I believe that I set this standard, when I named the individual image files on the English Wikipedia a few years ago. It is somewhat arbitrary, but I think that it's a good system (particularly distinguishing the consonant "j" from the vowel "y"). However, variation is certainly possible.
In origin, the "j" is not so much German as Latin. Latin also uses "th" this way— but not "ch". More than anything else, however, the transliterations follow Croatian, except with "h" to indicate a modified letter, which is the only part borrowed directly from English. (For example, Croatian "Ćerv" becomes "Cherv".) The name "thita", on the other hand, definitely follows Latin, as a transliteration of Greek; indeed, this letter was only ever used to transliterate Greek "θ" and has no analogue in (Latin-script) Croatian.