About Arthropod Appendages
One thing you can say about arthropods, is that they’ve got legs. Oh boy, do they have legs.
And other appendages. They have crazy numbers of legs, wings, antennae, various mouthparts, you name it. It makes us poor tetrapods, with only four main appendages, look kind of short in the limb department.
Which brings up the question, what is it that makes arthropods so appendage-happy? It’s almost as if their easiest evolutionary response to any change in their environment, is to sprout a couple more limbs.
And, in fact, it looks to me like that is exactly what happened, at least early on in their evolution. Arthropods are basically modular – a string of repeated segments, with each segment having a couple of appendages on it. And, at least in the early arthropods, it was a relatively minor and harmless mutation for them to just add or delete segments, which meant that they could quickly run the numbers of appendages up and down in response to whatever environment they found themselves in. The huge variety of arthropod forms then comes from adding segments, repurposing the appendages on specific segments to do specific things, merging segments together to get concentrations of appendages in one spot, or deleting appendages from selected segments where they are not needed.
Starting at the Beginning
This is all easier to think about if we return to the probable common ancestors of the arthropods, probably living about 1 billion years ago, well before the Cambrian period.
Let’s back up the family bush to well before the first insects, or even the first segmented organisms, to the ancestor of everything that resembles arthropods at all – the first Ecdysozoan, with a stiff armored skin that had to be shed periodically for them to grow (their key trait). This early form was probably basically just a stubby cylindrical creature with a mouth, and maybe light-sensitive patches near the head end.
Some of the Ecdysozoans pretty much stayed with this body plan, becoming long, thin, and practically appendage-free. These were the Nematodes, which have since diversified into perhaps as many as a million different species, and closely resemble worms. This is a good approach in some ways, as it has kept molting pretty simple for them. But, there are limits on what you can do when you are basically nothing more than a limbless tube, and so some of the Ecdysozoans took another path.
First of all, at some point the ancestors of arthopods developed both appendages and segmentation. Since we don’t have fossils of this particular ancestor, we don’t know whether it was the segments or the appendages that came first (but for our purposes, let’s assume that it was appendages). We also don’t know what the function of these appendages might have been. They may very well have had combined functions, acting as sensory organs (like antennae), grasping organs (like tentacles or mandibles), and locomotory organs (like legs) – and while they were better than nothing, they were probably stumpy, uncomplicated, and not all that good at any of those tasks.
Now, consider the body length. There are several reasons why a long body might be preferable to a short one. The longer you are, the longer your digestive tract can be, which makes digestion more efficient. Also, if you are moving through water, a long body is more hydrodynamic, allowing faster movement. And, if you are a burrower, a long body gives you more ability to push yourself through holes. Rather than getting longer by simply stretching out a single body unit like nematodes did, the ancestors of arthropods essentially got longer by duplicating the body, and tacking the duplicate onto the end of the “original” body. The overall progression was probably something like this; from blob, to one pair of appendages, to multiple appendages:
This duplication can occur through a a few types of simple mutations; a particular gene sequence gets copied twice during reproduction instead of once, resulting in two copies of the gene in the offspring where the parent had just had one; or the same gene sequence gets called multiple times instead of just once. Then, as the animal develops, a sequence of events that had formerly only happened once, now happens multiple times. If the sequence of events is “define a body and start it growing”, then having multiple copies of the sequence results in “define multiple identical bodies stuck together, and start them growing”.
A key point is that this is not creating a lot of new information, but it is creating a lot more animal body with a lot more scope for future variation. Basically the only new information coming into existence at this point is “take that thing you grew once, and grow it again”.
This results in a segmented body , with initially-identical pieces strung out to produce greater body length with minimal disruption to the animal. Which means that if the original body had two appendages on it, then all the subsequent segments will also have two appendages. We don’t have any confirmed fossils of this original common ancestor to all segmented arthropods, but based on the features that current arthropods currently have in common, I think it probably had an elongated, multi-segmented body, with each segment armored and having a pair of appendages on it. This lead to animals that we do have fossils for. These are the Lobopods, like Aysheaia, with the first two appendages reoriented and adapted for grasping and sensing, and the final two appendages becoming a bit rudimentary and devoted to purposes other than walking.
