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@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@gmail.com Cc: Steve Baskauf steve.baskauf@vanderbilt.edu, Ramona Walls < rlwalls2008@gmail.com>, TDWG Content Mailing List < tdwg-content@lists.tdwg.org>, Suzanna Lewis suzi@berkeleybop.org, Chris Mungall cjmungall@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).