Ungarbled version

Richard Pyle deepreef at BISHOPMUSEUM.ORG
Wed Oct 17 00:40:51 CDT 2001

> In other words, if author X says
> species A is in genus B of subfamily C of family D
> species J is in genus K of subfamily C of family D
> and author Y says
> species A is in genus B of subfamily C (same name but entirely
> different circumscription) of family E
> species J is in genus K of subfamily L of family D
> the assertion table would be
> name   parent
> A       B
> B       C
> C       D
> C       E
> J       K
> K       C
> K       L
> L       D

It's important to remember that the "Reference" is a key component to the
Assertions table. Also, although I didn't really clarify it in the Taxacom
post (but did in the private post), Rank is actually also part of the
assertion. So, using your example data, the Assertion table would look like
this (note that the "Rank" applies to the "Name", not the "Parent"):

ID  Cit.   Name  Rank      Parent
1    X      A    Species     B
2    X      B    Genus       C
3    X      C    Subfamily   D
4    X      D    Family      -
5    X      J    Species     K
6    X      K    Genus       C
7    Y      A    Species     B
8    Y      B    Genus       C
9    Y      C    Subfamily   E
10   Y      E    Family      -
11   Y      J    Species     K
12   Y      K    Genus       L
13   Y      L    Subfamily   D
14   Y      D    Family      -

Now, since I took the time to type all of this out, I'll use it to
illustrate several points.

First of all, each record in the Assertions table can be thought of as a
"Taxon sensu Author" record. So, in this dataset we have a "Species A sensu
X" record, and a "Species A sensu Y" record.  In this case, they're probably
more or less equivalent.  However, "Subfamily C sensu X" is not equivalent
to "Subfamily C sensu Y", because, as you stated, they represent quite
different circumscriptions.

Second, the reason that Rank is included in the Assertions table, rather
than the Names table, is that a given name can be used at different ranks,
depending on the circumscription (e.g., a genus name could be regarded as a
valid genus by some authors, or as a subgenus of another genus by other
authors, e.g.:

ID  Cit.   Name  Rank      Parent
15   Z      B    Subgenus    M
16   Z      M    Genus       C


Of course, there are any number of additional references with their
respective assertions about the same set of names, that can be added to the

> and the "assertion table" will not be able to reconstruct the fact
> that one should ONLY classify species A as belonging to family E *if*
> one defines C the way author Y does.

Sure it can, you just need to add one more record to the Assertions table,
which links genus B either directly to family E, or to a different subfamily
(other than C) that is placed within family E.  For example:

ID  Cit.   Name  Rank      Parent
17   T      B    Genus       E

Can't find a reference "T" that does this?  Then create one: Yanega, Pers.
Comm.  Seem too "artificial"?  Well, you're the one that set the parameters:

1. A is species of genus B
2. B is a genus within family E
3. C is a subfamily of family D (assuming that we're *not* following author
Y, as you suggest)

Well, given these premises (which represents the situation of your concern)
then neither of the available assertions about the status of genus B that
you provided are logically possible, given the 3 presmises above.  If you
want A to be in genus B, and you want to follow X for the subfamily C, then
you must have *some* reason why you want to put species A within Family E,
so there must be *some* assertion out there (even if it's your own) that
connects B to E via a route other than C.

For clarification to others following this thread, as I detailed in the
"Details" post I sent you, there is a field in the NAMES table that points
to the "correct" assertion; that is, the "accepted status" of the given
name. So, the NAMES table for the above would be as follows:

Name  CurrentStatus
 A         1
 B        17
 C         3
 D         4
 E        10
 J         5
 K         6
 L        13
 M        16

The value in the "CurrentStatus" field corresponds to the ID field of the
Assertions table, which means that we've decided to follow author X as
representing the "correct" circumscriptions for the names A, C, D, J, and K,
but we'd follow author Y for the correct circumscriptions of names E and L
We'd follow author T for name B, and Z for name M.

There's an entire thread of conversation that could take place eith respect
to whether the "Parent" field should point to the "Name" table, or th
"Assertion" table.  It's a complicated question, with fundamental
implications, but as this discussion is long enough on its own, I'll save
that thorn for another time.

> Further, it isn't fair to list
> both D and E as parents to C, because in neither author's
> classification can species J be considered a member of family E (that
> is, not all children of C are grandchildren of E).

I think I've addressed these in the examples above, but if not, please let
me know. It all depends on which Assertions you've chosen as "correct" for
each taxon name. There are rules to deal with inconsistencies, but one would
assume you would refrain from chosing inconsistent sets of assertions to
represent a particlar classification.

> Your designation
> of the "preferred" links certainly would help, but what happens when
> there are thrid and fourth and fifth authors' opinions that also use
> the same names in different ways?

See below.

> What would happen is you'd lose
> track of which authors' schemes grouped things which ways.

I think that's the fundamental point of miscommunication.  We keep track of
*all* authors' schemes independently. That's the objective part.  The
subjective part is erecting a "correct" classification, which involves
designating one assertion for each name to represent the "correct" current

> This kind
> of thing does actually happen. I still do feel that you can't
> reconstruct a single author's phylogeny by giving simple lists of
> parent-child linkages,

You're right, that's why we have complex parent-child linkages.  So, if we
wanted to know something about Subfamily C from the above data sets, we'd
see that we have a choice between two status options:

3: C sensu X, as a valid subfamily of family D
9: C sensu Y, as a valid subfamily of family E

To illustrate further the potential of this scheme, I'll add the following
additional options to the Assertions table:

18: C sensu Q, as a valid family of suborder N
19: C sensu R, as a junior synonym of family O
20: C sensu S, as a junior synonym of subfamily L

Thus, for the Taxon Name C, you'd want to select one of the five available
alternative Assertions about the status of C (3, 9, 18, 19 or 20) to
represent the "current status" of taxon name "C".


Richard L. Pyle
Ichthyology, Bishop Museum
1525 Bernice St., Honolulu, HI 96817
Ph: (808)848-4115, Fax: (808)847-8252
email: deepreef at bishopmuseum.org
"The views expressed are the author's, and not necessarily those of Bishop

More information about the Taxacom mailing list