This trend of using the first two appendages for non-walking purposes was continued in later arthropods, and today, surviving arthropods mostly either use those first appendages for sensing their environment (antennae) or for grasping prey (chelicerae in arachnids). After the first pair of appendages differentiated, the same sort of thing began to happen to the other appendages. So the main arthropod lineages started to differentiate their limbs like this: (diagram mostly based on this table of which appendages correspond to what in various arthropod lineages).
red = antennae; green = walking legs; black = mouthparts (mandibles, chelicerae, and maxillae); purple = claws or other manipulators (pedipalps on spiders, poison claws on centipedes, grasping and pinching claws on decapod crustaceans); orange = gills/swimmerettes.
I carried it out to 15 body segments because that seems to be a typical number, but many arthropods have more or fewer. And a lot of those segments end up fused together so that it is no longer obvious that they are distinct segments (like the five segments that fuse together to make an insect head, for example. Or however many segments go into making the not-obviously-segmented spider abdomen)
One point that I think is very interesting is that when you compare spiders with insects, all but the very last pair of spider legs correspond not to insect legs, but to the various appendages that insects have on their heads. And the centipedes are even more distinct from arachnids, with none of the centipede legs being homologous to arachnid legs!
The diverse nature of their limbs is apparently one of the things that has made arthropods so fantastically successful. Their large number of appendages gives them the ability to repurpose some of them for very specialized tasks (like shell-crackers, sensory organs, shovels, piercers, grabbers, wings, gills, jumpers, or even sacrificial decoys) while still retaining enough “conventional” limbs for prosaic purposes like standing up and walking around. This more than counters the downside of dealing with all those limbs every time they molt.
I could go on at much more length about arthropod limbs (and, in fact, other people have), and cover topics like the probable origins of insect wings, the debate about whether or not the original arthropod had biramous or uniramous limbs, or even stick in my oar about the Arthropod Head Problem, but I think that’s probably enough for now. Maybe next year.
 The Timetree of Life is a convenient way to see how long ago two types of organism actually shared a common ancestor. It looks like the various animal phyla, like Arthropoda and Chordata, diverged from each other around a billion years ago (give or take a few tens of millions of years), while the various arthropod subphyla separated about 600-800 million years ago. In general, the different biological classifications within the arthropod lineage have their last common ancestors spread out in time something like this:
Subphylum – 600 – 800 million years ago
Class – 400 – 650 million years ago
Order – 350 – 400 million years ago
Suborder – 275-300 million years ago
Superfamily – 230-240 million years ago
Family – 190-200 million years ago
Genus – 40-100 million years ago
Species – past few thousands to tens of millions of years ago
 When you think about it, the need to molt their outer skeleton periodically in order to grow is a feature of arthropods that seems to be at odds with all those appendages. They have to pull all of those fiddly appendages out of their old skins every time they molt, and they molt several times (usually around five or six times). If you ever watched an arthropod molt, this is clearly a tremendous hassle, with good odds of a limb getting bent, broken, or pulled off in the process. Molting goes a lot easier if you don’t have limbs, like snakes. So all these appendages really do seem to make an arthropod’s life more difficult than it seems it should need to be. The tradeoff is that those limbs are really very useful for doing a lot of things, so even with the difficulty they are still better off with them than without them. A lot of insects have come up with an inspired solution to this problem; for most of their lives they are things like caterpillars or grubs or maggots, with greatly reduced and simplified limbs compared to their adult forms. They save the fancy bits, like long legs and wings, for the final molt so they only have to deal with extracting them once.
 I was going to try drawing these, but then realized that modeling clay would be easier to work with. Except that our modeling clay was stuff that I’d had for close on 15 years, and it was like a rock even after warming it up. But, it turns out that you can re-soften modeling clay by working a bit of mineral oil into it. Baby oil works nicely. And as a bonus, it gives your modeling clay a nice perfume!
