The gemara says that R. Abba demanded that when two mules are tied to a wagon, they must both have a mother who is a horse and father who is a donkey, or the reverse. But if one would have a mother who is a donkey and the other a mother who is a horse it would be an issur of kelaim since we are machmir for the opinion of R. Yehuda who holds that we disregard the father and consider the child to be the species of the mother. However, R. Abba allowed them to figure out the species of the mother using the simanim of a short tale and long ears being indicative of a mother who is a donkey, and the opposite being indicative of a mother who is a horse. The gemara comments that his willingness to rely on these simanin indicate that he holds simanim are d'oraysa because otherwise he couldn't rely on them for kelayim which is essentially an issur d'oraysa. Rashi comments that whether simanim are indeed d'oraysa is a discussion in baba metziah and R. Abba would be holding that they are d'oraysa.
The Ramban asks that Rashi's connection to the gemara in Baba Metzia regarding identification of either an object or a dead husband, doesn't seem to be at all related to simanim in this context. The issue of using simanim of identification and questioning whether they are "d'oraysa" is because they aren't 100% conclusive since it is possible that another object may possess these simanim as well (whereas a siman muvhak that is conclusive is certainly d'oraysa). However, in our context chazal seemed to have a tradition that every child born from a mother donkey will have a short tail and long ears, and every child born from a mother horse will have a long tail and short ears. To rely on simanim in our context because they are d'oraysa, meaning a tradition from moshe m'sinai that they can be relied on for kelayim, wouldn't be any indication of simanim being d'oraysa when used as identification of an object or person.
It seems that the connection of Rashi forces us to say that the simanim used to identify whether the mother is a donkey or a horse are not a tradition from Moshe Mi'Sinai. Rather, Chazal must have noticed that statistically speaking one could safely assume that when the mule possesses simanim in the ears and tail of a donkey, it's mother is a donkey, and if it possesses simanim of a horse, it's mother is a horse. These aren't absolute, rather statistically correct. Therefore, Rashi holds that the simanim in our context are also identification type simanim rather than being absolute indicators of the species of the mother.