INTRODUCTION

The Fish Habitat Database, a document containing extensive information on habitat characteristics and requirements of several Great Lakes fish species, has potential to become a valuable source of detailed information on fish ecology. In order for this compilation to be truly useful, however, the information it contains must be accurate and reliable. Upon examination of randomly selected entries in the March 1993 Review Copy, the Great Lakes Fishery Commission found that some information in the database had been inaccurately reported from the original publication. These errors caused concern over the credibility of the database, and further examination of the cited information was necessary to determine the document's usefulness. In July 1993, an additional twenty-five randomly selected entries were checked for accuracy, giving the GLFC a better sense of the databases dependability.

METHODS

A system was developed to ensure that entries checked for accuracy would be randomly selected. First, a list of random numbers was developed using a random number generator on a scientific calculator. once this list of numbers was obtained,

these were used as the reference numbers by which entries were accessed. The calculator produced random numbers 0 through 999, therefore entries with reference numbers 1000 or greater were not selected.

Reference numbers were looked up in the appendix "List of Reference Numbers by Fish," a sequential listing of reference numbers indicating the fish species each number referred to. Once the type of fish was known, the section of the database pertaining to that fish was quickly scanned to locate the entry containing the specific reference number. After locating the entry, the information located under the heading "Details" was noted and checked against the source of that information.

The original publication was indicated under "Primary Reference Number," which contained a different number corresponding to the information's citation. Citations were listed in the appendix "Literature Cited (by [Primary] Reference Number)." If the journal or article referred to in the citation was available, the text was obtained and checked against the information in the database (research was done at the GLFC and the U.S. Fish and Wildlife Service, limiting the entries which could be checked to those whose source could be obtained at either of these locations). Any differences between the source and the database were noted. This process was repeated until twenty-five entries were checked against their original publication.

RESULTS

The majority of entries that were checked contained information that had been correctly reported from the original. Out of twenty-five entries whose information was checked, three were found to have questionable citations:

1) Reference number 69 contained primary reference (PR) number 386. The citation indicated by this PR number refers to an article that does not contain any information similar to that which is included in the entry.

2) Reference number 294 contained PR number 340. The journal article referred to by this citation contains no information similar to information contained in the entry.

It is possible that the PR number contained in the entry was mistyped, thereby preventing the correct citation from being located. It is also possible that the information in these entries can be located using the review number listed in the entry, which was 109 in both cases. Although it is not clear what function the review number serves, the book referred to by 109 was not available for clarification.

3) Reference Number 142 contained PR number 75. The article referred to in this citation, "Van Oosten, J. 1947. Mortality of Smelt, Osmerus mordax (Mitchell), in Lakes Huron and Michigan during the Fall and Winter of 1942-1943. Transacions of the American Fishery Society 76:283-289,11 was inaccurately reported. The "Mortality of Smelt... 11 article by Van Oosten actually appears in volume 74 of the same journal, pages 310-337.

The only other inconsistency between the sources and database information was in measurement units after conversion to the metric system. Although this is a minor point and the information is still technically correct, values are not consistently converted throughout the database (a quick scan of the database revealed some values reported in standard measure, instead of metric). It is also not a verbatim representation of the original author's research. Specifically, data in entries with reference numbers 958, 69, 13, 454, and 749 were found to be converted from the original publication.