Here are some interesting theories that I found online. Not sure if any of these are true, just something to think about. I did NOT write these theories.
1.) 'The Mix Theory'.... This is the most popular theory right now. This theory puts forth that merle is the result of the APBT being bred (mixed) with another breed. The breed most pointed to is the Catahoula Leopard Dog (CLD). There may be some validity to this theory, since it is known that rural hunters still today cross APBTs with CLDs to get a better hog hunter. The CLD was also crossed into the American Bulldogs. This cross began a breed called the Catahoula Bulldog, or CatBull for short. Alaphaha Blue Blood Bulldogs (ABBB), is a bulldog breed that also comes in merle.
Now, it is well known that in the early 1900s that APBT breeders would breed to any dog that was game. APBTs were crossed to English Bull Terriers, Boston Terriers & StaffyBulls, among others. It is completely plausable that if a game merle CatBull, ABBB or CLD/APBT mix were running around, that an APBT breeder out there past or present may have bred to it, thereby introducing merle into the APBT breed in the process. If the cross occured with a CatBull or ABBB, it may be unnoticeable since both the CatBull & ABBB are similar to the APBT in appearance.
2.) 'It's Always Been There Theory'.... This theory put forth that merle has always been in the APBT, but went unnoticed. This theory also has some substance to it. There are certain types of merles called "cryptic" or "phantom" merles. These are dogs are genetically merle, but appear to be non-merle. This happens most often in dogs that have a yellow-type coloration. It is already known that sable can "mask" merle. The reason is that merle acts on strongly on eumelanin & weakly (if at all) on phaelomelanin. Eumelonin is dark pigment (i.e. black, brown). Phaelomelanin is light pigment (i.e. yellow, red). This explains why a blue merle dog w/ tan points has merling on the parts of it's body that would otherwise be black, but the tan points remain un-merled.
Since merle does not affect phaelomelanin, an APBT that is yellow or cream in color can appear non-merle. However, these phantom merles are still genetically merle dogs & will produce merle pups. Phantom merles usually produce phantoms like themselves, so it is conceivable that merle could go unnoticed in the breed for generations.
OK, I can hear it now -- if it's always been here where did it come from? Either side. The Alaunt, the war-dog ancestor of the APBT is likely. The APBT shares several cousins through Alaunt blood, one being the Great Dane. Great Danes come in fawn, brindle, black, blue & harlequin - all colors that are also found in the APBT.
But on the flip side, there are also merle terriers. The Atlas Terrier is most notable & it coincidentally goes back to many of the same terriers the APBT does. So the merle allele could have come from either side of the house.
3.) 'The Spontanious Mutation Theory'.... This theory says that merle occured in the breed as the result of a spontanious mutation. Like the other two, this theory also has some merit. Merle is known as an unstable gene & it has been found that about 4% of the time merle mutates & for some reason it seems to be connected to the color black. All creatures exprience mutations & anything can cause it - stress, poor nutrition, etc. Bulldogs being a breed that suffered much from both stress & poor nutrition in it's early days, it would not surprise me if merle was the result of a mutation. In 99' one gamedog breeder had a blue merle pup show up in a litter between a sable bitch & a black male. I don't think this pup was the result of a either parent being phantom merle since he had bred these dogs before & never had a merle either before or since. He was also a veteran breeder of 20+ years & never had a merle out of the line he was running until that day. I believe the pup was the result of spontanious mutation.
(The original link is http://www.geocities.com/abkapbts1/merleorigin)