Relationship of taxonomy to specimens
“Taxonomy” for the purposes of this document and Arctos in general is “some formal naming system.” Arctos taxonomy is not necessarily hierarchical, it is not limited to biological taxonomies, we follow no organization or committee guidelines, and we make no distinction between taxonomy and nomenclature. We offer no judgements or guidelines regarding acceptability; if all or part of a system of names (and optionally metadata regarding those names) is accepted or endorsed by some user group, it’s acceptable for Arctos. If publication “A new species of critter” declares, however informally, that “Some critter” is a distinct “species” (in the loosest possible sense of the word), then “Some critter” may be a useful taxon in Arctos. If another publication declares “Some critterrrrr is obviously not a valid [taxon, biological distinction, etc.],” then “Some critterrrrr” may be a useful taxon name to include in Arctos for discovery purposes, even though it’s an obvious misspelling rejecting a name. In short, if a name might be useful in discovering specimens, it should be included in Arctos. (Remarks, annotated taxon relationships, and links to publications should be used to clarify.)
Arctos taxonomy consists of two tables: taxon_name is a list of “base names” (genera, species, kingdoms) and taxon_term contains metadata regarding a taxon_name, optionally organized in various ways. Informal terms, such as “Sorex sp.” or “Sorex cinereus or Sorex ugyunak” are Identifications, not taxonomy. (Identifications are generally drawn from and linked to taxonomy in various ways.)
The names in taxon_name are intended to be a formal taxonomic authority, and should all be drawn from relevant taxonomic publications. A name is not linked to any particular classification (rather, classifications are linked to names), and may be a homonym or hemihomonym. The namestring “Diptera” CORRECTLY includes classifications for flies and plants; this is not a mistake and should not be “corrected” to any particular point of view. Taxon names may be disambiguated at the specimen level via identification publications (“ID sensu”), and at the collection level by choosing and curating a classification. Database rules prevent change of used names. Names may be higher taxon terms, such as “Animalia,” ICZN-type genus-species-subspecies concatenations, ICBN-type genus-species-infraspecific rank-infraspecific epithet concatenations, or nonbiological taxonomy terms (“Granite”). The sole distinguishing feature of this field is that a Curator, usually a taxonomist, considers the values here formal taxonomy.
Taxon Term contains metadata regarding taxon names and a system for organizing such data. Every term may be arranged hierarchically within a classification, and hierarchical terms may optionally be ranked. Non-hierarchical terms must be ranked. Terms are further divided as:
- “Local data” is Curated from within Arctos and may be edited. A controlled vocabulary is required for “local” data.
- “Webservice data” is drawn from various sources via GlobalNames.org and is used primarily for locating specimens.
Terms which are organized hierarchically are “classification data” intended to be part of a classification system. These terms may be ranked or unranked.
Terms which are not organized hierarchically are intended for clarification, and are linked to but not part of classifications. These consist of things like HTML-formatted “display name” and author strings. A rank is required.
Taxonomy as a Hierarchy
While single classifications are hierarchical, taxonomy as a body of literature is anything but: Most name have a long history of “current family, according to….”, for example, and many refer to multiple concepts, such as plants and animals.
VARCHAR2 (20) null
Common Names are intended to help users find what they are looking for, and not to propagate any standard or system. A taxon may have several common names, in several languages and using several types of characters, or nothing. The same common name may apply to more than one taxon. For example, the term “common shrew” has been published for Sorex cinereus in North America and for Sorex araneus in Europe. Common names have not been capitalized except when they draw on a particular standard such as that of the American Ornithological Union (AOU Checklist). Adjectival forms of proper names are capitalized (e.g., “Alaska marmot”).
VARCHAR2 (30) not null
Taxon Relations are comprised of a relationship type, a related taxon, and an authority for the relationship. The related taxon is another record in the taxonomy table.
VARCHAR2 (45) null
Relationship Authority is an open text field, and it may be null. Presumably the Source Authority for an accepted taxon is adequate documentation, but if not, then Relationship Authority could cite a publication or the name of an expert to whom the relationship is attributed.
Most Taxon Relations represent synonymy among taxa. As evolutionary relationships and nomenclature are revisited, changes in taxonomy are suggested, and either accepted or rejected. Which changes are accepted, and by whom, is a routine issue. Therefore, keeping track of synonomy in the database can be important to users. If they cannot find material they are seeking under one name, they may find the name that they are using and its accepted synonym, or they may use a query which returns records from unaccepted synonyms.
Any number of taxa may be synonymous, but only one of these should have the value “accepted synonym of,” and it should have this value for each of its synonyms. Accepted synonyms should have a Valid Catalog Term flag of “yes.” The other synonyms should have the value “synonym of,” for each synonym, and their Valid Catalog Term flag should be “no.”
Named hybrids have a unique relationship to their parent taxa, and this is expressed by “hybrid offspring of.” Each named hybrid should have two such relationships.
