I take those concerns seriously, so I wish to respond to them. I will start with the latter point first, as it is easier to argue.
(1) In the foundation charter of the Lordship of Monmouth, for the Breton minor aristocrat Withenoc de la Boussac (see e.g. Wikipedia), King William is named but not present. Representing him is Alan Rufus ('Count Alan'). The only witnesses stated as present were Withenoc, his brother Baderon, and Count Alan.
The foundation charter of the Priory of Monmouth has the same three witnesses present.
As to Wikipedia articles, like any source they are as reliable and as comprehensive as the writer makes them. Furthermore, Wikipedia's prohibition against citing primary sources (such as charters) inhibits it from being more thorough.
(2) NPE rates are not uniform across all lines of ascent. Not to put too fine a point on it, royalty tends to have wives for breeding, and wetnurses for raising the children. That means that the wives have little time or opportunity to stray, and they bear so many children that a 'changeling' would immediately stand out.
Another example is the O'Driscoll lineage in Ireland, which reputedly dates back over 5200 years. NPEs have certainly occurred, many of which are due to adoption (a common event in, for example, both Roman and Russian history also). Still, over 60% of all O'Driscoll males have the rare I2a Y haplogroup, while the remainder have the form of R1b-L21 that is overwhelmingly typical of the Irish population and in particular of the O'Neill and other families who self-identify with the Eireann (non-Gaels) and are thus closely connected with the O'Driscolls.
You will be right to remark that this I2a haplotype could also be due to an NPE during the long lapse of time.
However, recent DNA analysis at Newgrange, the legendary resting place of the O'Driscoll male-line ancestors, has shown bodies buried there to date to about 3200 BC and to have I2a Y-chromosomes. This demonstrates that over 60% of O'Driscolls descend from the genetic 'stock' that produced the Newgrange bodies.
The moral of this is that NPE can be much less than the quoted rate of 2% per generation in some lineages. For the O'Driscolls it must be closer to ((5200/100)*3 = 156 generations, with 40% accumulated NPE from all causes, so (0.4/156)*100%) = 0.25% per generation.
If the Breton ducal lineage is more like the O'Driscolls, then over 2600 years we would expect a 20% accumulated NPE effect, and 80% of its living descendants to be in the 'official' male line.