[tdwg-content] Why it matters what kind of things we include in the definition of Individual

Steve Baskauf steve.baskauf at vanderbilt.edu
Sat Nov 6 04:17:40 CET 2010


What I think is getting lost in this attempt to define what is and what 
is not an Individual is that there is a clear and straightforward 
functional definition of Individual based on what it is intended to do:

An Individual serves as a resource relationship node that connects 
Occurrences to Identifications. 
(This is stated explicitly in the comment I included with the term 
definition.)

If you don't like the technical language, then look at the diagram: 
http://bioimages.vanderbilt.edu/pages/token-explicit.gif
which shows that there is a many-to-one relationship between Occurrence 
and Individual, and a one-to-many relationship between Individual and 
Identification. 

If you prefer it in layman's language: an Individual can connect many 
Occurrences to many Identifications.

If something that you want to call an Individual can't or doesn't do 
this, then it shouldn't be an Individual.  The purpose why I have asked 
for this class to be added to DwC is to be able to accomplish the 
purpose listed above, not to see how many things we can think of for 
which we have philosophical reasons to think that they should be called 
an "individual". 

We gain three clear benefits from being able to create instances of the 
Individual class:
Benefit 1. We can group Occurrences that document the same Individual 
over time (i.e. resampling).  This is exactly the reason why the present 
term dwc:individualID exists (read the definition at 
http://rs.tdwg.org/dwc/terms/index.htm#individualID).  That function is 
represented by the triangle on the left side of Individual in the 
diagram referenced above.
Benefit 2.  If there are multiple Identifications of an Individual, 
those identifications automatically are associated to all Occurrences 
that are associated with the Individual.  That function is represented 
by the triangle on the right side of the diagram.  If we connect several 
tokens to an Individual, those multiple Identifications are 
automatically associated with all of the tokens as well.
Benefit 3.  Individuals allow us to do semantic reasoning of a very 
primitive sort.  If an Occurrence A and the token that acts as its 
evidence are associated with Individual A having Identification A, and 
if Occurrence B and the token that acts as its evidence are associated 
with Individual B having Identification B, then if we discover that 
Individual A is the same as Individual B then we know that 
Identification B also applies to Occurrence A (and its documenting 
token) and that Identification A applies to Occurrence B (and its 
documenting token).  Writing it in this abstract way is a bit hard to 
follow, so I'll illustrate with two examples.  In a previous post, I 
mentioned a living individual (possibly the only one) of Crataegus 
harbisonii.  I have documented the Occurrence of this Individual on 
2008-10-31T09:49:29 at 36.07° latitude, -86.88° longitude by the token 
http://bioimages.vanderbilt.edu/baskauf/70915 (an image) and have 
applied an Identification of Crataegus harbisonii to that Individual.  
Ron Lance has also recorded the Occurrence of the same Individual at the 
same location around 2000 and documented it by propagating it by a 
cutting which is now a living specimen in the North Carolina Arboretum.  
If someone examines that living specimen and and applies an 
Identification of Crataegus somethingelse to the Individual from which 
it was collected, then I can infer automatically that his/her 
Identification of Crataegus somethingelse applies to my 2008 Occurrence 
and its associated image.  The person who looked at the living specimen 
would not need to look at my image for me to know that.  Another example 
happened when a taxonomist was looking at several bark and leaf images 
for a particular species I had photographed.  He wanted to know which 
flower images that I had taken came from the same tree as particular 
bark and leaf images.  He knew logically that if he could identify the 
Individual by its flower that by inference that Identification would 
also apply to the bark image even if he couldn't do the actual 
identification based on the bark alone.  A final application involves 
Identifications of "duplicates" found in different herbaria.  A 
taxonomist is doing a revision of a genus and borrows specimens of that 
genus from several herbaria.  Specimen A from herbarium A was identified 
as species A in the genus of interest.  Specimen B from herbarium B was 
identified as species B in the same genus.  By careful examination of 
the label records, the taxonomist is able to determine that the 
specimens are "duplicates" (i.e. they are from the same Individual).  By 
inference, the taxonomist knows that the identifications of species A 
and species B apply to both specimen A and specimen B because they are 
both from the same Individual. 

In my original thinking about what should constitute an instance of the 
class Individual, I only allowed actual biological individuals, or small 
localized populations that were so tightly linked that a taxonomist 
collecting specimens from it would call them "duplicates".  Under that 
definition of Individual, all three of the benefits listed above would 
apply.  My qualms about applying the term Individual to the various 
buckets of dead homogeneous and heterogeneous mixtures of organisms 
stems from loss of benefit number 1 in those cases.  Moving subsets of 
those dead organisms around and putting them into different jars has no 
aspect of resampling.  Sorting and re-assigning individualIDs to the 
various jars still only involves a single Occurrence, the one in which 
the trawler collected the original bucket from the ocean.  There are 
clever things we can do with multiple Identifications, but we've 
basically lost the triangle on the left side of Individual (no benefit 
#1).  My qualms about applying the term Individual to cut up pieces of 
organisms involves the triangle on the right side of Individual 
(connecting Individuals to Identifications).  If you chop up a fish into 
100 pieces of organs, tissues, DNA samples, etc. and call all of those 
pieces Individuals, there is no point in assigning separate 
Identifications to all of them.  Unless the original fish has had some 
kind of tricky human intervention like interspecific organ transplants, 
grafting, or creation of a chimera, it is a foregone conclusion that all 
of the parts of the individual fish have the same Identification.  
Assigning them all separate identifications would be a waste of time - 
no Benefit #2.  Finally, applying the term Individual to containers that 
we know to contain biological individuals that probably differ at lower 
taxonomic levels causes problems with Benefit #3.  Unless one has a way 
to specify that the Individual he is talking about is the kind of 
Individual that a taxonomist would take "duplicates" from (i.e. reliably 
a single taxon at a low level), it becomes difficult to be sure of the 
accuracy of the type of reasoning that I'd like us to be able to do 
based on Occurrences  and tokens documenting a common Individual. 

So what I've tried to do here is to explain why I'm opposed to 
broadening the definition of Individual to include all of the things 
that people have suggested it should include.  If the definition becomes 
so broad that we loose the benefits that were the reason for 
establishing the class Individual, then there is no point in having the 
class at all.  I think that if we stick to the definition that I 
proposed, we can at least get Benefits #1 and #2.  With the substitution 
of "taxon" for "species or lower...", I think to get benefit #3 we are 
going to need to also have the individualScope term that Rich proposed 
and it would need to include a value that indicated that the group of 
biological individuals were restricted to those that a taxonomist would 
call "duplicates". 

Steve

-- 
Steven J. Baskauf, Ph.D., Senior Lecturer
Vanderbilt University Dept. of Biological Sciences

postal mail address:
VU Station B 351634
Nashville, TN  37235-1634,  U.S.A.

delivery address:
2125 Stevenson Center
1161 21st Ave., S.
Nashville, TN 37235

office: 2128 Stevenson Center
phone: (615) 343-4582,  fax: (615) 343-6707
http://bioimages.vanderbilt.edu



More information about the tdwg-content mailing list