The sequences of SARS CoVs and other CoVs are readily available and there is no doubt that the differences between SARS CoVs and CoVs that cause colds and other respiratory diseases in animals are significant (with 20-30% homology). In contrast, the SARS CoV from masked palm civets or raccoon dogs are 99.8% homologous to the SARS CoVs from patients. These animal sequences represent the animal reservoir. Although it is possible that the sequences of virus isolated from live animal markets in Guangdong Province may be one step removed from the true reservoir, the sequence shows that any major recombination happened thousands or millions of years ago (which is why the homology with other CoVs is only 20-30%).
Antibodies in humans indicate that the jump from animals to humans was VERY recent (within the last year or two) and long after major recombinations among well studied animal CoVs. Moreover, animal traders with the highest frequency of SARS CoV antibodies are those that sell masked palm civets.
Thus, at a minimum, the civets facilitated the transfer of SARS CoV to humans, which is further supported with sequences from individual civet isolates.
I don't understand why the sequencing gives the date. I know glottochronology in linguistics depends on a demonstrably false assumption that languages change at a given rate.