The set of shared DNA amounts measured do not seem peculiar to me, even though they are not hitting the exact average amounts expected for the relationships shown. The highest variance from the average is for the relationships between Don and his three third cousins, but even there the amounts measured fall within the expected range that the Shared cM Project reports.
DNA inheritance is an extremely random process. When we use a summary statistic, like total cM matched, most of the variation present is hidden from us. This is good as it allows us to think about possible relationships easily, but it isn't adequate when we want to understand the details of what may have been inherited.
To understand what sort of thing is going on in the present case we need to explore deeper and think about what particular DNA segments may have been inherited; a total cM number isn't adequate. And in thinking about the process, it's important to remember that DNA isn't like sand or a liquid where a representative, evenly allocated half of a parent's DNA is passed down to the next generation. The DNA comes in chunks (segments), and one half of the somewhat large chunks are not passed down, potentially eliminating entire remaining segments from earlier ancestors (that is, not just passing down half-sized sub-segments of previously inherited segments).
Consider a halfway-plausible segment inheritance suggested by the capital letters shown in this version of the relationship diagram:
The above is showing Jenny as inheriting seven segments from Harold & Maude, namely A,B,C,W,X,Y, and Z. Don is shown as getting A,D,M, and for the three closer cousins: Teresa getting M,V,X, Dennis M,V,Y, and Jeremy V,W,Z.
So Jenny shares segment A with Don, X with Teresa, Y with Dennis, and segments W,Z with Jeremy. These segments (or collections of segments) are representing the inheritance of distinct pieces of DNA which total the 107 to 238 cM measurements of Jenny's sharing with each of the other tested people.
You'll see I've shown only one segment, M, being shared in common between Don and any of his third cousins, and shows that one of them, Jeremy, shares no segments with Don. This represents one possible way that a particular random arrangement of segments might have been inherited to cause the measured total shared cM numbers actually measured, 0 to 18 cM.
We can't completely know what segments the earlier generations had, but we can identify some segments that they must have had. Jenny's ancestor Betty must have also had all of Jenny's segments, ABCWXYZ, and Don's ancestor Fred must have had all of Don's segments from Harold & Maude, ADM.
Bill, the ancestor of the group of three cousins, must have had all of the segments that his descendants received, which would be MVWXYZ.
The foregoing is just an illustration of the sort of inheritance that must have taken place. Actual details might be available from the data the DNA testing company may provide. At least some companies (e.g. 23andMe) give identification of the segments that match. This takes the form of showing the chromosome, position, and length of each of the segments that matched. It is the total of all such segment lengths (in cM, not bases) that give the summary statistic of the total shared cM value.