While I suggest that Mary Ann Dixon might be Henry and Margaret Dixon's daughter, I also noted:
"There was also a Mary A. Dixon b. ca. 1824 living in St. Stephen [in 1871], but she seems unrelated. For example, she was Church of England, single and English."
Mary A. (47) lived with John (44) and Caroline (40) Smith in 1871. Perhaps Caroline was a sister. Miss Caroline Smith of St. Stephen Parish m. 12SEP1870 John Smith. Again, if both Caroline and Mary Dixon were 'of' St. Stephen Parish, then this suggests (in spite of the age difficulties) that they were Henry and Margaret's daughters. Even so, in 1851, in St. James Parish, Henry and Margaret have no such daughters. Nor can I find in the NB census of 1851 either Caroline nor Mary A. Dixon/Dickson.
John (54) and Carrie (50) Smith are in St. Stephen Parish in 1881. A daughter is Eveline Smith (10.)
Mary Ann age 48 (b. ca. 1833) was Howard's (probably 2nd) wife in 1881 as at:
This would also suggest Mary Ann was not b. 1MAR1824. Checking the image of the original, the surname is 'Douglass', 'Sarnbelle' is Laurabelle and otherwise the information is the same. The 1891 census of St, James Parish had Mary Ann Douglass a widow, age 67 years (so she was likely age 58 in 1881), b, NB and parents b. Ireland. Again, Henry and Margaret Dixon were b. Ireland. In 1891, in St. Stephen Parish, however, C. L. Smith (60), wife of John Smith (64) was b. NB ca. 1821 but her father was b. England and her mother b. Nova Scotia. This suggests that Mary A. Dixon (b. ca. 24 in 1871), sister of Caroline L. (Dixon) Smith and Mary Ann (b. ca. 1823-4), wife of Howard Douglas (whose parents were b. Ireland) are not the same person.
So you may be right, Henry and Margaret Dixon may not be Mary Ann's parents.
It may be a blind alley, but locating Caroline L. Dixon's parents might be useful. Also, the family seems to have used the spelling 'Douglass' fairly consistently.
Howard Douglass' daughter, Margaret, m. SEP1876 Warren Dinsmore.
Laura Bell Douglass m. St. James Parish 29DEC1883 Adam Dinsmore (13JUL1849-4MAY1932), s/o Adam and Elizabeth (Blair) Dinsmore. When Laura Bell Densmore (13AUG1860-26JUN1946) d., her mother was given as Mary Ann Nixon b. Ireland, her father as Howard Douglas (b. Scotland), and her husband as Adam. Rechecking the 1871 census, the family is:
Douglas Howard M 45 NB C of Eng Scotch Farmer M Douglas Mary Ann F 44 NB C of Eng Scotch M Douglas Sarah Jane F 19 NB C of Eng Scotch Douglas Mary E F 17 NB C of Eng Scotch Douglas George H M 15 NB C of Eng Scotch Douglas Margret C F 13 NB C of Eng Scotch Douglas Lonow ? F 10 NB C of Eng Scotch Douglas Sharlotte F 8 NB C of Eng Scotch
with 'Lonow?' apparently Laura Bell. Margaret is actually Mary Ann, and definitely (it would appear) not Caroline L. Dixon's sister. Also her maiden name may have been Nixon. With Nixon, there is a wider range of candidates for Mary Ann's parents. Notably the widow, Martha Nixon (b. Ireland ca. 1799) in St. James Parish in 1871. See: http://www.rootsweb.ancestry.com/~nbstjame/census/1871stjapt1pg22.htmhttp://www.rootsweb.ancestry.com/~nbstjame/census/1871stjapt1pg22.htm Martha Nixon (ca. 1800) was also a widow in 1851. Howard and Mary (b. ca. 1827) Douglas are already married (ca. 1843-4?) in 1851, so determining if she was, for example, Martha Nixon's daughter would be difficult. Perhaps probate records might be of help.
Martha Nixon d. Lynnfield, Charlotte Co., NB 29APR1885 age 85 years. She was originally from St. David Parish.