In 1986, the National Science Foundation funded (DEB 8601255) project to computerize the CAS Herpetology collection catalog did not allow the time to enter data from the field notes. Instead, data were entered directly from the handwritten catalog ledgers. Unfortunately, transcription errors between the original field notes and the catalog ledgers have rendered much of the catalog unreliable. These errors are compounded by the difficulties associated with interpreting the confusing organization of the field notes, unfamiliarity with Philippine localities, and an inconsistent field numbering system. Fortunately, the existence of the detailed original field notes creates an opportunity to correct these errors.
To understand fully the scope of the problem, we quantified and categorized the most common error types. To quantify the number of errors, we picked the first field number from the first 10 pages of 23 sets of field notes. Some of the field numbers picked did not match numbers within the CAS database (these could reflect specimens donated to other institutions or remaining in the Philippines). A total of 222 records matched CAS cataloged entries and were compared to the data given in the computerized database. Table 1 shows the number and percentage of errors encountered in four categories of data.
Table 1. Errors found in 222 randomly accessed data records.
Data Category |
No. of Records with Errors |
Percentage of Error Occurrence |
Locality data |
181 |
81.5 % |
Ecological Data |
155 |
69.8 % |
Collector(s) |
57 |
25.6 % |
Date |
21 |
9.5 % |
The most common error occurred in specific locality and usually consisted of missing data elements, misspelled locality names, and the omission of or incorrect elevation. In addition, 30% of the records contained interpretive errors in the hierarchical organization of the geographic data. The second most common error was the lack of specific microhabitat or ecological data. Collectors’ names were also erroneously attributed over 25% of the time, and occasionally, the date of collection conflicted with the original data. We address the main sources of error below.
Transcription Errors in Locality — Most errors occur in the specific locality description, which we attribute to transcription errors based on Brown’s method of having specimens cataloged. While he was alive, Brown was understandably protective of the huge amounts of data in his notes, and he usually preferred to extract and transcribe data himself rather than allow a curatorial assistant to enter the information directly from the field notes. Given that cataloging of the material began more than 40 years ago, perhaps part of the rationale for Brown summarizing only the “important” data was due to space constraints in the handwritten ledgers then in use. Brown also tended to have specimens cataloged by taxonomic group, based on revisions he was working on, so that different species from the same locality were often cataloged years or decades apart by different people. Brown’s transcriptions for the same locality over the years were rarely (if ever) identical and often included serious errors from misreading the complex field notes. Brown also generalized the locality information, giving a range of mileage offsets instead of transcribing the exact mileage distance as recorded in the notes. Specific place names for Barrio or Sitio were often omitted. He also converted specific elevation from an area into a range of elevations for all specimens collected in the general area, and the same was often done for dates. The microhabitat was often ignored completely. This created an additional, subjective level of confusion because it is not always clear if discrepancies between the original data in the notes and the transcribed version were due to informed modification of the original entries by Brown or careless errors, but our preliminary proofreading efforts indicate the latter.
Other transcription errors include the more common kind of misreading handwritten notes. Most of the original notes were written by Lawton Alcala or other members of Alcala’s field team. The handwriting can be difficult to decipher, and the organization is not always intuitive, both contributing significantly to transcription errors.
Confusion over the Organization of the Field Notes — Alcala and Brown’s field data span over 40 years and involve more than 23 discrete “sets” of notes, each roughly corresponding to a given trip. Most sets of notes consist of two primary data sources: Specimen Data Records, a catalog containing field number (not always in numerical sequence), taxon, specific locality, collector(s), date, and microhabitat; and Field Data Records, a more detailed locality summary with more general, inclusive ecological and geographical information for each site. These records also contain information on the time of day collections were made, forest conditions, soil type, temperature, weather conditions, and a list of other species observed. A unique site number links the Specimen and Field Data Records. Each set of notes contains shared and unique data elements so both must be used to extract full data for individual specimens. Lack of awareness of this organization and/or unavailability of both sets of notes has led to shortcomings in the handwritten ledgers and, now, the computerized data.
Complexity of Foreign Localities — The complexity of the Philippine geopolitical infrastructure contributes to problems interpreting specific localities. A given Philippine locality can have six (or more) higher geographic distinctions (i.e., Province, Island, District, Municipality, Barangay/Town and Barrio and/or Sitio), all of which are useful and often essential to locate properly the site. As in many former Spanish colonies, duplication of place names is common at all levels of the administrative hierarchy. With commendable rigor, the Alcala and Brown field teams duly recorded all available locality data, but they did not explicitly identify the administrative level (e.g., whether municipality, barrio, sitio etc.), perhaps because it was obvious to most Filipinos. This has led to some fundamental misinterpretation and inconsistency of the basic locality information in the database. The locality data for small islands are particularly chaotic, due in large part to this geopolitical complexity. These sparsely populated islands are often included in the administrative municipality on the nearest larger island. For example, Sicogon Island is part of the Carles Municipality of Panay Island , but the specific locality field in the ledgers and database often, but inconsistently, list “Carles” without mention of Municipality, implying that it is a place name on Sicogon Island itself.
Problems with an Inconsistent Field Numbering System — Another source of error is the field numbering system used by the field teams, an outcome of improvising with scarce resources rather than carelessness. Blank field tags were hand-numbered until sets of printed, pre-numbered SU tags became available. Because several field teams often worked in different areas at the same time, each group, unknown to the others, duplicated field numbers in the series. Even after extensive use of the field notes, simply identifying the proper notes for each of the duplicate series can be challenging. Our proofing efforts, to date, routinely yield 3 to 5 duplicates, with as many as 8, for many field numbers in the 3- and 4-digit field number range. Brown was also confused by these duplicate numbers and constructed a “key” to the numbering system, based on characteristics of the tags (e.g. , “in pencil,” “pre-printed SU fish collection,” “SU amphibian collection tags,” etc.). To resolve many of the data conflicts, specimens and their field tags must be examined to determine the proper origin of the specimen data. Both duplication of number series and the misreading the hand-written tags have led to erroneous localities in the database (e.g., species recorded from islands where they assuredly do not occur).
Project Goals— In 2005, we were awarded a National Science Foundation grant (0447211) to accomplish three main tasks: 1) correct the data in our computerized catalog by detailed review and retrospective capture of the data as they appear in the original field notes; 2) archive and preserve the fragile field notes by digitizing and indexing them; and 3) georeference the corrected localities in accordance with NSF-funded HerpNET standards and protocols. We propose to broaden access to the Philippine collection by posting to the CAS website the corrected catalog with georeferenced localities, digitized field notes (linked to specimen records), a checklist to the amphibians and reptiles of the Philippines, a distributional checklist of known species from each island, and an annotated type catalog of the CAS Philippine holdings.