Taxon relations may also represent hierarchical relationships between taxa. So far, this is included only for the purpose of constructing botanical trinomens with author text for both the species and the infraspecific category. For example, Trichophorum pumilum (M. Vahl) Schinz & Thell var.Rollandii (Fern.) Hult. would be constructed from the “parent” binomen, Trichophorum pumilum (M. Vahl) Schinz & Thell plus the infraspecific rank, “subspecies,” and author text from the “child” trinomen, Trichophorum pumilum var. Rollandii (Fern.) Hult.
There are various “taxonomy-like” search fields on SpecimenSearch, and they query different data.
“Identification” queries the identification applied to specimens, and includes the option to query non-current identifications. Identifications are usually based in taxonomy, but are not wholly taxonomy. Examples of strings that will currently find specimens include:
- Sorex yukonicus (a taxon name)
- Sorex cinereus or Sorex ugyunak (an indeterminate identification)
- Monticola rufiventris and Cuculus optatus (a cataloged item consisting of multiple taxa)
- Diadocidia sp. nov. cf. borealis (a “working name” not yet formalized as taxonomy)
- Iron-rich mudstone (a non-Linnean term identifying a non-biological specimen)
“Taxon Name” queries all taxonomy classifications, and classifications related to taxa attached to specimens. For example, querying MVZ Mammals for Muridae returns Euryoryzomys nitidus, even though MVZ’s preferred classification uses Cricetidae as the family of that name, because Freebase asserts that the unranked term “Muridae” is between “Muroidea” and “Sigmodontinae.”
Each ranked term – Kingdom, Family, etc., – query only the collection’s preferred classification. “Muridae” in Family will NOT return specimens identified as Euryoryzomys nitidus; only Family=Cricetidae will locate specimens identified using that taxon name, which may be viewed as “Cricetidae according to MVZ.”
Details and more examples
Each collection “claims” one taxon term source, and data from that source, when available, are stored with the specimen record. For example, if CollectionX claims the FAKE classification source, which contain the following data for namestring “Sorex cinereus”:
and CollectionY claims the NORANKS source:
then searching for “Mammalia” in the “Class” field from SpecimenSearch will find CollectionX specimens but will NOT find CollectionY specimens.
Searching the “Any taxon term” field will (by way of both example classifications above) find specimens from both collections. Additionally, if there is an additional BADEXAMPLE classification for Sorex cinereus:
then searching the Family field for “Shrewidae” will find no specimens, as neither collection prefers the BADEXAMPLE classification, but searching the “any taxon term” field for Shrewidae will find Sorex cinereus from both collections, based on the assertions made by the BADEXAMPLE classification.
In other words, the individual ranked fields (Family, Kingdom, etc.) represent the current assertions of the collections; current taxonomy as defined by taxonomists. These fields are appropriate for asking pointed questions: “how many shrews does the XYZ collection hold?” The “any taxon term” field represents classifications according to some provider of GlobalNames.org. This field is a very large net, and most queries using it will return false positives; specimens that do not fit with modern, consensus ideas of the term searched. This field is appropriate for exploratory searching: “what genera did someone at some point consider to be members of taxon BLA?”
Q: Why not hierarchies/thesauri/some other model?
A: The taxonomy model is primarily designed for flexibility, an idea somewhat incompatible with more-structured data. This model is designed to work with changes, uncertainty, unranked terms, and even nonbiological taxonomies, and to communicate with services such as globalnames.
Q: Why does the editing interface not ….. ?
A: The editing interfaces are designed to get us by until someone does a better job, and are a reflection of the model in being very generic (and therefore also very limited and scary in terms of multi-record updates). We hope that most taxonomy will eventually be managed in remote, limited-scope applications (that is, applications that do not have to worry about historical names and homonyms and all the other reality that is taxonomy at the scale of Arctos) rather than in Arctos (which can now leverage any number of those services to compile a complete picture).
Q: How does the model assert “acceptedness”?
A: All names exist because someone with the proper credentials loaded them, and presumably because there exists “appropriate” (a term which varies by discipline, tradition, and time) literature creating or supporting them. Collections choose a classification, which asserts “classification according to this collection.”
Create a Name
There are two ways by which taxon names may be created:
- Click the “create name” link and enter a namestring.
- Click the “cloning classification as a new new name” link.
Search Arctos before creating names. Do not try to create duplicate namestrings, even in the case of homonyms, hemihomonyms, committee rulings, or for any other reason.
Create a Classification
Classifications may be created by pulling from globalnames, cloning an existing classification into a local source, or manually creating a local source.
There is currently no classification bulkloader. One can be created if necessary.
The pre-migration Arctos data have been pushed to source “Arctos.” These data are the default classification for all collections, and generally represent a ranked singular taxonomic viewpoint.
Taxonomy search examples
Note values that pop into search form.
