[tdwg-content] Fwd: Darwin Core Proposal - environment terms (biome)

[Ramona Walls]

Forwarding Pier's response, because he is not on the TDWG list.

Thanks, Pier!
------------------------------------------------------
Ramona L. Walls, Ph.D.
Scientific Analyst, The iPlant Collaborative, University of Arizona
Research Associate, Bio5 Institute, University of Arizona
Laboratory Research Associate, New York Botanical Garden

---------- Forwarded message ----------
From: Pier Luigi Buttigieg <pbuttigi at mpi-bremen.de>
Date: Fri, May 15, 2015 at 7:51 PM
Subject: Re: [tdwg-content] Darwin Core Proposal - environment terms (biome)
To: John Deck <jdeck88 at gmail.com>
Cc: Steve Baskauf <steve.baskauf at vanderbilt.edu>, Ramona Walls <
rlwalls2008 at gmail.com>, TDWG Content Mailing List <
tdwg-content at lists.tdwg.org>, Suzanna Lewis <suzi at berkeleybop.org>, Chris
Mungall <cjmungall at lbl.gov>


Hi all,

I've restricted my responses (in line, below) to the usage of ENVO classes
as I can't really comment on DwC strategy.

> John Deck wrote:
[...]
> The ENVO definition of biome is : "A biome is an environmental system to
> which resident ecological communities have evolved adaptations." (
> http://purl.obolibrary.org/obo/ENVO_00000428)

We try to be agnostic to spatial scale with this definition as we
encountered numerous instances of the term being used outside of
'classical' biome classification systems with reasonable rationale (e.g.
many environments - such as marine and lacustrine environments - are not
adequately covered by existing schemes). Even within classical systems,
identifying the scale threshold is hardly precise (if anyone knows of
anything that defines this, please let me know) and definitions are revised
from time to time as new technologies (e.g. remote sensing) emerge. Some
have suggested using prevailing climate as a way to stabilise 'large'
scales, but this is problematic as microclimates (e.g. near large water
bodies) can result in the emergence of different biomes (in the classical
sense) existing at comparatively small regions. The presence of an
ecological community which has adapted to a given environment seems to be
the common theme. A successful, in situ adaptation process indicates that
the environment a) can sustain viable populations over multiple generations
and b) persists long enough for these populations to undergo evolutionarily
consequential changes, distinguishing it from other environment types. If
users wish to use a 'classical' biome type, they have access to an adapted
version of the WWF classification (see below).

> A resident ecological community from the perspective of a microbe likely
> does not care about the large-scale plant and animal communities, so it
is
> a matter of perspective taken from the point of view of the subject.  To
> that end, leaf litter as the biome seems entirely reasonable if the
> microbes resident there have evolved adaptations to leaf litter.
[...]
>> Steve Baskauf wrote:
>>  I haven't looked at the definition given to "biome" in ENVO, but based
>> on what I believe is the common consensus on what a biome is (a major,
>> large-scale set of plant and animal communities occupying a geographic
>> region), it doesn't seem right to apply that term to "leaf litter".
[...]
>> There are a number of standard lists of the world's biomes and they
>> include large-scale regions like "temperate deciduous forest", not
>> small-scale features.

ENVO includes a representation of the WWF biome classification system. At
one stage, the Udvardy, WWF, and Bailey systems were all in ENVO (which was
quite confusing). We opted to use the WWF system (obsoleting the other
classes) as it was the most current and had the highest global coverage.
The WWF categories were modified to make them more suitable for an ontology
(e.g. compound classes were split).

As a microbial ecologist, I think that the position outlined by John is a
valid one. The scales used in the 'classical' definition are, ultimately, a
function of our own observational capacities and various forms of ecosystem
can be nested across scales. However, Steve's right in saying that this is
a fairly profound change in the usage of a well-established term (with a
substantial literature base behind it). Further, just because a microbial
ecologist (or anyone else, but I'm going with this example) declares
something to be a biome, doesn't make it so: the communities of microbes
living in leaf litter may not have evolved in that particular environment,
they may simply have adaptations to other environments that allow them to
colonise a one with sufficient similarities.

In our annotation guidelines
(http://www.environmentontology.org/annotation-guidelines) we do ask that
such "small scale" biomes are requested with reference to some form of
empirical data supporting the notion that the communities have adapted to
that particular environment. ENVO doesn't try to dictate what is "right"
here, but attempts to represent how different communities (who are creating
new conventions which reflect their phenomena they study) are talking about
environments. Whether they turn out to be correct in their usage of a given
term is a somewhat different question and we're always happy to receive
critiques and input.

I think it's best if we declare or produce subset of biome classes that are
approved by a certain body (e.g. the WWF). Conversely, classes that are
somehow 'nascent' or 'experimental' can also be marked. Plans to produce
subsets of ENVO that are relevant to specific working bodies are already
queued.

