The focus is on dissolution of first unions, and Ménard displays the data as a survival graph, showing what fraction of unions have not dissolved after a certain number of years. Below is the graph of interest, which is for Canada ex-Quebec (see here for breakdown of cohorts)
The three lines indicate the fraction of unions still together after a certain number of years of their start, assigned into three cohorts. These are first marriages only, not included are those of non-first unions. Look at the first five year time horizon -- even for the best-performing cohort ("marriage without prior cohabitation") the survival rate is 95%, or 5% of those marriages dissolve. After 30 years 25% of that cohort dissolves.
How does this relate to housing? Marriage dissolution is an event that is difficult to budget at the household level but causes significant impairment to the reduced ability to service mortgages that are dependent upon multiple sources of income. These data highlight a small part of the inherent risks of property ownership, especially through bouts of reduced liquidity and/or depressed prices, not necessarily related to credit rating. While those entering into housing investments are doing so with the intention of never needing to sell, the data paint a different picture of what the reality will be for some.