 For a more detailed (but still written for non-geneticists) explanation of how this actually works, see the book Endless Forms Most Beautiful, by Sean Carroll. If I understand correctly what he writes about arthropod development, it works something like this:
The genetic code for animal development is somewhat hierarchal, with a set of “toolbox” genes whose function is to organize when, where, and how the other genes are expressed. These toolbox genes are fairly simple and are present with very little variation in all animals. The first set of genes lays out the body orientation. They organize the first few cells of the developing embryo into top, bottom, right, and left, basically instructing each of the cells as to what general part of the body they will form, and in the case of arthropods the body forms into a tube. As cells continue dividing and growing, a second set of genes (the pair-rule and segmentation genes) lays out the body plan. For arthropods, these define the body segments, with some subset of the Homeobox (Hox) genes switched on or off for each segment.
The Hox genes are switches that control (a) whether or not a given segment develops limbs, and (b) which developmental code will be used to produce these limbs. Carroll says that arthropods have ten different Hox genes, and each segment may have any number of them active. The thing that makes each Hox gene different from the others is that it activates a particular group of genes for appendage development. It is these developmental genes that have the large differences that lead to huge variations in appendage shape and function, the Hox genes simply control which set of developmental genes is going to be used for the appendages on each segment. So if a segment has Hox 1 active, then appendages grow one way; if it has Hox 2 active then they grow in a different way; and if both Hox 1 and Hox 2 are active, then the appendage grows in yet another way. So by having various combinations of the ten Hox genes, we can see that there are a lot of different kinds of appendages than any given arthropod can grow on a single body. In principle, given ten Hox genes, and 0, 1, or 2 of each of these genes active in each body segment, a single animal could produce at least 55 different kinds of appendage (or no appendage, for 56 possibilities). If one allows up to 3 Hox genes to be active per segment, then there are obviously even more possibilities. It is therefore straightforward for each body segment to have its own particular, unique appendage.
 A key point about DNA and how it controls living things: DNA consists of instructions for making stretches of RNA, which in turn make proteins. That is it. The RNA and proteins then have to do everything. When a gene is referred to as a “switch”, that means it is a DNA strand that makes RNA which makes a protein which then binds to other DNA strands in particular spots. The presence of this protein bound to another DNA strand then either forces the other strand to make its respective product, or to refrain from making it, or maybe even to select among several products that it could make. If that other strand of DNA is also a switch, then it in turn is responsible for turning other DNA strands on or off. So, the development process for animals appears to begin with a whole series of cascading DNA switches, turning portions of the DNA in specific cells on or off depending on their locations in the developing embryo. Ultimately, this causes the cells to differentiate from each other (nerves, muscles, exoskeleton, digestion, other organs, etc.) and to grow in different patterns (long legs, short legs, wings, mandibles, etc.) depending on how their DNA switches become set.
This is also what makes cloning difficult in most animals. If you, say, take a sample of cheek tissue from someone with the intent of cloning them from it, the cells you collected already have all their switches set to define them as cheek wall tissue. If you cultivate them in a petri dish, they might survive and grow for a while, but it just gives you a lot of cheek instead of an embryo. To actually develop a new individual, you’d need to reset all the switches, recreate the environment of an egg to prompt the correct switch-setting cascade, and start over.
 The source table also included trilobites, but I didn’t include them as they used the same system as the hypothetical ancestral arthropod: a pair of antennae, and other limbs are basically identical walking/swimming legs. Which brings up a point: if they didn’t have any appendages devoted to being mouthparts, how did they eat? The answer is apparently that all of their legs sort of doubled as mouthparts: at the base of each leg, there was a jagged projection called a “gnathobase” that they could use to tear apart their food and carry it to their mouths. If humans ate the same way, we would have teeth on our shoulders and we would chew up our food by shrugging.
 As a matter of fact, a year after writing this page, I did review the various hypothesis about the origin of insect wings, if you care to see it.