Names that contain the string “microtus”
Names that ARE Microtus:
Names that contain microtus AND have an attribute of “mammalia”:
Names that contain microtus AND have an attribute of value “mammalia” and rank (=term type) “phylclass”
Names that contain microtus AND have an unranked attribute of value “mammalia”
Microtus according to Arctos
Genus microtus according to Arctos
Relationship of taxonomy to specimens
The link between taxonomy and specimens is through Identification and Identification_Taxonomy. However, Taxonomy no longer resolves to a single set of ranked assertions (eg, row in table). Collections chose a single preferred source (in manage collection), and the data in that classification become available for various “specimen-taxonomy” purposes (such as asserting ranked terms via DWC exports, searching for specimens, and printing labels). Some taxonomy searches may produce unexpected results for the naive user. For example, given the following:
- Collection1 prefers Source1
- Collection2 prefers Source2
- Source1 places NameN in Family1
- Source2 places NameN in Family2
- Source3 places NameN in Family3
- Collecton1 and Collection2 both have specimens identified as NameN.
the following are true:
- Searching Identification (scientific name) for NameN will find NameN specimens from Collection1 and Collection2
- Searching Identification for Family1 will find nothing; the Identification is to a name (string), not a “taxon concept” (data object including metadata).
- Searching Taxon Name (any taxon name) for Family1, Family2, or Family3 will find NameN specimens from Collection1 and Collection2
- Searching Family for Family1 will find only specimens from Collection1
- Searching Family for Family2 will find only specimens from Collection2
- Searching Family for Family3 will find nothing
- Collection1 is asserting that “the Family of NameN is Family1.”
- Collection2 is asserting that “the Family of NameN is Family2.”
- Something else (via GlobalNames, perhaps) is asserting that “the Family of NameN is Family3.”
Possible reasons for this include but are not limited to the following:
- NameN is a hemihomonym. Collection1 is a vertebrate collection, and Collection2 is an invertebrate collection.
- One collection has accepted a new taxonomic opinion, and the other simply hasn’t got around to updating their classifications.
- One collection has accepted a controversial taxonomic hypothesis with which the other collection disagrees.
- One collection has confounded taxonomy and storage location, and cannot update their classification until shelves are relabeled and specimens are moved.
Note that the sums of the family-according-to-collection searches do NOT necessarily add up to the “family-according-to-someone” search total; a collection may use a classification that does not include a term ranked “family,” and doing so will return no specimens with queries which include family-according-to-collection.
Names should generally not be edited. Do not create garbage names, and contact the Advisory Committee if you find a garbage name.
It is generally not possible to edit remote (e.g., pulled from GlobalNames) classifications; they are automatically maintained, and edits would be lost with the next update. Please note that this does not preclude using those classifications as preferred, nor from editing them via any tools (including collaboration with sources such as ITIS) before they are pushed to GlobalName and then Arctos. This is in fact our (unfortunately theoretical and untested) preferred approach: Manage data in specialized (e.g., mammals), collaborative, and funded tools, where things like homonyms and alternate classifications are easily dealt with (often by exclusion), and push those data to Arctos via GlobalNames.
There are no “batch” tools in Arctos, because the data scope and structure does not support them. “UPDATE….WHERE GENUS=’Philometra’….” cannot do what is probably intended; Philometra is a homonym, and may additionally be recorded as a subgenus, unranked term, etc., and so the update is very likely to both miss intended data and update unintended and unrelated data. (This is one reason we encourage external, limited-scope tools or collaborations to manage taxonomy; Philometra will likely mean exactly one thing in “Lepidoptera of Europe” or “Parasites of North America,” but both of those things and much more are included in Arctos.)
The classification editing form is in two parts.
“Non-classification terms” are data which is not part of the classification but which are important for various curatorial tasks or for understanding why a specimen has been assigned to a name, or why a name has been assigned to a classification. Some terms are especially important:
- display_name is the HTML-formatted namestring. This is included in various places on Arctos forms, and in FLAT and DWC as FORMATTED_SCIENTIFIC_NAME. The form will suggest values for display_name; click to accept the suggestions, or type your own, including any necessary HTML markup.
- nomenclatural_code controls how display_name is suggested, and helps format display. Actionable values are “ICZN” and “ICNB.”
- author_text is the author of ICZN names, or the species author of ICBN names
- infraspecific_author is the author of the infraspecific epithet in ICBN names
The ordering of these terms is unimportant, and un-paired terms (e.g., display_name=NULL or NULL=”Sorex cinereus”) will be ignored (that is, deleted at save).
Classification terms are ordered taxonomic data (drag rows to order). Ranks are optional; the following mixture of ordered ranked and unranked terms is possible (and common):
The following ranks are used in building “specimen data” (which is used for things like printing labels, locating specimens from the ranked search terms such as Family, and as terms in specimen summary):
- class (in FLAT as “phylclass” due to RDBMS reserved words)
- order (in FLAT as “phylorder” due to RDBMS reserved words)
- “subspecies” (for ICBN data, this is “infraspecific term” and ranked as “infraspecific rank” which includes subspecies, subsp., variety, var., varietas, subvar., etc.)
Infrageneric terms should be supplied as multinomials: species=Sorex cinereus, never only cinereus.
http://arctos.database.museum/name/Carex%20brunnescens%20subsp.%20alaskana#Arctos is an example of a botanical name which includes a subspecific epithet and two authors.