Representatives from most Year-1 HerpNET participants met on Friday, April 11, at the MVZ in Berkeley to discuss progress, problems, and strategies for accomplishing our georeferencing and networking tasks. Several veteran MaNIS participants, including John Wieczorek who has overseen and developed software for georeferencing in that project, offered valuable advice based on their collective experience. The most critical parts of the discussion are summarized below.
1. Automated, Web-based Georeferencing
Substantial progress has been made in the development of automated, web-based georeferencing systems—viz., BioGeoMancer, by Reed Beaman (Yale Univ.) and Georeferencing Calculator by John Wieczorek (Univ. California, Berkeley). Although these programs are available now, John and Reed expect to have a greatly improved version available in a few months. The significance to HerpNET participants is as follows—participants can submit their geographic data for automatic geocoding.
N.B.— BioGeoMancer is in development pending support and currently does not support uncertainty calculations. The Georeferencing Calculator has been tested and in production use for more than a year (http://bnhm.berkeley.museum/manis/GC.html). Updates to the software occur occasionally to add Datums and map-scale hitherto not encountered, but the algorithms remain unchanged and follow the methods described in the MaNIS Georeferencing Guidelines (http://dlp.cs.berkeley.edu/manis/GeorefGuide.html).
• The results will be best for North American localities; participants can expect about 30%–40% of their localities to be properly geocoded with appropriate error estimations, but these results will have to be verified.
• Owing to ambiguities in, or incompleteness of, locality information, probably 25%–30% will return more than one geocoded locality; in these cases, someone will have to select and verify the correct locality.
• There will be some localities that cannot be geocoded automatically. These will have to be georeferenced manually with tools on the Internet (e.g., TopoZone, MapQuest, Alexandria Map Library), available software (e.g., StreetAtlas), and paper maps. Georeferencing guidelines to be followed by all participating institutions will be provided.
Although automation greatly speeds up the geocoding process, the results must be verified. Therefore, participants should plan on time and personnel for this task. The process of verification will be explained later.
3. Locality Data Needed from All Participating Institutions
To facilitate georeferencing, we would like all participating institutions to submit their locality data as soon as possible. Note that we are requesting only locality data, which will be formatted for automatic georeferencing and the results returned to the participants. Year-1 participants have completed this task. We will send instructions to all other participants.
4. Global Localities
Because BioGeoMancer will greatly speed up georeferencing U.S., Canadian, and Mexican localities, we have the opportunity to begin georeferencing global localities. For the time being, institutions from Year 1 should claim Central American and Caribbean localities first. Secondary choices of countries to georeference elsewhere in the world should reflect the strengths of an institution’s resources and/or holdings from that country.
5. Complementary Georeferencing Efforts
After considerable discussion, we agreed that it makes sense for us to divide the localities to be georeferenced by geographic regions and to have institutions georeference areas in which they have expertise and resources. A checklist of geographic regions will be set up that indicates the institution that has taken responsibility for (claimed) the region to be georeferenced.
6. Distributed Network
We plan to initiate the distributed network of collection data as early as June or July of 2003, and add institutions as they are ready.
7. REU Proposals
All participating institutions are encouraged to submit REU proposals as supplements to the HerpNET grant. This may provide a way for institutions to hire needed personnel for the project. Procedural details will follow in another communication.
8. Money for Servers
The $5,000 allocated for purchase of a server may be used to buy hardware and software that is configured into a "system" for use in this project. The $5,000 cannot be used for any other purpose.
9. Next Meeting of Participants
The next meeting of participants is scheduled tentatively for 12 December 2003 at KU. This will include participants from Years 1 and 2.
10. Discussion Board
The HerpNET coordinator at Univ. of California, Berkeley, with help from the architect of AmphibiaWeb at the UC Berkeley Digital Library Project, will develop a discussion board suited for the needs of HerpNET. The discussion board will be the main communication site and will contain individualized information for specific institutions, as well as general information for all participants. An e-mail notification of new postings also will be part of the discussion board.
11. Maps and Georeferencing Resources
Institutions should purchase maps and other resources after consulting with the PIs. This measure will ensure that there are no unnecessary duplicate purchases and that any resource acquired will be for the benefit of the entire HerpNET community.
12. Concept Checklist
Darwin Core is a standard for natural history collections that defines searchable fields from a database. All institutions need to review the list of fields used for MaNIS (http://elib.cs.berkeley.edu/manis/darwin2ConceptInfo030315jrw.htm) and decide what additional fields specific to reptiles and amphibians HerpNET may require. One example mentioned was "Age class."