Salvete!
My understanding is that with Filius/Filia (Son/Daughter) the use of ‘Filiabus’ is used in the dative and ablative plural to differentiate it from the masculine ‘filiis’.
However what happens to other nouns like Discipulus/Discipula? For both of these the dative and ablative plural is discipulis For example a sentence such as “discipulis pecuniam do”; how do you know if the students are male or female? 
Valete
Do you have examples where “Discipula” is used in the dative or ablative and causes ambiguity? I take it your “discipulis pecuniam do” is one you made up and so it doesnt really help. Real examples without any context would show that there was a problem. I would have thought that, unless ambiguity was intended, the context would have made it clear whether it was men or women.
My guess is that the alternative ending for filia is because filia and filius are common words in everyday speech and latin would have evolved to overcome any possible ambiguity.
I’m really disappointed that both Neue-Wagener and even ThLL are pretty unhelpful on this question. Or is their unhelpfulness due to lack of suitable evidence? It would seem, though, that no *discipulabus are attested in ancient sources, since they don’t cite it. That is of course if we rely on the aforesaid authorities.
Discipulus apparently has once a genetive plural discipulum, so we (or I at least) have learnt something… 
Thanks Timothée I just quickly looked at the entries in L&S and couldn’t find a dative or ablative pl.
I don’t know about ancient sources, but in Renaissance texts etc it’s a fairly common occurrence to generalize the “filiis/filiabus” pattern to all nouns where confusion might occur. So that you’ll see discipulabus, magistrabus, etc.
The merging occurred because of phonological developments:
*-āis > *-ais > -eis > -īs
*-ōis > *-ois > -eis > -īs
-ais is attested in Oscan, -eis in Archaic Latin (mainly in inscriptions). At least the following per se ambiguous plural dative-ablatives are contextually attested referring to the feminine: famulis, captiuis, amiculis, sponsis, Bacchis, asinis. Ennius’ Andromeda (about Nereids): “Filiis propter te obiecta sum innocens Nerei.” Ovid’s Metamorphoses: “Euboea duabus et totidem natis Andros fraterna petita est.” Animis (f.) is the most common of the an sich ambiguous dat.-abl.
-ābus developed (at least partly) on the influence of duabus and ambabus. It first appeared to disambiguate between word pairs particularly in wills (filiis filiabusque, liberis liberabusque). It is mostly used in words denoting women. Dis deabus(que) go early together (Cicero). These three are the most common -abus words.
Words in -abus include: deabus, filiabus (suabus/gemellabus), libertabus, [g]natabus, Cassiabus, pupillabus, amicabus, feminabus (inconfusable!), alumnabus, equabus, asinabus, mulabus, animabus. Provincial goddesses include (influenced by deabus) Dominabus, Siluanabus, Nymphabus, Fatabus, matronabus, matronis Aufaniabus, Gauadiabus, Glanicabus. Parcabus. Atiliabus. Suleuiabus. From 3rd declension there are matrabus and Caelestabus.
There are also following: eabus, ex raptabus, cum aliis paucabus, pro duabus pubicabus, manibus dextrabus, portabus, oleabus, uillabus, horabus.
Priscian also says: “Incipiemus dicere Romanabus, si enim dixerimus Romanis, masculinum sexum intellegimus.” Donatus agrees.
We also have: ipsabus, istabus, mimabus, ursabus, famulabus, diuis diuabusque, conseruabus, alumnabus, agnabus, ceruabus, puellabus, monachabus.
This list is quite comprehensive—I should think!—and it has nothing on discipula. So what should one do? I suggest that you could write discipulis if the context makes it clear that you speak of discipula. If you combine it with discipulus, you should probably write discipulis discipulabusque.
This message does not include post-classical attestations (except discipula, obviously). I collected these from Leumann and NW mentioned earlier in the thread.
The faux-heteroclite filiabus looks perfectly monstrous to me, but lawyers will go to any lengths to avoid ambiguity, and no doubt it was used (at least originally) only in circumstances where gender discrimination was crucial. Same with deabus, of course, which it’s on a par with. “Gods and goddesses” often go in tandem, esp. in ritual formulas, and given e.g. deorum dearumque (and duabus vs. duobus), how to form dat.-abl. other than by deis(/dis/diis) deabusque? Filiabus is to be explained similarly. Donatus’ Romanabus is an artificial and howl-worthy analogical extension.
discipulum for –orum, cf. e.g. deum, liberum, etc.
Morphological developments often come about by analogy. But I’ve been taken to task for saying that analogy can motivate linguistic change, and I don’t quite understand why. In this case of filiabus I suppose the motivation could be said to be the desire for gender disambiguation, but in more ordinary cases of analogical leveling (generalization of weak aorists in Greek, say, or of weak aorist endings) is the objection simply that analogy in and of itself has no explanatory power, that the motivation is rather the impulse (or whatever) towards regularization, and analogy merely a means of effecting it? If that’s it, it seems a rather pettifogging objection to me, but I’m not much of a linguist, and maybe the point was something else.
EDIT. I think I get it. Formations like puellabus, for instance, are not motivated by analogy—nothing is motivated by analogy—, they’re motivated by a felt need for a feminine declension distinct from the masculine(/neuter) throughout. (But I’m guessing the feminine-looking endings of 1st-decl. masculines—all of them agent nouns, nomina agentis?—were still acceptable.) Yes? So I withdraw pettifogging (but isn’t it a lovely word?). The OP’s heading was spot on.