Ramona,
Thanks for engaging. My thinking is this:
Our goal is to address the mess that is “habitat”. So much - environmental conditions, associated taxa, geography, geology - gets crammed into that one poor term.
We want to either replace or augment or subclass “habitat” with some new terms, with each new term capturing some key dimension of information that “people” are interested in. Before talking specifically about the terms under consideration, let me disclose some of my biases:
<statement of biases> i. I’d like to see terms that will help organize the thousands of habitat terms used by the authors of taxonomic treatments in the Flora of North America. You can see some of these terms here - http://bit.ly/1Fnow3E
We’re currently categorizing these terms to enable semantic search over habitat; as we do so, we are also thinking about the user interface of treatment templates for future contributors. Our opinion is that we should look at existing habitat terms, answer the question “what have authors been trying to tell us?”, and then design interfaces that make it easier for them to tell us those things. (This isn’t to say that we shouldn’t also coax new types of information out of treatment authors, based on our understanding of what data consumers want to know.)
The habitat dimensions that, for plants, keep coming up are: wet vs dry; open canopy vs closed canopy; soil type; sloped vs. flat; and associated taxa. We also see a range of geography/geology related dimensions.
ii. I think the semantics of Darwin Core terms should be clear from their natural language definitions. Ontologies that can be used to provide values for the terms should not be relied upon as documentation for the terms themselves.
iii. I’ve long argued that Darwin Core should not make commitments to any particular upper ontologies. I explained some of my reasons in my 2013 TDWG talk (http://bit.ly/1FjCimJ). I recall that you spoke with me about this last year, and expressed support for the idea of creating ontological layers on top of Darwin Core. I’ve since tried to do this for BCO, but struggled. I’ve been meaning to ask for your help with this. </statement of biases>
Back to the proposed terms:
Biome. As has been noted by others, the currently proposed definition is at odds with common understanding of the term, and is somewhat confusing due to its dependance on evolution. I think I understand what the definition is getting at - namely, expanding the traditional notion of biome to include microbiomes.
Environmental feature. We often see habitat terms such as “rocky outcrops”; “arroyos”; “bogs”; etc. So I can see a lot of utility in a “feature” term. My preference would be to call it “geographic feature” or “geologic feature”, since I think that’s how most people think of such features (see, e.g. geonames). The currently proposed definition is “A material entity which determines an environmental system.” This requirement that the feature “determine” the environmental system strikes me as too strong.
Environmental material. The currently proposed definition is “A portion of environmental material is a fiat object which forms the medium or part of the medium of an environmental system.” A much clearer definition is one that I’ve heard Pierre give, which was along the lines of “The substance that was displaced by the sample prior to its being removed from its environment”. In addition to being clearer, this second definition has the advantage of not relying on the BFO notion of “fiat object”.
Happy Victoria Day to all my fellow British subjects! Joel.
On Wed, 13 May 2015, Ramona Walls wrote:
This is my first chance to reply to this thread, but I think several of Joel's comments need to be addressed.
- re.: ENVO terms: "As far as I can tell, no one knows how to use these
them." -- I know how to use them, and I know a community of people who know how to use them. True, that just like Darwin Core, they are often used incorrectly, but further documentation and outreach can help with that.
- "There was a lot of confusion over whether particular aspects of an
environment constituted an environmental feature, an environmental material, or a biome. The correct answer was often dependent on context. For example if a small mammal were found in leaf litter, then "leaf litter" would be the environmental material, and the biome would be "forest". But if a microbe were sampled from the same leaf litter, then "leaf litter" would be the biome, and I'm not sure what the environmental material would be." -- 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. -- Joel and I seem to remember outcomes of that RCN meeting quite differently (probably we were in different break-out groups). As I recall, the major problem was that people couldn't use ENVO because the classes they needed were not in there, not because they didn't know how. This is a problem that would actually be helped by DwC adopting ENVO, because it would create more users, and therefore more contributors to the ontology. Another major problem was that people often want to describe environments in terms of parameters like light level, salinity, temperature, etc. ENVO does not currently include classes like this, but a movement is underfoot to perhaps add such a branch to ENVO.
- "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." -- True, there are cases when you cannot specify a biome, feature, and material for an organism, but usually you can provide at least one of two of them, which goes a long way toward standardizing environmental records and making large-scale queries possible. I have not yet seen a better way to classify environment on this scale. As I mentioned above, when it comes to describing environments in terms of their physico-chemical paramaters, ENVO does not serve, but that does not negate the utility of ENVO-style descriptions. Furthermore, as with most DwC terms, these are optional, and people/institutions who don't have to provide them if they are not relevant.
