Evaluate semantics of "ecosystem functional groups"

[pbuttigieg]

CC @timalamenciak

#1556 noted the IUCN term "ecosystem functional group".

What are the differentia of this category ?

Do we need a class for this ?

[timalamenciak]

Yes, I think we do need a class. My latest comment on #1571 suggests three classes for ecological classifications:

biogeographic realm - A division which considers geophysical barriers
bioregion - A division which considers geophysical barriers AND climatic conditions
ecoregion - A division which considers distinct species assemblage AND geophysical barriers AND climatic conditions

In this way "ecosystem functional group" would be subclass of ecoregion: "An ecoregion division of ecosystems that considers the functional traits of ecosystems." This also means defining functional traits:

functional traits = defn "Morphological, biochemical, physiological, structural, phenological, or behavioral characteristics that are expressed in phenotypes of individual organisms and are considered relevant to the response of such organisms to the environment or their effects on ecosystem properties." source: SER standards

[pbuttigieg]

In this way "ecosystem functional group" would be subclass of ecoregion: "An ecoregion division of ecosystems that considers the functional traits of ecosystems." This also means defining functional traits:

I don't think this works. This is the definition of a classification rationale, not the things in the world it classifies. I very much doubt an EFG is a type of ecoregion.

What, metaphysically, is the thing that is delineated by this approach? what are its boundaries? what are its differentia?

functional traits = defn "Morphological, biochemical, physiological, structural, phenological, or behavioral characteristics that are expressed in phenotypes of individual organisms and are considered relevant to the response of such organisms to the environment or their effects on ecosystem properties." source: SER standards

We'll have to unpack this so it's more logically coherent. I see what they are attempting to do, but these traits are anchored to organisms, not ecosystems, which puts it at odds with the EFG def above.

[pbuttigieg]

Strawman definitions to iterate upon, which assume that the boundaries of EFGs are asserted by human convention (i.e. by fiat):

'ecosystem group' =def. "A fiat part of an astronomical object which is composed of at least two ecosystems."

'ecosystem functional group' =def. "An ecosystem group which has fiat boundaries determined by thresholds detected or asserted in the variability of phenotypic traits."

[timalamenciak]

I don't think this works. This is the definition of a classification rationale, not the things in the world it classifies.

I think this is the core issue - are we trying to describe ecosystems, or to describe the classification schemes that people use to classify the ecosystems? I think there is value in both, but if we are trying to describe ecosystems, we may be reinventing the wheel a bit and it might be simpler to pick one classification scheme and hew to that.

'ecosystem group' =def. "A fiat part of an astronomical object which is composed of at least two ecosystems."

I believe that the "group" refers to a grouping of functional traits within a particular area.

'ecosystem functional group' =def. "An ecosystem group which has fiat boundaries determined by thresholds detected or asserted in the variability of phenotypic traits."

It might be helpful to take T2.2 Deciduous Temperate Forests (pg. 45 of PDF) as an example:

Each page has these flowcharts describing the ecosystem but ecological traits are at the core. Perhaps then:

'ecosystem functional group' =def. "An ecosystem which has fiat boundaries determined by the composition of ecological traits."

Then we would need to define ecological traits.

[cmungall]

I think this is the core issue - are we trying to describe ecosystems, or to describe the classification schemes that people use to classify the ecosystems?

You have hit the nail on the head. I think a lot of OBO could be a lot more up-front about the fact it is geared entirely towards the former, with the latter relegated to ad-hoc metaclass systems. This is not necessarily bad, we just need to be more transparent about it.

This is touched on in this general OBO-wide ticket:
OBOFoundry/OBOFoundry.github.io#2454

I think there is value in both, but if we are trying to describe ecosystems, we may be reinventing the wheel a bit and it might be simpler to pick one classification scheme and hew to that.

Agree 100%. In addition, there could be a push to represent alternative classification schemes using best OWL practice, and have SSSOM mappings between them.

But yes, ontologies work best (and in particular OBO-style ontologies) if they try for a boring theory-free simple consensus classification system that is broadly understandable by a wide range of users, following common-sense best practice

[pbuttigieg]

From Table 2 of the PDF source above

Ecosystem Functional Group: A group of related ecosystems within a biome that share common ecological drivers promoting convergence of biotic traits that characterise the group. Functional groups are derived from the top-down by subdivision of biomes (Level 2).

It is a group of ecosystems that are parts of a biome (the same instance?) within which organisms share similar traits due to similar selective pressures

[pbuttigieg]

Looking at the instances of EFGs, I don't think we need a new class. The actual "grouping" occurs at the instance level, not at the class level.

As @cmungall and @ben-norton #1556 (comment) noted, the organising levels of the IUCN scheme are more convenience / theory (in an informal sense) driven hierarchies than metaphysical ones.

This means that we can extend ENVOs range of ecosystems to cover what wasn't present in the WWF (e.g. Temperate pyric humid forests), and map to the IUCN with annotation properties (I don't think there are any PIDs for us to build SSSOM matrices )

[pbuttigieg]

@cmungall @matentzn is there some recent documentation on how to set up a ROBOT spreadsheet ?

@timalamenciak can then use that to prep definitions and annotation properties in a batch

[dr-shorthair]

the organising levels of the IUCN scheme are more convenience / theory (in an informal sense) driven hierarchies than metaphysical ones.

Indeed. I'm generally skeptical of hierarchies with a fixed number of levels having special names or tags, unless they are clearly tied to some concrete differentiation.
People always later seem to need sub- and super-levels between the levels in the original hierarchy - see Taxonomy (the traditional biological kind) and the Chronostratigraphic timescale (geology) for a couple of well-known examples.

[pbuttigieg]

Thanks for the analysis everyone

I'll close this issue and create a new one for adding the IUCN ecosystem classes to the existing framework

@timalamenciak be sure to add @cmungall @dr-shorthair in separate dc:contributor annotation properties on this bunch

I'll create a GitHub discussion on the categorisation schemes vs metaphysical entities with excerpts from this thread (perhaps similar threads if i can find them) for posterity .

[timalamenciak]

I think I've got a reasonable handle on the ROBOT template syntax. Just to be clear: we are minting each of the realms, biomes and ecosystem functional groups as instances.

• each realm will be type = biogeographic realm (ENVO:03620000)
• each biome will be type = biome (ENVO:00000428) & part_of = realm_INSTANCE where the realm_INSTANCE is replaced with the instance of the actual realm it is in.
• each ecosystem functional group will be type = ecoregion ENVO:01000276 & part of = biome_INSTANCE where the biome_INSTANCE is replaced with the one it is part_of.

[pbuttigieg]

porting to #1579