Create design patterns for biome and ecosystem

[cmungall]

Currently there is inconsistent classification between biome and ecosystem, exemplified here:

• terrestrial biome should be a subclass of terrestrial ecosystem #1208.

terrestrial biome =
biome
and ('determined by' some
('planetary landmass' or ('part of' some 'planetary landmass')))

def: "A biome which is primarily or completely situated on a landmass."

Note the large deviation between textual def and logical def, this is a red flag see S11 text and logical defs should match

(FWIW it seems the logical definition is problematic)

terrestial ecosystem has no logical definition and no textual definition.

My understanding is that "X biome" should always be classified under "X ecosystem", and what differentiates the child from the parent is that the child has reached climax state. See #787 for full discussion.

If we are agreed on this (can you confirm @pbuttigieg), I recommend:

• creating a characteristic "in climax state". This could live in ENVO or PATO
• creating a DP for "X biome = X ecosystem and has-characteristic some in-climax-state"
• making the textual definitions reflect this

I also recommend regularizing the DPs for ecosystems but this can be discussed in a separate ticket.

I also recommend avoid single-children and naming the non-climax state siblings but we can also discuss this in a separate ticket

Note: my personal preference is to only have a single concept and if people need to annotate an ecosystem as being in a climax state they do this with a field in their database rather than pre-composing shadow classes. However, my current proposal above is conservative and allows ENVO to retain the dual ecosystem and biome classes.

Related issues

• definition of environmental system/environment #1051.
• NTR: rainforest/rainforest biome and subtypes #1174.

[pbuttigieg]

My understanding is that "X biome" should always be classified under "X ecosystem", and what differentiates the child from the parent is that the child has reached climax state. See #787 for full discussion.

If we are agreed on this (can you confirm @pbuttigieg ),

Yes - there's a series of successional ecosystem states with different community compositions. When the communities stabilise and the succession "ends" (roughly, something is always changing in nature), then the ecosystem is populated with climax communities, and can be considered a biome by most definitions (many add size thresholds, but microbiomes have made things generalize)

I recommend:
• creating a characteristic "in climax state". This could live in ENVO or PATO
• creating a DP for "X biome = X ecosystem and has-characteristic some in-climax-state"
• making the textual definitions reflect this

Is characteristic the new quality? Is this a COB thing? This should stay in ENVO and probably be reworded / added to a family of characteristics (in pioneering state, etc)

At any rate, we can try it, although it's more accurate to say that the ecosystem has dominant ecological communities that have reached their climax successional state.

[pbuttigieg]

PS: the textual definitions cannot be so easily normalised while maintaining alignment to the WWF definitions. We'll have to preserve that link in a comment or gloss

also recommend regularizing the DPs for ecosystems but this can be discussed in a separate ticket.

Yes.

I also recommend avoid single-children and naming the non-climax state siblings

As noted above, there are a few other states, but they are usually not named.

Note: my personal preference is to only have a single concept and if people need to annotate an ecosystem as being in a climax state they do this with a field in their database rather than pre-composing shadow classes. However, my current proposal above is conservative and allows ENVO to retain the dual ecosystem and biome classes.

Biomes exist, and ecologists will continue to reference them. The ecosystem + successional state code makes sense and may improve consistency, but removing biomes and decoupling from the WWF classification wouldn't be wise IMO. We can live with some single child hierarchies.

[wdduncan]

characteristic of is included in RO. When we update the RO import, it will be included.

Probably of no consequence, but you should be aware that characteristic of is a functional property. So, given:

:x :characteristic-of :y .
:z  :characteristic-of :y .

results in the inference that :x same-as :z.

[cmungall]

Is characteristic the new quality? Is this a COB thing?

characteristic of is included in RO. When we update the RO import, it will be included.

Correct, the proposal to rename inheres-in to characteristic-of was approved at the 2018 RO meeting, it's independent of and preddated COB, but COB is consistent with this. As Bill says this will be reflected next time we refresh the RO import.

This should stay in ENVO and probably be reworded / added to a family of characteristics (in pioneering state, etc)

@wdduncan can you make a PR where we introduce such a hierarchy? I suggest to start:

• ecosystem characteristic [ENVO:NEW] a characteristic of an ecosystem
  • ecosystem successional state [ENVO:NEW] an ecosystem characteristic that has its basis in the process of change in species structure in the community that inhabits that ecosystem (https://en.wikipedia.org/wiki/Ecological_succession)
    • ecosystem climax state [ENVO:NEW] an ecosystem succession state that holds when the dominant ecological communities that have reached a steady state. (https://en.wikipedia.org/wiki/Climax_community)

Better definitions can be tweaked in the PR with full credit. We can flesh out siblings in that PR, or in future PRs.

I am OK with these having ENVO IDs for now, but there is a larger issue to resolve here about modularity and terms in PCO, ENVO, PATO, and ECOCORE. cc @ramonawalls @diatomsRcool. If someone later wants a term "climax community" where does it go? But we can resolve some of these later.

After that we can try making some logical defs

PS: the textual definitions cannot be so easily normalised while maintaining alignment to the WWF definitions. We'll have to preserve that link in a comment or gloss

good point. For now let's definitely preserve a definition sourced from WWF. Though I am not seeing that many WWF derived definitions? E.g.

vs https://www.worldwildlife.org/biomes/mangroves

WWF issues could perhaps be split into a separate ticket

[cmungall]

WWF issues here: #658

[pbuttigieg]

I am OK with these having ENVO IDs for now, but there is a larger issue to resolve here about modularity and terms in PCO, ENVO, PATO, and ECOCORE. cc @ramonawalls @diatomsRcool. If someone later wants a term "climax community" where does it go? But we can resolve some of these later.

I'd keep these in ENVO indefinitely as its the only ontology using them for the moment and we can spare ourselves another import chain. If there's other usage, ECOCORE would be the target.

good point. For now let's definitely preserve a definition sourced from WWF. Though I am not seeing that many WWF derived definitions? E.g.

It's not really a "for now" thing - we need to base these defs on a systematic approach of biome classification, even if it's not semantically clean. We can't design-pattern everything

[cmungall]

Do we have sufficient agreement that I can go ahead and start implementing a pattern that defines biomes in terms of ecosystems?

Structurally this would be:

X biome = X ecosystem and has-characteristic some 'ecosystem climax state'

@pbuttigieg, you said:

we need to base these defs on a systematic approach of biome classification, even if it's not semantically clean. We can't design-pattern everything

I think what we are proposing here is a systematic approach. Of course, I agree that we can't design pattern everything. In fact, I think ENVO already has way too many OWL equivalence axioms that need to be removed or weakened. However, I would propose that as a default if we have a concept X, and labels "X biome" and "X ecosystem" already in ENVO, then a reasonable assumption is that these are discriminated on the basis of climax state as you propose. If you are aware of a large number of exceptions it would be great to articulate these here before we start making any changes.

I also want to state here that this is far from my preferred solution. I have observed many many ecologists and biologists use and query ENVO and everyone is confused by the duplication of concepts across biome and ecosystem. My preferred solution is to reduce the number of concepts and have annotators describe climax state via an additional post-composed annotation. However, I trust there are users who need these distinct terms, so I think the DP solution here is the best compromise.