- "The terms "env_biome", "env_feature", and "env_material" already exist
in the MIxS Sample extension to Darwin Core (along with "submitted to INSDC", etc.). Why do they need to be moved into the core?" -- The main reason I can see is that they have a lot of applicability outside of MIXS, that is, for occurrences that do not have any sequences associated with them, and should not be hidden away in a place that suggests they can only be applied to sequence data.
Ramona
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
On Fri, Apr 24, 2015 at 3:00 AM, tdwg-content-request@lists.tdwg.org wrote: Send tdwg-content mailing list submissions to tdwg-content@lists.tdwg.org
Today's Topics: 1. Re: Darwin Core Proposal - environment terms (joel sachs) ---------------------------------------------------------------------- Message: 1 Date: Thu, 23 Apr 2015 13:29:47 -0400 (EDT) From: joel sachs <jsachs@csee.umbc.edu> Subject: Re: [tdwg-content] Darwin Core Proposal - environment terms To: John Wieczorek <tuco@berkeley.edu> Cc: TDWG Content Mailing List <tdwg-content@lists.tdwg.org> Message-ID: <Pine.LNX.4.64.1504231321240.18117@linuxserver1.cs.umbc.edu> Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed John, I have some concerns with these terms. As far as I can tell, no one knows how to use these them. I was at a phenotype RCN meeting last year where the theme was environmental ontologies. The attendees were pretty savvy in terms of both ontologies, and environmental terminology. We were given an overview of ENVO, and then, as an experiment, we broke into groups, and each group tried to use ENVO to describe particular environments. I don't recall any group being successful. There was a lot of confusion over whether particular aspects of an environment constituted an environmental feature, an environmental material, or a biome. The correct answer was often dependent on context. For example if a small mammal were found in leaf litter, then "leaf litter" would be the environmental material, and the biome would be "forest". But if a microbe were sampled from the same leaf litter, then "leaf litter" would be the biome, and I'm not sure what the environmental material would be. Due to the confusion, Pier Luigi gave us a more in-depth tutorial when we re-convened. We didnt break back out into groups, but I wish we had, because I wonder if we would have had much more success. 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. But the question isn't "What should MIxS do four years ago?", but "What should TDWG do now?". One wrinkle is that dwc:Habitat already exists. Will it stay in the core? Is the idea to create usage guides that explain when to use dwc:Habitat and when and how to use biome, feature, and material? Such an approach could work, but 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.) So I'm not necessarily opposed to the addition of these terms, but I do wonder why we need them. You wrote that "there is currently no possibility of a Darwin Core PreservedSpecimen or MaterialSample record to meet the minimum requirements of a Mimarks Specimen record[6], as there is currently no way to share required environment terms." But MIMARKS specimen records are also required to have the fields "Submitted to INSDC", "Investigation-type", "Project name", "Nucleic acid sequence source", "Target gene or locus", and "Sequencing method". So won't it still be the case that there will be no possibility of a Darwin Core record being MIMARKS compliant, without appropriate augmentation? The terms "env_biome", "env_feature", and "env_material" already exist in the MIxS Sample extension to Darwin Core (along with "submitted to INSDC", etc.). Why do they need to be moved into the core? Cheers, Joel. On Thu, 26 Mar 2015, John Wieczorek wrote: > Dear all, > > This message pertains to a proposal[1] set forth in September 2013 > concerning the environment terms biome, environmentalFeature, and > environmentalMaterial. I'm renewing the proposal because so much time has > passed and the original proposal was not carried through to completion. > There were no objections to the addition of those terms during the initial > public commentary. Discussion revolved around how the recommendations for > how to populate them. > > The recommendations for all three terms will suggest using a controlled > vocabulary such as ENVO. The examples will be based on the set of > subclasses of the corresponding ENVO terms for biome[2], > environmentalFeature[3], and environmentalMaterial[4]. As with all Darwin > Core terms, the constraints on content are not part of the definition - > they are only illustrative recommendations. > > The importance of these