This may be compared to the emergence of macrophagy in abyssal tunicates, also accompanied by a reduction of the filtering system although in Cladorhizidae the result is more extreme, with a main body plan different from Porifera and resembling no other modern anatomical design.In cases like that above, the lack attribution of phylum rank for these 'sponges' hides the fact that the Porifera may very well have given rise to an independent phyla.Such a unique body plan would deserve recognition as a distinct phylum, if these animals were not so evidently close relatives of Porifera. Their siliceous spicules show clear similarities to several families of poecilosclerid Demospongiae.
The web page points out clear evidence that some distinct phyla did in fact originate from others. (Arthropods from worms, for instance. This should be obvious to anyone who has ever seen a maggot or a caterpillar.) Furthermore, many of the first appearance dates in the fossil record are long after the Cambrian.
Indeed, there are certain changes in body plans whose description as a new phyla or not seems to be rather subjective. A breed of humans with three eyes, an exoskeleton, antenna and wings like a bumble bee might be difficult to classify - but I'd probably call it a new body plan.
But the point I was raising earlier about the no new phyla after the Cambrian Explosion goes to a different aspect what appears to be a resistance to certain mutations in master control genes.
In the Gehrig example, the lowest life forms have no eyes and the eye gene with many sites has exponential possible combinations of amino acids. And yet the same combination of amino acids are selected for vision in all animals - vertebrates and invertebrates - across phyla. The gene experiments between the mouse and fly make the point rather vividly.
In the NASA article, the author suggests that microevolution cannot account for the fossil record. He discusses alternative models for macroevolution (genetic drift, etc.) - but strongly suggests that environmentally linked hormonal induced changes in the control genes can create the observed effect of quickly emerging and successful phyla.
When I take the Gehrig view in combination with the NASA article, it makes sense to me that the phenomenon of no new body plans could be easily explained by a stability or immutability of control genes setting in after the Cambrian. Without such an immutability considering the intervening mass extinctions - one is left contemplating why no new body plans. The interest by the NASA scientist (and others) indicates this is serious inquiry.
If the NASA scientist is correct, then the environmental conditions which caused sweeping hormonal changes during the Cambrian explosion have not since recurred. There ought to be some evidence of such a one-time event either in the cosmos or the geologic record.
But with or without such evidence, highly immutable control genes not only help to explain what is observed in the fossil record and in the laboratory but also indicate that evolution is not directionless. Direction in evolution goes against the notion of random mutations in evolution theory. Without evidence of natural causation, direction (immutability of control genes) also can be argued by intelligent design supporters.