This may not be much help, but I just noticed that on the 1881 Canadian census, both before and after the household of Dominick Surat, are households of Melancons. If you can accept the spelling variations, in my experience the names that seem to be related and are directly next door are, in fact, related. Does it seem like a stretch to think that Melancon and Melanson are the same name?
Also, I know Wikitree stresses that one shouldn't create duplicate profiles, but (if you believe that the family on the 1881 Canadian census IS the family you are working on), then you can use the estimated year of birth (1825) and use the "estimated/uncertain" option for the birth date. You can use the census for the source.
If you start to create a profile for Marie (AKA MARY) as spouse of Dominique/Dominick, while you are entering information for Marie, if the Wikitree system finds other profiles that seem similar, those profiles will pop up as ones you should closely compare with the information you have. That way they are presented to you methodically, and you can either reject them or-- if you think you find a match, accept the proposed match as your Marie.
Don't be so afraid of creating a duplicate that you completely prevent yourself from progressing. Sometimes the possible matches are extremely vague, with no (or few) dates, places or family members, and there is no way to tell if they represent the same person or not. In that case, continue with your own profile, and just don't completely reject the profiles that are too vague to be sure aren't a match.
If you DO unintentially create a duplicate profile, it can be merged later, when it is possible to make an informed comparison.
That's what I would probably do. I wish I could be more help. If you want to get some other opinions, you might want to "answer" yourself again to push this question back up to the top of the G2G feed and see if anyone else has differing, (or more helpful) advice.