terms was recognized anew at a Darwin Core and MIxS > Hackathon in Florence in Sep 2014[5]. One important outcome of that > workshop was the the realization that there is currently no possibility of > a Darwin Core PreservedSpecimen or MaterialSample record to meet the > minimum requirements of a Mimarks Specimen record[6], as there is currently > no way to share required environment terms. This creates a huge and easy to > solve barrier to integration of data across the collection, sample, and > sequence realms. > > This proposal is not substantively different from the one discussed in > 2013. It differs from the final amended previous proposal in two ways, 1) > only the three terms biome, environmentalFeature, and environmentalMaterial > are proposed here (the proposal to change to the term 'habitat' has been > dropped), and 2) the term definitions have been updated to agree with those > in ENVO. The terms will be in the Darwin Core namespace (following the TDWG > community consensus in the previous discussion as well the consensus to > coin the MaterialSample class in the Darwin Core namespace rather than use > obi:specimen, with the equivalency being made on the ontology side in > BCO[7]). > > The complete definitions of the three proposed terms is given below the > following references. This reopens the 30-day public commentary period for > the addition of new terms as described in the Darwin Core Namespace > Policy[8]. > > [1] Original tdwg-content proposal for environment terms. > http://lists.tdwg.org/pipermail/tdwg-content/2013-September/003066.html > [2] ENVO biome. http://purl.obolibrary.org/obo/ENVO_00000428 > [3] ENVO environmentalFeature. http://purl.obolibrary.org/obo/ENVO_00002297 > [4] ENVO environmentalMaterial. http://purl.obolibrary.org/obo/ENVO_00010483 > [5] DwC MIxS Meeting Notes. >https://docs.google.com/document/d/1Zexgsiol6WC83vDzMTCF3uUB7DcFmKL15DFEPbw 5w6c/edit?usp=sharing > [6] Table of the core items of Mimarks checklists. > http://www.nature.com/nbt/journal/v29/n5/fig_tab/nbt.1823_T1.html > [7] Biological Collections Ontology. https://github.com/tucotuco/bco > [8] Darwin Core Namespace Policy. > http://rs.tdwg.org/dwc/terms/namespace/index.htm#classesofchanges > > > Term Name: biome > Identifier: http://rs.tdwg.org/dwc/terms/biome > Namespace: http://rs.tdwg.org/dwc/terms/ > Label: Biome > Definition: An environmental system to which resident ecological > communities have evolved adaptations. > Comment: Recommended best practice is to use a controlled vocabulary such > as defined by the biome class of the Environment Ontology (ENVO). Examples: > "flooded grassland biome", > "http://purl.obolibrary.org/obo/ENVO_01000195". > Type of Term: http://www.w3.org/1999/02/22-rdf-syntax-ns#Property > Refines: > Status: proposed > Date Issued: 2013-09-26 > Date Modified: 2015-03-26 > Has Domain: > Has Range: > Refines: > Version: biome-2015-03-26 > Replaces: > IsReplaceBy: > Class: http://rs.tdwg.org/dwc/terms/Event > ABCD 2.0.6: not in ABCD > > Term Name: environmentalFeature > Identifier: http://rs.tdwg.org/dwc/terms/environmentalFeature > Namespace: http://rs.tdwg.org/dwc/terms/ > Label: Environmental Feature > Definition: A material entity which determines an environmental system. > Comment: Recommended best practice is to use a controlled vocabulary such > as defined by the environmental feature class of the Environment Ontology > (ENVO). Examples: "meadow", > "http://purl.obolibrary.org/obo/ENVO_00000108". > Type of Term: http://www.w3.org/1999/02/22-rdf-syntax-ns#Property > Refines: > Status: proposed > Date Issued: 2013-09-26 > Date Modified: 2015-03-26 > Has Domain: > Has Range: > Refines: > Version: environmentalFeature-2015-03-26 > Replaces: > IsReplaceBy: > Class: http://rs.tdwg.org/dwc/terms/Event > ABCD 2.0.6: not in ABCD > > Term Name: environmentalMaterial > Identifier: http://rs.tdwg.org/dwc/terms/environmentalMaterial > Namespace: http://rs.tdwg.org/dwc/terms/ > Label: Environmental Material > Definition: A portion of environmental material is a fiat object which > forms the medium or part of the medium of an environmental system. > Comment: Recommended best practice is to use a controlled vocabulary such > as defined by the environmental feature class of the Environment Ontology > (ENVO). Examples: "scum", > "http://purl.obolibrary.org/obo/ENVO_00003930". > Type of Term: http://www.w3.org/1999/02/22-rdf-syntax-ns#Property > Refines: > Status: proposed > Date Issued: 2013-09-26 > Date Modified: 2015-03-26 > Has Domain: > Has Range: > Refines: > Version: environmentalMaterial-2015-03-26 > Replaces: > IsReplaceBy: > Class: http://rs.tdwg.org/dwc/terms/Event > ABCD 2.0.6: not in ABCD >