Hi, Chris. On the face of it, my guess is that's the only scenario that explains the results: the two daughters' X-chromosome segments that match your dad were inherited from their father. But the X-chromosome has something of a mind of its own, and what we see is not always what we expect.
I started to dive into stuff about the X-chromosome being something of a SNP desert, with relatively few tested SNPs along its length of 156 million base pairs, and the volatility of the pseudo-autosomal regions at the start and end of the chromosome, and what happens if you phase a male's kit at GEDmatch, and yada yada. I'll save it for a blog post. I get gripes about the length of my G2G stuff as it is. But Rob would have beat me to it, anyway. ;-)
To cut to the chase, here's an important observation from an 18-month-old research paper, "A Genealogical Look at Shared Ancestry on the X Chromosome," from the journal Genetics: "Unfortunately, the X chromosome is short, such that the chance of any signal of recent ancestry on the X decays rather quickly."
There's a lot of information out there that indicates segments of the X can be quite old, that they can pass down for multiple generations--more likely to occur the more males, proportionately, are in that particular line of inheritance--and that a seeming X-chromosome match has the possibility of not aligning with a suspected MRCA at all, but many generations earlier. Conversely, I've never seen any peer-reviewed studies that indicate any degree of assurance of genealogical matching using the X except in a limited fashion. Opinions differ widely.
Roberta Estes wrote just last month, "In general, due to the thin spread of SNPs on the X chromosome, you will need, on average, a cM match that is twice as large as on other chromosomes to be considered of equal weight. In other words, a 10 cM match on the X chromosome would only be genealogically equivalent to approximately a 5 cM match on any other chromosome." And my personal opinion is that the Poisson approximation used in that Genetics paper is likely too generous when modeling possible ancestry; it doesn't adequately take into account that the result of crossing over during meiosis in the X doesn't behave as predictably as do the autosomal chromosomes.
Bottom line, do these results mean that the X-chromosome segment the two daughters' show as a match to your dad mean that the segment was inherited from their father? Yep; almost a certainty.
Does it mean that the daughters' mother and your dad are related? Maybe...but not closely. That 16cM amount, if valid IBD, would probably be around the 5th cousin level, so MRCA around 4g- or 5g-grandparents. But the segment might be IBS and not valid; triangulation work would be required to validate.
Does it mean that the daughters' mother and her husband are related? Maybe, but probably not. I did a check for runs of homozygosity on GEDmatch against the daughters' kits, and there are some possible, tiny half-IBD or half-IBS areas, but none at the 7cM threshold. If there's a relationship from pedigree collapse, it's too far back to be determinable from the current DNA data.
Does it then mean that your dad and the daughters' father are likely to be recent cousins? Nope; doesn't imply that at all. If the father's test results become available, I'll bet that segment in the X is apparent...but it may be just about the only thing that's shared; shared autosomal DNA segments would be needed corroborate. The two men may have a common ancestor, but she may be many generations ago, perhaps before the genealogical timeframe; the segment may be an artifact that's been transmitted down the line for generations. That's why--despite how valuable and useful the X is in DNA analysis for genealogy--none of the major testing companies report matches based solely on the X-chromosome.
I think these words of wisdom from Jared Smith are a good go-by for dealing with the X-chromosome:
- "Use the amount of autosomal DNA match to determine relatedness, and X-match data to narrow down to particular family lines."
- "X-matches that do not share measurable autosomal DNA would generally be distantly related."