A “holy grail” for biodiversity informatics is a suitably quality controlled, human- and machine-queryable list of all the world’s species, preferably arranged in a suitable taxonomic hierarchy such as kingdom-phylum-class-order-family-genus or other. To make it truly comprehensive we need fossils as well as extant taxa (dinosaurs as well as dinoflagellates) and to cover all groups from viruses to vertebrates (possibly prions as well, which are, well, virus-like). Linnaeus had some pretty good attempts in his day, and in the internet age the challenge has been taken up by a succession of projects such as the “NODC Taxonomic Code” (a precursor to ITIS, the Integrated Taxonomic Information System - currently 722,000 scientific names), the Species 2000 consortium, and the combined ITIS+SP2000 product “Catalogue of Life”, now in its 16th annual edition, with current holdings of 1,635,250 living and 5,719 extinct valid (“accepted”) species, plus an additional 1,460,644 synonyms (information from http://www.catalogueoflife.org/annual-checklist/2016/info/ac). This looks pretty good until one realises that as well as the estimated “target” of 1.9 million valid extant species there are probably a further 200,000-300,000 described fossils, all with maybe as many synonyms again, making a grand total of at least 5 million published species names to acquire into a central “quality assured” system, a task which will take some time yet.
Ten years ago, in 2006, the author participated in a regular meeting of the steering committee for OBIS, the Ocean Biogeographic Information System which, like GBIF, aggregates species distribution data (for marine species in this context) from multiple providers into a single central search point. OBIS was using the Catalogue of Life (CoL) as its “taxonomic backbone” (method for organising its data holdings) and, again like GBIF, had come up against the problem of what to do with names not recognised in the then-latest edition of CoL, which was at the time less than 50% complete (information on 884,552 species). A solution occurred to me that since genus names are maybe only 10% as numerous as species names, and every species name includes its containing genus as the first portion of its binomial name (thanks, Linnaeus!), an all-genera index might be a tractable task (within a reasonable time frame) where an all-species index was not, and still be useful for allocating incoming “not previously seen” species names to an appropriate position in the taxonomic hierarchy. OBIS, in particular, also wished to know if species (or more exactly, their parent genera) were marine (to be displayed) or nonmarine (hide), similar with extant versus fossil taxa. Sensing a challenge, I offered to produce such a list, in my mind estimating that it might require 3 months full-time, or the equivalent 6 months in part time effort to complete and supply back to OBIS for their use.
To cut a long story short… the project, which I christened the Interim Register of Marine and Nonmarine Genera or IRMNG (originally at CSIRO in Australia, now hosted on its own domain “www.irmng.org” and located at VLIZ in Belgium) has successfully acquired over 480,000 published genus names, including valid names, synonyms and a subset of published misspellings, all allocated to families (most) or higher ranks (remainder) in an internally coherent taxonomic structure, most with marine/nonmarine and extant/fossil flags, all with the source from which I acquired them, sources for the flags, and more; also for perhaps 50% of genera, lists of associated species from wherever it has been convenient to acquire them (Catalogue of Life 2006 being a major source, but many others also used). My estimated 6 months has turned into 10 years and counting, but I do figure that the bulk of the basic “names acquisition” has been done for all groups (my estimate: over 95% complete) and it is rare (although not completely unknown) for me to come across genus names not yet held, at least for the period 1753-2014 which is the present coverage of IRMNG; present effort is therefore concentrated on correcting internal errors and inconsistencies, and upgrading the taxonomic placement (to family) for the around 100,000 names where this is not yet held (also establishing the valid name/synonym status of a similar number of presently “unresolved” generic names).
With the move of the system from Australia to VLIZ, completed within the last couple of months, there is the facility to utilise all of the software and features presently developed at VLIZ that currently runs WoRMS, the World Register of Marine Species and its many associated subsidiary databases, as well as (potentially) look at forming a distributed editing network for IRMNG in the future, as already is the case for WoRMS, presuming that others are see a value in maintaining IRMNG as a useful resource e.g. for taxonomic name resolution, detection of potential homonyms both within and across kingdoms, and generally acting as a hierarchical view of “all life” to at least genus level. A recently implemented addition to IRMNG is to hold ION identifiers (also used in BioNames), for the subset of names where ION holds the original publication details, enabling “deep links” to both ION and BioNames wherein the original publication can often be displayed, as previously described elsewhere in this Blog. Similar identifiers for plants are not yet held in the system but could be, (for example Index Fungorum identifiers for fungi), for cases where the potential linked system adds value in giving, for example, original publication details and onward links to the primary literature.
All in all I feel that the exercise has been of value not only to OBIS (the original “client”) but also to other informatics systems such as GBIF, Encyclopedia of Life, Atlas of Living Australia, Open Tree of Life and others who have all taken advantage of IRMNG data to add to their systems, either for the marine/nonmarine and extant/fossil flags or as an addition to their primary taxonomic backbones, or both. In addition it has allowed myself, the founding IRMNG compiler, to “scratch the taxonomic itch” and finally flesh out what is meant by statements that a certain group contains x families or y genera, and what these actually might be. Finally, many users of the system via its web interface have commented over time on how useful it is to be able to input “any” name, known or unknown, with a good chance that IRMNG can tell them something about the genus (or genus possible options, in the case of homonyms) as well as the species, in many cases, as well as discriminate extant from fossil taxon names, something not yet offered to any significant extent by the current Catalogue of Life.
Readers of iPhylo are encouraged to try IRMNG as a “taxonomic name resolution service” by visiting www.irmng.org and of course, welcome to contact me with suggestions of missed names (concentrating at genus level at the present time) or any other ideas for improvement (which I can then submit for consideration to the team at VLIZ who now provide the technical support for the system).
