Dear All,
Completely agree with Rich’s
analysis.
May I coin two new terms? But before, I
think we should separate the definition of occurrences from their use, which
remove many questions evocated in previous messages):
Occurrence:
At least a triplet (a taxon name, a
location, a time); whatever the precision of each member of the triplet
is.
The difference is in only in the use that
we can do of an occurrence depending on the respective precision of each member
of the triplet. Ontologies by definition should reflect the patterns not the
processes (although technically I suppose that processes can be described by
ontologies … but it is an extension of the meaning of the word).
Name:
From “living organism” (extant
or fossil) down to infrasubspecific rank if needed. Can it be a common name?
Yes, it may decrease the precision or even the accuracy, that is all.
Location:
From Earth/continent/ocean/catchment down
to precise geocoordinates. Earth is always implicit and by default until we
find life out in space.
Time:
From 4.5 billion year range / geological
era down to a precise date/time stamp.
Here are the two new terms I propose (and
more could be coined using the same way):
Geoccurrence: an occurrence with
geocoordinates.
Loccurrence: an occurrence with only a
locality/geographic name.
Should we coin terms for occurrences resulting
from modeling?
Another consideration: species distribution
modeling is a rationalization of the production of distribution maps, just like
cladistics is a rationalization of the production of phylogenies.
For cladistics, in essence we sample
individuals in the real genealogic tree (= tokogenetic tree of Hennig): but can
we say that actually all characters used in cladistics lead back to a given
individual? Maybe true for molecular data but this statement needs more
thinking; I don’t think it is true for morphology, and it is the same way
for synthetic descriptions and older works, as Rich described as using all imprecise
old records.
Likewise for distribution, we sample individuals,
and also use the best of loccurrences based on imprecise location (cf. Jeremy
Jackson work on historical records and trends).
As for recording nativeness, I would
suggest that it is a general issue for all controlled vocabularies that try to establish
categories over a continuum: the only way to get rid of all these problematic
definitions, and most probably incl. occurrences, is to express them with fuzzy
logic: we can say that a species is more or less native, especially if the
abundance is gradient from a center to peripheral areas, and then it could
derive from species distribution modeling based on geoccurrences and loccurrences
expressed as fuzzy functions.
So the next step is to include fuzzy logic
in ontologies ;-). And TDWG becoming a fuzzy think tank ;-).
BW
Nicolas.
From:
tdwg-content-bounces@lists.tdwg.org
[mailto:tdwg-content-bounces@lists.tdwg.org] On
Behalf Of Richard Pyle
Sent: Wednesday 13 October 2010
07:08
To: tuco@berkeley.edu
Cc: tdwg-content@lists.tdwg.org;
tdwg-bioblitz@googlegroups.com
Subject: Re: [tdwg-content] How to
record "Nativeness"?
Thanks, John. I agree that there
will be little value in trying to define and name distinct units of space and
time, but there may be value in defining units along the taxonomic axis.
However, we should first come to a community consensus on what the maximum
scope of each axis is.
My sense is that the maximum scope of
space is "Earth" (at least until we begin
documenting populations of extraterrestrial life).
My sense is that the maximum scope of time
is effectively "any window of time during the past 4 billion years or
so".
But I don't have a clear
sense for what the maximum scope of "one or more
organisms" ought to be. I'm content with extending it to
"populations" as a unit of "organisms", because I see a
smooth transition from two individual organisms all the way up to a population
of organisms. But should we accept taxonConcept (which can be
thought of as an implied set of populations) as an extension of
"organisms"? If so, then "Animalia Occurred on Earth
sometime during the past 2 billion years" is a legitimate Occurrence
record (pretty damn useless...but still legitimate).
I think it matters, and is relevant to
this exchange -- both because of Steve's point about more clearly defining what
an "Occurrence" can be, and because we still don't have a good idea
of how and where to score "nativeness" (for which there is clearly an
expressed need).
I agree that fitness-for-use should be
determined from the content of the records, but coming back to Donald's (and
others') point about filtering "non-native" records, there needs to
be a way to include this information in the content of the records in
order to determine fitness-for-use. I believe that a controlled
vocabulary for establishmentMeans will probably be all we have to
do to satisy 95% of the user need. But before we can nail down what
that controlled vocabulary would encompass, I think we need to come to some
sort of consensus on the issues that Steve has articulated.
Aloha,
Rich