trait list: morphometric traits

[fdschneider]

Do we try to provide a complete list of morphometric traits? I imagine it is quite difficult. Depending on taxon, the ways how legs are measured and named differ very much.
Some measurements are relative lengths, i.e. ratios of length and body length. Some are established length/width ratios or wing segment lengths.

Which resolution do we offer?

[fdschneider]

regarding legs:

I suggest we should be consistent with offering a trait for each major leg segment (trochanter, femur, tibia, tarsus, full leg) and differentiate by position (front, mid, hind). How do people measure spiders legs (numbered)? I guess we can pool left and right legs. If data owners differentiate that, they could add a note.

[nadjasimons]

I agree with not having different traits for left and right legs and I also think that we don't need to differentiate between the first, second, etc. leg. Those should be captured due to the fact that each measurement is one row in the data and different measurements on the same individual can be linked through specimen_ID,. Data owners can then specify this in the comments.

[fdschneider]

Okay for now, but comments are not machine readable. A solution could be to allow an expansion of the traitlist to -- besides the indifferent trait 'leg_length' -- also include sub-traits at a lower hierarchical level (#7): 'length_first_leg_left', 'length_second_leg_left', ...

It depends on how well the trait list is conceptualised as a semantic ontology: If each trait definition is globally well defined, e.g. by an unambiguous URI reference, then other trait definitions could link to it upstream, e.g. a trait ontology B could refer all of its more detailled traits 'length_first_leg_left', 'length_second_leg_left' etc to the single trait 'leg_length' in trait ontology A. Or ontology C, which is specialised on morphometrics (see #11) could define the distance between its traits 'landmark13' and 'landmark15' as being equal to 'A::leg_length'. But this is up to the traitlist authors.

[nadjasimons]

Yes, this sounds useful. So this would mean that for our first traitlist, we would create the ontology A in your example and while defining it make sure that more detailed ontologies can relate to ours?

[fdschneider]

Yes. that is what we should do.
I'll close this as answered. see #18 for a more general discussion how this may be implemented.