Thanks colleagues, foor all your reactions and advice.
There is a lot to discuss about this subject, that is clear. I am no
specialist in this matter, but my intuition that the local water must
be key player, is kind of supported through several of the comments.
After all, it needs not be water that is directly consumed, but water
that is conatained in the food (grass, vegetables, even grain
contains a large fraction of water that is close to the local
Meteoric water). Only under real arid conditions the water produced
in the food "combustion" process may play a dominant role.
But let me just be more specific about my data.
The last two cases I had at hand were wild Cape gannets (South
African West Coast and Namibia) and Dutch laboratory mice. First the
I expect their body water to be pretty close to sea water. After all,
they eat fish, and fish have no way to escape a VSMOW-like imprint on
their body water and tissue (seems to me, at least!). And indeed, as
far as 18O is concerned, a total of 20 background values of a total
of 16 individuals show a de18O of close to zero, with a spread of
about +/- 1 per mil. de2H, however, shows about half of the values
close to zero, and half considerably above zero (the highest being up
to almost 100o/oo !). Looking closer (after the suggestions on
Isogeochem) I saw that some of the duplicate mm show one de2H close
to zero, and one off by 20-50 per mil.
The laboratory mice samples show a similar behaviour: de18O at about
-4 per mil (Dutch tap water is around -6 per mil, as well as all the
regional vegetation their food is supposedly coming from), de2H at
about -25, but with some large excursions to more positive values.
So after all, I think Dachun Zhang's observation of some kind of
H-exchange in stored blood might be the answer. The samples we
process are taken in the field, stored in flame-off glass capilaries
and sometimes take many months before they are analysed. The analysis
starts with microdistillation, but if exchange has been going on
before that time, the damage has been done. As it is so variable, it
might have something to do with the leaktightness of the capilary?
It is something to worry about in the field of Doubly labelled water
for energy expenditure, though, since it can just as well affect the
Marilyn, as for your nice series of background values on humans, I
suggest you contact Klaas Westerterp (university of Maastricht,
Netherlands, <[log in to unmask]>), who has an impressive
track record of using DLW techniques on humans. My guess is he is
interested, and he can tell you in how far your series contains
Thanks once more, and if anyone has a concrete idea about a possible
exchange process, I'll be eager to know.