Security reasons.
The token that replaces your actual card card number for the transaction (the "Device Account Number") is known only to your iPhone and the issuing bank. Therefore, each bank has to not only sign up, but modify their back end processing to support it.
Logistical and acceptance issues aside, this is a good thing. CurrentC, by comparison, only tokenizes the data as far as the merchant, where your bank account data will be stored. These are the same stores that in 3 cases so far have public breaches; the banks thus far have a better record. Not spotless (see JP Morgan Chase), but not as many yet-publicized breaches. Further, having the data revealed only at the bank (at the end of the payment chain) limits the amount of potential intercept points that hackers can target.
So it's far from a seamless integration, to be sure, but at least it doesn't require the merchants to do much more than they should already be contemplating already, especially since most have to replace their existing terminals with EMV-capable terminals next year anyway.
I predict slow but steady adoption in the beginning as long as there aren't any vulnerabilities discovered in Apple Pay. If everything works well, expect more rapid adoption by banks and merchants. Some may be fence-sitting and waiting to see how it works, which isn't a particularly bad idea when you think about it.