I presume it was not d but rather dh. The Indo-European aspirated consonsants bh-, dh-, gh-, gwh- became f-, f-, h-, f respectively at the beginning of a word. This is important because Latin was always stressed on the first syllable during the earliest stages of its history, it was only during the Late Archaic, Early Classical periods that the stress was moved to where it was during Cicero’s times. As for the reasons for this process I would bet on the spirantization triggered by the presence of the aspiration. The Spanish intervocalic -g-, -d-, or -b- (as in pagar, nada, or pruebo) are not exactly pronounced as in English, French, German or Italian, even though I am not sure why it happened in Spanish. In the pre-written stage of Latin these sounds were possibly pronounced in a similar manner (e.g. d- could be pronounced as the English voice -th- in bathe), and it is not really that far to v- and f- from there (‘drink miwk in Suvvern London’). The fact that this sound change was not triggered in other positions within the words can be accounted for by the word-initial stress that ‘drew’ articulatory focus away from the consonants. Anyway, this is all conjecture 
2 Likes