>> Ramona Walls wrote:
>>
[...]
>>  -- ENVO very clearly distinguishes between a biome, a feature, and a
>> material. It is never the case that the same ENVO class can be use as
>> both a biome and a feature or a feature and a material. Although the
same
>> entity, depending on its role, may serve as either a biome or material
>> (or feature for that matter), in that case, it would be an instance of
two
>> different classes in ENVO. Take the leaf litter example. A correct
>> annotation would need to point to both a "leaf litter biome" class and a
>> "leaf litter material" class. It is really crucial not to confuse
>> material
>> entities in world with the roles they take on as instances of classes in
>> ENVO.

A "leaf litter biome" would, roughly, refer to the environmental system
that is determined by (~ must include) the community of organisms that have
adapted to the conditions in leaf litter. As noted above, there should be
some sort of evidence that this environment-specific adaptation occurred.

As a material, "leaf litter" is referring to some portion of 'stuff'
primarily composed of (but not necessarily limited to) fallen, dead or
dying plant material. As another example, when you use ENVO:water you
roughly mean "a volume of material primarily composed of H2O, but which is
likely to include stuff other than H2O found in some environment".

I'm not sure that leaf litter works as a feature as it doesn't seem to have
countable parts that would be called, e.g. "pieces of leaf litter" (does
it?). One would rather say "dead leaf" or "dead twig". As an alternative
example, "rock", as a mass noun, is a material, but a "piece of rock" can
be a feature.

>>> Joel sachs wrote:
[...]
>>> I have some concerns with these terms. As far as I can tell, no one
>>> knows how to use these them.
[...]

I feel that creating interfaces for annotators to use ontologies without
delving too deeply into (the individual) ontology are sorely needed. Some
of us have discussed something like a GUI-based wizard to help people use
ENVO (gamifying it to increase 'uptake' and annotation accuracy), but
haven't had the time to put it together.
In the meantime, I can certainly help write more sets of annotation
guidelines for different communities (linking to them from the ENVO website
to show that there are multiple ways to use the ontology).

>>> Creating tripartite (biome/feature/material) decompositions of habitats
>>> sometimes makes sense. Certainly, it made sense for some of the early
>>> metagenomic assays that gave rise to ENVO. But it doesn't always make
>>> sense, and there are often better ways to characterize an environment.
I
>>> think it was a mistake for these terms to be made mandatory in
>>> MIxS/MIMARKS.

The main arguments for using the tripartite annotation (and its mandatory
status) were: 1) many of the better ways of describing environments (e.g.
hard data) were non-recoverable and 2) adding more than one term for each
of ENVO's main hierarchies would add too much to the already long
checklist. Even when other data is missing, there is usually enough
information around to compose a 'three-phase zoom in' (from biome to
material) on an entity's environment. This way, at least rough comparative
studies could be performed using an ontology (or, at the very least, a
controlled vocabulary). It's clear, however, that many MIxS report
submitters don't use ENVO very well, even after directed to the annotation
guidelines. Again, some sort of nifty annotation interface would probably
make this more successful.

>>> ... I'd like to see our usage
>>> guides differ from current ENVO/MIxS guidelines which mandate one and
>>> only one value for each of the terms. "Environmental feature",
>>> in particular, often merits multiple uses within the same record,
>>> and I think disallowing such usage would impede uptake of the term
>>> set. (As far as I can see from browsing metagenomic sampling metadata,
>>> it *has* impeded uptake of the term set.)
ENVO's guidelines suggest that there should be *at least* one class from
each hierarchy used.
Indeed, multiple feature and material classes can and should be used to
fully characterise an entity's environment. There is certainly more than
one feature that is likely to exert a strong
causal influence on (i.e. determine) an entity's environment and all those
that are deemed relevant should all be recorded. Ideally, they would be
'ranked', but this requires some further thinking and implementation. For
materials, entities can be partially surrounded by multiple materials
(Chris' duck swimming in water example, for instance).

>>> So I'm not necessarily opposed to the addition of these terms, but I do
>>> wonder why we need them.
I think there are some good reasons to use some form of ontology in
annotations to enhance comparative power across granularities and shades of
meaning. Naturally, ontologies like ENVO are constantly developing and if
they don't meet a community's needs, there are usually ways to either
report and discuss issues (e.g.
https://github.com/EnvironmentOntology/envo/issues) or become a
co-developer.
[...]

I hope this has helped rather clarify our thinking. As always, we're very
interested in insight (especially on our issue tracker) to help enhance the
usefulness of the ontology.
Best,
Pier

PS: For general interest, I'll be meeting some urban environment
specialists next week and intend to add more city-based environment classes
(e.g. "urban prairies" such as those proliferating in Detroit).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.tdwg.org/pipermail/tdwg-content/attachments/20150516/9ee481ee/attachment.html